Corporate Governance and Corporate Finance, a European Perspective.Pdf

Corporate Governance and Corporate Finance

This impressive collection defines the current status of corporate governance and corporate finance from a European perspective. The book brings together a comprehensive range of contemporary and classic articles, with a major emphasis on recent research, accompanied by a new and authoritative editorial commentary. Topics discussed include:

● Alternative perspectives on corporate governance systems ● Equity ownership structure and control ● Corporate governance, underperformance and management turnover ● Directors’ remuneration ● Governance, performance and financial strategy ● The market for corporate control

Addressing both the theory and practice of corporate governance and corporate finance, this book is an invaluable resource for scholars and students engaged with managerial finance, financial economics and business law, as well as the role of the corporation in the modern economy. It is also fascinating reading for practitioners interested in managerial finance.

Ruud A. I. van Frederikslust is Associate Professor of Finance at Erasmus University, the Netherlands.

James S. Ang is Professor of Finance at Florida State University, USA.

P. Sudi Sudarsanam is Professor of Finance and Corporate Control at Cranfield School of Management, UK.

Corporate Governance and Corporate Finance A European perspective

Edited by Ruud A. I. van Frederikslust, James S. Ang and P.Sudi Sudarsanam First published 2008 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York, NY 10016 This edition published in the Taylor & Francis e-Library, 2007. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” Routledge is an imprint of the Taylor & Francis Group, an informa business © 2008 Selection and editorial matter, Ruud A. I. van Frederikslust, James S. Ang and P.Sudi Sudarsanam; individual chapters, the contributors

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Corporate governance and corporate finance : a European perspective / edited by Ruud A. I. van Frederikslust, James S. Ang and P.Sudi Sudarsanam. p. cm. Includes bibliographical references and index. 1. Corporate governance – Europe. 2. Corporations – Europe – Finance. I. Van Frederikslust, R. A. I. II. Ang, James S. III. Sudarsanam, P.S. HD2741.C77485 2007 658.15094 – dc22 2007007573

ISBN 0-203-94013-X Master e-book ISBN

ISBN10: 0–415–40531–9 (hbk) ISBN10: 0–415–40532–7 (hbk) ISBN10: 0–203–94013–X (hbk)

ISBN13: 978–0–415–40531–7 (hbk) ISBN13: 978–0–415–40532–4 (pbk) ISBN13: 978–0–203–94013–6 (ebk) Contents

List of tables ix List of figures xv About the editors xvi Preface by Floris Maljers xviii Acknowledgements xx

General Introduction 1

PART 1 Alternative perspectives on corporate governance systems 7

1 Michael C. Jensen 11 THE MODERN INDUSTRIAL REVOLUTION, EXIT, AND THE FAILURE OF INTERNAL CONTROL SYSTEMS

2 Andrei Shleifer and Robert W. Vishny 52 A SURVEY OF CORPORATE GOVERNANCE

3 Rafael La Porta, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert Vishny 91 INVESTOR PROTECTION AND CORPORATE GOVERNANCE

4 James S. Ang, Rebel Cole, and James Wuh Lin 111 AGENCY COSTS AND OWNERSHIP STRUCTURE

PART 2 Equity ownership structure and control 133

5 Hiroshi Osano 135 SECURITY DESIGN, INSIDER MONITORING, AND FINANCIAL MARKET EQUILIBRIUM

6 Marco Becht and Ailisa Röell 157 BLOCKHOLDINGS IN EUROPE: AN INTERNATIONAL COMPARISON vi CORPORATE GOVERNANCE AND CORPORATE FINANCE 7 Mara Faccio and Larry H. P.Lang 163 THE ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS

8 Marc Goergen and Luc Renneboog 191 WHY ARE THE LEVELS OF CONTROLS (SO) DIFFERENT IN GERMAN AND UK COMPANIES? EVIDENCE FROM INITIAL PUBLIC OFFERINGS

9 Mara Faccio and M. Ameziane Lasfer 221 DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES IN WHICH THEY HOLD LARGE STAKES?

PART 3 Corporate governance, underperformance and management turnover 255

10 Paolo F.Volpin 257 GOVERNANCE WITH POOR INVESTOR PROTECTION: EVIDENCE FROM TOP EXECUTIVE TURNOVER IN ITALY

11 Luc Renneboog 285 OWNERSHIP, MANAGERIAL CONTROL AND THE GOVERNANCE OF COMPANIES LISTED ON THE BRUSSELS STOCK EXCHANGE

12 Julian Franks, Colin Mayer and Luc Renneboog 313 WHO DISCIPLINES MANAGEMENT IN POORLY PERFORMING COMPANIES?

13 Jay Dahya, John J.McConnell and Nickolaos Travlos 347 THE CADBURY COMMITTEE, CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER

PART 4 Directors’ remuneration 367

14 Martin J. Conyon and Kevin J. Murphy 371 THE PRINCE AND THE PAUPER? CEO PAY IN THE UNITED STATES AND UNITED KINGDOM

15 Wayne R. Guay 398 THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK: AN ANALYSIS OF THE MAGNITUDE AND DETERMINANTS CONTENTS vii 16 N. K. Chidambaran and Nagpurnanand R. Prabhala 419 EXECUTIVE STOCK OPTION REPRICING, INTERNAL GOVERNANCE MECHANISMS, AND MANAGEMENT TURNOVER

17 Julie Ann Elston and Lawrence G. Goldberg 450 EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY

18 John E. Core and David F.Larcker 467 PERFORMANCE CONSEQUENCES OF MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP

PART 5 Governance, performance and financial strategy 487

19 Erik Lehmann and Jürgen Weigand 491 DOES THE GOVERNED CORPORATION PERFORM BETTER? GOVERNANCE STRUCTURES AND CORPORATE PERFORMANCE IN GERMANY

20 Paul Gompers, Joy Ishii and Andrew Metrick 523 CORPORATE GOVERNANCE AND EQUITY PRICES

21 Gertjan Schut and Ruud van Frederikslust 557 SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES

22 Jim Lai and Sudi Sudarsanam 568 CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE: IMPACT OF OWNERSHIP, GOVERNANCE AND LENDERS

23 Jorge Farinha 599 DIVIDEND POLICY, CORPORATE GOVERNANCE AND THE MANAGERIAL ENTRENCHMENT HYPOTHESIS: AN EMPIRICAL ANALYSIS

24 Christian Leuz, Dhananjay Nanda and Peter D. Wysocki 623 EARNINGS MANAGEMENT AND INVESTOR PROTECTION: AN INTERNATIONAL COMPARISON

PART 6 On takeover as disciplinary mechanism 645

25 Rezaul Kabir, Dolph Cantrijn and Andreas Jeunink 647 TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS IN THE NETHERLANDS: AN EMPIRICAL ANALYSIS viii CORPORATE GOVERNANCE AND CORPORATE FINANCE 26 Tim Jenkinson and Alexander Ljungqvist 662 THE ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE

27 Jens Köke 708 THE MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY: A GOVERNANCE DEVICE?

28 Julian Franks and Colin Mayer 734 HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE

Name Index 749 Tables

1.1 Labor force and manufacturing wage estimates of various countries and areas playing an actual or potential role in international trade in the past and in the future 20 1.2 Total R&D and capital expenditures for selected companies and the venture capital industry, 1980–1990 29 1.3 Benefit-cost analysis of corporate R&D and investment programs 30 1.4 Difference between value of benchmark strategy for investing R&D and net capital expenditure 32 1.5 Summary statistics on R&D, capital expenditures, and performance measures for 432 firms with sales greater than $250 million in the period 1980–1990 r = 10 percent 33 3.1 Legal origin and investors rights 96 4.1 Agency costs, ownership structure, and managerial alignment with shareholders 121 4.2 Descriptive statistics for variables used to analyze agency costs 122 4.3 Determinants of agency costs at small corporations 124 4.4 Determinants of agency costs at small corporations 127 6.1 Median size (%) of largest ultimate outside voting block for listed industrial companies 159 6.2 Voting power concentration, percentage of companies for which largest voting power stake lies in (in and below) various ranges 160 7.1 List of country specific data sources used to collect data on direct ownership 166 7.2 Screenings used to identify and remove corporations without reliable ownership data 168 7.3 Ultimate control of publicly traded firms 174 7.4 Concentration of control: financial versus non financial firms 176 7.5 Concentration of control and firm size 178 7.6 Legal restrictions on issuing dual class shares 180 7.7 Percentage of firms adopting control-enhancing devices 183 7.8 Means of enhancing control in Europe by types of controlling owners 184 7.9 Cash-flow and control rights 186 x CORPORATE GOVERNANCE AND CORPORATE FINANCE 7.10 How concentrated is control by families/unlisted firms? 187 8.1 Comparison of regulatory issues 195 8.2 Summary statistics of independent variables 202 8.3 Proportion of voting rights held by initial shareholders, by new large shareholders, and by small shareholders in recent German and U.K. IPOs 205 8.4 State of control of IPOs six years after flotation 206 8.5 Voting rights in excess of 25%, six years after flotation 207 8.6 Determinants of the evolution of control concentration in recent German and U.K. IPOs 208 8.7 Prediction of state of control for German and U.K. IPOs 212 8.A1 Pearson correlation coefficients of independent variables 214 9.1 Description of proxy variables 231 9.2 Distribution of pension funds holdings in test companies 232 9.3 Annual changes in occupational pension fund holdings and list of occupational pension funds investing in their own companies 234 9.4 Descriptive statistics on means of selected data on the test and control firms 236 9.5 Logit regressions of the probability that pension fund holdings exceeds 3% of shares 95–96 238 9.6 Determinants of board structure for low and high growth firms 240 9.7 Pearson correlation coefficients between the variables used 242 9.8 Relationship between firm value and pension fund holdings 244 9.9 Pension fund holdings and long term accounting and stock price performance 246 10.1 Descriptive statistics: sample of all firms traded on the Milan Stock Exchange 262 10.2 Turnover and performance: family-, state-, foreign-, and bank-controlled firms 265 10.3 Turnover and performance: regression analysis of the whole sample 266 10.4 Turnover and performance in family-controlled firms 267 10.5 Analysis of the firm’s Q ratio 269 and 284 10.6 Family executives versus other executives 271 10.7 International comparison 272 10.8 Separation of ownership from control in family-controlled firms 273 10.9 Executive turnover and pyramidal groups 275 10.10 Internal and external governance forces 276 10.11 Q ratio and pyramidal groups 277 10.12 Matrix of correlations 280 11.1 Blocking minority, majority and supermajority shareholdings 294 11.2 Largest direct and ultimate (direct and indirect) levered shareholdings, and the control leverage factor 295 11.3 The market in share stakes over the period 1989–1994 296 11.4 Tobit model of the determinants of executive board restructuring in listed industrial and commercial companies 298 TABLES xi 11.5 Tobit model of the determinants of executive board restructuring in listed holding companies 300 12.1 Annual Board turnover in 1990–1992 partitioned by decile of abnormal share price performance calculated in 1990 for a sample of 243 UK quoted companies 318 12.2 Size and category of the largest shareholder in the sample of UK quoted companies, 1988 to 1993 320 12.3 Concentration of ownership in worst and best performing companies, partitioned by size of equity market capitalization 322 12.4 Purchases of share blocks for 243 companies for the period 1991 to 1993 323 12.5 Board characteristics and firm performance for 243 companies 325 12.6 The relation between Board turnover and leverage for companies in the lowest decile of performance 327 12.7 Regressions of executive Board turnover on governance and performance for the total sample, 1988–1993 330 12.8 Regressions of executive Board turnover on governance and performance for the worst performing companies, 1988–1993 336 13.1 Financial, ownership, and Board characteristics for 460 U.K. industrial firms over the period 1989 through 1996 351 13.2 Incidence and rates of CEO turnover in 460 U.K. industrial firms, 1989 through 1996 354 13.3 Forced CEO turnover in 460 U.K. industrial firms grouped by quartiles of performance, 1989 through 1996 355 13.4 Logit regressions of CEO turnover on IAROA and status of Cadbury compliance, 1989 through 1996 358 13.5 Logit regressions of CEO turnover on ISAR and status of Cadbury compliance, 1989 through 1996 360 14.1 Summary statistics for 1997 CEO total compensation, by company size and industry 375 14.2 Summary statistics for components of CEO pay, by company size and industry 376 14.3 Estimated elasticities of CEO compensation with respect to firm revenues 380 14.4 Explanatory regressions for 1997 CEO compensation 381 14.5 Summary statistics for stock-based CEO incentives, by company size and industry 383 14.6 Explanatory regressions for stock-based pay-performance sensitivity 385 14.7 CEO pay-performance elasticities for salary and bonus, by industry 387 14.8 CEO turnover-performance regressions, by industry 388 15.1 Summary statistics for CEOs’ stock option portfolios, common stockholdings, and cash compensation 403 15.2 The sensitivity of CEOs’ wealth to equity risk and stock price 406 15.3 Summary firm characteristics 408 15.4 The relation between investment opportunities and the sensitivity of CEOs’ wealth to equity risk 409 xii CORPORATE GOVERNANCE AND CORPORATE FINANCE 15.5 The relation between stock-return volatility and the sensitivity of CEOs’ wealth to equity risk 412 15.6 Description of parameters used in Black–Scholes computations for common stock 415 16.1 Repricing announcements by year 423 16.2 Repricing announcements by industry 424 16.3 Prior returns of repricers 425 16.4 Operating performance of repricers, non-repricers and control firms 426 16.5 Compensation level, structure, and changes 428 16.6 Cross-sectional characteristics of repricers 434 16.7 Multivariate analysis of repricing firms 439 16.8 Characteristics of CEO repricers and non-CEO repricers 442 16.9 Relation between CEO repricing and turnover 445 17.1 Descriptive statistics 1970–1986 458 17.2 OLS fixed-effects regressions of (Panel A) MB salaries and (Panel B) firm-level SB salaries 459 17.3 Managing Board (MB) salaries 460 17.4 Supervisory Board (SB) salary 461 17.5 Managing Board (MB) salaries by industry 462 18.1 Industry composition of firms adopting target ownership plans 471 18.2 Descriptive statistics 473 18.3 OLS regression models of log(stock value/salary)t 476 18.4 Determinants of the decision to adopt a target ownership plan 477 18.5 Two-year post-adoption excess operating and stock performance for target ownership firms 480 19.1 Ownership structures: sample of 361 German manufacturing firms, 1991–1996 497 19.2A Summary statistics: sample of 361 German manufacturing firms, 1991–1996 499 19.2B Summary statistics: sample of 361 German manufacturing firms, 1991–1996 500 19.3A Panel regression estimates of the return on total assets equation, 1992–1996 505 19.3B Panel regression estimates of the ownership concentration equation, 1992–1996 509 19.4A Robust estimates of the return on total assets equation, 1992–1996 512 19.4B Robust estimates of the return on total assets equation, 1992–1996 513 19.5 OLS estimates of the firm-specific fixed effects equation, 1992–1996 514 20.1 Governance provisions 526 20.2 The governance index 528 20.3 Correlations between the subindices 529 20.4 The largest firms in the extreme portfolios 530 20.5 Summary statistics 531 TABLES xiii 20.6 Performance-attribution regressions for decile portfolios 533 20.7 Performance-attribution regressions under alternative portfolio constructions 534 20.8 Q regressions 536 20.9 Operating performance 537 20.10 Capital expenditure 540 20.11 Acquisitions 542 20.12 Insider trading 543 20.13 Fama-MacBeth return regressions 545 21.1 Location of the joint venture and nationality of the partners 558 21.2 Overview of the results found in the literature 560 21.3 Test results of CAR and SCAR 561 21.4 Estimated functions and test results 563 22.1 Definition of restructuring strategies 577 22.2 Definition of explanatory variables 578 22.3 Descriptive statistics for stock returns and accounting performance 580 22.4 Descriptive statistics for agency and control variables 582 22.5 Descriptive statistics for restructuring strategies and frequency of restructuring strategies pursued by U.K. firms during 1989–1994 583 22.6 Stakeholder dominance and choice of restructuring strategy 586 22.7 Logistic regression of restructuring strategies on agency and control variables (year of decline) 588 22.8 Logistic regression of restructuring strategies on agency and control variables (year 1 after decline) 589 22.9 Logistic regression of restructuring strategies on agency and control variables (year 2 after decline) 590 22.10 Summary of the effect of each explanatory variable on the choice of restructuring strategies 592 22.11 Joint impact of explanatory variables on individual restructuring strategy choice 594 23.1 Sector distribution of sample according to AIC-actuaries industry classification codes 607 23.2 Summary descriptive statistics 608 23.3 Beneficial and non-beneficial insider ownership statistics 610 23.4 OLS regression results 612 23.5 Critical entrenchment levels 614 23.6 OLS regression results with insider ownership defined as total 616 24.1 Descriptive statistics of sample firms and countries 629 24.2 The variables are computed from 70,955 firm-year observations for fiscal years 1990 to 1999 across 31 countries and 8,616 non-financial firms 630 24.3 Earnings management and institutional clusters 635 24.4 Earnings management, outside investor protection and private control benefits 638 24.5 Earnings management and outside investor protection 639 xiv CORPORATE GOVERNANCE AND CORPORATE FINANCE 25.1 Number of takeover defenses adopted by Dutch companies 651 25.2 Distribution of different takeover defenses 651 25.3 Means, medians, standard deviations, and correlations 653 25.4 The difference in ownership concentration of firms with cumulative takeover defenses 654 25.5 Estimates of logistic regressions relating the likelihood of adopting a specific takeover defense mechanism to ownership concentration 655 25.6 Cumulative abnormal returns around the announcement of defense with preferred share issue 656 26.1 Control rights in Germany 665 26.2 Ownership of German stock exchange listed companies (1991) 668 26.3 Defensive share structure of German stock exchange listed companies (1991) 669 26.4 Potential targets for hostile stakebuilders 670 26.5 Stakebuilders and targets 674 26.6 Pre-contest target company ownership structure, takeover defences, and performance 676 26.7 Stakebuilding strategy 678 26.8 Defensive actions 682 26.9 Outcomes 684 27.1 Characteristics of listed and nonlisted firms 713 27.2 Characteristics of entering and exiting firms 714 27.3 Size distribution of blocks purchased by new shareholders 718 27.4 Causes of control changes: univariate results 719 27.5 Causes of control changes: multivariate results 722 27.6 Causes of control changes: robustness tests 723 27.7 Corporate restructuring following control changes 724 27.8 Performance following control changes 725 27.9 Logistic regressions predicting CEO turnover 727 27.10 Data selection procedure 729 27.11 Definition of variables 730 28.1 Proportion of directors who resigned in the targets of accepted and successful and unsuccessful hostile bids 737 28.2 Asset disposals for a five-year period around the bid for unsuccessful bids occurring in 1985 and 1986 738 28.3 Bid premiums for three samples: targets of accepted bids, successful hostile bids, and unsuccessful hostile bids 739 28.4 Abnormal share price returns for the period up to five years before the bid for various samples of merging and nonmerging firms 740 28.5 Proportion of companies omitting or changing dividends in accepted and hostile bids 741 28.6 Cash flow to asset ratios in 1980 and 1985/1986 for various samples of merging and nonmerging firms 742 28.7 Tobin’s Q for 1980 (or first year of listing) and 1985/1986 (or last year of listing/annual report) for various samples of merging and nonmerging firms 743 28.8 Number of bids subject to revision partitioned by hostility of bid 745 28.9 Tobin’s Q, abnormal returns, and cash flow rates of return 746 Figures

4.1 Operating expense-to-sales ratio by one-digit SIC for a sample of 1,708 small corporations 116 4.2 Sales-to-asset ratio by one-digit SIC for a sample of 1,708 small corporations 117 4.3 Operating expense-to-sales ratio by sales quartile for a sample of 1,708 small corporations 118 4.4 Sales-to-asset ratio by sales quartile for a sample of 1,708 small corporations 119 5.1 Sequence of the events 141 5.2 Four possible case of F˜ 143 6.1 Direct stakes and voting blocks 159 6.2 U.K.: Percentile plot of largest ultimate voting block 161 6.3 Germany: percentile plot of largest ultimate voting block 161 7.1 The Nordström family group (Sweden) 171 7.2 Unicem (Italy) 173 9.1 Structure of block share ownership in our sample firms 226 10.1 Structure of the Pesenti’s group as of 12/31/1995 263 14.1 Median cash compensation of US and UK CEOs, 1989–1997 377 14.2 Prevalence of stock option plans, 1979–1997 378 16.1 Model of the repricing decision as a sequential two-stage process 429 16.2 Median annual values by year for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP 429 16.3 Median annual values by size decile for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP 430 16.4 Median annual values by industry for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP 430 18.1 Timeline 472 21.1 Motive for the joint ventures by sector 558 21.2 Autonomous relationships 562 27.1 Ownership structure of Dornier in 1992 716 27.2 Ownership structure of Boge AG in 1990, 1991, and 1992 717 About the editors

Ruud A. I. van Frederikslust, until 1 January 2007 has held various positions at RSM Erasmus University since the graduate management programme was first established in 1970. His most recent position was Associate of Professor of Corporate Finance. He is author of the work Predictability of Corporate Failure (Kluwer Academic Publishers), and editor-in-chief of the volumes: Science, Management and Entrepreneurship (Kluwer Academic Publishers), Mergers & Acquisitions and Corporate Restructuring and Recovery (Elsevier Business Intelligence). He has participated in the organisation of leading conferences in Europe and the USA and also presented numerous research papers at the conferences. He has published in leading journals such as the Multinational Finance Journal and the Journal of Financial Transformation. He is a former member of the Board of the European Finance Association.

James S. Ang, Bank of America Eminent Scholar, Professor of Finance, College of Business, Florida State University. He joined the College of Business of Florida State University as the William O. Cullom Chair Professor of Finance in 1980 from Purdue University. His main areas of research interest are amongst others, corporate restructuring, corporate governance and control. He has published extensively in leading academic journals such as Journal of Finance, Journal of Financial Economics, Journal of Financial and Quantitative Analysis, Journal of Business, Journal of Corporate Finance, The Bell Journal of Economics, Journal of Money, Credit and Banking and The Review of Economics and Statistics. He was a two term Editor of the journal Financial Management, and a past president of the Financial Management Association International. He is a member (current and past) of the Editorial Board of several Journals. He was a member of the Board of Directors of the European Financial Management Association, and a Visiting Professor at the London Business School.

P. Sudi Sudarsanam, Professor of Finance and Corporate Control, School of Management, Cranfield University. Formerly Professor of Finance and Accounting at the Cass Business School in London, he has also been a Visiting Professor at universities in Vienna, Innsbruck, Athens, Szczecin and Chapel Hill, North Carolina. His main areas of research interest are corporate restructuring, mergers and acquisitions, executive compensation, corporate behavioral finance and corporate governance. He is one of the leading authorities on mergers and acquisitions in Europe and author of The Essence of Mergers and Acquisitions (Prentice Hall, 1995), which has been translated into five European and Asian languages. His more recently published second book, Creating Value from Mergers and Acquisitions: The Challenges, an International and Integrated Perspective (FT Prentice Hall, 2003) has been acclaimed by both practitioners and academics. He has ABOUT THE EDITORS xvii published extensively in leading academic journals such as Financial Management, Journal of Banking & Finance, European Finance Review, European Financial Management and Journal of Industrial Economics. He has presented numerous research papers at leading conferences in Europe and the USA. He has also contributed to Acquisitions Monthly. He is on the editorial board of the Journal of Business Finance & Accounting. He is a member of the UK Competition Commission and of its Expert Committee on Cost of Capital. Preface*

The last few years have probably seen more articles and books on both Corporate Governance and Corporate Finance than ever before.Whilst much has happened in the field of Corporate Governance following a number of serious accidents such as Enron and Ahold, there are still some very fundamental issues, which will have to be addressed. Moreover this is a field where globalisation of rules and attitudes has not, or not yet, been achieved. First of all there is the fundamental difference between the US approach of strict and often detailed legislation on the one side and the European attitude that a Company Board can deviate from the required – or, more appropriately, recommended – course of action by giving an explanation which then has to be accepted by the shareholders. Or to put it in the vernacular, the well known ‘comply or explain’ possibility most European codes offer has a counterpart in ‘comply or go to jail’ on the other side of the Atlantic. Sarbanes Oxley is a case in point and European companies with a New York listing frequently are confronted with this. A second issue is the question of whether more supervision by independents (however defined) also means more effective supervision. The problem remains that in the end the supervision of a group of relative outsiders can never be complete but will depend on the work of other outsiders, notably the external auditors. As an aside I would like to mention that the almost endemic inclination to change accounting conventions and definitions with disturbing regularity has been a further impediment to clarity in reporting. Even the most active and experienced Audit Committee will have to accept that there are limits to what they can control. In other words, in the end the supervisors will have, above all, to judge the integrity of the company management and in case of doubt be prepared to take action. The third fundamental question is whether more and better supervision leads to better company results. Here some more research seems to be in order but the few articles I have seen are inconclusive.That is a serious matter because the new approach to corporate government, notably Sarbanes Oxley, allegedly generates very high costs and amounts of a $100 million and more are mentioned as the cost of introducing the system in major multinational companies. Many publications address other important issues, such as the legitimacy of Boards, whom they represent, how they should be nominated/elected or how their independence can be assured. Over the past decade the Anglo-American tradition of giving primacy of place to the shareholder has by now in fact been accepted in most European countries, with the probable exception of Germany, where codetermination still has in important place in corporate governance. Another topic for discussion is the question of the requirements for a good Board member. Integrity, transparency and activity – meaning the willingness to take action when necessary – are probably the most important for independent Board members. For the sake of completeness I mention the European debate on the PREFACE xix one-tier/two-tiers Board, an issue that, in my view, will gradually disappear with the de facto convergence between the two systems, which one can see in practice. This would especially be the case for UK Boards where the recommended split of the CEO and the Board Chairmanship (duality) in practice means that the board functioning is in many respects comparable to the Rhineland model. Corporate Finance and notably the merger and acquisition aspect of it, has also recently drawn more attention from the public, the experts and the lawmakers.This is not the place to discuss the dangers and merits of the growth of private equity groups and hedge funds, two phenomena that are rather different, though sometimes confused in popular perception. Suffice it here to mention that one of the reasons for the rapid growth of private equity, which is mentioned with increasing frequency, is the increasing burden of corporate governance requirements as imposed by the authorities. There is obviously another more direct relationship between Corporate Governance and Corporate Finance, certainly in the case of a takeover by a financial – as opposed to an industrial – party.The financial party will also be inclined, at least initially, to be hostile, although in the course of the process this may gradually change by applying techniques such as the ‘bear hug’. A threat of a takeover or acquisition by a financial consortium will often – though not always – be a reflection of the market judgement of the performance of the management. The bidders feel, rightly or wrongly, that they can use the company’s assets more productively than the present management. In effect the hostile bidders criticise by implication the attitude and judgement of the independent directors.They are seen as responsible for maintaining and tolerating a performance that outsiders apparently consider below the required standards. A new owner, allegedly, would be able to improve profitability by changing the existing situation, probably by first replacing (part of) the management and then proceeding by hiving off non-core assets, however defined. A further step is splitting the company in parts and selling these to interested parties, the ‘buy and break’ scenario. Burdening the acquired company with considerable loans at relatively heavy costs would then usually finance the investment. The risk of an undesired offer from a hostile financial partner increases therefore if the Board has not been able or willing to exercise a proper level of quality control on the performance of the senior executives.To avoid this, the non-executive Board members have to be absolutely independent and sufficiently courageous to take action if and when necessary. Related to this is the problem of formulating and implementing a fair remuneration policy, which encourages and rewards good performance but does not accept compromises if the target results are not achieved.The recent reports on the development of some practices with the pricing of options in the US demonstrate that not every Board is sufficiently aware of its responsibility to hold the management to established targets. The subjects of Corporate Governance and Corporate Finance are therefore related in a number of interesting and often complex ways and this book makes a timely contribution to the discussion.

FLORIS MALJERS

NOTE

* Floris Maljers is former Chairman of the Advisory Board for RSM Erasmus University and former CEO of Unilever. With his outstanding business experience and contacts he still helps companies like Philips, KLM, Guinness and BP in a series of non-executive roles. He is emeritus professor of strategic management at RSM Erasmus University. Acknowledgements

The editors and publishers wish to thank the authors and the following publishers who have kindly given permission for the use of copyright material.

Academic Press Inc. for article: Julian Franks, Colin Mayer and Luc Renneboog (2001), ‘Who disciplines management in poorly performing companies?’, Journal of Financial Intermediation, 10(3–4), 209–248.

Blackwell Publishing Ltd for articles: Michael C. Jensen (1993),’The modern industrial revolution, exit and the failure of internal control systems’, Journal of Finance, 48(3), 831–880; Andrei Shleifer and Robert W. Vishny (1997), ‘A survey of corporate governance’, Journal of Finance, 52(2), 737–784; James S. Ang, Rebel Cole, and James Wuh Lin (2000), ‘Agency costs and ownership structure’, Journal of Finance, 55(1), 81–106; Martin J. Conyon and Kevin J. Murphy (2000), ‘The prince and the pauper? CEO pay in the United States and United Kingdom’, The Economic Journal, 110(Nov), 640–671; Jay Dahya, John J. McConnell and Nickolaos Travlos (2002), ‘The Cadbury Committee, corporate performance and management turnover’, Journal of Finance, 57(1), 461–483; Jorge Farinha (2003), ‘Dividend policy, corporate governance and the managerial entrenchment hypothesis: an empirical analysis’, Journal of Business Finance & Accounting, 30(9–10), 1173–1210. Elsevier Ltd for articles: Julian Franks and Colin Mayer (1996), ‘Hostile takeovers and the correction of managerial failure’, Journal of Financial Economics, 40(1), 163–181; Wayne R. Guay (1999), ‘The sensitivity of CEO wealth to equity risk: an analysis of the magnitude and determinants’, Journal of Financial Economics, 53(1), 43–71; Marco Becht and Ailisa Röell (1999), ‘Blockholdings in Europe: an international comparison’, European Economic Review, 43(4–6), 1049–1056; Rafael La Porta, Florencio Lopez- de-Silanes, Andrei Shleifer and Robert Vishny (2000), ‘Investor protection and corporate governance’, Journal of Financial Economics, 58(1–2), 3–27; Mara Faccio and M. Ameziane Lasfer (2000), ‘Do occupational pension funds monitor companies in which they hold large stakes?’, Journal of Corporate Finance, 6(1), 71–110; Luc Renneboog (2000), ‘Ownership, managerial control and the governance of companies listed on the Brussels stock exchange’, Journal of Banking & Finance, 24(12), 1959–1995; John E. Core and David F.Larcker (2001),‘Performance consequences of mandatory increases in executive stock ownership’, Journal of Financial Economics, 64(3), 317–340;Tim Jenkinson and Alexander Ljungqvist (2001), ‘The role of hostile stakes in German corporate governance’, Journal of Corporate Finance, 7(4), 397–446; Paolo F. Volpin (2002), ‘Governance with poor investor protection: evidence from top executive turnover in Italy’, Journal of Financial Economics, 64(1), 61–90; Mara Faccio and Larry H. P.Lang (2002), ACKNOWLEDGEMENTS xxi ‘The ultimate ownership of Western European corporations’, Journal of Financial Economics, 65(3), 365–395; N. K. Chidambaran and Nagpurnanand R. Prabhala, (2003),‘Executive stock option repricing, internal governance mechanisms, and management turnover’, Journal of Financial Economics, 69(1), 153–189; Julie Ann Elston and Lawrence G. Goldberg (2003), ‘Executive compensation and agency costs in Germany’, Journal of Banking & Finance, 27(7), 1391–1410; Christian Leuz, Dhananjay Nanda and Peter D. Wysocki (2003), ‘Earnings management and investor protection: an international comparison’, Journal of Financial Economics, 69(3), 505–527; Jens Köke (2004), ‘The market for corporate control in a bank-based economy: a governance device?’ Journal of Corporate Finance, 10(1), 53–80.

John Wiley & Sons Ltd for article: Rezaul Kabir, Dolph Cantrijn and Andreas Jeunink (1997), ‘Takeover defences, ownership structure and stock returns: an empirical analysis with Dutch data’, Strategic Management Journal, 18(2), 97–109.

MIT Press Journals for article: Paul Gompers, Joy Ishii and Andrew Metrick (2003), ‘Corporate governance and equity prices’, The Quarterly Journal of Economics, 118(1), 107–155.

Multinational Finance Society for article: Gertjan Schut and Ruud van Frederikslust (2004), ‘Shareholder wealth effects of joint venture strategies’, Multinational Finance Journal, 8(3–4), 211–225.

Oxford University Press for article: Marc Goergen and Luc Renneboog (2003), ‘Why are the levels of controls so different in German and UK companies? Evidence from initial public offerings’, Journal of Law, Economics and Organization, 19(1), 141–175.

Springer Science B.V. for articles: Jim Lai and Sudi Sudarsanam (1997), ‘Corporate restructuring in response to performance decline: impact of ownership, governance and lenders’, European Finance Review, 1(2), 197–233; Hiroshi Osano (1999), ‘Security design, insider monitoring, and financial market equilibrium’, European Finance Review, 2(3), 273–302; Erik Lehmann and Jürgen Weigand (2000), ‘Does the governed corporation perform better? Governance structures and corporate performance in Germany’, European Finance Review, 4(2), 157–195. In addition the editors and publishers wish to thank the Library of RSM Erasmus University for their assistance in obtaining these articles.

General introduction1

LTHOUGH MUCH HAS BEEN PUBLISHED ON corporate governance in recent Ayears, the predominant paradigm is a US one. Yet the European context is highly significant in both its differences and similarities to the US experience. Continental Europe is characterised by a governance system in which large block shareholders play a key role, whereas the US system is driven by the needs of dispersed shareholders articulated through the stock market.The UK, however, shares much similarities with the US system although it has its own idiosyncratic features and institutional characteristics. There has been much debate over the relative merits of the block holder-dominated and stock market-dominated systems in promoting economic efficiency and growth, but this remain an inconclusive debate.There is also a trend towards greater convergence of some, or all, aspects of corporate governance driven in part by the forces of globalisation of product and capital markets but also inspired by political visions of greater harmonisation within Europe and between Europe and the rest of the world. Thus corporate governance, far from being monochromatic, presents a rich variety of laws, practices and institutional structures. An understanding of this variety is a necessary precondition to understanding how corporate governance is likely to evolve in the future and affect the functioning of companies and capital markets, as well as influence the practice of corporate finance. The architecture of corporate governance differs across countries but consists broadly of the internal managerial control and monitoring mechanisms and the external monitoring and control mechanisms.The former include the board of structures, the presence of block shareholders, managerial incentive contracts and other devices to ensure alignment of shareholder and manager interests. The external mechanisms include the stock market and the market for corporate control (i.e. hostile takeovers).The relative balance among these mechanisms and their relative efficiencies differ among countries. The present collection of papers focuses on corporate governance and its impact in several European countries. This volume brings together many of the most important classic and contemporary published articles on the main elements of corporate governance and their impact on various aspects of corporate finance within a European perspective. This collection illustrates and explains the differing perspectives on corporate governance frameworks and the interrelated set of mechanisms that constitute the corporate governance architecture. Some papers discuss the structural design features such as the ownership structure (e.g. institutional shareholders and block shareholders), financial structure, and structure of the board of directors. Other papers deal with the governance processes (e.g. executive compensation arrangements and pay-for-performance sensitivities). The role of takeovers as a managerial control device is also discussed.The impact of corporate governance structure as well as the related processes on certain corporate strategic and financing policies is the subject of a few other papers. 2 CORPORATE GOVERNANCE AND CORPORATE FINANCE The articles in this volume are organised under six subject headings:

1. Alternative perspectives on corporate governance systems 2. Equity ownership structure and control 3. Corporate governance, underperformance and management turnover 4. Directors’ remuneration 5. Governance, performance and financial strategy 6. On takeover as disciplinary mechanism

BROAD OVERVIEW OF THE COLLECTED PAPERS

Part 1 Alternative perspectives on corporate governance systems This part deals with corporate governance systems and presents alternative typologies of such systems based on the institutional character (i.e. bank-centred or stock market-centred), the legal traditions of countries and the extent of legal protection of investor rights against firms. The legal traditions are also shown as the outcome of long political and economic processes in different countries. The papers provide the historical backdrop to the observed differences in corporate governance systems. The relationship between corporate governance systems on the one hand and cost of capital and sources of corporate finance on the other is explored. One element of the cost of capital in the form of agency costs is shown to depend on the ownership structure of firms and the extent of owner–manager alignment (Ang et al., 2000).

Part 2 Equity ownership structure and control The efficacy of a corporate governance system requires effective monitoring of managers by the shareholders. Shleifer and Vishny (1997) have argued that large shareholders can perform this monitoring function.They also acknowledge that this role imposes costs such as free riding by non-block shareholders who may avoid the monitoring effort but reap the benefits of the block shareholder’s monitoring. Large shareholders in this event may only be willing to undertake monitoring if they can expect to receive compensation for their efforts including an appropriate share of the added value that monitoring generates.They may also look to design the securities they hold in the firms with features that guarantee such reward. In the first paper of this part, Hiroshi Osano (1999) addresses the problem of security design that will satisfy a large investor and motivate her to undertake monitoring.The author shows that the optimal security is a debt-like security such as standard debt with a positive probability of default, or debt with call options. Such a design also leads to the financial market equilibrium being Pareto optimal. The remaining three papers by Becht and Röell (1999), Faccio and Lang (2002) and Goergen and Renneboog (2003) present empirical evidence on the ownership structure of European corporations with particular focus on block shareholdings and control exercised through pyramidal structures. Pyramidal structures allow block holders to control vast groups of companies with relatively small investment capital.

Part 3 Corporate governance, underperformance and management turnover Part 3 comprises papers that deal with the effectiveness of corporate governance mechanisms in disciplining top managers. A measure of such discipline is the removal or turnover of such managers in response to poor corporate performance. Where such managers are entrenched owing to their ownership of the company’s voting shares, connections to controlling shareholders or both, this form of disciplining is unlikely and the corporate GENERAL INTRODUCTION 3 governance mechanism is rendered ineffective.The controlling large block shareholder may in theory have an incentive to improve corporate performance, if necessary by disciplining top mangers of underperformers but this concern may be overridden by their enjoyment of private benefits of control. Minority shareholders, however,do not enjoy such countervailing private benefits. Large shareholding may thus align the interest of managers and large shareholders but create another agency problem (i.e. between large, controlling shareholders and minority shareholders). The phenomenon whereby controlling shareholders manage to receive private benefits at the expense of minority shareholders is called ‘tunnelling’.

Part 4 Directors’ remuneration One of the important tools in the armoury of the board of directors to align the interests of shareholders and mangers is the executive compensation contract for the CEO and other executive managers. In theory, a properly written contract should achieve such alignment by making level of compensation conditional upon performance. However, in practice, there are problems in both contract writing and enforcement and in specifying the appropriate performance measures and the time horizon for the award of compensation. Many scholars (e.g. Bebchuk and Fried, 2004) have argued that entrenched top managers capture the pay setting process and manipulate the boards into awarding them excessive pay.They argue that such managers often get rewarded for failure rather than success in delivering performance and shareholder value. Thus, far from being a solution to the agency problem, executive compensation may itself be a manifestation of an agency problem and reflect failure of the governance arrangements in firms. The sensitivity of pay to performance is therefore an important test of governance effectiveness. Executive compensation arrangements may vary from country to country depending on governmental and public attitudes to ‘high’ managerial remuneration. The papers in this part deal with many of these issues and provide empirical evidence.

Part 5 Governance, performance and financial strategy The papers in this part deal with the impact of governance structure on corporate performance and its financial strategy.

Corporate governance and performance Although there has been an intensive debate on the relative merits of different systems of corporate governance, empirical evidence on the link between corporate governance and firm performance almost exclusively refers to the market-oriented Anglo-American system. The paper of Lehmann and Weigand (2000) therefore investigates the more network- or bank-oriented German system. In panel regressions for 361 German corporations over the time period 1991 to 1996, significantly, the authors find ownership concentration to negatively affect profitability. However, this effect depends intricately on stock market exposure, the location of control rights, and the time horizon (short-run vs. long-run).They conclude that (1) the presence of large shareholders does not necessarily enhance profitability, (2) ownership concentration seems to be sub-optimal for many German corporations, and, finally, (3) having financial institutions as largest shareholders of traded corporations improves corporate performance.

Corporate governance and corporate financial strategy The governance structure of strategic investments and its impact on both the investment itself and the investing firms is a matter of interest to firms. Often strategic investments 4 CORPORATE GOVERNANCE AND CORPORATE FINANCE may fail due to poor governance structures as evidenced by numerous studies on post-merger integration (Sudarsanam, 2003, chapter 22). Such governance problems also plague joint ventures and strategic alliances (Sudarsanam, 2003, chapter 10). Schut and Van Frederikslust (2004) show that strategic intention, the context in which the strategy is unfolded and the extent to which the company has ownership and managerial control over the implementation of the joint venture, strongly explains the extent to which it can create value. Financial distress is another context in which the influence of corporate governance on the strategic and other decisions a firm makes can be studied. Firms in performance decline may choose a variety of restructuring strategies for recovery with conflicting welfare implications for different stakeholders such as shareholders, lenders and managers. Choice of recovery strategies is therefore determined by the complex interplay of ownership structure, corporate governance and lender monitoring of such firms.The paper of Lai and Sudarsanam (1997) points to the rich and complex ways in which different corporate governance mechanisms interact, sometimes conflicting with and, at others, complementing one another.

Part 6 On takeover as disciplinary mechanism In Part 3 we collected papers that examine the relationship between different corporate control devices and top management turnover as a manifestation of their disciplinary effect.The control devices are generally internal to the control or ownership structure of the underperforming firms. In his part, we present papers that deal with an external control device. This is the market for control in which management teams compete for control of corporate assets. Manne (1995) was the first to conceptualise the market for corporate control as a managerial disciplinary device with underperforming firms becoming targets of bidders with presumably greater ability to correct underperformance and create greater value for target shareholders. Hostile takeover is the means by which corporate control is wrested from the underperforming target managers. In theory such an external disciplinary device may be considered redundant if the internal control mechanisms are effective.The incidence of a hostile takeover may thus be an indictment of the failure of the internal controls. This argument pre-supposes that internal controls and hostile takeover are substitutes. A contrary perspective is that efficient internal controls facilitate hostile takeovers by preventing the entrenched incumbent management of the target firm from raising the barricades against managerial change. In his view hostile takeover and robust internal controls are complementary tools serving the same purpose.The papers in this part deal with the role of hostile takeovers as a disciplinary device and how governance regimes in certain countries substitute for them or facilitate them. This volume brings together key articles, which represent the current state of theories and practices of corporate governance and corporate finance, ensuring that this collection will be an invaluable resource for scholars, students and practitioners in business law, managerial finance, and financial economics.

NOTE

1 We acknowledge comments and suggestions on prior drafts by W. Koppelman and S. Witteman.

REFERENCES

Ang, J.S., Cole, R. and Wuh Lin, J. (2000), ‘Agency costs and ownership structure’, Journal of Finance, 55(1): 81–106. GENERAL INTRODUCTION 5 Bebchuk, L. and Fried, J. (2004), Pay without performance, Cambridge, Mass: Harvard University Press. Becht, M. and Röell, A. (1999), ‘Block holdings in Europe: an international comparison’, European Economic Review, 43(4–6): 1049–1056. Faccio M. and Lang, Larry H.P.(2002),‘The ultimate ownership of Western European corporations’, Journal of Financial Economics, 65(3): 365–395. Goergen, M. and Renneboog, L. (2003), ‘Why are the levels of controls (so) different in German and UK companies? Evidence from initial public offerings’, Journal of Law, Economics and Organization, 19(1): 141–175. Lai, J. and Sudarsanam, S. (1997), ‘Corporate restructuring in response to performance decline: impact of ownership, governance and lenders’, European Finance Review, 1(2): 197–233. Lehmann, E. and Weigand, J. (2000), ‘Does the governed corporation perform better? Governance structures and corporate performance in Germany’, European Finance Review, 4(2): 157–195. Manne, H.G. (1965),‘Merger and the market for corporate control’, Journal of Political Economy, 73: 110–120. Osano, H. (1999), ‘Security design, insider monitoring, and financial market equilibrium’, European Finance Review, 2(3): 273–302. Schut G.J. and van Frederikslust R. (2004),‘Shareholder wealth effects of joint venture strategies’, Multinational Finance Journal, 8(3–4): 211–225. Shleifer A. and Vishny R.W. (1997),‘A survey of corporate governance’, Journal of Finance, 52(2): 737–783. Sundarsanam, P.S. (2003), Creating value from mergers and acquisitions. The challenges, an international and integrated perspective, New York: FT Prentice Hall.

Part 1 Alternative perspectives on corporate governance systems

INTRODUCTION

ART 1 DEALS WITH CORPORATE GOVERNANCE SYSTEMS, and presents Palternative typologies of such systems based on the institutional character (i.e. bank- centred or stock market-centred), the legal traditions of countries and the extent of legal protection of investor rights against firms.The legal traditions are also shown as the outcome of long political and economic processes in different countries. The papers provide the historical backdrop to the observed differences in corporate governance systems. The relationship between corporate governance systems on the one hand and cost of capital and sources of corporate finance on the other is explored. One element of the cost of capital in the form of agency costs is shown to depend on the ownership structure of firms and the extent of owner-manager alignment. In the first paper Jensen (Ch.1) identifies alternative corporate control mechanisms and how failures in the internal control mechanism (i.e. the internal corporate governance failed to prevent excessive and unprofitable investments, for example, in R&D). He compares the 1980s restructuring of US industry with a similar restructuring of the 1890s. He traces the failure of the internal governance mechanism to the processes and characteristics of the board of directors and argues that the market for corporate control played a key role in the 1980s in unravelling the excess capacity built up in earlier decades. In his view corporate boards in the US have been ‘captured’ by management. Jensen argues for the organisational and governance innovations introduced by the venture capital and leveraged buyout firms as more effective corporate control mechanisms. The paper provides a long historical perspective on the evolution of alternative corporate control mechanisms and how they complement, or substitute for, one another.The role of corporate governance in corporate restructuring and in turn in promoting efficient allocation of resources is an important focus of this seminal paper. Shleifer and Vishny (Ch.2) focus on legal protection of shareholder rights as a way of preventing managers from stealing the funds entrusted to them by shareholders and ensuring that managers deliver adequate returns on their investments. While legal protection may solve extreme cases of expropriation and stealing, it may be inadequate in ensuring that managers make investment decisions in the shareholders’ interests and not in their own. Concentrated ownership may provide a solution to this agency problem by allowing large shareholders to monitor and control managers but this solution creates problems of its own (i.e. excessive risk exposure, potential expropriation of dispersed owners by the large shareholder in collusion with managers, and enhancing their own private benefits rather than the firm value). Moreover, incompetence of entrenched managers may be costlier to shareholders than managerial malfeasance or self-interest pursuit. 8 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Shleifer and Vishny survey corporate governance systems around the world covering shareholders’ legal rights and their enforcement regime, the monitoring role of banks as creditors, and concentrated ownership structures that result from leveraged buyouts (LBOs). They identify three principal governance systems – US and the UK with strong shareholder rights, Germany with permanent large shareholders and weak shareholder rights and Japan that falls between the other two. Comparing corporate governance systems around the world, Shleifer and Vishny conclude that legal protection of shareholder rights in combination with large shareholders provides the elements of effective corporate governance and cite the US, UK, Germany and Japan as examples of such optimal combination. They note, in contrast, how weak shareholder rights and concentrated ownership by family interests have reduced Italian firms’ access to stock and loan markets. In Chapter 3, La Porta, Lopez-de-Silanes, Shleifer and Vishny follow the same legalistic line of exposition by elevating shareholder rights protection as a central framework for examining corporate governance systems in different countries. They define corporate governance as a set of mechanisms by which outside investors – shareholders and creditors – protect themselves against expropriation by insiders (i.e. large controlling shareholders and managers). By rendering expropriation technology expensive or ineffective, the legal mechanism increases the access corporations have to external finance. La Porta et al. compare the efficiency of private contracts subject to enforcement by the court (the Coase theorem, 1937) to that of laws and regulations that explicitly limit the scope for expropriation and argue that even in countries with a judiciary that can arbitrate private contracts, such laws and regulations can be efficient and add value.They classify countries into four groups based on their legal origin – common law in the UK, the US and the British Commonwealth, French civil law, German civil law and Scandinavian civil law. They find that common law countries afford better investor protection than civil law countries while, within civil law countries, Germany is more efficient than France in enforcement of investor rights. La Porta et al. explore the impact of investor protection on ownership structure of firms, financial development and resource allocation and conclude that the impact is positive.They also find the classification of governance systems as bank-centred or stock market-centred is unhelpful in evaluating their relative efficiencies or in explaining the prevalent patterns of corporate financing. They advocate strong legal protection of investor rights, a reliable enforcement mechanism and, in the absence of these, a robust market regulator such as the SEC in the US to enhance corporate governance that promotes financial development. The paper thus clearly establishes the linkage between shareholder rights as an element of corporate governance and corporate finance. While agency cost resulting from the divorce of managerial control of a firm from its ownership has emerged, since Jensen and Meckling’s seminal paper in 1976, as a central problem of corporate governance and its reduction a measure of the efficiency of a corporate governance structure, plausible proxies for agency cost have eluded empirical researchers. As Ang et al. (Ch.4) argue, the base case of a firm free of any agency cost is one owned and managed by the same person/s. However empirical data on such firms have been difficult to obtain since such firms are not publicly listed and are small. Using unique data from the National Survey of Small Business Finances in the US, the authors’ test many of the implications of the Jensen and Meckling model.They are able to use proxies for agency costs between owner-managed firms and firms with nonmanaging shareholders. They find that the former are significantly more efficient in using their assets (i.e. they display less shirking and incur significantly lower levels of operating costs, that is, they display less perquisite consumption). Ang et al. also report that agency costs decrease as ownership becomes more concentrated and creditors provide additional, but much weaker,monitoring of managers.These results are consistent with the benign role of large ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 9 shareholders advocated by Shleifer and Vishny, and of leverage recommended by Jensen, in the first three papers in this part.

REFERENCES

Coase, Ronald (1937) ‘The nature of the firm’, Economica, 4: 386–405. Jensen, Michael and William Meekling (1976) ‘Theory of the firm: managerial behavior, agency costs, and ownership structure’, Journal of Financial Economics, 3: 305–360.

Chapter 1

Michael C. Jensen THE MODERN INDUSTRIAL REVOLUTION, EXIT, AND THE FAILURE OF INTERNAL CONTROL SYSTEMS

Source: Journal of Finance, 48(3) (1993): 831–880.

ABSTRACT Since 1973 technological, political, regulatory, and economic forces have been changing the worldwide economy in a fashion comparable to the changes experienced during the nineteenth century Industrial Revolution. As in the nineteenth century, we are experiencing declining costs, increasing average (but decreasing marginal) productivity of labor, reduced growth rates of labor income, excess capacity, and the requirement for downsizing and exit. The last two decades indicate corporate internal control systems have failed to deal effectively with these changes, especially slow growth and the requirement for exit. The next several decades pose a major challenge for Western firms and political systems as these forces continue to work their way through the worldwide economy.

I. INTRODUCTION change leading to declining costs, increasing average but decreasing marginal productivity Parallels between the modern and of labor, reduced growth rates in labor historical industrial revolutions income, excess capacity, and—ultimately— downsizing and exit. FUNDAMENTAL TECHNOLOGICAL, POLITICAL, The capital markets played a major role regulatory, and economic forces are radi- in eliminating excess capacity both in the cally changing the worldwide competitive nineteenth century and in the 1980s. The environment. We have not seen such a merger boom of the 1890s brought about a metamorphosis of the economic landscape massive consolidation of independent firms since the Industrial Revolution of the and the closure of marginal facilities. In the nineteenth century. The scope and pace of 1980s the capital markets helped eliminate the changes over the past two decades excess capacity through leveraged acquisi- qualify this period as a modern industrial tions, stock buybacks, hostile takeovers, revolution, and I predict it will take leveraged buyouts, and divisional sales. Just decades for these forces to be fully worked as the takeover specialists of the 1980s out in the worldwide economy. were disparaged by managers, policymakers, Although the current and historical eco- and the press, the so-called Robber Barons nomic transformations occurred a century were criticized in the nineteenth century. In apart, the parallels between the two are both cases the criticism was followed by strikingly similar: most notably, the wide- public policy changes that restricted the spread technological and organizational capital markets: in the nineteenth century 12 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS the passage of antitrust laws restricting material ruin. ...The fruits of the toil of combinations, and in the late 1980s the millions are boldly stolen to build up colos- reregulation of the credit markets, anti- sal fortunes for the few, unprecedented in takeover legislation, and court decisions the history of mankind; and the posses- that restricted the market for corporate sors of these in turn despise the republic control. and endanger liberty. From the same Although the vast increases in productivity prolific womb of government injustice are associated with the nineteenth century bred two great classes of tramps and industrial revolution increased aggregate millionaires. (McMurray (1929), p. 7). welfare, the large costs associated with the obsolescence of human and physical Technological and other developments capital generated substantial hardship, mis- that began in the mid-twentieth century understanding, and bitterness. As noted in have culminated in the past two decades in 1873 by Henry Ward Beecher,a well-known a similar situation: rapidly improving pro- commentator and influential clergyman of ductivity, the creation of overcapacity and, the time, consequently, the requirement for exit. Although efficient exit—because of the The present period will always be memo- ramifications it has on productivity and rable in the dark days of commerce in human welfare—remains an issue of great America. We have had commercial dark- importance, research on the topic has been ness at other times. There have been relatively sparse since the 1942 publication these depressions, but none so obstinate of Schumpeter’s insights on creative and none so universal . . . Great Britain destruction.1 These insights will almost cer- has felt it; France has felt it; all Austria tainly receive renewed attention in the and her neighborhood has experienced it. coming decade: It is cosmopolitan. It is distinguished by its obstinacy from former like periods of Every piece of business strategy acquires commercial depression. Remedies have its true significance only against the no effect. Party confidence, all stimulat- background of that process and within ing persuasion, have not lifted the pall, the situation created by it. It must be and practical men have waited, feeling seen in its role in the perennial gale of that if they could tide over a year they creative destruction; it cannot be under- could get along; but they could not tide stood irrespective of it or, in fact, on the over the year. If only one or two years hypothesis that there is a perennial could elapse they could save themselves. lull ...The usual theorist’s paper and The years have lapsed, and they were the usual government commission’s worse off than they were before. What is report practically never try to see that the matter? What has happened? Why, behavior, on the one hand, as a result of from the very height of prosperity with- a piece of past history and, on the other out any visible warning, without even a hand, as an attempt to deal with a situa- cloud the size of a man’s hand visible on tion that is sure to change presently—as the horizon, has the cloud gathered, as it an attempt by those firms to keep on were, from the center first, spreading all their feet, on ground that is slipping over the sky? (Price (1933), p. 6). away from under them. In other words, the problem that is usually being visual- On July 4, 1892, the Populist Party plat- ized is how capitalism administers existing form adopted at the party’s first conven- structures, whereas the relevant problem tion in Omaha reflected similar discontent is how it creates and destroys them. and conflict: (Schumpeter (1976), p. 83).

We meet in the midst of a nation brought Current technological and political to the verge of moral, political, and changes are bringing this issue to the THE MODERN INDUSTRIAL REVOLUTION 13 forefront. It is important for managers, excellent discussions of the period by policymakers, and researchers to under- Chandler (1977, 1990, 1992), McCraw stand the magnitude and generality of the (1981, 1992), and Lamoreaux (1985).) implications of these forces. Originating in Britain in the late eighteenth century, the First Industrial Revolution— as Chandler (1990, p. 250) labels it—wit- Outline of the paper nessed the application of new energy In this paper,I review the industrial revolu- sources to methods of production.The mid- tions of the nineteenth century and draw on nineteenth century witnessed another wave these experiences to enlighten our under- of massive change with the birth of modern standing of current economic trends. transportation and communication facili- Drawing parallels to the 1800s, I discuss in ties, including the railroad, telegraph, some detail the changes that mandate exit steamship, and cable systems. Coupled with in today’s economy. I address those factors the invention of high-speed consumer pack- that hinder efficient exit, and outline the aging technology, these innovations gave control forces acting on the corporation to rise to the mass production and distribution eventually overcome these barriers. systems of the late nineteenth and early Specifically, I describe the role of the twentieth centuries—the Second Industrial market for corporate control in affecting Revolution (Chandler (1990), p. 62). efficient exit, and how the shutdown of the The dramatic changes that occurred capital markets has, to a great extent, from the middle to the end of the century transferred this challenge to corporate clearly warranted the term “revolution.” internal control mechanisms. I summarize The invention of the McCormick reaper evidence, however, indicating that internal (1830s), the sewing machine (1844), and control systems have largely failed in high-volume canning and packaging devices bringing about timely exit and downsizing, (mid-1880s) exemplified a worldwide leaving only the product market or surge in productivity that “substituted legal/political/regulatory system to resolve machine tools for human craftsmen, excess capacity. Although overcapacity will interchangeable parts for hand-tooled in the end be eliminated by product market components, and the energy of coal for that forces, this solution generates large, of wood, water, and animals” (McCraw unnecessary costs. I discuss the forces that (1981), p. 3). New technology in the paper render internal control mechanisms inef- industry allowed wood pulp to replace rags fective and offer suggestions for their as the primary input material (Lamoreaux reform. Lastly, I address the challenge this (1985), p. 41). Continuous rod rolling modern industrial revolution poses for transformed the wire industry: within a finance professionals; that is, the changes decade, wire nails replaced cut nails as the that we too must undergo to aid in the main source of supply (Lamoreaux (1985), learning and adjustments that must occur p. 64). Worsted textiles resulting from over the next several decades. advances in combining technology changed the woolen textile industry (Lamoreaux (1985), p. 98). Between 1869 and 1899, the capital invested per American manu- II. THE SECOND INDUSTRIAL facturer grew from about $700 to $2,000; REVOLUTION in the period 1889 to 1919, the annual growth of total factor productivity was The Industrial Revolution was distin- almost six times higher than that which guished by a shift to capital-intensive pro- had occurred for most of the nineteenth duction, rapid growth in productivity and century (McCraw (1981), p. 3). living standards, the formation of large As productivity climbed steadily, produc- corporate hierarchies, overcapacity, and, tion costs and prices fell dramatically. The eventually, closure of facilities. (See the 1882 formation of the Standard Oil Trust, 14 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS which concentrated nearly 25 percent of marginal facilities in the merged entities. the world’s kerosene production into three Between 1895 and 1904, over 1,800 firms refineries, reduced the average cost of a were bought or combined by merger into gallon of kerosene by 70 percent between 157 firms (Lamoreaux (1985), p. i.). 1882 and 1885. In tobacco, the invention of the Bonsack machine in the early 1880s reduced the labor costs of cigarette III. THE MODERN INDUSTRIAL production 98.5 percent (Chandler (1992), REVOLUTION p. 5).The Bessemer process reduced the cost of steel rails by 88 percent from the early The major restructuring of the American 1870s to the late 1890s, and the business community that began in the electrolytic refining process invented in the 1970s and is continuing in the 1990s is 1880s reduced the price of a kilo of being brought about by a variety of factors, aluminum by 96 percent between 1888 and including changes in physical and manage- 1895 (Chandler (1992), pp. 4–6). In chem- ment technology, global competition, icals, the mass production of synthetic regulation, taxes, and the conversion of dyes, alkalis, nitrates, fibers, plastics, and formerly closed, centrally planned socialist film occurred rapidly after 1880. and communist economies to capitalism, Production costs of synthetic blue dye, for along with open participation in interna- example, fell by 95 percent from the 1870s tional trade. These changes are significant to 1886 (Chandler (1992), p. 5). New lost- in scope and effect; indeed, they are bring- cost sources of superphosphate rock and ing about the Third Industrial Revolution. the manufacture of superphosphates To understand fully the challenges that changed the fertilizer industry. In sugar current control systems face in light of this refining, technological changes dramati- change, we must understand more about cally lowered the costs of sugar production these general forces sweeping the world and changed the industry (Lamoreaux economy, and why they are generating (1985), p. 99). excess capacity and thus the requirement Lamoreaux (1985) discusses other cases for exit. where various stimuli led to major What has generally been referred to as increases in demand and, in turn, expansion the “decade of the 80s” in the United that led to excess capacity (the page States actually began in the early 1970s numbers in parentheses reference her with the ten-fold increase in energy prices discussions). This growth occurred in cere- from 1973 to 1979, and the emergence of als (when “Schumacher broke down the the modern market for corporate control, American prejudice against eating oats” and high-yield nonrated bonds in the mid- (p. 98)), whisky (when crop failures in 1970s. These events, among others, were Europe created a sudden large demand for associated with the beginnings of the Third U.S. producers (p. 99)), and tin plate Industrial Revolution which—if I were to (when the McKinley tariff raised domestic pick a particular date—would be the time demand and prices (p. 97)). of the oil price increases beginning in 1973. The surplus capacity developed during the period was exacerbated by the fall in demand brought about by the recession and The decade of the 80s: capital panic of 1893. Although attempts were markets provided an early response made to eliminate overcapacity through to the modern industrial revolution pools, associations, and cartels (p. 100), not until the capital markets motivated exit The macroeconomic data available for the in the 1890s’ mergers and acquisitions 1980s shows major productivity gains (M&A) boom was the problem substan- (Jensen (1991)). 1981 was in fact a tially resolved. Capacity was reduced watershed year: Total factor productivity through the consolidation and closure of growth in the manufacturing sector more THE MODERN INDUSTRIAL REVOLUTION 15 than doubled after 1981 from 1.4 percent Indeed, Marty Lipton, prominent defender per year in the period 1950 to 1981 to of American CEOs, expresses a common 3.3 percent in the period 1981 to 1990.2 view of the 1980s when he states that “the Nominal unit labor costs stopped their takeover activity in the United States has 17-year rise, and real unit labor costs imposed short-term profit maximization declined by 25 percent. These lower labor strategies on American Business at the costs came not from reduced wages or expense of research, development and employment, but from increased productivity: capital investment. This is minimizing our Nominal and real hourly compensation ability to compete in world markets and increased by a total of 4.2 and 0.3 percent still maintain a growing standard of living per year respectively over the 1981 to at home” (Lipton (1989), p. 2). 1989 period.3 Manufacturing employment On average, selling-firm shareholders in reached a low in 1983, but by 1989 had all M&A transactions in the period 1976 to experienced a small cumulative increase of 1990 were paid premiums over market 5.5 percent.4 Meanwhile, the annual value of 41 percent,9 and total M&A trans- growth in labor productivity increased from actions generated $750 billion in gains to 2.3 percent between 1950 and 1981 to target firms’ shareholders (measured in 3.8 percent between 1981 and 1990, while 1992 dollars).10 This value change repre- a 30-year decline in capital productivity sents the minimum forecast value change was reversed when the annual change in by the buyer (the amount the buyer is willing the productivity of capital increased from to pay the seller), and does not include 1.03 percent between 1950 and 1981 to further gains (or losses) reaped by the 2.03 percent between 1981 and 1990.5 buyer after execution of the transaction.11 During the 1980s, the real value of It includes synergy gains from combining public firms’ equity more than doubled the assets of two or more organizations from $1.4 to $3 trillion.6 In addition, real and the gains from replacing inefficient median income increased at the rate of governance systems, as well as possible 1.8 percent per year between 1982 and wealth transfers from employees, commu- 1989, reversing the 1.0 percent per year nities, and bondholders.12 As Shleifer and decrease that occurred from 1973 to Summers (1988) point out, if the value 1982.7 Contrary to generally held beliefs, gains are merely transfers of wealth from real research and development (R&D) creditors, employees, suppliers, or commu- expenditures set record levels every year nities, they do not represent efficiency from 1975 to 1990, growing at an average improvements. Thus far, however, little annual rate of 5.8 percent.8 The Economist evidence has been found to support (1990), in one of the media’s few accurate substantial wealth transfers from any portrayals of this period, noted that from group,13 and it appears that most of these 1980 to 1985 “American industry went on gains represent increases in efficiency. an R&D spending spree, with few big Part of the attack on M&A transactions successes to show for it.” was centered on the high-yield (or so-called Regardless of the gains in productivity, “junk”) bond market, which eliminated efficiency, and welfare, the 1980s are mere size as an effective deterrent against generally portrayed by politicians, the takeover. This opened the management of media, and others as a “decade of greed America’s largest corporations to monitoring and excess.” In particular, criticism was and discipline from the capital markets. It centered on M&A transactions, 35,000 of also helped provide capital for newcomers which occurred from 1976 to 1990, with a to compete with existing firms in the total value of $2.6 trillion (1992 dollars). product markets. Contrary to common beliefs, only 364 of High-yield bonds opened the public these offers were contested, and of those capital markets to small, risky, and unrated only 172 resulted in successful hostile firms across the country, and made it pos- takeovers (Mergerstat Review (1991)). sible for some of the country’s largest firms 16 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS to be taken over. The sentiment of reaching nature of the restructuring, it J. Richard Munro (1989, p. 472), would be surprising if none occurred. Chairman and CEO of Time Inc., exemplifies However,the negative assessment character- the critical appraisal of their role: istic of general opinion is inconsistent with both the empirical evidence and the almost Notwithstanding television ads to the universal opinion of finance scholars who contrary, junk bonds are designed as the have studied the phenomenon. In fact, currency of “casino economics” . . . takeover activities were addressing an they’ve been used not to create new important set of problems in corporate plants or jobs or products but to do the America, and doing it before the companies opposite: to dismantle existing compa- faced serious trouble in the product markets. nies so the players can make their They were, in effect, providing an early profit....This isn’t the Seventh Cavalry warning system that motivated healthy coming to the rescue. It’s a scalping adjustments to the excess capacity that began party. to proliferate in the worldwide economy.

The high leverage incurred in the eighties Causes of excess capacity contributed to an increase in the bankruptcy rate of large firms in the early Excess capacity can arise in at least four 1990s. That increase was also encouraged ways, the most obvious of which occurs by the recession (which in turn was at least when market demand falls below the level partly caused by the restriction in the required to yield returns that will support credit markets implemented in late 1989 the currently installed production capacity. and 1990 to offset the trend toward higher This demand-reduction scenario is most leverage), and the revisions in bankruptcy familiarly associated with recession procedures and the tax code (which made episodes in the business cycle. it much more difficult to restructure finan- Excess capacity can also arise from two cially distressed firms outside the courts, types of technological change. The first see Wruck (1990)). The unwise public type, capacity-expanding technological policy and court decisions that contributed change, increases the output of a given significantly to hampering private adjust- capital stock and organization. An example ment to this financial distress seemed to be of the capacity-expanding type of change is at least partially motivated by the general the Reduced Instruction Set CPU (RISC) antagonism towards the control market at processor innovation in the computer work- the time. Even given the difficulties, the station market. RISC processors bring general effects of financial distress in the about a ten-fold increase in power, but can high-yield markets were greatly overem- be produced by adapting the current pro- phasized, and the high-yield bond market duction technology. With no increase in the has recently experienced near-record levels quantity demanded, this change implies of new issues. While precise numbers are that production capacity must fall by difficult to come by, I estimate the total 90 percent. Price declines increase the bankruptcy losses to junk bond and bank quantity demanded in these situations, and HLT (highly levered transaction) loans therefore reduce the capacity adjustment from inception of the market in the mid- that would otherwise be required. If 1970s through 1990 amounted to less than demand is elastic, output of the higher- $50 billion (Jensen (1991), footnote 9). In powered units will grow as it did for much comparison, IBM alone lost $51 billion of the computing industry’s history; now, (almost 65 percent of the total market however, the new workstation technology value of its equity) from its 1991 high to is reducing the demand for mainframe its 1992 close.14 computers. Mistakes were made in the takeover The second type is obsolescence-creating activity of the 1980s; indeed, given the far change—that is, one that obsoletes the THE MODERN INDUSTRIAL REVOLUTION 17 current capital stock and organization. they been stated explicitly, would not have Wal-Mart and the wholesale clubs that are been acceptable to the rational investor.” revolutionizing retailing are examples of There are clues in the history of the nine- such change. These new, focused, large teenth century that similar overshooting scale, low-cost retailers are dominating occurred then as well. In Jensen (1991), I old-line department stores which can no analyze the incentive, information, and longer compete. Building these new low-cost contracting problems that cause this over- stores means much current retail capacity shooting and argue that these problems of becomes obsolete—when Wal-Mart enters boom-bust cycles are general in venture a new market total retail capacity expands, markets—but that they can be corrected by and it is common for some of the existing reforming contracts that currently pay pro- high-cost retail capacity to go out of busi- moters for doing deals, rather than for ness.15 More intensive use of information doing successful deals. and other technologies, direct dealing with manufacturers, and the replacement of high-cost, restrictive work-rule union labor Current forces leading to excess are several sources of the competitive capacity and exit advantage of these new organizations. The ten-fold increase in crude oil prices Finally, excess capacity also results when between 1973 and 1979 had ubiquitous many competitors simultaneously rush to effects, forcing contraction in oil, chemi- implement new, highly productive technologies cals, steel, aluminum, and international without considering whether the aggregate shipping, among other industries. In addi- effects of all such investment will be tion, the sharp crude oil price increases greater capacity than can be supported by that motivated major changes to econo- demand in the final product market. mize on energy had other, perhaps even Sahlman and Stevenson’s (1985) analysis larger,implications. I believe the reevaluation of the winchester disk drive industry of organizational processes and procedures provides an example of this phenomenon. stimulated by the oil shock also generated Between 1977 and 1984, venture capitalists dramatic increases in efficiency beyond the invested over $400 million in 43 different original pure energy-saving projects. The manufacturers of winchester disk drives; original energy-motivated reexamination of initial public offerings of common stock corporate processes helped initiate a major infused additional capital in excess of $800 reengineering of company practices and million. In mid-1983, the capital markets procedures that still continues to accelerate assigned a value of $5.4 billion to 12 pub- throughout the world. licly traded, venture-capital-backed hard Since the oil price increases of the disk drive manufacturers—yet by the end 1970s, we again have seen systematic of 1984, the value assigned to those com- overcapacity problems in many industries panies had plummeted to $1.4 billion. In similar to those of the nineteenth century. his study of the industry, Christensen While the reasons for this overcapacity (1993) finds that over 138 firms entered nominally differ among industries, I doubt the industry in the period from its invention they are independent phenomena. We do in 1956 to 1990, and of these 103 subse- not yet fully understand all the causes pro- quently failed and six were acquired. pelling the rise in excess capacity in the Sahlman and Stevenson (p. 7) emphasize 1980s, yet I believe there were a few basic the lack of foresight in the industry: “The forces in operation. investment mania visited on the hard disk industry contained inherent assumptions about the long-run industry size Macro policies and profitability and about future growth, profitability and access to capital for each Major deregulation of the American econ- individual company.These assumptions, had omy (including trucking, rail, airlines, 18 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS telecommunications, banking and financial produce boxes with 1,000 times the capacity services industries) under President Carter for a given price. Consequently, computers contributed to the requirements for exit in are becoming commonplace—in cars, these industries,16 as did important toasters, cameras, stereos, ovens, and so on. changes in the U.S. tax laws that reduced Nevertheless, the increase in quantity tax advantages to real estate development, demanded has not been sufficient to avoid construction, and other activities.The end of overcapacity, and we are therefore witness- the cold war has had obvious ramifications ing a dramatic shutdown of production for the defense industry, as well as less lines in the industry—a force that has direct effects on the industry’s suppliers. wracked IBM as a high-cost producer. In addition, two generations of managerial A change of similar magnitude in auto pro- focus on growth as a recipe for success duction technology would have reduced the caused many firms, I believe, to overshoot price of a $20,000 auto in 1980 to under their optimal capacity, setting the stage for $20 today. Such increases in capacity and cutbacks, especially in white collar corporate productivity in a basic technology have bureaucracies. Specifically, in the decade unavoidably massive implications for the from 1979 to 1989 the Fortune 100 firms organization of work and society. lost 1.5 million employees, or 14 percent of Fiberoptic and other telecommunica- their workforce.17 tions technologies such as compression algorithms are bringing about similarly vast increases in worldwide capacity and Technology functionality. A Bell Laboratories study of Massive changes in technology are clearly excess capacity indicates, for example, that part of the cause of the current industrial given three years and an additional expen- revolution and its associated excess diture of $3.1 billion, three of AT&T’s new capacity. Both within and across industries, competitors (MCI, Sprint, and National technological developments have had far- Telecommunications Network) would be reaching impact. To give some examples, able to absorb the entire long distance the widespread acceptance of radial tires switched service that was supplied by (lasting three to five times longer than the AT&T in 1990 (Federal Communications older bias ply technology and providing Commission (1991), p. 1140). better gas mileage) caused excess capacity in the tire industry; the personal computer Organizational innovation revolution forced contraction of the market for mainframes; the advent of aluminum Overcapacity can be caused not only by and plastic alternatives reduced demand changes in the physical technology, but for steel and glass containers; and fiberop- also by changes in organizational prac- tic, satellite, digital (ISDN), and new tices and management technology. The compression technologies dramatically vast improvements in telecommunications, increased capacity in telecommunication. including computer networks, electronic Wireless personal communication such as mail, teleconferencing, and facsimile cellular phones and their replacements transmission are changing the workplace in promise to further extend this dramatic major ways that affect the manner in which change. people work and interact. It is far less The changes in computer technology, valuable for people to be in the same including miniaturization, have not only geographical location to work together revamped the computer industry, but also effectively, and this is encouraging smaller, redefined the capabilities of countless other more efficient, entrepreneurial organizing industries. Some estimates indicate the units that cooperate through technology.19 price of computing capacity fell by a factor This encourages even more fundamental of 1,000 over the last decade.18 This means changes. Through competition “virtual that computer production lines now organizations”—networked or transitory THE MODERN INDUSTRIAL REVOLUTION 19 organizations where people come together 509,000 to 252,000.23 From 1985 to temporarily to complete a task, then 1989 multifactor productivity in the separate to pursue their individual industry increased at an annual rate of specialties—are changing the structure of 5.3 percent compared to 1.3 percent for the standard large bureaucratic organiza- the period 1958 to 1989 (Burnham tion and contributing to its shrinkage. (1993),Table 1 and p. 15). Virtual organizations tap talented special- The entry of Japan and other Pacific ists, avoid many of the regulatory costs Rim countries such as Hong Kong, Taiwan, imposed on permanent structures, and Singapore, Thailand, Korea, Malaysia, and bypass the inefficient work rules and high China into worldwide product markets has wages imposed by unions. In doing so, they contributed to the required adjustments in increase efficiency and thereby further Western economies over the last several contribute to excess capacity. decades. Moreover, competition from new In addition, Japanese management tech- entrants to the world product markets niques such as total quality management, promises to get considerably more intense. just-in-time production, and flexible manufacturing have significantly increased the efficiency of organizations where they have Revolution in political economy been successfully implemented throughout The movement of formerly closed the world. Some experts argue that, properly communist and socialist centrally planned implemented, these new management economies to more market-oriented open techniques can reduce defects and spoilage capitalist economies is likely to generate by an order of magnitude.These changes in huge changes in the world economy over managing and organizing principles have the next several decades. These changes contributed significantly to the productivity promise to cause much conflict, pain, and of the world’s capital stock and economized suffering as world markets adjust, but also on the use of labor and raw materials, thus large profit opportunities. also contributing to the excess capacity 20 More specifically, the rapid pace of problems. development of capitalism, the opening of closed economies, and the dismantlement Globalization of trade of central control in communist and socialist states is occurring to various degrees in With the globalization of markets, excess China, India, Indonesia, Pakistan, other capacity tends to occur worldwide. Japan, Asian economies, and Africa.This evolution for example, is currently in the midst of will place a potential labor force of almost substantial excess capacity caused, at least a billion people—whose current average partially, by the breakdown in its own income is less than $2 per day—on world corporate control system;21 it is now in the markets.24,25 Table 1.1 summarizes some process of a massive restructuring of its of the population and labor force estimates economy.22 Yet even if the requirement for relevant to this issue. The opening of exit were isolated in the United States, the Mexico and other Latin American countries interdependency of today’s world economy and the transition of communist and socialist would ensure that such overcapacity would central and eastern European economies to have reverberating, global implications. For open capitalist systems (at least some of example, the rise of efficient high-quality which will make the transition in some producers of steel and autos in Japan and form) could add almost 200 million laborers Korea has contributed to excess capacity in with average incomes of less than $10 per those industries worldwide. Between 1973 day to the world market. and 1990 total capacity in the U.S. For perspective,Table 1.1 shows that the steel industry fell by 38 percent from average daily U.S. income per worker is 156.7 million tons to 97 million tons, and slightly over $90, and the total labor force total employment fell over 50 percent from numbers about 117 million, and the 20 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Table 1.1 Labor Force and Manufacturing Wage Estimates of Various Countries and Areas Playing an Actual or Potential Role in International Trade in the Past and in the Future

Total Potential Average Daily Populationa Labor Forceb Earningsc Country/Area (Millions) (Millions) (U.S.$)

Major potential entrants from Asia China 1,155.8 464.4 $1.53 India 849.6 341.4 $2.46 Indonesia 187.8 75.4 NAd Pakistan 115.5 46.4 $3.12 Sri Lanka 17.2 6.9 $1.25 Thailand 56.9 23.0 $1.49 Vietnam 68.2 27.4 NA Total/Average:Total pop./labor force & average earnings 2,451.0 984.9 $1.97e Potential entrant under NAFTA Mexico 87.8 35.5 $10.29 Major potential entrants from central and eastern Europe Czechoslovakia 15.6 6.3 $6.45 Hungary 10.3 4.2 $9.25 Poland 38.2 15.4 $6.14 Romania 23.2 9.4 $8.98 Yugoslavia 23.8 9.6 NA Former U.S.S.R. 286.7 115.8 $6.69 Total/Average: Mexico, central & eastern Europe 485.6 196.2 $7.49 Previous world market entrants from Asia Hong Kong 5.8 2.3 $25.79 Japan 123.9 50.1 $146.97 Korea 43.3 17.5 $45.37 Malaysia 17.9 7.4 NA Singapore 2.8 1.1 $27.86 Taiwan 20.7 8.4 NA Total/Average 214.4 86.8 $116.16 U.S. and E.E.C. for comparison United States 252.7 117.3 $92.24 European Economic Community 658.4 129.7 $78.34 Total/Average 911.1 246.7 $84.93 a Population statistics from Monthly Bulletin of Statistics (United Nations, 1993), 1991 data. b Potential labor force estimated by applying the 40.4 percent labor force participation rate in the European Economic Community to the 1991 population estimates, using the most recent employment estimates (Statistical Yearbook, United Nations, 1992) for each member country. c Unless otherwise noted, refers to 1991 earnings from the Monthly Bulletin of Statistics (United Nations, 1993) or earnings from Statistical Yearbook (United Nations, 1992) adjusted to 1991 levels using the Consumer Price Index. Earnings for Poland were calculated using 1986 earnings and 1986 year-end exchange rate, while earnings for Romania were calculated using 1985 earnings and 1985 exchange rate. An approximation for the former U.S.S.R. was made using 1987 data for daily earnings in the U.S.S.R. and the estimated 1991 exchange rate for the former U.S.S.R. from the Monthly Bulletin of Statistics. dNA Not available. In the case of Yugoslavia, inflation and currency changes made estimates unreliable. For Indonesia, Vietnam, Malaysia, and Taiwan data on earnings in manufacturing are unavailable. e Average daily wage weighted according to projected labor force in each grouping. THE MODERN INDUSTRIAL REVOLUTION 21 European Economic Community average Act, which will remove trade barriers wage is about $80 per day with a total between Canada, the United States, and labor force of about 130 million.The labor Mexico) is but one general example of forces that have affected world trade conflicts that are also occurring in the extensively in the last several decades total steel, automobile, computer chip, computer only about 90 million (Hong Kong, Japan, screen, and textile industries. In addition it Korea, Malaysia, Singapore, and would not be surprising to see a return to Taiwan).26 demands for protection from even domestic While the changes associated with bringing competition. This is currently underway in a potential 1.2 billion low-cost laborers the deregulated airline industry, an industry onto world markets will significantly that is faced with significant excess capacity. increase average living standards through- We should not underestimate the strains out the world, they will also bring massive this continuing change will place on obsolescence of capital (manifested in the worldwide social and political systems. In form of excess capacity) in Western both the First and Second Industrial economies as the adjustments sweep Revolutions, the demands for protection through the system. Western managers from competition and for redistribution of cannot count on the backward nature of income became intense. It is conceivable these economies to limit competition from that Western nations could face the modern these new human resources. Experience in equivalent of the English Luddites who China and elsewhere indicates the problems destroyed industrial machinery (primarily associated with bringing relatively current knitting frames) in the period 1811 to technology on line with labor forces in these 1816, and were eventually subdued by the areas is possible with fewer difficulties than militia (Watson (1993)). In the United one might anticipate.27 States during the early 1890s, large groups One can confidently forecast that the of unemployed men (along with some transition to open capitalist economies will vagrants and criminals), banding together generate great conflict over international under different leaders in the West, trade as special interests in individual Midwest, and East, wandered cross-country countries try to insulate themselves from in a march on Congress. These “industrial competition and the required exit.The tran- armies” formed to demand relief from “the sition of these economies will require large evils of murderous competition; the sup- redirection of Western labor and capital to planting of manual labor by machinery; the activities where it has a comparative excessive Mongolian and pauper immigration; advantage. While the opposition to global the curse of alien landlordism . . .” competition will be strong, the forces are (McMurray (1929), p. 128). Although the likely to be irresistible in this day of rapid armies received widespread attention and and inexpensive communication, trans- enthusiasm at the onset, the groups were portation, miniaturization, and migration. soon seen as implicit threats as they The bottom line, of course, is that with roamed from town to town, often stealing even more excess capacity and the require- trains and provisions as they went. Of the ment for additional exit, the strains put on 100,000 men anticipated by Coxey, only the internal control mechanisms of Western 1,000 actually arrived in Washington to corporations are likely to worsen for protest on May 1, 1893. At the request of decades to come. the local authorities, these protesters In the 1980s managers and employees disbanded and dispersed after submitting a demanded protection from the capital petition to Congress (McMurray (1929), markets. Many are now demanding protec- pp. 253–262). tion from international competition in the We need look no further than central and product markets (often under the guise of eastern Europe or Asia to see the effects of protecting jobs). The current dispute over policies that protect organizations from the NAFTA (North American Free Trade foreign and domestic competition. Hundreds 22 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS of millions of people have been condemned this way, exit is significantly delayed at to poverty as a result of governmental substantial cost of real resources to soci- policies that protect firms from competition ety. The tire industry is an example. in the product markets (both domestic and Widespread consumer acceptance of radial foreign) and attempt to ensure prosperity tires meant that worldwide tire capacity and jobs by protecting organizations had to shrink by two-thirds (because against decline and exit. Such policies are radials last three to five times longer than self-defeating, as employees of state-owned bias ply tires). Nonetheless, the response by factories in these areas are now finding. the managers of individual companies was Indeed, Porter (1990) finds that the most often equivalent to: “This business is going successful economies are those blessed through some rough times. We have to with intense internal competition that make major investments so that we will forces efficiency through survival of the have a chair when the music stops.” fittest. A.William Reynolds (1988), Chairman and Our own experience in the 1980s demon- CEO of GenCorp (maker of General Tires), strated that the capital markets can also illustrates this reaction in his testimony play an important role—that capital before the Subcommittee on Oversight and market pressures, while not perfect, can Investigations (February 18, 1988), U.S. significantly increase efficiency by bringing House Committee on Energy and about earlier adjustments. Earlier Commerce: adjustments avoid much of the waste generated when failure in the product The tire business was the largest piece of markets forces exit. GenCorp, both in terms of annual revenues and its asset base.Yet General Tire was not GenCorp’s strongest performer. IV. THE DIFFICULTY OF EXIT Its relatively poor earnings performance was due in part to conditions affecting The asymmetry between growth and all of the tire industry....In 1985 decline worldwide tire manufacturing capacity substantially exceeded demand. At the Exit problems appear to be particularly same time, due to a series of technological severe in companies that for long periods improvements in the design of tires and enjoyed rapid growth, commanding market the materials used to make them, the positions, and high cash flow and profits. In product life of tires had lengthened these situations, the culture of the organi- significantly. General Tire, and its com- zation and the mindset of managers seem petitors, faced an increasing imbalance to make it extremely difficult for adjust- between supply and demand. The eco- ment to take place until long after the nomic pressure on our tire business was problems have become severe, and in some substantial. Because our unit volume cases even unsolvable. In a fundamental was far below others in the industry, we sense, there is an asymmetry between the had less competitive flexibility....We growth stage and the contraction made several moves to improve our stage over the life of a firm. We have spent competitive position: We increased our little time thinking about how to manage investment in research and development. the contracting stage efficiently, or more We increased our involvement in the high importantly how to manage the growth performance and light truck tire cate- stage to avoid sowing the seeds of decline. gories, two market segments which In industry after industry with excess offered faster growth opportunities. We capacity, managers fail to recognize that developed new tire products for those they themselves must downsize; instead segments and invested heavily in an they leave the exit to others while they con- aggressive marketing program designed tinue to invest. When all managers behave to enhance our presence in both markets. THE MODERN INDUSTRIAL REVOLUTION 23 We made the difficult decision to reduce Contracting problems our overall manufacturing capacity by closing one of our older, less modern Explicit and implicit contracts in the plants in Waco,TX ...I believe that the organization can become major obstacles General Tire example illustrates that we to efficient exit. Unionization, restrictive were taking a rational, long-term work rules, and lucrative employee approach to improving GenCorp’s overall compensation and benefits are other ways performance and shareholder in which the agency costs of free cash flow value. . . . As a result of the takeover can manifest themselves in a growing, attempt, . . . [and] to meet the principal cash-rich organization. Formerly dominant and interest payments on our vastly firms became unionized in their heyday (or increased corporate debt, GenCorp had effectively unionized in organizations like to quickly sell off valuable assets and IBM and Kodak) when managers spent abruptly lay-off approximately 550 some of the organization’s free cash flow to important emloyees. buy labor peace. Faced with technical innovation and worldwide competition GenCorp sold its General Tire subsidiary (often from new, more flexible, and to Continental AG of Hannover, West nonunion organizations), these dominant Germany for approximately $625 million. firms cannot adjust fast enough to maintain Despite Reynolds’s good intentions and their market dominance (see DeAngelo and efforts, Gen Corp’s increased investment DeAngelo (1991) and Burnham (1993)). seems not to be a socially optimal response Part of the problem is managerial and for managers in a declining industry with organizational defensiveness that inhibits excess capacity. learning and prevents managers from changing their model of the business (see Information problems Argyris (1990)). Implicit contracts with unions, other Information problems hinder exit because employees, suppliers, and communities add the high-cost capacity in the industry must to formal union barriers to change by be eliminated if resources are to be used reinforcing organizational defensiveness efficiently. Firms often do not have good and inhibiting change long beyond the information on their own costs, much less optimal time—even beyond the survival point the costs of their competitors; it is there- for the organization. In an environment like fore sometimes unclear to managers that this a shock must occur to bring about they are the high-cost firm which should effective change. We must ask why we exit the industry.28 Even when managers do cannot design systems that can adjust acknowledge the requirement for exit, it is more continuously, and therefore more often difficult for them to accept and efficiently. initiate the shutdown decision. For the The security of property rights and the managers who must implement these enforceability of contracts are extremely decisions, shutting plants or liquidating the important to the growth of real output, firm causes personal pain, creates efficiency, and wealth. Much press coverage uncertainty, and interrupts or sidetracks and official policy seems to be based on the careers. Rather than confronting this pain, notion that all implicit contracts should be managers generally resist such actions as unchangeable and rigidly enforced. Yet it is long as they have the cash flow to subsidize clear that, given the occurrence of the losing operations. Indeed, firms with unexpected events, not all contracts, large positive cash flow will often invest in whether explicit or implicit can (or even even more money-losing capacity— should) be fulfilled. Implicit contracts, in situations that illustrate vividly what I call addition to avoiding the costs incurred in the agency costs of free cash flow (Jensen the writing process, provide opportunity to (1986)). revise the obligation if circumstances change; 24 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS presumably, this is a major reason for their hostile tender offers in the 1970s. Prior to existence. the 1970s capital market discipline took Indeed the gradual abrogation of the place primarily through the proxy process. legal notion of “at will” employment is (Pound (1993) analyzes the history of the coming close to granting property rights in political model of corporate control.) jobs to all employees.29 While casual The legal/political/regulatory system is breach of implicit contracts will destroy far too blunt an instrument to handle the trust in an organization and seriously problems of wasteful managerial behavior reduce efficiency, all organizations must effectively. (The breakup and deregulation evolve a way to change contracts that are of AT&T, however, is one of the court no longer optimal. For example, bank- system’s outstanding successes. As we shall ruptcy is essentially a state-supervised see below, it helped create over $125 billion system for breaking (or more politely, of increased value between AT&T and the rewriting) contracts that are mutually Baby Bells.) inconsistent and therefore, unenforceable. While the product and factor markets are All developed economies evolve such a slow to act as a control force, their system. Yet, the problem is a very general discipline is inevitable—firms that do one, given that the optimality of changing not supply the product that customers contracts must be one of the major reasons desire at a competitive price cannot survive. for leaving many of them implicit. Research Unfortunately, when product and factor into the optimal breach of contracts, and the market disciplines take effect it can often bonding against opportunistic behavior be too late to save much of the enterprise. that must accompany it, is an important topic To avoid this waste of resources, it is impor- that has received considerable attention tant for us to learn how to make the other in the law and economics literature (see three organizational control forces more Polinsky (1989) ) but is deserving of more expedient and efficient. attention by organization theorists. Substantial data support the proposition that the internal control systems of publicly held corporations have generally failed to V. THE ROLE OF THE MARKET FOR cause managers to maximize efficiency and 30 CORPORATE CONTROL value. More persuasive than the formal statistical evidence is the fact that few firms ever restructure themselves or The four control forces operating on engage in a major strategic redirection the corporation without a crisis either in the capital There are only four control forces operating markets, the legal/political/regulatory on the corporation to resolve the problems system, or the product/factor markets. But caused by a divergence between managers’ there are firms that have proved to be flex- decisions and those that are optimal from ible in their responses to changing market society’s standpoint.They are the conditions in an evolutionary way. For example, investment banking firms and ● capital markets, consulting firms seem to be better at ● legal/political/regulatory system, responding to changing market conditions. ● product and factor markets, and ● internal control system headed by Capital markets and the market for the board of directors. corporate control As explained elsewhere (Jensen (1989a, The capital markets provided one mecha- 1989b, 1991), Roe (1990, 1991)), the nism for accomplishing change before capital markets were relatively constrained losses in the product markets generate a by law and regulatory practice from about crisis. While the corporate control activity 1940 until their resurrection through of the 1980s has been widely criticized as THE MODERN INDUSTRIAL REVOLUTION 25 counterproductive to American industry, efficient reduction of capacity by one of the few have recognized that many of these major firms in the industry. Also, by elimi- transactions were necessary to accomplish nating some of the cash resources from the exit over the objections of current man- oil and tobacco industries, these capital agers and other constituencies of the firm market transactions promote an environ- such as employees and communities. For ment that reduces the rate of growth of example, the solution to excess capacity in human resources in the industries or even the tire industry came about through the promotes outright reduction when that is market for corporate control. Every major the optimal policy. U.S. tire firm was either taken over or The era of the control market came to an restructured in the 1980s.31 In total, 37 end, however, in late 1989 and 1990. tire plants were shut down in the period Intense controversy and opposition from 1977 to 1987 and total employment in the corporate managers, assisted by charges of industry fell by over 40 percent. (U.S. fraud, the increase in default and bankruptcy Bureau of the Census (1987),Table 1a-1.) rates, and insider trading prosecutions, The pattern in the U.S. tire industry is caused the shutdown of the control market repeated elsewhere among the crown jewels through court decisions, state antitakeover of American business. amendments, and regulatory restrictions Capital market and corporate control on the availability of financing (see Swartz transactions such as the repurchase of (1992), and Comment and Schwert stock (or the purchase of another com- (1993)). In 1991, the total value of trans- pany) for cash or debt creates exit of actions fell to $96 billion from $340 billion resources in a very direct way. When in 1988.32 LBOs and management buyouts Chevron acquired Gulf for $13.2 billion in fell to slightly over $1 billion in 1991 from cash and debt in 1984, the net assets $80 billion in 1988.33 The demise of the devoted to the oil industry fell by $13.2 control market as an effective influence on billion as soon as the checks were mailed American corporations has not ended the out. In the 1980s the oil industry had to restructuring, but it has meant that organ- shrink to accommodate the reduction in the izations have typically postponed addressing quantity of oil demanded and the reduced the problems they face until forced to by rate of growth of demand. This meant pay- financial difficulties generated by the ing out to shareholders its huge cash product markets. Unfortunately the delay inflows, reducing exploration and develop- means that some of these organizations ment expenditures to bring reserves in line will not survive—or will survive as mere with reduced demands, and closing refining shadows of their former selves. and distribution facilities. The leveraged acquisitions and equity repurchases helped accomplish this end for virtually all major VI. THE FAILURE OF CORPORATE U.S. oil firms (see Jensen (1986b, 1988)). INTERNAL CONTROL SYSTEMS Exit also resulted when Kohlberg, Kravis, and Roberts (KKR) acquired RJR- With the shutdown of the capital markets Nabisco for $25 billion in cash and debt in as an effective mechanism for motivating its 1986 leveraged buyout (LBO). Given change, renewal, and exit, we are left to the change in smoking habits in response to depend on the internal control system to consumer awareness of cancer threats, the act to preserve organizational assets, both tobacco industry must shrink, and the pay- human and nonhuman. Throughout corpo- out of RJR’s cash accomplished this to rate America, the problems that motivated some extent. Furthermore, the LBO debt much of the control activity of the 1980s prohibits RJR from continuing to squander are now reflected in lackluster perform- its cash flows on the wasteful projects it ance, financial distress, and pressures for had undertaken prior to the buyout. Thus, restructuring. Kodak, IBM, Xerox, ITT, and the buyout laid the groundwork for the many others have faced or are now facing 26 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS severe challenges in the product markets. (Unfortunately he resigned only several We therefore must understand why these months later—after, according to press internal control systems have failed and reports, running into resistance from the learn how to make them work. current management and board about the By nature, organizations abhor control necessity for dramatic change.) systems, and ineffective governance is a General Electric (GE) under Jack Welch, major part of the problem with internal who has been CEO since 1981, is a coun- control mechanisms. They seldom respond terexample to my proposition about the in the absence of a crisis. The recent GM failure of corporate internal control board revolt (as the press has called it) systems. GE has accomplished a major which resulted in the firing of CEO Robert strategic redirection, eliminating 104,000 Stempel exemplifies the failure, not the suc- of its 402,000 person workforce (through cess, of GM’s governance system. General layoffs or sales of divisions) in the period Motors, one of the world’s high-cost produc- 1980 to 1990 without the motivation of a ers in a market with substantial excess threat from capital or product markets.34 capacity, avoided making major changes in But there is little evidence to indicate this is its strategy for over a decade. The revolt due to anything more than the vision and came too late: the board acted to remove the persuasive powers of Jack Welch rather than CEO only in 1992, after the company had the influence of GE’s governance system. reported losses of $6.5 billion in 1990 and General Dynamics (GD) provides another 1991 and (as we shall see in the next sec- counterexample. The appointment of tion) an opportunity loss of over $100 billion William Anders as CEO in September 1991 in its R&D and capital expenditure program (coupled with large changes in its manage- over the eleven-year period 1980 to 1990. ment compensation system which tied Moreover, the changes to date are still too bonuses to increases in stock value) resulted small to resolve the company’s problems. in its rapid adjustment to excess capacity in Unfortunately, GM is not an isolated the defense industry—again with no apparent example. IBM is another testimony to the threat from any outside force. GD gen- failure of internal control systems: it failed erated $3.4 billion of increased value on a to adjust to the substitution away from its $1 billion company in just over two years mainframe business following the revolu- (see Murphy and Dial (1992)). Sealed Air tion in the workstation and personal (Wruck (1992)) is another particularly computer market—ironically enough a interesting example of a company that revolution that it helped launch with the restructured itself without the threat of an invention of the RISC technology in 1974 immediate crisis. CEO Dermot Dumphy (Loomis (1993)). Like GM, IBM is a high- recognized the necessity for redirection, and cost producer in a market with substantial after several attempts to rejuvenate the excess capacity. It too began to change its company to avoid future competitive prob- strategy significantly and removed its CEO lems in the product markets, created a crisis only after reporting losses of $2.8 billion in by voluntarily using the capital markets in a 1991 and further losses in 1992 while los- leveraged restructuring. Its value more than ing almost 65 percent of its equity value. tripled over a three-year period. I hold these Eastman Kodak, another major U.S. companies up as examples of successes of company formerly dominant in its market, the internal control systems, because each also failed to adjust to competition and has redirection was initiated without immediate performed poorly. Its $37 share price in crises in the product or factor markets, 1992 was roughly unchanged from 1981. the capital markets, or in the legal/political/ After several reorganizations, it only regulatory system. The problem is that they recently began to seriously change its are far too rare. incentives and strategy, and it appointed a Although the strategic redirection of chief financial officer well-known for General Mills provides another counterex- turning around troubled companies. ample (Donaldson (1990)), the fact that it THE MODERN INDUSTRIAL REVOLUTION 27 took more than ten years to accomplish the generated cash flows. For example, consider change leaves serious questions about the a firm that provides dividends plus capital social costs of continuing the waste caused gains to its shareholders over a ten-year by ineffective control. It appears that period that equal the cost of capital on the internal control systems have two faults. beginning of period share value. Suppose, They react too late, and they take too long however,that management spent $30 billion to effect major change. Changes motivated of internally generated cash flow on R&D by the capital market are generally accom- and capital expenditures that generated no plished quickly—within one and a half to returns. In this case the firm’s shareholders three years. As yet no one has demon- suffered an opportunity loss equal to the strated the social benefit from relying on value that could have been created if the purely internally motivated change that firm had paid the funds out to them and they offsets the costs of the decade-long delay had invested it in equivalently risky projects. exhibited by General Mills. The opportunity cost of R&D and capital In summary,it appears that the infrequency expenditures thus can be thought of as the with which large corporate organizations returns that would have been earned by an restructure or redirect themselves solely investment in equivalent-risk assets over on the basis of the internal control the same time period. We don’t know mechanisms in the absence of crises in the exactly what the risk is, nor what the product, factor, or capital markets or expected returns would be, but we can the regulatory sector is strong testimony to make a range of assumptions. A simple the inadequacy of these control mechanisms. measure of performance would be the difference between the total value of the R&D plus capital and acquisition expenditures VII. DIRECT EVIDENCE OF THE invested in a benchmark strategy and the FAILURE OF INTERNAL CONTROL total value the firm actually created with SYSTEMS its investment strategy. The benchmark strategy can be thought of as the ending The productivity of R&D and capital value of a comparable-risk bank account (with an expected return of 10 percent) expenditures into which the R&D and capital expendi- The control market, corporate restructurings, tures in excess of depreciation (hereafter and financial distress provide substantial referred to as net capital expenditures) had evidence on the failings of corporate internal been deposited instead of invested in real control systems. My purpose in this section projects. For simplicity I call this the is to provide another and more direct benchmark strategy. The technical details estimate of the effectiveness of internal con- of the model are given in the Appendix.The trol systems by measuring the productivity of calculation of the performance measure corporate R&D and capital expenditures. takes account of all stock splits, stock The results reaffirm that many corporate dividends, equity issues and repurchases, control systems are not functioning well. dividends, debt issues and payments, and While it is impossible to get an unambigu- interest. ous measure of the productivity of R&D and capital expenditures, by using a period Three measures of the productivity of as long as a decade we can get some R&D and net capital expenditures approximations. We cannot simply measure the performance of a corporation by Measure 1 the change in its market value over time (more precisely the returns to its share- Consider an alternative strategy which pays holders) because this measure does not the same dividends and stock repurchases take account of the efficiency with which as the firm actually paid (and raises the the management team manages internally same outside capital) and puts the R&D 28 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS and capital and acquisition expenditures like a one-horse shay, the firm arrives at the (in excess of depreciation) in marketable end of the period still generating cash securities of the same risk as the R&D and returns, but then collapses with no addi- capital expenditures, yielding expected tional cash payments to equityholders, and returns equal to their cost of capital, i. equity value of zero as of the horizon date. Under the assumption that the zero investment and R&D strategy yields a terminal Measure 3 value of the firm equal to the ending debt plus the beginning value of equity (that is, To allow for the effects of the reduced investment equal to depreciation is suffi- investment and R&D on intermediate cash cient to maintain the original equity value flows my third measure assumes that all of the firm), Measure 1 is the difference intermediate cash flows are reduced in the between the actual ending total value of benchmark investment strategy by the firm and the value of the benchmark. the amount paid out to shareholders in The exact equation is given in the Appendix the form of dividends and net share for this measure as well as the next two repurchases and that the original value of measures of performance. the equity is maintained. This measure is Unless capital and R&D expenditures likely to yield an upward biased estimate of are completely unproductive, this first the productivity of R&D and capital crude measure of the productivity of R&D expenditures. and capital expenditures will be biased downward. I define two additional measures that use different assumptions about The data and results the effect of the reduced R&D and capital expenditures on the ending value of the The data for this analysis consist of all 432 firm’s equity and on the ability of the firm firms on COMPUSTAT with 1989 sales of to make the intermediate cash dividend and $250 million or more for which complete stock repurchase payouts to shareholders. data on R&D, capital expenditures, depre- If R&D is required to maintain a competitive ciation, dividends, and market value were position in the industry, the ending value of available for the period December 31, 1979 the equity in the benchmark strategy is through December 31, 1990.The estimates likely to be less than the initial value of of the productivity of R&D are likely to be equity even though nominal depreciation of upward biased because the selection crite- the capital stock is being replaced. ria use only firms that managed to survive Moreover, with no R&D and maintenance through the period and eliminate those that only of the nominal value of the capital failed. I have calculated results for various stock, the annual cash flows from opera- rates of interest but report only those using tions are also likely to be lower than those a 10 percent rate of return.This rate is prob- actually realized (because organizational ably lower than the cost of capital for R&D efficiency and product improvement will expenditures at the beginning of the period lag competitors, and new product introduc- when interest rates were in the high teens, tion will be lower). Therefore I use two and probably about right or on the high side more conservative measures that will yield at the end of the period when the cost of higher estimates of the productivity of capital was probably on the order of 8 to 10 these expenditures. percent. A low approximation of the cost of capital appropriate to R&D and capital expenditures will bias the performance Measure 2 measures up, so I am reasonably comfort- able with these conservative assumptions. The second measure assumes that Because they are interesting in their own replacement of depreciation and zero right,Table 1.2 presents the data on annual expenditures on R&D are sufficient to R&D and capital expenditures of nine maintain the intermediate cash flows but, selected Fortune 500 corporations and the THE MODERN INDUSTRIAL REVOLUTION 29 Table 1.2 Total R&D and Capital Expenditures for Selected Companies and the Venture Capital Industry, 1980–1990 ($ Billions)

Venture Capital Year GM IBM Xerox Kodak Intel GE Industry Merck AT&T Total R&D Expenditures

1980 2.2 1.5 0.4 0.5 0.1 0.8 0.6 0.2 0.4 1981 2.2 1.6 0.5 0.6 0.1 0.8 1.2 0.3 0.5 1982 2.2 2.1 0.6 0.7 0.1 0.8 1.5 0.3 0.6 1983 2.6 2.5 0.6 0.7 0.1 0.9 2.6 0.4 0.9 1984 3.1 3.1 0.6 0.8 0.2 1.0 2.8 0.4 2.4 1985 4.0 3.5 0.6 1.0 0.2 1.1 2.7 0.4 2.2 1986 4.6 4.0 0.7 1.1 0.2 1.3 3.2 0.5 2.3 1987 4.8 4.0 0.7 1.0 0.3 1.2 4.0 0.6 2.5 1988 5.3 4.4 0.8 1.1 0.3 1.2 3.9 0.7 2.6 1989 5.8 5.2 0.8 1.3 0.4 1.3 3.4 0.8 2.7 1990 5.9 4.9 0.9 1.3 0.5 1.5 1.9 0.9 2.4 Total 42.7 36.8 7.1 10.1 2.5 11.9 27.8 5.4 19.3 Total Capital Expenditures 1980 7.8 6.6 1.3 0.9 0.2 2.0 NA 0.3 17.0 1981 9.7 6.8 1.4 1.2 0.2 2.0 NA 0.3 17.8 1982 6.2 6.7 1.2 1.5 0.1 1.6 NA 0.3 16.5 1983 4.0 4.9 1.1 0.9 0.1 1.7 NA 0.3 13.8 1984 6.0 5.5 1.3 1.0 0.4 2.5 NA 0.3 3.5 1985 9.2 6.4 1.0 1.5 0.2 2.0 NA 0.2 4.2 1986 11.7 4.7 1.0 1.4 0.2 2.0 NA 0.2 3.6 1987 7.1 4.3 0.3 1.7 0.3 1.8 NA 0.3 3.7 1988 6.6 5.4 0.5 1.9 0.5 3.7 NA 0.4 4.0 1989 9.1 6.4 0.4 2.1 0.4 5.5 NA 0.4 3.5 1990 10.1 6.5 0.4 2.0 0.7 2.1 NA 0.7 3.7 Total 87.5 64.2 9.9 16.1 3.3 27.0 NA 3.7 91.2 Net Capital Expenditures (Capital Expenditures less Depreciation) 1980 3.6 3.8 0.5 0.5 0.1 1.2 NA 0.2 9.9 1981 5.3 3.5 0.6 0.7 0.1 1.1 NA 0.2 9.9 1982 1.7 3.1 0.4 0.9 0.1 0.6 NA 0.2 7.7 1983 1.1 1.3 0.3 0.2 0.1 0.6 NA 0.1 3.9 1984 1.1 2.3 0.5 0.2 0.3 1.3 NA 0.1 0.7 1985 3.5 3.4 1.2 0.6 0.1 0.8 NA 0.07 0.9 1986 5.6 1.3 0.2 0.5 0.0 0.6 NA 0.04 3.2 1987 0.8 0.7 0.3 0.6 0.1 0.2 NA 0.07 0.6 1988 0.7 1.5 0.2 0.7 0.3 0.3 NA 0.2 5.9 1989 2.0 2.2 0.2 0.8 0.2 0.7 NA 0.2 1.1 1990 2.7 2.3 0.3 0.7 0.4 0.6 NA 0.4 0.3 Total 24.5 25.4 2.7 6.4 1.8 8.0 NA 1.8 24.7 Total Value of R&D plus Net Capital Expenditures 67.2 62.2 9.8 16.7 4.3 19.9 27.8 7.2 44.0 Ending Equity Value of the Company, 12/90 26.2 64.6 3.2 13.5 13.5 50.0 60 34.8 32.9

NA Not available. Source: Annual reports, COMPUSTAT, Business Week R&D Scoreboard, William Sahlman. Venture Economics for total disbursements by industry. Capital expenditures for the venture capital industry are included in the R&D expenditures which are the total actual disbursements by the industry. 30 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS total venture capital industry from January had done this (and not changed the companies 1, 1980 through December 31, 1990.Table in any way), GM would have owned two of 1.3 contains calculations that provide some the world’s low-cost, high-quality automobile benchmarks for evaluating the productivity producers. of these expenditures. As Table 1.3 shows, the difference Total R&D expenditures over the eleven- between the value of GM’s actual strategy year period range from $42.7 billion for and the value of the equivalent-risk bank General Motors to $5.4 billion for Merck. account strategy amounts to $ –100.7 The individual R&D expenditures of GM billion by Measure 2 (which assumes the and IBM were significantly greater than the ending value of the company given no R&D $27.8 billion spent by the entire U.S. venture or net capital expenditures is zero in the capital industry over the eleven-year period. benchmark strategy), $ –115.2 billion for Because venture capital data include both Measure 1 (which assumes the original the R&D component and capital expendi- value of the equity is maintained), and tures, we must add in corporate capital $ –90 billion by Measure 3 (which assumes expenditures to get a proper comparison to cash flows fall by the amount of all interme- the venture industry figures. Total capital diate cash outflows to shareholders and expenditures range from $91.2 billion for debtholders). I concentrate on Measure 2 AT&T and $87.5 billion for GM to $3.7 which I believe is the best measure of the billion for Merck. Capital expenditures net three. By this measure, IBM lost over $11 of depreciation range from $25.4 billion for billion relative to the benchmark strategy IBM to $1.8 billion for Merck. (and this is prior to the $50 billion decline in It is clear that GM’s R&D and invest- its equity value in 1991 and 1992), while ment program produced massive losses.The Xerox and Kodak were down $8.4 billion company spent a total of $67.2 billion in and $4.6 billion respectively. GE and Merck excess of depreciation in the period and were major success stories, with value produced a firm with total ending value of creation in excess of the benchmark strategy equity (including the E and H shares) of of $29.9 billion and $28 billion respectively. $26.2 billion. Ironically, its expenditures AT&T gained $2.1 billion over the bench- were more than enough to pay for the mark strategy, after having gone through the entire equity value of Toyota and Honda, court-ordered breakup and deregulation of which in 1985 totaled $21.5 billion. If it the Bell system in 1984. The value gains of

Table 1.3 Benefit-Cost Analysis of Corporate R&D and Investment Programs: Actual Total Value of Company at 12/31/90 Less Total Value of the Benchmark Strategy r 10 percent (billions of dollars)

Venture Capital GM IBM Xerox Kodak Intel GE Industry Merck AT&T

Measure 1: Gain (Loss) [Assumes beginning value of equity is maintained]

($115.2) ($49.4) ($13.6) ($12.4) $1.8 $18.4 $17 $22.6 ($34.5)

Measure 2: [Assumes ending equity value is zero]

($100.7) ($11.8) ($8.4) ($4.6) $3.2 $29.9 $17 $28.0 $2.1

Measure 3: [Assumes ending equity value equals beginning value and intermediate cash flows are smaller by the amount paid to equity under company’s strategy]

($90.0) ($5.4) ($8.0) ($1.8) $1.8 $36.4 $17 $28.1 $21.3 THE MODERN INDUSTRIAL REVOLUTION 31 the seven Baby Bells totaled $125 billion by Table 1.5 provides summary statistics Measure 2 (not shown in the table), making (including the minimum, mean, five frac- the breakup and deregulation a nontrivial tiles of the distribution, maximum, and success given that prices to consumers have standard deviation) on R&D expenditures, generally fallen in the interim. net capital expenditures, and the three The value created by the venture capital performance measures. The mean ten-year industry is difficult to estimate. We would R&D and net capital expenditures are like to have estimates of the 1990 total $1.296 billion and $1.367 billion respectively; end-of-year value of all companies funded the medians are $146 million and $233 during the eleven-year period. This value is million. The average of Measure 2 over all not available so I have relied on the $60 432 firms is slightly over $1 billion with a billion estimate of the total value of all t-value of 3.0, indicating that on average IPOs during the period. This overcounts this sample of firms created value above those firms that were funded prior to 1980 that of the benchmark strategy. The and counts as zero all those firms that had average for Measures 1 and 2 are $ –221 not yet come public as of 1990. Because of million and $1.086 billion respectively. All the pattern of increasing investment over productivity measures are upward biased the period from the mid-1970s, the over- because failed firms are omitted from the counting problem is not likely to be as sample, and because the decade of the severe as the undercounting problem.Thus, eighties was an historical outlier in stock the value added by the industry over the market performance. The median perform- bank account strategy is most probably ance measures are $24 million, $200 greater than $17 billion as shown in Table million, and $206 million respectively. The 1.3. Since the venture capital industry is in maximum performance measures range Table 1.3 as another potential source of from $37.7 billion to $47 billion. comparison, and since virtually its entire Although the average performance value creation is reflected in its ending measures are positive, well-functioning equity value, I have recorded its value cre- internal control systems would substan- ation under each measure as the actual tially truncate the lower tail of the distri- estimate of greater than $17 billion. bution. And given that the sample is subject Because the extreme observations in the to survivorship bias,36 and that the period distribution are the most interesting, Table was one in which stock prices performed 1.4 gives the three performance measures historically above average, the results for the 35 companies at the bottom of the demonstrate major inefficiencies in the list of 432 firms ranked in reverse order on capital expenditure and R&D spending Measure 2 (Panel A), and on the 35 com- decisions of a substantial number of panies at the top of the ranked list (Panel firms.37 I believe we can improve these B), also in reverse order. As the tables control systems substantially, but to do so show, GM ranked at the bottom of the we must attain a detailed understanding of performance list, preceded by Ford, British how they work and the factors that lead to Petroleum, Chevron, Du Pont, IBM, Unisys, their success or failure. United Technologies, and Xerox. Obviously many of the United States’ largest and best-known companies appear on this list VIII. REVIVING INTERNAL (including GD prior to its recent turn- CORPORATE CONTROL SYSTEMS around), along with Japan’s Honda Motor company. Panel B shows that Philip Morris Remaking the board as an effective created the most value in excess of control mechanism the benchmark strategy, followed by Wal-Mart, Bristol Myers, GE, Loews, The problems with corporate internal Merck, Bellsouth, Bell Atlantic, Procter & control systems start with the board of Gamble, Ameritech, and Southwestern Bell.35 directors. The board, at the apex of the 32 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Table 1.4 Difference between Value of Benchmark Strategy for Investing R&D and Net Capital Expenditure and Actual Strategy under Three Assumptions regarding Ending Value of Equity and Intermediate Cash Flows for Benchmark Strategy (Performance Measures 1–3)

Panel A: Performance measures for the 35 companies at the bottom of the ranked list of 432 companies in the period 1980–1990 on performance measure 2. r 10 percent

Performance Measure (Millions) Rank Company 1 2 3

432 General Motors Corp. (115,188) (100,720) (90,024) 431 Ford Motor Co. (29,304) (25,447) (20,392) 430 British Petroleum P.L.C.(ADR) (35,585) (23,699) (19,958) 429 Chevron Corp. (25,497) (15,859) (10,586) 428 Du Pont (E.I.) de Nemours (21,122) (15,279) (8,535) 427 Intl. Business Machines Corp. (49,395) (11,826) (5,394) 426 Unisys Corp. (14,655) (11,427) (11,899) 425 United Technologies Corp. (10,843) (9,032) (7,048) 424 Xerox Corp. (13,636) (8,409) (7,978) 423 Allied Signal Inc. (8,869) (7,454) (5,002) 422 Hewlett-Packard Co. (9,493) (6,373) (8,605) 421 ITT Corp. (9,099) (6,147) (3,611) 420 Union Carbide Corp. (8,673) (5,893) (3,341) 419 Honeywell Inc. (7,212) (5,361) (5,677) 418 Lockheed Corp. (5,744) (5,339) (5,149) 417 Digital Equipment (7,346) (5,082) (7,346) 416 Penn Central Corp. (5,381) (4,846) (4,938) 415 Eastman Kodak Co. (12,397) (4,630) (1,762) 414 Chrysler Corp. (5,054) (4,604) (3,041) 413 Atlantic Richfield Co. (13,239) (3,977) (1,321) 412 Northrop Corp. (4,489) (3,904) (3,743) 411 Goodyear Tire & Rubber Co. (4,728) (3,805) (2,532) 410 Phillips Petroleum Co. (11,027) (3,614) (5,427) 409 Honda Motor Ltd. (Amer. shares) (4,880) (3,435) (3,898) 408 Texaco Inc. (11,192) (3,354) 3,830 407 Texas Instruments Inc. (5,359) (3,350) (4,276) 406 NEC Corp. (ADR) (4,803) (3,326) (3,736) 405 National Semiconductor Corp. (3,705) (3,246) (3,632) 404 General Dynamics Corp. (4,576) (2,966) (3,783) 403 Grace (W.R.) & Co. (4,599) (2,776) (2,314) 402 Imperial Chem. Inds. P.L.C.(ADR) (7,223) (2,575) (1,287) 401 Tektronix Inc. (3,414) (2,500) (3,070) 400 Advanced Micro Devices (2,647) (2,419) (2,603) 399 Wang Laboratories (CLB) (2,815) (2,368) (2,564) 398 Motorola Inc. (3,863) (2,270) (2,588)

Panel B: Performance measures for the 35 companies ranked at the top of the list of 432 companies in the period 1980–1990 on performance measure 2. r 10 percent.

Performance Measure (Millions) Rank Company 1 2 3

35 Kellogg Co. 5,747 7,190 8,245 34 Pfizer Inc. 4,607 7,477 8,650 33 General Mills Inc. 6,205 7,605 8,256 32 Kyocera Corp. (ADR) 7,194 7,959 7,709 31 Minnesota Mining & Mfg. Co. 2,375 8,270 10,008 30 Canon Inc. (ADR) 7,717 8,326 8,285 THE MODERN INDUSTRIAL REVOLUTION 33

Performance Measure (Millions) Rank Company 1 2 3

29 Matsushita Electric (ADR) 5,356 8,694 6,999 28 Tele-Communications (CLA) 8,692 8,998 8,698 27 Kubota Corp. (ADR) 7,383 9,246 8,280 26 Marion Merrell Dow Inc. 9,489 9,606 9,865 25 Unilever N.V. (N.Y. shares) 8,642 10,574 12,510 24 Fuji Photo Film (ADR) 10,102 10,858 10,518 23 Hitachi Ltd. (ADR) 8,412 10,863 10,800 22 Amoco Corp. (331) 11,437 14,838 21 Sony Corp. (Amer. shares) 10,001 11,591 11,019 20 Lilly (Eli) & Co. 7,462 11,818 12,001 19 Ito Yokado Co. Ltd. (ADR) 12,415 13,178 12,982 18 Abbot Laboratories 11,076 13,555 14,043 17 Johnson & Johnson 8,945 13,796 13,199 16 Nynex Corp. 7,971 13,975 14,856 15 U.S. West Inc. 8,696 14,398 14,430 14 Exxon Corp. (9,213) 14,976 40,096 13 Pacific Telesis Group 11,530 16,846 17,648 12 Glaxo Holdings P.L.C.(ADR) 16,215 17,007 18,348 11 Southwestern Bell Corp. 11,807 17,702 17,880 10 Ameritech Corp. 11,883 18,453 18,516 9 Procter & Gamble Co. 12,900 19,247 20,293 8 Bell Atlantic Corp. 13,146 19,921 20,235 7 Bellsouth Corp. 15,205 23,921 24,644 6 Merck & Co. 22,606 28,045 28,092 5 Loews Corp. 28,540 29,265 29,579 4 General Electric Co. 18,411 29,945 36,363 3 Bristol Myers Squibb 27,899 30,321 33,296 2 Wal-Mart Stores 37,701 38,239 38,486 1 Philip Morris Cos. Inc. 37,548 42,032 47,029

Table 1.5 Summary Statistics on R&D, Capital Expenditures, and Performance Measures for 432 Firms with Sales Greater than $250 Million in the Period 1980–1990 r 10 percent (millions of dollars)

R&D Net Capital Performance Performance Performance Statistic Expenditures Expenditures Measure 1 Measure 2 Measure 3

Mean 1,296 1,367 221 1,086 1,480 Minimum 0 377 115,188 100,720 90,023 0.1 fractile 0 23 3,015 1,388 1,283 0.25 fractile 19 66 693 109 103 0.5 fractile 146 233 24 200 206 0.75 fractile 771 1012 577 1,220 1,386 0.9 fractile 3,295 3,267 3,817 5,610 6,385 Maximum 42,742 34,456 37,701 42,032 47,029 Standard Deviation 3,838 3,613 8,471 7,618 7,676 internal control system, has the final and to provide high-level counsel. Few responsibility for the functioning of the boards in the past decades have done this firm. Most importantly, it sets the rules of job well in the absence of external crises. the game for the CEO.The job of the board This is particularly unfortunate given that is to hire, fire, and compensate the CEO, the very purpose of the internal control 34 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS mechanism is to provide an early warning officer.’ I should have known right there system to put the organization back on that he wasn’t going to pay a goddamn track before difficulties reach a crisis bit of attention to anything I said.” So it stage. The reasons for the failure of the turned out, and after a year Hanley quit board are not completely understood, but the board in disgust. we are making progress toward understanding these complex issues. The The result is a continuing cycle of available evidence does suggest that CEOs ineffectiveness: by rewarding consent and are removed after poor performance,38 but discouraging conflicts, CEOs have the the effect, while statistically significant, power to control the board, which in turn seems too late and too small to meet the ultimately reduces the CEO’s and the obligations of the board. I believe bad company’s performance. This downward systems or rules, not bad people, underlie spiral makes the resulting difficulties the general failings of boards of directors. likely to be a crisis rather than a series of Some caution is advisable here because small problems met by a continuous while resolving problems with boards can self-correcting mechanism. The culture of cure the difficulties associated with a boards will not change simply in response nonfunctioning court of last resort, this to calls for change from policy makers, alone cannot solve all the problems with the press, or academics. It only will defective internal control systems. I resist follow, or be associated with, general the temptation in an already lengthy paper recognition that past practices have to launch into a discussion of other organi- resulted in major failures and substantive zational and strategic issues that must be changes in the rules and practices governing attacked. A well-functioning board, however, the system. is capable of providing the organizational culture and supporting environment for a Information problems continuing attack on these issues. Serious information problems limit the Board culture effectiveness of board members in the typical large corporation. For example, the CEO Board culture is an important component almost always determines the agenda and of board failure. The great emphasis on the information given to the board. This politeness and courtesy at the expense of limitation on information severely hinders truth and frankness in boardrooms is both the ability of even highly talented board a symptom and cause of failure in the members to contribute effectively to the control system. CEOs have the same inse- monitoring and evaluation of the CEO and curities and defense mechanisms as other the company’s strategy. human beings; few will accept, much less Moreover, the board requires expertise seek, the monitoring and criticism of an to provide input into the financial aspects active and attentive board. Magnet (1992, of planning—especially in forming the p. 86) gives an example of this environ- corporate objective and determining the ment. John Hanley, retired Monsanto CEO, factors which affect corporate value. accepted an invitation from a CEO Yet such financial expertise is generally lacking on today’s boards. Consequently, to join his board—subject, Hanley wrote, boards (and management) often fail to to meeting with the company’s general understand why long-run market value counsel and outside accountants as a maximization is generally the privately and kind of directorial due diligence. Says socially optimal corporate objective, and Hanley: “At the first board dinner the they often fail to understand how to CEO got up and said, ‘I think Jack was a translate this objective into a feasible foun- little bit confused whether we wanted him dation for corporate strategy and operating to be a director or the chief executive policy. THE MODERN INDUSTRIAL REVOLUTION 35 Legal liability company.While the initial investment could vary, it should seldom be less than The factors that motivate modern boards $100,000 from the new board member’s are generally inadequate. Boards are personal funds; this investment would force often motivated by substantial legal liabili- new board members to recognize from the ties through class action suits initiated by outset that their decisions affect their own shareholders, the plaintiff’s bar, and wealth as well as that of remote sharehold- others—lawsuits which are often triggered ers. Over the long term the investment can by unexpected declines in stock price.These be made much larger by options or other legal incentives are more often consistent stock-based compensation. The recent with minimizing downside risk rather than trend to pay some board member fees in maximizing value. Boards are also moti- stock or options is a move in the right vated by threats of adverse publicity from direction. Discouraging board members the media or from the political/regulatory from selling this equity is important so that authorities. Again, while these incentives holdings will accumulate to a significant often provide motivation for board mem- size over time. bers to cover their own interests, they do not necessarily provide proper incentives to take actions that create efficiency and Oversized boards value for the company. Keeping boards small can help improve their performance. When boards get Lack of management and board member beyond seven or eight people they are less equity holdings likely to function effectively and are easier for the CEO to control.40 Since the possibility for Many problems arise from the fact animosity and retribution from the CEO is that neither managers nor nonmanager too great, it is almost impossible for those board members typically own substantial who report directly to the CEO to participate fractions of their firm’s equity. While the openly and critically in effective evaluation average CEO of the 1,000 largest firms and monitoring of the CEO. Therefore, (measured by market value of equity) the only inside board member should be the holds 2.7 percent of his or her firm’s CEO. Insiders other than the CEO can equity in 1991, the median holding is only be regularly invited to attend board 0.2 percent and 75 percent of CEOs own meetings in an ex officio capacity. Indeed, 39 less than 1.2 percent (Murphy (1992)). board members should be given regular Encouraging outside board members to opportunities to meet with and observe hold substantial equity interests would executives below the CEO—both to expand provide better incentives. Stewart (1990) their knowledge of the company and CEO outlines a useful approach using levered succession candidates, and to increase equity purchase plans or the sale of in-the- other top-level executives’ understanding of money options to executives to resolve this the thinking of the board and the board problem in large firms, where achieving process. significant ownership would require huge dollar outlays by managers or board members. By requiring significant outlays by Attempts to model the process after managers for the purchase of these quasi political democracy equity interests, Stewart’s approach reduces the incentive problems created by Suggestions to model the board process the asymmetry of payoffs in the typical after a democratic political model in which option plan. various constituencies are represented are Boards should have an implicit under- likely to make the process even weaker.To standing or explicit requirement that new see this we need look no farther than members must invest in the stock of the the inefficiency of representative political 36 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS democracies (whether at the local, state, or process of hiring, firing, evaluating, and federal level), or at their management of compensating the CEO. Clearly, the CEO quasi-business organizations such as the cannot perform this function apart from his Post Office, schools, or power generation or her personal interest. Without the direc- entities such as the TVA. This does not mean, tion of an independent leader, it is much however, that the current corporate system more difficult for the board to perform its is satisfactory as it stands; indeed there is critical function.Therefore, for the board to significant room for rethinking and revision. be effective, it is important to separate the For example, proxy regulations by the CEO and chairman positions.41 The inde- SEC make the current process far less pendent chairman should, at a minimum, be efficient than it otherwise could be. given the rights to initiate board appoint- Specifically, it has been illegal for any ments, board committee assignments, and shareholder to discuss company matters (jointly with the CEO) the setting of the with more than ten other shareholders board’s agenda. All these recommenda- without prior filing with, and approval of, tions, of course, will be made conditional on the SEC. The November 1992 relaxation the ratification of the board. of this restriction allows an investor to An effective board will often evidence communicate with an unlimited number of tension among its members as well as with other stockholders provided the investor the CEO. But I hasten to add that I am not owns less than 5 percent of the shares, has advocating continuous war in the boardroom. no special interest in the issue being In fact, in well-functioning organizations the discussed, and is not seeking proxy authority. board will generally be relatively inactive These restrictions still have obvious short- and will exhibit little conflict. It becomes comings that limit effective institutional important primarily when the rest of the action by those shareholders most likely to internal control system is failing, and this pursue an issue. should be a relatively rare event. The As equity holdings become concentrated challenge is to create a system that will not in institutional hands, it is easier to resolve fall into complacency and inactivity during some of the free-rider problems that limit periods of prosperity and high-quality the ability of thousands of individual share- management, and therefore be unable to rise holders to engage in effective collective early to the challenge of correcting a failing action. In principle such institutions can management system. This is a difficult task therefore begin to exercise corporate because there are strong tendencies for control rights more effectively. Legal and boards to evolve a culture and social norms regulatory restrictions, however, have pre- that reflect optimal behavior under prosper- vented financial institutions from playing a ity, and these norms make it extremely major corporate monitoring role. (Roe difficult for the board to respond early to (1990, 1991), Black (1990), and Pound failure in its top management team.42 (1991) provide an excellent historical review of these restrictions.) Therefore, if institutions are to aid in effective gover- Resurrecting active investors nance, we must continue to dismantle the A major set of problems with internal rules and regulations that have prevented control systems are associated with the them and other large investors from curbing of what I call active investors accomplishing this coordination. (Jensen (1989a, 1989b)). Active investors are individuals or institutions that simulta- The CEO as chairman of the board neously hold large debt and/or equity positions in a company and actively partic- It is common in U.S. corporations for the ipate in its strategic direction. Active CEO to also hold the position of chairman investors are important to a well-functioning of the board. The function of the chairman governance system because they have the is to run board meetings and oversee the financial interest and independence to view THE MODERN INDUSTRIAL REVOLUTION 37 firm management and policies in an unbiased the Modigliani-Miller (M&M) theorems on way. They have the incentives to buck the the independence of firm value, leverage, system to correct problems early rather and payout policy have been extremely than late when the problems are obvious but productive in helping the finance profession difficult to correct. Financial institutions structure the logic of many valuation issues. such as banks, pensions funds, insurance The 1980s control activities, however, have companies, mutual funds, and money demonstrated that the M & M theorems managers are natural active investors, but (while logically sound) are empirically they have been shut out of board rooms incorrect. The evidence from LBOs, lever- and firm strategy by the legal structure, by aged restructurings, takeovers, and venture custom, and by their own practices.43 capital firms has demonstrated dramatically Active investors are important to a well- that leverage, payout policy, and ownership functioning governance system, and there is structure (that is, who owns the firm’s much we can do to dismantle the web of securities) do in fact affect organizational legal, tax, and regulatory apparatus that efficiency, cash flow, and, therefore, value.45 severely limits the scope of active investors Such organizational changes show these in this country.44 But even absent these effects are especially important in low- regulatory changes, CEOs and boards can growth or declining firms where the agency take actions to encourage investors to hold costs of free cash flow are large.46 large positions in their debt and equity and to play an active role in the strategic direc- Evidence from LBOs tion of the firm and in monitoring the CEO. Wise CEOs can recruit large block LBOs provide a good source of estimates of investors to serve on the board, even selling value gain from changing leverage, payout new equity or debt to them to induce their policies, and the control and governance commitment to the firm. Lazard Freres system because, to a first approximation, the Corporate Partners Fund is an example of company has the same managers and the an institution set up specifically to perform same assets, but a different financial policy this function, making new funds available and control system after the transaction.47 to the firm and taking a board seat to Leverage increases from about 18 percent advise and monitor management perform- of value to 90 percent, large payouts to prior ance. Warren Buffet’s activity through shareholders occur, equity becomes concen- Berkshire Hathaway provides another trated in the hands of managers (over example of a well-known active investor.He 20 percent on average) and the board played an important role in helping (about 20 and 60 percent on average, Salomon Brothers through its recent legal respectively), boards shrink to about seven and organizational difficulties following the or eight people, the sensitivity of managerial government bond bidding scandal. pay to performance rises, and the companies’ Dobrzynski (1993) discusses many varieties equity usually become nonpublicly traded of this phenomenon (which she calls rela- (although debt is often publicly traded). tionship investing) that are currently arising The evidence of DeAngelo, DeAngelo, both in the United States and abroad. and Rice (1984), Kaplan (1989), Smith (1990), and others indicates that premi- Using LBOs and venture capital firms ums to selling-firm shareholders are as models of successful organization, roughly 40 to 50 percent of the prebuyout governance, and control market value, cash flows increase by 96 percent from the year before the buyout Organizational experimentation in the to three years after the buyout, and value 1980s increases by 235 percent (96 percent market adjusted) from two months prior to the Founded on the assumption that firm cash buyout offer to the time of going public, flows are independent of financial policy, sale, or recapitalization about three years 38 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS later on average.48 Palepu and Wruck consisting of no more than eight (1992) show that large value increases also people, occur in voluntary recapitalizations where e. CEOs who are typically the only the company stays public but buys back a insider on the board, and finally significant fraction of its equity or pays out f. CEOs who are seldom the chairman a significant dividend. Clinical studies of of the board. individual cases demonstrate that these changes in financial and governance policies LBO associations and venture funds also generate value-creating changes in behavior solve many of the information problems of managers and employees.49 facing typical boards of directors. First, as a result of the due diligence process at the time the deal is done, both the managers A proven model of governance structure and the LBO and venture partners have LBO associations and venture capital funds extensive and detailed knowledge of virtu- provide a blueprint for managers and ally all aspects of the business. In addition, boards who wish to revamp their top-level these boards have frequent contact with control systems to make them more effi- management, often weekly or even daily cient. LBOs and venture capital funds are, during times of difficult challenges. This of course, the preeminent examples of contact and information flow is facilitated by active investors in recent U.S. history, and the fact that LBO associations and venture they serve as excellent models that can be funds both have their own staff. They also emulated in part or in total by virtually any often perform the corporate finance function corporation. The two have similar gover- for the operating companies, providing the nance structures, and have been successful major interface with the capital markets and in resolving the governance problems of investment banking communities. both slow growth or declining firms (LBO Finally, the close relationship between associations) and high growth entrepre- the LBO partners or venture fund partners neurial firms (venture capital funds).50 and the operating companies facilitates the Both LBO associations and venture cap- infusion of expertise from the board during ital funds, of which KKR and Kleiner times of crisis. It is not unusual for a part- Perkins are prominent examples, tend to be ner to join the management team, even as organized as limited partnerships. In effect, CEO, to help an organization through such the institutions which contribute the funds emergencies. Very importantly, there are to these organizations are delegating the market forces that operate to limit the task of being active investors to the general human tendency to micromanage and partners of the organizations. Both gover- thereby overcentralize management in the nance systems are characterized by: headquarters staff. If headquarters develops a reputation for abusing the relationship a. limited partnership agreements at with the CEO, the LBO or venture organiza- the top level that prohibit headquar- tion will find it more difficult to complete ters from cross-subsidizing one new deals (which frequently depend on the division with the cash from another, CEO being willing to sell the company to the b. high equity ownership on the part of LBO fund or on the new entrepreneur being managers and board members, willing to sell an equity interest in the new c. board members (who are mostly the venture to the venture capital organization). LBO association partners or the venture capitalists) who in their funds directly represent a large IX. IMPLICATIONS FOR THE fraction of the equity owners of FINANCE PROFESSION each subsidiary company, d. small boards of directors (of the One implication of the foregoing discussion operating companies) typically is that finance has failed to provide firms THE MODERN INDUSTRIAL REVOLUTION 39 with an effective mechanism to achieve structure of financial claims on the efficient corporate investment. While firm’s cash flows (e.g., equity, modern capital-budgeting procedures are bond, preferred stock, and warrant implemented by virtually all large corpora- claims).52 tions, it appears that the net present value 2. Governance—the top-level control (or more generally, value-maximizing) rule structure, consisting of the decision imbedded in these procedures is far rights possessed by the board of from universally followed by operating directors and the CEO, the proce- managers. In particular, the acceptance of dures for changing them, the size negative-value projects tends to be com- and membership of the board, and mon in organizations with substantial the compensation and equity hold- amounts of free cash flow (cash flow in ings of managers and the board.53 excess of that required to fund all value- 3. Organization—the nexus of contracts increasing investment projects) and in defining the internal “rules of the particular in firms and industries where game” (the performance measure- downsizing and exit are required. The ment and evaluation system, finance profession has concentrated on how the reward and punishment system, capital investment decisions should be and the system for allocating made, with little systematic study of how decision rights to agents in the they actually are made in practice.51 This organization).54 narrowly normative view of investment decisions has led the profession to ignore The close interrelationships between what has become a major worldwide these factors have dragged finance scholars efficiency problem that will be with us for into the analysis of governance and organi- several decades to come. zation theory.55 In addition, the perceived Agency theory (the study of the “excesses of the 1980s” have generated inevitable conflicts of interest that occur major reregulation of financial markets in when individuals engage in cooperative the United States affecting the control behavior) has fundamentally changed cor- market, credit markets (especially the porate finance and organization theory, but banking, thrift, and insurance industries), it has yet to affect substantially research and market microstructure.56 These on capital-budgeting procedures. No longer changes have highlighted the importance of can we assume managers automatically act the political and regulatory environment to (in opposition to their own best interests) financial, organizational, and governance to maximize firm value. policies, and generated a new interest in Conflicts between managers and the what I call the “politics of finance.”57 firm’s financial claimants were brought to The dramatic growth of these new center stage by the market for corporate research areas has fragmented the finance control in the last two decades.This market profession, which can no longer be divided brought widespread experimentation, simply into the study of capital markets teaching us not only about corporate and corporate finance. Finance is now finance, but also about the effects of much less an exercise in valuing a given leverage, governance arrangements, and stream of cash flows (although this is still active investors on incentives and organiza- important) and much more the study of tional efficiency. These events have taught how to increase those cash flows—an us much about the interdependencies effort that goes far beyond the capital among the implicit and explicit contracts asset-pricing model, Modigliani and Miller specifying the following three elements of irrelevance propositions, and capital budg- organizations: eting. This fragmentation is evidence of progress, not failure; but the inability to 1. Finance—I use this term narrowly understand this maturation causes conflict here to refer to the definition and in those quarters where research is judged 40 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS and certified, including the academic — the nature of implicit contracts, journals and university departments. the optimal degree to which Specialists in different subfields have private contracts should be tended to react by labeling research in left open to abrogation or areas other than their own as “low-quality” change, and how to bond or and “illegitimate.” Acknowledging this monitor to limit opportunistic separation and nurturing communication behavior regarding those among the subfields will help avoid this implicit contracts. intellectual warfare with substantial benefit to the progress of the profession. ● how politics, the press, and public My review of macro and organizational opinion affect the types of gover- trends in the previous pages has highlighted nance, financial, and organizational many areas for future research for finance policies that firms adopt. scholars.They include understanding: — how capital market forces can ● the implications of the modern (or be made a politically and Third) Industrial Revolution, and economically efficient part of how it will affect financial, product, corporate control mechanisms. and labor markets, as well as the — how active investors can be level and distribution of worldwide resurrected and reconciled with income and wealth. a legal structure that currently favors liquid and anonymous — how industry-wide excess markets over the intimate illiq- capacity arises, how markets uid market relations that seem and firms respond to such mar- to be required for efficient ket pressures, and why exit is so governance. difficult for organizations to deal with. — the implications of new X. CONCLUSION technology for organizational downsizing. For those with a normative bent, making — the financial policies appropriate the internal control systems of corporations for the new virtual or network work is the major challenge facing econo- organizations that are arising. mists and management scholars in the 1990s. For those who choose to take a ● the weaknesses that cause internal purely positive approach, the major chal- corporate control systems to fail lenge is understanding how these systems and how to correct them. work, and how they interact with the other control forces (in particular the product — the reasons for the asymmetry and factor markets, legal, political, and reg- between corporate growth and ulatory systems, and the capital markets) decline, and how to limit the impinging on the corporation. I believe the organizational and strategic reason we have an interest in developing inefficiencies that seem to creep positive theories of the world is so that we into highly successful rapidly can understand how to make things work growing organizations. more efficiently. Without accurate positive — how capital budgeting decisions theories of cause and effect relationships, are actually made and how normative propositions and decisions based organizational practices can be on them will be wrong. Therefore, the two implemented that will reduce objectives are completely consistent. the tendency to accept negative Financial economists have a unique value projects. advantage in working on these control and THE MODERN INDUSTRIAL REVOLUTION 41 organizational problems because we plus stock repurchases plus the ending understand what determines value, and we value of the interest payments plus debt know how to think about uncertainty and payments objective functions. To do this we have to nt nt understand even better than we do now the VT Vn dt(1 ) bt(1 r) , factors leading to organizational past failures (and successes): we have to break where the investor is assumed to reinvest open the black box called the firm, and this all intermediate payouts from the firm means understanding how organizations at the cost of equity and debt, and r and the people in them work. In short, respectively. we’re facing the problem of developing a Consider an alternative strategy which viable theory of organizations. To be pays the same dividends and stock repur- successful we must continue to broaden our chases, dt, (and raises the same outside thinking to new topics and to learn and capital) and puts the R&D and capital and develop new analytical tools. acquisition expenditures (in excess of This research effort is a very profitable depreciation) in marketable securities of venture. I commend it to you. the same risk as the R&D and capital expenditures, yielding expected returns equal to their cost of capital, i. Under the APPENDIX: DIRECT ESTIMATES OF assumption that the zero investment and THE PRODUCTIVITY OF R&D—THE R&D strategy yields a terminal value, Vn, MODEL equal to the ending debt, Bn, plus the beginning value of equity, S0, (that is, investment Consider a firm in period t with cash flow equal to depreciation is sufficient to main- from operations, Ct, available for: tain the original equity value of the firm), the value created by this strategy is Rt R&D expenditures, K capital investment, t d payments to shareholders in the V T S0 Bn (Kt Rt at) t nt nt form of dividends and net share (1 i) dt(1 ) n t repurchases, bt(1 r) bt interest and net debt payments, at acquisitions net of asset sales. The difference between the terminal values d 0, b 0, a 0 mean respectively of the two strategies is that a new equity is raised in the form of capital contributions from VT VT Vn Vn (Kt Rt at) equityholders, net bond issues (1 i)n t exceed interest and debt repay- Sn S0 (Kt Rt at) ments, and asset sales exceed (1 i)n t (1) acquisitions.

By definition Ct Rt Kt dt bt at This is my first crude measure of The initial value of the firm equals the sum the productivity of R&D and capital of the market values of equity and debt, expenditures. Unless capital and R&D V0 S0 B0 and the final value at the expenditures are completely unproductive, end of period n, is Vn. Assume for simplic- this measure will be biased downward. ity that taxes are zero, and debt is riskless. Therefore I define two more conservative If r is the riskless interest rate and is the measures that will yield higher estimates of cost of equity capital, the total value, VT, the productivity of these expenditures. The created by the firm’s investment, R&D, and second assumes that replacement of payout policy measured at the future hori- depreciation and zero expenditures on R&D zon date n, is the final value of the firm plus are sufficient to maintain the intermediate the ending value of the dividend payments cash flows but at the end of the period the 42 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS firm has equity value of zero. This second 1980) conducts detailed firm and industry studies. measure is: See also Hirschman’s (1970) work on exit. 2 Measured by multifactor productivity, U.S. Department of Labor (USDL) (1990, Table 3). VT VT Vn Bn (Kt Rt at) n t See Jensen (1991) for a summary. Multifactor (1 i) (2) productivity showed no growth between 1973 and 1980 and grew at the rate of 1.9 per- Alternatively, to allow for the effects of cent per year between 1950 and 1973. the reduced investment and R&D on inter- Manufacturing labor productivity grew at an mediate cash flows my third measure annual rate of 2.3 percent in the period 1950 to assumes that all intermediate cash flows 1981 and at 3.8 percent in 1981 to 1990 are reduced in the benchmark investment (USDL, 1990, Table 3). Using data recently revised by the Bureau of Economic Analysis strategy by the amount paid out to share- from 1977 to 1990, the growth rate in the holders in the form of dividends and net earlier period was 2.2 and 3.0 percent in the share repurchases and that the original 1981 to 1990 period (USDL, 1991, Table 1). value of the equity is maintained. This Productivity growth in the nonfarm business measure is likely to yield an upward biased sector fell from 1.9 percent in the 1950 to 1981 estimate of the productivity of R&D and period to 1.1 percent in the 1981 to 1990 capital expenditures.The measure is:58 period (USDL, 1990, Table 2). The reason for the fall apparently lies in the relatively large V V V d r n t growth in the service sector relative to the T T n t (1 ) manufacturing sector and the low measured S0 Bn (Kt Rt at) productivity growth in services. (1 i)n t (3) There is considerable controversy over the adequacy of the measurement of productivity in the service sector.The USDL has no productivity measures for services employing nearly NOTES 70 percent of service workers, including, among others, health care, real estate, and securities * Harvard Business School. Presidential brokerage. In addition, many believe that service Address to the American Finance Association, sector productivity growth measures are downward January 1993, Anaheim, California. I appreci- biased. Service sector price measurements, for ate the research assistance of Chris Allen, Brian example, take no account of the improved Barry, Susan Brumfield, Karin Monsler,and par- productivity and lower prices of discount outlet ticularly Donna Feinberg, the support of the clubs such as Sam’s Club. The Commerce Division of Research of the Harvard Business Department measures output of financial School, and the comments of and discussions services as the value of labor used to produce it. with George Baker, Carliss Baldwin, Joe Bower, Since labor productivity is defined as the value Alfred Chandler, Harry and Linda DeAngelo, of total output divided by total labor inputs it is Ben Esty, Takashi Hikino, Steve Kaplan, Nancy impossible for measured productivity to grow. Koehn, Claudio Loderer, George Lodge, John Between 1973 and 1987 total equity shares Long, Kevin Murphy, Malcolm Salter, René traded daily grew from 5.7 million to 63.8 Stulz, Richard Tedlow, and especially Richard million, while employment only doubled—implying Hackman, Richard Hall, and Karen Wruck on considerably more productivity growth than many of these ideas. that reflected in the statistics. Other factors, 1 In a rare finance study of exit, DeAngelo and however, contribute to potential overestimates DeAngelo (1991) analyze the retrenchment of of productivity growth in the manufacturing the U.S. steel industry in the 1980s. Ghemawat sector. See Malabre and Clark (1992) and and Nalebuff (1985) have an interesting paper Richman (1993). entitled “Exit,” and Anderson (1986) provides 3 Nominal and real hourly compensation, a detailed comparison of U.S. and Japanese Economic Report of the President, Table B42 retrenchment in the 1970s and early 1980s and (1993). their respective political and regulatory policies 4 USDL (1991). toward the issues. Bower (1984, 1986) analyzes 5 USDL (1990). Trends in U.S. producivity the private and political responses to decline have been controversial issues in academic and in the petrochemical industry. Harrigan (1988, policy circles in the last decade. One reason, THE MODERN INDUSTRIAL REVOLUTION 43 I believe, is that it takes time for these 15 Zellner (1992) discusses the difficulties complicated changes to show up in the traditional retailers have in meeting Wal-Mart’s aggregate statistics. In their recent book prices. Baumol, Blackman, and Wolff (1989, pp. ix-x) 16 Vietor, Forthcoming. changed their formerly pessimistic position. In 17 Source: COMPUSTAT. their words: “This book is perhaps most easily 18 “In 1980 IBM’s top-of-the-line computers summed up as a compendium of evidence provided 4.5 MIPS (millions of instructions per demonstrating the error of our previous second) for $4.5 million. By 1990, the cost of a ways. ...The main change that was forced MIP on a personal computer had dropped to upon our views by careful examination of the $1,000 . . .” (Keene (1991)), p. 110). By 1993 long-run data was abandonment of our earlier the price had dropped to under $100. The gloomy assessment of American productivity technological progress in personal computers performance. It has been replaced by the has itself been stunning. Intel’s Pentium (586) guarded optimism that pervades this book. This chip, introduced in 1993, has a capacity of 100 does not mean that we believe retention of MIPS—100 times the capacity of its 286 chip American leadership will be automatic or easy. introduced in 1982 (Brandt (1993)). In Yet the statistical evidence did drive us to con- addition, the progress of storage, printing, and clude that the many writers who have suggested other related technology has also been rapid that the demise of America’s traditional posi- (Christensen (1993)). tion has already occurred or was close at hand 19 The Journal of Financial Economics which were, like the author of Mark Twain’s obituary, I have been editing with several others since a bit premature. . . . It should, incidentially, be 1973 is an example. The JFE is now edited by acknowledged that a number of distinguished seven faculty members with offices at three economists have also been driven to a similar universities in different states and the main evaluation . . . ” editorial administrative office is located in yet 6 As measured by the Wilshire 5,000 index of another state. North Holland, the publisher, is all publicly held equities. located in Amsterdam, the printing is done in 7 Bureau of the Census (1991). India, and mailing and billing is executed in 8 Business Week Annual R&D Scoreboard. Switzerland. This “networked organization” 9 Annual premiums reported by Mergerstat would have been extremely inefficient two Review (1991, fig. 5) weighted by value of decades ago without fax machines, high-speed transaction in the year for this estimate. modems, electronic mail, and overnight delivery 10 I assume that all transactions without services. publicly disclosed prices have a value equal to 20 Wruck and Jensen (1993) provide an 20 percent of the value of the average publicly analysis of the critical organizational innovations disclosed transaction in the same year, and that that total quality management is bringing to the they have average premiums equal to those for technology of management. publicly disclosed transactions. 21 A collapse I predicted in Jensen (1989a). 11 In some cases buyers overpay, perhaps 22 See Neff, Holyoke, Gross, and Miller because of mistakes or because of agency (1993). problems with their own shareholders. Such 23 Steel industry employment is now down to overpayment represents only a wealth transfer 160,000 from its peak of 600,000 in 1953 from the buying firm’s claimants to those of the (Fader (1993)). selling firm and not an efficiency gain. 24 I am indebted to Steven Cheung for 12 Healy, Palepu, and Ruback (1992) esti- discussions on these issues. mate the total gains to buying- and selling-firm 25 Although migration will play a role it will shareholders in the 50 largest mergers in the be relatively small compared to the export of period 1979 to 1984 at 9.1 percent. They also the labor in products and services. Swissair’s find a strong positive cross-sectional relation 1987 transfer of part of its reservation system between the value change and the cash flow to Bombay and its 1991 announcement of plans changes resulting from the merger. to transfer 150 accounting jobs to the same city 13 See Kaplan (1989), Jensen, Kaplan, and are small examples (Economist Intelligence Stiglin (1989), Pontiff, Shleifer, and Weisbach Unit (1991)). (1990), Asquith and Wizman (1990), and 26 Thailand and China have played a role in Rosett (1990). the world markets in the last decade, but since 14 Its high of $139.50 occurred on 2/19/91 it has been such a small part of their potential I and it closed at $50.38 at the end of 1992. have left them in the potential entrant category. 44 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 27 In a recent article focusing on the (1990), Tedlow (1991), Tiemann (1990), prospects for textile manufacturer investment in Wruck and Stephens (1992a, 1992b), Wruck Central European countries, van Delden (1993, (1990, 1991, 1992), Wruck and Palepu p. 43) reports: “When major French group (1992). Rhone Poulenc’s fibres division started a 31 In May 1985, Uniroyal approved an LBO discussion for a formal joint venture in 1991, proposal to block hostile advances by Carl they discovered an example of astonishing Icahn. About the same time, BF Goodrich began competitiveness. Workers—whose qualifications diversifying out of the tire business. In January matched those normal in the West—cost only 1986, Goodrich and Uniroyal independently 8% of their West European counterparts, and spun off their tire divisions and together, in a yet achieved productivity rates of between 60% 50–50 joint venture, formed the Uniroyal- and 75% compared to EC level. Moreover, Goodrich Tire Company. By December 1987, energy costs of the integrated power station are Goodrich had sold its interest in the venture to 50% below West German costs, and all of this Clayton and Dubilier; Uniroyal followed soon is complemented by extremely competitive raw after. Similarly, General tire moved away from material prices.” tires: the company, renamed GenCorp in 1984, The textile industry illustrates the problems sold its tire division to Continental in 1987. with chronic excess capacity brought on by a Other takeovers in the industry during this situation where the worldwide demand for period include the sale of Firestone to textiles grows fairly constantly, but growth in the Bridgestone and Pirelli’s purchase of the productivity of textile machinery through tech- Armstrong Tire Company. By 1991, Goodyear nological improvements is greater. Moreover, was the only remaining major American tire additional capacity is being created because manufacturer.Yet it too faced challenges in the new entrants to the global textile market must control market: in 1986, following three years upgrade outdated (and less productive) weaving of unprofitable diversifying investments, machinery with new technology to meet mini- Goodyear initiated a major leveraged stock mum global quality standards. This means repurchase and restructuring to defend itself excess capacity is likely to be a continuing from a hostile takeover from Sir James problem in the industry and that adjustment will Goldsmith. Uniroyal-Goodrich was purchased by have to occur through exit of capacity in Michelin in 1990. See Tedlow (1991). high-cost Western textile mills. 32 In 1992 dollars, calculated from 28 Total quality management programs Mergerstat Review, 1991, p. 100f. strongly encourage managers to benchmark 33 In 1992 dollars, Mergerstat Review, their firm’s operations against the most 1991, figs. 29 and 38. successful worldwide competitors, and good 34 Source: GE annual reports. cost systems and competitive benchmarking are 35 Because of the sharp decline in Japanese becoming more common in well-managed firms. stock prices, the Japanese firms ranked in the 29 Shleifer and Summers (1988) seem to top 35 firms would have performed less well if take the position that all implicit contracts the period since 1990 had been included. should be enforced rigidly and never be 36 I am in the process of creating a database breached. that avoids the survivorship bias. Hall (1993a, 30 A partial list of references is: Dann and 1993b) in a large sample free of survivorship DeAngelo (1988), Mann and Sicherman bias finds lower market valuation of R&D in the (1991), Baker and Wruck (1989), Berger and 1980s and hypothesizes that this is due to a Ofek (1993), Bhide (1993), Brickley, Jarrell, higher depreciation rate for R&D capital. The and Netter (1988), Denis (1992), Donaldson Stern Stewart Performance 1000 (1992) ranks (1990), DeAngelo and DeAngelo (1991), companies by a measure of the economic value DeAngelo, DeAngelo, and Rice (1984), Esty added by management decisions that is an (1992, 1993), Grundfest (1990), Holderness alternative to performance measures 1–3 sum- and Sheehan (1991), Jensen (1986a, 1986b, marized in Table 1.4 GM also ranks at the 1988, 1989a, 1989b, 1991), Kaplan (1989a, bottom of this list. 1989b, 1992), Lang, Poulsen, and Stulz 37 Changes in market expectations about the (1992), Lang, Stulz, and Walkling (1991), prospects for a firm (and therefore changes in Lewellen, Loderer, and Martin (1987), its market value) obviously can affect the Lichtenberg (1992), Lichtenberg and Siegel interpretation of the performance measures. (1990), Ofek (1993), Palepu (1990), Pound Other than using a long period of time there is (1988, 1991, 1992), Roe (1990, 1991), Smith no simple way to handle this problem.The large THE MODERN INDUSTRIAL REVOLUTION 45 increase in stock prices in the 1980s would 1992b), Wruck and Stephens (1992a, 1992b), indicate that expectations were generally being Jensen and Barry (1992), Jensen, Burkhardt, revised upward. and Barry (1992), Jensen, Dial, and Barry 38 CEO turnover approximately doubles from (1992), Lang and Stultz (1992), Denis (1992). 3 to 6 percent after two years of poor perform- 46 See Jensen (1986), and the references in ance (stock returns less than 50 percent below the previous footnote. equivalent-risk market returns, Weisbach 47 Assets do change somewhat after an LBO (1988)), or increases from 8.3 to 13.9 percent because such firms often engage in asset sales from the highest to the lowest performing decile after the transaction to pay down debt and to of firms (Warner, Watts, and Wruck (1988)). get rid of assets that are peripheral to the See also DeAngelo (1888) and DeAngelo and organization’s core focus. DeAngelo (1989). 48 See Palepu (1990) for a review of 39 See also Jensen and Murphy (1990a, research on LBOs, their governance changes, 1990b) for similar estimates based on earlier and their productivity effects. Kaplan and data. Stein (1993) show similar results in more 40 In their excellent analysis of boards, recent data. Lipton and Lorsch (1992) also criticize the 49 See references in footnote 45 above. In a functioning of traditionally configured boards, counterexample, Healy and Palepu argue in recommend limiting membership to seven or their study of CUC that the value increase fol- eight people, and encourage equity ownership by lowing its recapitalization occurred not because board members. Research supports the proposi- of incentive effects of the deal, but because of tion that as groups increase in size they become the information the recapitalization provided to less effective because the coordination and the capital markets about the nature of the process problems overwhelm the advantages company’s business and profitability. gained from having more people to draw on (see 50 Jensen (1989a, 1989b) and Sahlman Steiner (1972) and Hackman (1990)). (1990) analyze the LBO association and 41 Lipton and Lorsch (1992) stop short of venture capital funds respectively. recommending appointment of an independent 51 Counterexamples are Bower (1970), chairman, recommending instead the appoint- Baldwin and Clark (1992), Baldwin (1982, ment of a “lead director” whose functions 1988, 1991), Baldwin and Trigeorgis (1992), would be to coordinate board activities. and Shleifer and Vishny (1989). 42 Gersick and Hackman (1990) and 52 See Harris and Raviv (1988, 1991), and Hackman (1993) study a similar problem: the Stulz (1990). issues associated with habitual behavior 53 Jensen (1989a, 1989b), Kester (1991), routines in groups to understand how to create Sahlman (1990), Pound (1988, 1991, 1992). more productive environments. They apply the 54 Jensen (1983), Jensen and Meckling analysis to airline cockpit crews. (1992). 43 See Roe (1990, 1991), Black (1990), and 55 Examples of this work include Gilson, Pound (1991). Lang, and John (1990), Wruck (1990, 1991), 44 See Porter (1992a, 1992b, 1992c). Hall Lang and Stulz (1992), Lang, and Poulsen, and and Hall provide excellent empirical tests of the Stulz (1992) on bankruptcy and financial myopic capital market hypothesis on which distress, Warner, Watts, and Wruck (1989), much of debate on the functioning of U.S. Weisbach (1988), Jensen and Murphy (1990a, capital markets rests. 1990b) and Gibbons and Murphy (1990) on 45 See Kaplan (1989a, 1989b, 1989c, executive turnover, compensation, and organiza- 1992), Smith (1990), Wruck (1990), tional performance, Esty (1992, 1993) on the Lichtenberg (1992), Lichtenberg and Siegel effects of organizational form on thrift losses, (1990), Healy, Palepu, and Ruback (1992), and Gilson and Kraakman (1991) on governance, Ofek (1993). Brickley and Dark (1987) on franchising, There have now been a number of detailed Boycko, Shleifer, and Vishny (1993), Kaplan clinical and case studies of these transactions (1989a, 1989b, 1989c, 1992), Smith (1990), that document the effects of the changes on Kaplan and Stein (1993), Palepu (1990), and incentives and organizational effectiveness as Sahlman (1990) on leverage buyouts and well as the risks of bankruptcy from overlever- venture capital organizations. aging. See Baker and Wruck (1989), Wruck 56 The 1989 Financial Institutions Reform, (1991, 1992a), Holderness and Sheehan Recovery, and Enforcement Act increased (1988, 1991), Wruck and Keating (1992a, federal authority and sanctions by shifting 46 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS regulation of the S&L industry from the FDIC Baker, George, and Karen Wruck, 1989, to the Treasury, and insurance of the industry Organizational changes and value creation in from FSLIC to the Federal Home Loan Bank leveraged buyouts: The case of O. M. Scott Board.The act banned thrift investment in high- and Sons Company, Journal of Financial yield bonds, raised capital ratios and insurance Economics 25, 163–190. premiums. The 1990 Comprehensive Thrift and Baldwin, Carliss Y., 1982, Optimal sequential Bank Fraud Prosecution and Tax Payer Recovery Act increased criminal penalties for investment when capital is not readily financial institution-related crimes. The 1991 reversible, Journal of Finance 37, 763–782. FDIC Improvement Act tightened examination ——, 1988, Time inconsistency in capital and auditing standards, recapitalized the Bank budgeting, Working paper, Harvard Business Insurance Fund and limited foreign bank pow- School. ers. The Truth in Banking Act of 1992 required ——, 1991, How capital budgeting deters stricter disclosure of interest rates and fees on innovation—and what companies can do deposit accounts and tightened advertising about it, Research-Technology Management, guidelines. The National Association of November-December, 39–45. Insurance Commissioners (NAIC) substantially ——and Kim B. Clark, 1992. Capabilities and restricted the ability of insurance companies to invest in high-yield debt in 1990 to 1991. capital investment: New perspectives on cap- 57 See Jensen (1989b, 1991), Roe (1990, ital budgeting, Journal of Applied Corporate 1991), Grundfest (1990), Bhide (1993), Black Finance 5, 67–82. (1990), Pound (1988, 1991, 1992), and —— and Lenos Trigeorgis, 1992, Toward DeAngleo, DeAngelo, and Gilson (1993). remedying the underinvestment problem: 58 Most conservatively, we could assume that Competitiveness, real options, capabilities, and the cutback in R&D and capital expenditures TQM,Working paper,Harvard Business School. under the alternative strategy results in a reduc- Baumol, William, Sue Anne Beattey Blackman, tion in intermediate cash flows by the amount of and Edward Wolff, 1989, Productivity and the net cash paid to shareholders in the form of American Leadership (MIT Press, Boston). dividends and share repurchases and a final equity value of zero. Berger, Philip, and Eli Ofek, 1993, Leverage and value: The role of agency costs and taxes, Unpublished manuscript,The Wharton School. V V V d (1 r)n t B T T n t n Bhide, Amar, 1993, The hidden costs of stock n t (Kt Rt)(1 i) (4) market liquidity, Journal of Financial Economics 34, 31–55. I expect this measure provides an unreasonably Black, Bernard S., 1990, Shareholder passivity high estimate of the productivity of R&D and reexamined, Michigan Law Review 89, investment expenditures and therefore do not report it. 520–608. Bower,Joseph L., 1970, Managing the Resource Allocation Process: A Study of Corporate Planning and Investment (Division of REFERENCES Research, Graduate School of Business Administration, Harvard University, Boston). Anderson, Douglas, 1986, Managing retreat: ——, 1984, Restructuring petrochemicals: A Disinvestment policy, in Thomas K. comparative study of business and govern- McCraw, ed.: America versus Japan ment to deal with a declining sector of the (Harvard Business School Press, Boston), economy, in Bruce R. Scott and George C. pp. 337–372. Lodge, eds.: U.S. Competitiveness in the Argyris, Chris, 1990, Overcoming Organizational World Economy (Harvard Business School Defenses (Allyn and Bacon, Needham, Press, Boston). Mass.). ——, 1986, When Markets Quake (Harvard Asquith, Paul, and Thierry A. Wizman, 1990, Business School Press, Boston). Event risk, convenants, and bondholder Boycko, Maxim, Andrei Shleifer, and Robert W. returns in leveraged buyouts, Journal of Vishny, 1993, Voucher privatization, Financial Economics 27, 195–213. Unpublished manuscript, Harvard University. THE MODERN INDUSTRIAL REVOLUTION 47 Brandt, Richard, 1993, Tiny transistors and —— and Edward Rice, 1984, Going private: gold pizza, Business Week, March 29. Minority freezeouts and stockholder wealth, Brickley, James A., and Frederick H. Dark, Journal of Law and Economics 27, 367–401. 1987, The choice of organizational form: The —— and Stuart C. Gilson, 1993,The collapse of case of franchising, Journal of Financial First Executive Corporation: Junk bonds, Economics 18, 401–420. adverse publicity, and the “run on the bank” ——, Gregg A. Jarrell, and Jeffrey M. Netter, phenomenon, Unpublished manuscript, 1988, The market for corporate control: The University of Southern California. empirical evidence since 1980, Journal of DeAngelo, Linda, 1988, Managerial competition, Economic Perspectives 2, 49–68. information costs, and corporate governance: ——, Ronald C. Lease, and Clifford W. Smith, The use of accounting performance measures Jr., 1988, Ownership structure and voting on in proxy contests, Journal of Accounting and antitakeover amendments, Journal of Economics 10, January, 3–36. Financial Economics 20, 267–291. Denis, David J., 1992, Organizational form and Burnham, James D., 1993, Changes and the consequences of highly leveraged transac- Challenges: The Transformation of the U.S. tions: Kroger’s recapitalization and Steel Industry, Policy Study No. 115 (Center Safeway’s LBO, Unpublished manuscript, for the Study of American Business, Virginia Polytechnic Institute. Washington University, St. Louis, Mo.). Dobrzynski, Judith, 1993, Relationship investing, Chandler, Alfred D., Jr., 1977, The Visible Business Week, March 15. Hand: The Managerial Revolution in Donaldson, Gordon, 1990,Voluntary restructur- American Business (Harvard University ing: The case of General Mills, Journal of Press, Cambridge, Mass.). Financial Economics 27, 117–141. ——, 1990, Scale and Scope, The Dynamics of Economic Report of the President, 1993 (U.S. Industrial Capitalism (The Belknapp Press of Government Printing Office,Washington, D.C.). Harvard University Press, Cambridge, Economist Intelligence Unit, 1991, Switzerland Mass.). Country Report, No. 3. ——, 1992, The emergence of managerial capi- Economist, 1990, Out of the ivory tower, talism, Harvard Business School Case No. February 3. 9–384–081, revised by Thomas J. McCraw. Esty, Benjamin C., 1992, Organizational form, Christensen, Clayton, 1993, The rigid disk drive leverage and incentives: A study of risk- industry, 1956–1990: A history of commer- taking in the S&L industry, Unpublished cial and technological turbulence, Working manuscript, Harvard Business School. paper, Harvard Business School. ——, 1993, A tale of two thrifts, Unpublished Comment, Robert, and G.William Schwert, 1993, manuscript, Harvard Business School. Poison or placebo? Evidence on the deterrent Fader, Barnaby J., 1993, Struggle to survive in and wealth effects of modern antitakeover the town that steel forgot, New York Times, measures. Unpublished manuscript, Simon April 27. School of Business, University of Rochester. Federal Communications Commission, 1991, Dann, Larry Y., and Harry DeAngelo, 1988, Competition in the Interstate Interexchange Corporate financial policy and corporate con- Marketplace, FCC 91–251, Sept. 16. trol: A study of defensive adjustments in asset Gersick, Connie J., and J. Richard Hackman, and ownership structure, Journal of Financial 1990, Habitual routines in task-performing Economics 20, 87–127. teams, Organizational Behavior and Human DeAngelo, Harry, and Linda DeAngelo, 1989, Decision Processes 47, 65–97. Proxy contests and the governance of publicly Ghemawat, Pankaj, and Barry Nalebuff, 1985, held corporations, Journal of Financial Exit, Rand Journal of Economics 16, 184–194. Economics 23, 29–59. Gibbons, Robert, and Kevin J. Murphy, 1990, ——, 1991, Union negotiations and corporate Relative performance evaluation for chief policy: A study of labor concessions in the executive officers, Industrial and Labor domestic steel industry during the 1980s, Relations Review 43, 30S–51S (Special Journal of Financial Economics 30, 3–43. issue, Cornell University). 48 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Gilson, Ronald, and Reinier Kraakman, 1991, Organizations and States (Harvard Reinventing the outside director: An agenda University Press, Cambridge, Mass.). for institutional investors, Stanford Law Holderness, Clifford G., and Dennis P. Sheehan, Review 43, 863–900. 1988,The role of majority shareholders in pub- Gilson, Stuart C., Kose John, and Larry H. P. licly held corporations: An exploratory analysis, Lang, 1990, Trouble debt restructurings: An Journal of Financial Economics 20, 317–346. empirical study of private reorganization of ——, 1991, Monitoring an owner: The case of firms in default, Journal of Financial Turner Broadcasting, Journal of Financial Economics 27, 315–353. Economics 30, 325–346. Grossman, Sanford J., and Oliver D. Hart, 1988, Jensen, Michael C., 1983, Organization theory One share/one vote and the market for corpo- and methodology, Accounting Review 58, rate control, Journal of Financial Economics 319–339. 20, 175–202. ——, 1986a,The agency costs of free cash flow: Grundfest, Joseph, 1990, Subordination of Corporate finance and takeovers, American American capital, Journal of Financial Economic Review 76, 323–329. Economics 27, 89–117. ——, 1986b,The takeover controversy: Analysis Hackman, J. Richard, ed., 1990, Groups That and evidence, Midland Corporate Finance Work (Jossey-Bass, San Francisco). Journal 4(2), 6–32. ——, 1993, Teams, leaders, and organizations: ——, 1988, Takeovers: Their causes and conse- New directions for crew-oriented flight quences, Journal of Economic Perspectives training, in E. L. Weiner, B. G. Kanki, and 2, 21–48. R. L. Helmreich, eds.: Cockpit Resource ——, 1989a, Eclipse of the public corporation, Management (Academic Press, Orlando, Fla.). Harvard Business Review 67(5), 61–74. Hall, Bronwyn H., 1993a, The stock market’s ——, 1989b, Active investors, LBOs, and the pri- valuation of R&D investment during the vatization of bankruptcy, Journal of Applied 1980’s, American Economic Review, 83, Corporate Finance 2, 35–44. 259–264. ——, 1991, Corporate control and the politics of ——, 1993b, New evidence on the impacts of finance, Journal of Applied Corporate research and development, Unpublished Finance 4, 13–33. manuscript, Stanford University. —— and Brian Barry, 1992, Gordon Cain and —— and Robert E. Hall, 1993, The value and the Sterling Group (A) and (B), Harvard performance of U.S. corporations, Business School Case Nos. 9–942–021 and Unpublished manuscript, Stanford University. 9–942–022. Harrigan, Kathryn R., 1980, Strategies for Jensen, Michael C.,Willy Burkhardt, and Brian K. Declining Businesses (Lexington Books, Barry, 1992, Wisconsin Central Ltd. Railroad Lexington, Mass.). and Berkshire Partners (A): Leverage buyouts ——, 1988, Managing Maturing Businesses: and financial distress, Harvard Business School Restructuring Declining Industries and Case No. 9–190–062. Revitalizing Troubled Operations (Lexington Jensen, Michael C., Jay Dial, and Brian K. Books, Lexington, Mass.). Barry, 1992, Wisconsin Central Ltd. Railroad Harris, Milton, and Artur Raviv, 1988, and Berkshire Partners (B): LBO associa- Corporate governance: Voting rights and tions and corporate governance, Harvard majority rules, Journal of Financial Business School Case No. 9–190–070. Economics 20, 203–235. Jensen, Michael C., Steven Kaplan, and Laura ——, 1991, The theory of capital structure, Stiglin, 1989, Effects of LBOs on tax rev- Journal of Finance 46, 297–355. enues of the U.S. Treasury, Tax Notes 42, Healy, Paul M., Krishna G. Palepu, and Richard 727–733. S. Ruback, 1992, Does corporate perform- Jensen, Michael C., and William H. Meckling, ance improve after mergers?, Journal of 1992, Knowledge, control and organizational Financial Economics 31, 135–175. structure, in Lars Werin and Hans Wijkander, Hirschman, Albert, 1970, Exit, Voice and eds.: Current Economics (Blackwell, Oxford), Loyalty: Responses to Decline in Firms, pp. 251–274. THE MODERN INDUSTRIAL REVOLUTION 49 Jensen, Michael C., and Kevin J. Murphy, —— and Ralph A. Walkling, 1991, A test of the 1990a, Performance pay and top-management free cash flow hypothesis: The case of bidder incentives, Journal of Political Economy 98, returns, Journal of Financial Economics 29, 225–264. 315–335. ——, 1990b, CEO incentives—It’s not how Lewellen, Wilbur, Claudio Loderer, and Kenneth much you pay, but how, Harvard Business Martin, 1987, Executive compensation and Review, 68(3), 138–153. executive incentive problems: An empirical Jensen, Michael C., and Jerold B.Warner,1988, analysis, Journal of Accounting and The distribution of power among corporate Economics 9, 287–310. managers, shareholders, and directors. Lichtenberg, Frank R., 1992, Corporate Journal of Financial Economics 20, 3–24. Takeovers and Productivity (M.I.T. Press: Kaplan, Steven N., 1989a, The effects of man- Cambridge, Mass.). agement buyouts on operating performance —— and Donald Siegel, 1990, The effects of and value, Journal of Financial Economics leveraged buyouts on productivity and related 24, 581–618. aspects of firm behavior, Journal of Financial ——, 1989b, Campeau’s acquisition of Economics 27, 165–194. Federated: Value added or destroyed, Journal Lipton, Martin, 1989, Corporate governance: of Financial Economics 25, 191–212. Major issues for the 1990’s, Address to the ——, 1989c, Management buyouts: Evidence on Third Annual Corporate Finance Forum at taxes as a source of value, Journal of Finance the J. Ira Harris Center for the Study of 44, 611–632. Corporate Finance, University of Michigan ——, 1992, Campeau’s acquisition of School of Business, April 6. Federated: A post-petition post-mortem, —— and Jay Lorsch, 1992, A modest proposal working paper, University of Chicago. for improved corporate governance, Business —— and Jeremy Stein, 1990, How risky is the Lawyer 48, 59–77. debt in highly leveraged transactions?, Loomis, Carol J., 1993, Dinosaurs?, Fortune, Journal of Financial Economics 27, June 15, 85–000. 215–245. McCraw,Thomas K., 1981, Rethinking the trust ——, 1993, The evolution of buyout pricing and question, in T. McCraw, ed.: Regulation in financial structure in the 1980s, Quarterly Perspective (Division of Research, Harvard Journal of Economics 108, 313–358. University, Graduate School of Business, dis- Keene, Peter G., 1991, Every Manager’s Guide tributed by Harvard University Press, to Information Technology (Harvard Boston), pp. 1–55. University Press, Boston). ——, 1992, Antitrust:The perceptions and real- Kester, Carl, 1991, Japanese Takeovers: The ity in coping with big business, Harvard Global Contest for Corporate Control Business School Case No. 9–391–292. (Harvard Business School Press, Cambridge, McMurray, Donald L., 1929, Coxey’s Army: A Mass.). Study of the Industrial Army Movement of Lamoreaux, Naomi, R., 1985, The Great 1894 (Little, Brown, Boston). Merger Movement in American Business, Magnet, Myron, 1992, Directors, wake up! 1895–1904 (Cambridge University Press, Fortune, June 15, 85–92. Cambridge). Malabre, Alfred L., Jr., and Lindley H. Clark, Lang, Larry H. P., Annette Poulsen, and Rene Jr., 1992, Dubious figures: Productivity M. Stulz, 1992, Asset sales, leverage, and the statistics for the service sector may agency costs of managerial discretion, understate gains, Wall Street Journal, Unpublished manuscript, Ohio State August 12. University. Mann, Steven V., and Neil W. Sicherman, 1991, Lang, Larry H. P., and Rene M. Stulz, 1992, The agency costs of free cash flow: Contagion and competitive intra-industry Acquisition activity, and equity issues, effects of bankruptcy announcements: An Journal of Business 64, 213–227. empirical analysis, Journal of Financial Mergerstat Review, 1991 (Merrill Lynch, Economics 32, 45–60. Schaumburg, Illinois). 50 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Munro, J. Richard, 1989, Takeovers: The myths Price, Walter W., 1933, We Have Recovered behind the mystique, Vital Speeches, May 15. Before! (Harper & Brothers, New York). Murphy, Kevin, 1992, Executive Compensation Reynolds, A.William, 1988, Chairman and Chief in Corporate America, 1992 (United Executive Officer, GenCorp Inc., before Shareholders Association, Washington, D.C.). the Subcommittee on Oversight and —— and Jay Dial, 1992, Compensation and Investigations, U.S. House Committee on strategy at General Dynamics (A) and (B), Energy and Commerce, February 8. Harvard Business School Case Nos. Richman, Louis S., 1993, Why the economic N9–493–032 and N9–493–033. data mislead us, Fortune, March 8. Neff, Robert, Larry Holyoke, Neil Gross, and Roe, Mark J., 1990, Political and legal Karen Lowry Miller, 1993, Fixing Japan, restraints on ownership and control of public Business Week, March 29. companies, Journal of Financial Economics Ofek, Eli, 1993, Capital structure and firm 27, 7–42. response to poor performance: An empirical ——, 1991, a political theory of American analysis, Journal of Financial Economics corporate finance, Columbia Law Review 91, 34, 3–30. 10–67. Palepu, Krishna G., 1990, Consequences of Rosett, Joshua G., 1990, Do union wealth leveraged buyouts, Journal of Financial concessions explain takeover premiums? The Economics 27, 247–262. evidence on contract wages, Journal of Polinsky, A. Mitchell, 1989, An Introduction to Financial Economics 27, 263–282. Law and Economics (Little, Brown & Co., Sahlman, William A., 1990, The structure and Boston). governance of venture-capital organizations. Pontiff, Jeffrey, Andrei Shleifer, and Michael S. Journal of Financial Economics 27, 473–521. Weisbach, 1990, Reversions of excess pension Sahlman, William A., and Howard H. assets after takeovers, Rand Journal of Stevenson, 1985, Capital market myopia, Economics 21, 600–613. Journal of Business Venturing 1, 7–30. Porter,Michael E., 1990, Competitive Advantage Schumpeter, Joseph A., 1976, Capitalism of Nations (Free Press, New York). Socialism, and Democracy (Harper & Row, ——, 1992a, Capital choices: Changing the way New York). America invests in industry, Journal of Shleifer, Andrei, and Larry Summers, 1988, Applied Corporate Finance 5, 4–16. Breach of trust in hostile takeovers, in ——, 1992b. Capital choices: Changing the way A. Auerbach, ed.: Corporate Takeovers: America invests in industry, A Research Causes and Consequences (University of Report presented to The Council on Chicago Press, Chicago), pp. 33–56. Competitiveness and co-sponsored by Shleifer, Andrei, and Robert W. Vishny, 1989, Harvard Business School. Management entrenchment: The case of ——, 1992c, Capital disadvantage: America’s manager-specific investments, Journal of failing capital investment system, Harvard Financial Economics 25, 123–139. Business Review 70(5), 65–82. Smith, Abbie J., 1990, Corporate ownership Pound, John, 1988, Proxy contests and the structure and performance: The case of efficiency of shareholder oversight, Journal of management buyouts, Journal of Financial Financial Economics 20, 237–265. Economics 27, 143–164. ——, 1991, Proxy voting and the SEC: Investor Steiner, I. D., 1972, Group Process and protection versus market efficiency, Journal Productivity (Academic Press, New York). of Financial Economics 29, 241–285. Stern Stewart Performance 1000, 1992, ——, 1992, Beyond takeovers: Politics comes to Corporate Finance, October/November,33–45. corporate control, Harvard Business Review Stewart, G. Bennett, III, 1990, Remaking the 70(2), 83–93. public corporation from within, Harvard ——, 1993, The rise of the political model of Business Review 68(4), 126–137. corporate governance and corporate control, Stulz, Rene M., 1990, Managerial discretion Unpublished manuscript, Harvard University, and optimal financing policies, Journal of Kennedy School of Government. Financial Economics 26, 3–00. THE MODERN INDUSTRIAL REVOLUTION 51 Swartz, L. Mick, 1992, The impact of third Watson, Bruce, 1993, For a while the Luddites generation antitakeover laws on security had a smashing success, Smithsonian, April, value, Unpublished manuscript, University of pp. 140–154. Manitoba. Weisbach, Michael S., 1988, Outside directors Tedlow, Richard, 1991, Hitting the skids: Tires and CEO turnovers, Journal of Financial and time horizons, Unpublished manuscript, Economics 20, 431–460. Harvard Business School. Wruck, Karen H. 1990, Financial distress, Tiemann, Jonathan, 1990,The economics of exit reorganization, and organizational efficiency, and restructuring: The Pabst Brewing Journal of Financial Economics 27, Company, Unpublished manuscript, Harvard 420–444. Business School. ——, 1991, What really went wrong at Revco?, United Nations, 1993, Monthly Bulletin of Journal of Applied Corporate Finance 4, Statistics, February (United Nations, New 79–92. York). ——, 1992, Changing the rules of the game: United Nations, 1992, 1988/89 Statistical Sealed Air Corporation’s leveraged recapital- yearbook, thirty-seventh issue (United ization, Working paper, Harvard Business Nations, New York). School. U.S. Bureau of the Census, 1987, Census of —— and Michael C. Jensen, 1993, Science, Manufacturing (U.S. Government Printing specific knowledge, and total quality manage- Office, Washington, D.C.) ment, Unpublished manuscript, Harvard U.S. Bureau of the Census, Housing and Business School. Household Economic Statistics Division, Wruck, Karen H., and A. Scott Keating, 1992a, 1991 (U.S. Government Printing Office, Sterling Chemicals, Inc., quality and produc- Washington, D.C.). tivity improvement program, Harvard U.S. Department of Labor, Bureau Business School Case No. 9–493–026. of Labor Statistics, Office of Productivity ——, 1992b, Sterling Chemicals, Inc., Harvard and Technology, 1990, Multifactor Business School Case No. 9–493–025. Productivity Measures, Report No. USDL Wruck, Karen H., and Krishna Palepu, 1992, 91–412. Consequences of leveraged shareholder payouts: ——, 1991, International Comparisons of Defensive versus voluntary recapitalizations, Manufacturing Productivity and Unit Labor Working paper, Harvard Business School. Cost Trends, Report No. USDL 92–752. Wruck, Karen H. and Steve-Anna Stephens, van Delden, Hendrik H., 1993, Czech and 1992a, Leveraged buyouts and restructuring: Slovak Republics: Lands of opportunity, ITS The case of Safeway, Inc., Harvard Business Textile Leader, Spring, 43–48. School Case No. 192–095. Vietor, Richard, 1994, Contrived Competition ——, 1992b, Leveraged buyouts and restructur- (Harvard University Press, Cambridge, ing: The case of Safeway, Inc.: Media Mass.) Forthcoming. response, Harvard Business School Case No. Warner, Jerold B., Ross L. Watts, and Karen H. 192–094. Wruck, 1989, Stock prices and top manage- Zellner, Wendy, 1992, Not everybody loves ment changes, Journal of Financial Wal-Mart’s low prices, Wall Street Journal, Economics 20, 461–492. October 12. Chapter 2

Andrei Shleifer and Robert W.Vishny* A SURVEY OF CORPORATE GOVERNANCE

Source: Journal of Finance, 52(2) (1997): 737–784.

ABSTRACT This article surveys research on corporate governance, with special attention to the importance of legal protection of investors and of ownership concentration in corporate governance systems around the world.

CORPORATE GOVERNANCE DEALS is a great deal of disagreement on how WITH the ways in which suppliers of good or bad the existing governance mech- finance to corporations assure themselves anisms are. For example, Easterbrook and of getting a return on their investment. Fischel (1991) and Romano (1993a) How do the suppliers of finance get make a very optimistic assessment of managers to return some of the profits to the United States corporate governance them? How do they make sure that managers system, whereas Jensen (1989a, 1993) do not steal the capital they supply or believes that it is deeply flawed and that a invest it in bad projects? How do suppliers major move from the current corporate of finance control managers? form to much more highly leveraged organ- At first glance, it is not entirely obvious izations, similar to LBOs, is in order.There why the suppliers of capital get anything is also constant talk of replacing the Anglo- back. After all, they part with their money, Saxon corporate governance systems with and have little to contribute to the enter- those patterned after Germany and Japan prise afterward.The professional managers (see, for example, Roe (1993) and or entrepreneurs who run the firms might Charkham (1994)). But the United States, as well abscond with the money. Although Germany, Japan, and the United Kingdom they sometimes do, usually they do not. have some of the best corporate governance Most advanced market economies have systems in the world, and the differences solved the problem of corporate gover- between them are probably small relative nance at least reasonably well, in that they to their differences from other countries. have assured the flows of enormous amounts According to Barca (1995) and Pagano, of capital to firms, and actual repatriation Panetta, and Zingales (1995), Italian cor- of profits to the providers of finance. But porate governance mechanisms are so this does not imply that they have solved the undeveloped as to substantially retard the corporate governance problem perfectly, or flow of external capital to firms. In less that the corporate governance mechanisms developed countries, including some of the cannot be improved. transition economies, corporate governance In fact, the subject of corporate gover- mechanisms are practically nonexistent. In nance is of enormous practical importance. Russia the weakness of corporate gover- Even in advanced market economies, there nance mechanisms leads to substantial A SURVEY OF CORPORATE GOVERNANCE 53 diversion of assets by managers of many sometimes referred to as separation of privatized firms, and the virtual nonexis- ownership and control. We want to know tence of external capital supply to firms how investors get the managers to give (Boycko, Shleifer, and Vishny (1995)). them back their money.To begin, Section I Understanding corporate governance not outlines the nature of the agency problem, only enlightens the discussion of perhaps and discusses some standard models of marginal improvements in rich economies, agency. It also focuses on incentive con- but can also stimulate major institutional tracts as a possible solution to the agency changes in places where they need to problem. Finally, Section I summarizes be made. some evidence pointing to the large magni- Corporate governance mechanisms are tude of this problem even in advanced economic and legal institutions that can be market economies. altered through the political process— Sections II through IV outline, in broad sometimes for the better. One could take a terms, the various ways in which firms can view that we should not worry about attract capital despite the agency problem. governance reform, since, in the long run, Section II briefly examines how firms can product market competition would force raise money without giving suppliers of firms to minimize costs, and as part of this capital any real power. Specifically, we cost minimization to adopt rules, includ- consider reputation-building in the capital ing corporate governance mechanisms, market and excessive investor optimism, enabling them to raise external capital at and conclude that these are unlikely to be the lowest cost. On this evolutionary theory the only reasons why investors entrust of economic change (Alchian (1950), capital to firms. Stigler (1958)), competition would take Sections III and IV then turn to the two care of corporate governance. most common approaches to corporate While we agree that product market governance, both of which rely on giving competition is probably the most powerful investors some power.The first approach is force toward economic efficiency in the to give investors power through legal world, we are skeptical that it alone can protection from expropriation by managers. solve the problem of corporate governance. Protection of minority rights and legal One could imagine a scenario in which prohibitions against managerial self-dealing entrepreneurs rent labor and capital on the are examples of such mechanisms. The spot market every minute at a competitive second major approach is ownership by price, and hence have no resources left over large investors (concentrated ownership): to divert to their own use. But in actual matching significant control rights with practice, production capital is highly spe- significant cash flow rights. Most corporate cific and sunk, and entrepreneurs cannot governance mechanisms used in the rent it every minute. As a result, the people world—including large share holdings, who sink the capital need to be assured relationship banking, and even takeovers— that they get back the return on this capi- can be viewed as examples of large tal. The corporate governance mechanisms investors exercising their power.We discuss provide this assurance. Product market how large investors reduce agency costs. competition may reduce the returns on While large investors still rely on the legal capital and hence cut the amount that system, they do not need as many rights as managers can possibly expropriate, but it the small investors do to protect their does not prevent the managers from expro- interests. For this reason, corporate gover- priating the competitive return after the nance is typically exercised by large capital is sunk. Solving that problem investors. requires something more than competition, Despite its common use, concentrated as we show in this survey. ownership has its costs as well, which can Our perspective on corporate governance be best described as potential expropria- is a straightforward agency perspective, tion by large investors of other investors 54 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS and stakeholders in the firm. In Section V, to be some of the major unresolved we focus on these potential costs of owner- puzzles in the analysis of corporate ship by large investors. governance. In Section VI, we turn to several specific Before proceeding, we should mention examples of widely used corporate several important topics closely related to governance mechanisms, which illustrate corporate governance that our article does the roles of legal protection and concen- not deal with, as well as some of the refer- trated ownership in corporate governance. ences on these topics. Our article does not We begin by discussing debt governance deal with foundations of contract theory; and equity governance as alternative for that, see Hart and Holmstrom (1987), approaches to addressing the agency Hart (1995, part I), and Tirole (1994). problem. We then turn to a brief discussion Second, we do not deal with some of the of a hybrid form—the leveraged buy out— basic elements of the theory of the firm, which reveals both the benefits and the such as the make or buy decision (vertical costs of concentrated ownership. Finally, integration). On this topic, see Williamson we look at state enterprises as a manifes- (1985), Holmstrom and Tirole (1989), tation of a radical failure of corporate and Hart (1995, part I). Third, while we governance. pay some attention to cooperatives, we do In Section VII, we bring sections III not focus on a broad variety of noncapital- through VI together by asking: which ist ownership patterns, such as worker system is the best? We argue that a good ownership or nonprofit organizations. A corporate governance system should major new treatise on this subject is combine some type of large investors with Hansmann (1996). Finally, although we legal protection of both their rights and talk about the role of financial intermedi- those of small investors. Indeed, corpora- aries in governance, we ignore their tions in successful market economies, such function as collectors of savings from the as the United States, Germany, and Japan, public. For recent overviews of intermedia- are governed through somewhat different tion, see Allen and Gale (1994), combinations of legal protection and Dewatripont and Tirole (1995) and concentrated ownership. Because all these Hellwig (1994). In sum, this survey deals economies have the essential elements of with the separation of financing and man- a good governance system, the available agement of firms, and tries to discuss how evidence does not tell us which one of this separation is dealt with in theory and their governance systems is the best. In in practice. contrast, corporate governance systems in The last preliminary point is on the selec- most other countries, ranging from poor tion of countries we talk about. Most of the developing countries, to transition available empirical evidence in the English economies, to some rich European coun- language comes from the United States, tries such as Italy, lack some essential which therefore receives the most attention elements of a good system. In most cases, in this article. More recently, there has been in fact, they lack mechanisms for legal a great surge of work on Japan, and to a protection of investors. Our analysis lesser extent on Germany, Italy, and suggests that the principal practical ques- Sweden. In addition, we frequently refer to tion in designing a corporate governance the recent experience of privatized firms in system is not whether to emulate the Russia, with which we are familiar from our United States, Germany, or Japan, but advisory work, even though there is little rather how to introduce, significant legal systematic research on Russia’s corporate protection of at least some investors so governance. Unfortunately, except for the that mechanisms of extensive outside countries just mentioned, there has been financing can develop. extremely little research done on corporate Finally, in Section VIII, we summarize governance around the world, and this our argument and present what we take dearth of research is reflected in our survey. A SURVEY OF CORPORATE GOVERNANCE 55 I. THE AGENCY PROBLEM how these residual control rights are allocated efficiently. A. Contracts In principle, one could imagine a contract in which the financiers give funds The agency problem is an essential element to the manager on the condition that they of the so-called contractual view of the retain all the residual control rights. Any firm, developed by Coase (1937), Jensen time something unexpected happens, they and Meckling (1976), and Fama and get to decide what to do. But this does not Jensen (1983a,b). The essence of the quite work, for the simple reason that the agency problem is the separation of man- financiers are not qualified or informed agement and finance, or—in more standard enough to decide what to do—the very terminology—of ownership and control. An reason they hired the manager in the first entrepreneur, or a manager, raises funds place. As a consequence, the manager ends up from investors either to put them to pro- with substantial residual control rights and ductive use or to cash out his holdings in therefore discretion to allocate funds as he the firm.The financiers need the manager’s chooses.There may be limits on this discre- specialized human capital to generate tion specified in the contract—and much of returns on their funds. The manager needs corporate governance deals with these the financiers’ funds, since he either does limits, but the fact is that managers do not have enough capital of his own to invest have most of the residual control rights. or else wants to cash out his holdings. But In practice, the situation is more how can financiers be sure that, once they complicated. First, the contracts that the sink their funds, they get anything but a managers and investors sign cannot require worthless piece of paper back from the too much interpretation if they are to manager? The agency problem in this context be enforced by outside courts. In the United refers to the difficulties financiers have in States, the role of courts is more extensive assuring that their funds are not expropriated than anywhere else in the world, but even or wasted on unattractive projects. there the so-called business judgment rule In most general terms, the financiers keeps the courts out of the affairs of com- and the manager sign a contract that panies. In much of the rest of the world, specifies what the manager does with the courts only get involved in massive viola- funds, and how the returns are divided tions by managers of investors’ rights between him and the financiers. Ideally, (e.g., erasing shareholders’ names from the they would sign a complete contract, that register). Second, in the cases where specifies exactly what the manager does in financing requires collection of funds from all states of the world, and how the profits many investors, these investors themselves are allocated. The trouble is, most future are often small and too poorly informed to contingencies are hard to describe and exercise even the control rights that they foresee, and as a result, complete contracts actually have.The free rider problem faced are technologically infeasible.This problem by individual investors makes it uninterest- would not be avoided even if the manager ing for them to learn about the firms they is motivated to raise as much funds as he have financed, or even to participate in the can, and so tries hard to accommodate the governance, just as it may not pay citizens financiers by developing a complete to get informed about political candidates contract. Because of these problems in and vote (Downs (1957)). As a result, designing their contract, the manager and the effective control rights of the the financier have to allocate residual managers—and hence the room they have control rights—i.e., the rights to make for discretionary allocation of funds—end decisions in circumstances not fully fore- up being much more extensive than they seen by the contract (Grossman and Hart would have been if courts or providers of (1986), Hart and Moore (1990)).The the- finance became actively involved in detailed ory of ownership addresses the question of contract enforcement. 56 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS B. Management discretion compensation or transfer pricing that have a bad smell. For example, Victor Posner, a The upshot of this is that managers end up Miami financier, received in 1985 over $8 with significant control rights (discretion) million in salary from DWG; a public over how to allocate investors’ funds. To company he controlled, at the time the begin, they can expropriate them. In many company was losing money (New York pyramid schemes, for example, the organizers Times, June 23, 1986). Because such end up absconding with the money. expropriation of investors by managers is Managerial expropriation of funds can also generally kept down by the courts in the take more elaborate forms than just taking United States, more typically managers use the cash out, such as transfer pricing. For their discretion to allocate investors’ funds example, managers can set up independent for less direct personal benefits. The least companies that they own personally, and costly of this is probably consumption of sell the output of the main company they perquisites, such as plush carpets and run to the independent firms at below company airplanes (Burrough and Helyar market prices. In the Russian oil industry, 1990). Greater costs are incurred when such sales of oil to manager-owned trading managers have an interest in expanding the companies (which often do not even pay for firm beyond what is rational, reinvesting the oil) are evidently common. An even the free cash, pursuing pet projects, and so more dramatic alternative is to sell the on. A vast managerialist literature explains assets, and not just the output, of the company how managers use their effective control to other manager-owned businesses at rights to pursue projects that benefit them below market prices. For example, the rather than investors (Baumol (1959), Economist (June 1995) reports that Marris (1964), Williamson (1964), Jensen Korean chaebol sometimes sell their (1986), etc.). Grossman and Hart (1988) subsidiaries to the relatives of the chaebol aptly describe these benefits as the private founder at low prices. Zingales (1994) benefits of control. describes an episode in which one state- Finally, and perhaps most important, controlled Italian firm sold some assets to managers can expropriate shareholders by another at an excessively high price. The entrenching themselves and staying on the buying firm, unlike the selling firm, had a job even if they are no longer competent or large number of minority shareholders, and qualified to run the firm (Shleifer and these shareholders got significantly diluted Vishny (1989)). As argued in Jensen and by the transaction. In short, straight-out Ruback (1983), poor managers who resist expropriation is a frequent manifestation being replaced might be the costliest of the agency problem that financiers need manifestation of the agency problem. to address. Finally,before the reader dismisses Managerial opportunism, whether in the the importance of such expropriation, we form of expropriation of investors or of point out that much of the corporate law misallocation of company funds, reduces development in the 18th and 19th centuries the amount of resources that investors are in Britain, Continental Europe, and Russia willing to put up ex ante to finance the firm focused precisely on addressing the problem (Williamson (1985), Grossman and Hart of managerial theft rather than that of (1986)). Much of the subject of corporate shirking or even empire-building (Hunt governance deals with constraints that (1936), Owen (1991)). managers put on themselves, or that In many countries today, the law investors put on managers, to reduce the protects investors better than it does in ex post misallocation and thus to induce Russia, Korea, or Italy. In the United investors to provide more funds ex ante. States, for example, courts try to control Even with these constraints, the outcome is managerial diversion of company assets to in general less efficient than would occur if themselves, although even in the United the manager financed the firm with his States there are cases of executive own funds. A SURVEY OF CORPORATE GOVERNANCE 57 An equally interesting problem concerns if it prevents efficient ex post bargaining the efficiency of the ex post resource between managers and shareholders. The allocation, after investors have put up their reason for introducing the duty of loyalty is funds. Suppose that the manager of a firm probably to avoid the situation in which cannot expropriate resources outright, but managers constantly threaten shareholders, has some freedom not to return the money in circumstances that have not been to investors. The manager contemplates specified in the contract, to take ever less going ahead with an investment project efficient actions unless they are bribed not that will give him $10 of personal benefits, to. It is better for shareholders to avoid but will cost his investors $20 in foregone bargaining altogether than to expose them- wealth. Suppose for simplicity that the selves to constant threats.This argument is manager owns no equity in the firm.Then, as similar to that of why corruption in general argued by Jensen and Meckling (1976), the is not legal, even if ex post it improves the manager will undertake the project, resulting resource allocation: the public does not in an ex post inefficiency (and of course an want to give the bureaucrats incentives to ex ante inefficiency as investors cut down come up with ever increasing obstacles to finance to such a firm). private activity solely to create corruption The Jensen-Meckling scenario raises the opportunities (Shleifer and Vishny obvious point: why don’t investors try to (1993)). But the consequence is that, with bribe the manager with cash, say $11, not limited corruption, not all the efficient to undertake the inefficient project? This bargains are actually realized ex post. would be what the Coase (1960) Theorem Similarly, if the duty of loyalty to share- predicts should happen, and what holders prevents the managers from being Grossman and Hart (1986) presume actu- paid off for not taking self-interested ally happens ex post. In some cases, such as actions, then such actions will be taken golden parachutes that convince managers even when they benefit managers less than to accept hostile takeover bids, we actually they cost shareholders. observe these bribes (Walkling and Long (1984), Lambert and Larcker (1985)). C. Incentive contracts More commonly, investors do not pay managers for individual actions and In the previous section, we discussed the therefore do not seem to arrive at efficient agency problem when complete, contingent outcomes ex post. The Jensen-Meckling contracts are infeasible. When contracts view is empirically accurate and the Coase are incomplete and managers possess more Theorem does not seem to apply. Moreover, expertise than shareholders, managers typ- the traditional reason for the failure of the ically end up with the residual rights of Coase Theorem, namely that numerous control, giving them enormous latitude for investors need to agree in order to bribe the self-interested behavior. In some cases, this manager, does not seem relevant, since the results in managers taking highly ineffi- manager needs only to agree on his bribe cient actions, which cost investors far more with a small board of directors. than the personal benefits to the managers. The reason we do not observe managers Moreover, the managers’ fiduciary duty to threatening shareholders and being bribed shareholders makes it difficult to contract not to take inefficient actions is that such around this inefficiency ex post. threats would violate the managers’ legal A better solution is to grant a manager a “duty of loyalty” to shareholders. While it highly contingent, long term incentive con- is difficult to describe exactly what this tract ex ante to align his interests with duty obligates the managers to do (Clark those of investors. While in some future (1985)), threats to take value-reducing contingencies the marginal value of the actions unless one is paid off would surely personal benefits of control may exceed violate this duty. But this only raises the the marginal value of the manager’s con- question of why this legal duty exists at all tingent compensation, such instances will 58 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS be relatively rare if the incentive component Similarly to Berle and Means, Jensen and of pay is substantial. In this way, incentive Murphy interpret their findings as evidence contracts can induce the manager to act in of inefficient compensation arrangements, investors’ interest without encouraging although in their view these arrangements blackmail, although such contracts may be are driven by politically motivated restric- expensive if the personal benefits of control tions on extremely high levels of pay. are high and there is a lower bound on the Kaplan (1994a,b) shows that the manager’s compensation in the bad states sensitivity of pay (and dismissal) to per- of the world. Typically, to make such formance is similar in the United States, contracts feasible, some measure of per- Germany, and Japan, although average formance that is highly correlated with the levels of pay are the highest in the United quality of the manager’s decision must be States. The question is whether there is a verifiable in court. In some cases, the cred- similar failure to pay for performance in all ibility of an implicit threat or promise from countries, or,alternatively, the results found the investors to take action based on an by Jensen and Murphy are not so counter- observable, but not verifiable, signal may intuitive. In particular, even the sensitivity also suffice. Incentive contracts can take a of pay to performance that Jensen and variety of forms, including share ownership, Murphy find would generate enormous stock options, or a threat of dismissal if swings in executive wealth, which require income is low (Jensen and Meckling considerable risk tolerance. More sensitivity (1976), Fama (1980)). The optimal may not be efficient for risk-averse execu- incentive contract is determined by the tives (Haubrich (1994)). manager’s risk aversion, the importance of The more serious problem with high pow- his decisions, and his ability to pay for the ered incentive contracts is that they create cash flow ownership up front (Ross enormous opportunities for self-dealing for (1973), Stiglitz (1975), Mirrlees (1976), the managers, especially if these contracts Holmstrom (1979, 1982)). are negotiated with poorly motivated Incentive contracts are indeed common boards of directors rather than with large in practice. A vast empirical literature on investors. Managers may negotiate for incentive contracts in general and manage- themselves such contracts when they know ment ownership in particular dates back at that earnings or stock price are likely to least to Berle and Means (1932), who rise, or even manipulate accounting num- argue that management ownership in large bers and investment policy to increase their firms is too small to make managers inter- pay. For example, Yermack (1997) finds ested in profit maximization. Some of the that managers receive stock option grants early studies take issue with Berle and shortly before good news announcements Means by documenting a positive relation- and delay such grants until after bad ship between pay and performance, and news announcements. His results suggest thus rejecting the extreme hypothesis of that options are often not so much an complete separation of ownership and incentive device as a somewhat covert control (Murphy (1985), Coughlan and mechanism of self-dealing. Schmidt (1985), Benston (1985)). More Given the self-dealing opportunities in recently, Jensen and Murphy (1990) look high powered incentive contracts, it is not at the sensitivity of pay of American execu- surprising that courts and regulators have tives to performance. In addition to looking looked at them with suspicion. After all, at salary and bonuses, Jensen and Murphy the business judgment rule that governs the also examine stock options and the effects attitude of American courts toward agency on pay of potential dismissal after poor problems keeps the courts out of corporate performance. Jensen and Murphy arrive at decisions except in the matters of executive a striking number that executive pay rises pay and self-dealing. These legal and polit- (and falls) by about $3 per every $1000 ical factors, which appear to be common in change in the wealth of a firm’s shareholders. other countries as well as in the United A SURVEY OF CORPORATE GOVERNANCE 59 States, have probably played an important more generally at announcement effects of role in keeping down the sensitivity of investment projects of oil and other firms, executive pay to performance (Shleifer and and find negative returns on such Vishny (1988), Jensen and Murphy announcements in the oil industry, although (1990)).While it is a mistake to jump from not in others. The study of investment this evidence to the conclusion that announcements is complicated by the fact managers do not care about performance that managers in general are not obligated at all, it is equally problematic to argue to make such announcements, and hence that incentive contracts completely solve those that they do make are likely to be the agency problem. better news than the average one. Still, the managers in the oil industry announce even D. Evidence on agency costs the bad news. The announcement selection problem In the last ten years, a considerable amount does not arise in the case of a particular of evidence has documented the prevalence kind of investment, namely acquisitions, of managerial behavior that does not serve since almost all acquisitions of public the interests of investors, particularly companies are publicly announced. shareholders. Most of this evidence comes Some of the clearest evidence on agency from the capital market in the form of problems therefore comes from acquisition “event” studies.The idea is that if the stock announcements. Many studies show that price falls when managers announce a bidder returns on the announcement of particular action, then this action must acquisitions are often negative (Roll serve the interests of managers rather than (1986) surveys this evidence). Lewellen, those of the shareholders. While in some Loderer, and Rosenfeld (1985) find that circumstances this inference is not justified negative returns are most common for bid- because the managerial action, while ders in which their managers hold little serving the interests of shareholders, equity, suggesting that agency problems inadvertently conveys to the market some can be ameliorated with incentives. Morck, unrelated bad news about the firm Shleifer,and Vishny (1990) find that bidder (Shleifer and Vishny (1986a)), in general returns tend to be the lowest when bidders such event study analysis is fairly com- diversify or when they buy rapidly growing pelling. It has surely become the most firms. Bhagat, Shleifer, and Vishny (1990), common empirical methodology of corpo- Lang and Stulz (1994), and Comment and rate governance and finance (see Fama, Jarrell (1995) find related evidence of Fisher,Jensen, and Roll (1969) for the first adverse effects of diversification on com- event study). pany valuation. Diversification and growth We have pointed out above that manage- are among the most commonly cited rial investment decisions may reflect their managerial, as opposed to shareholder, personal interests rather than those of the objectives. Kaplan and Weisbach (1992) investors. In his free cash flow theory, document the poor history of diversification Jensen (1986) argues that managers by the U.S. firms and the common incidence choose to reinvest the free cash rather than of subsequent divestitures. Finally, Lang, return it to investors. Jensen uses the Stulz, and Walkling (1991) find that bidder example of the oil industry, where in the returns are the lowest among firms with low mid-1980s integrated oil producers spent Tobin’s Qs and high cash flows.Their result roughly $20 per barrel to explore for new supports Jensen’s (1986) version of agency oil reserves (and thus maintain their large theory, in which the worst agency problems oil exploration activities), rather than occur in firms with poor investment oppor- return their profits to shareholders or even tunities and excess cash. In sum, quite a bit buy proven oil reserves that sold in the of evidence points to the dominance of marketplace for around $6 per barrel. managerial rather than shareholder motives McConnell and Muscarella (1986) look in firms’ acquisition decisions. 60 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Even clearer evidence of agency control diminishes after the deaths of problems is revealed by the studies that powerful managers. focus on managers directly threatened with There is also a great deal of evidence the loss of private benefits of control.These that control is valued, which would not be are the studies of management resistance the case if controlling managers (or share- to takeovers, which are now too numerous holders) received the same benefits as the to survey completely. Walkling and Long other investors. Barclay and Holderness (1984) find that managerial resistance to (1989, 1992) find that, in the United value-increasing takeovers is less likely States, large blocks of equity trade at a when top managers have a direct financial substantial premium to the posttrade price interest in the deal going through via share of minority shares, indicating that the ownership or golden parachutes, or when buyers of the blocks that may have a top managers are more likely to keep their controlling influence receive special jobs. Another set of studies finds that, when benefits. Several studies compare the managers take anti-takeover actions, prices of shares with identical dividend shareholders lose. For example, DeAngelo rights, but differential voting rights. Lease, and Rice (1983) and Jarrell and Poulsen McConnell, and Mikkelson (1983, 1984), (1988a) find that public announcements of DeAngelo and DeAngelo (1985), and certain anti-takeover amendments to Zingales (1995) all show that, in the corporate charters, such as super-majority United States, shares with superior voting provisions requiring more than 50 percent rights trade at a premium. On average, this of the votes to change corporate boards, premium is very small, but Zingales (1995) reduce shareholder wealth. Ryngaert shows that it rises sharply in situations (1988) and Malatesta and Walkling where control over firms is contested, (1988) find that, for firms who have indicating yet again that controlling man- experienced challenges to management agement teams earn benefits that are not control, the adoption of poison pills— available to minority investors. which are devices to make takeovers Even more dramatic evidence comes extremely costly without target manage- from other countries. Levy (1982) finds the ment’s consent—also reduce shareholder average voting premium of 45.5 percent in wealth. Comment and Schwert (1995), Israel, Rydqvist (1987) reports 6.5 percent however, question the event study evidence for Sweden, Horner (1988) shows about given the higher frequency of takeovers 20 percent for Switzerland, and, most among firms with poison pills in place. recently, Zingales (1994) reports the Taken as a whole, the evidence suggests 82 percent voting premium on the Milan that managers resist takeovers to protect Stock Exchange. Zingales (1994) and their private benefits of control rather than to Barca (1995) suggest that managers in serve shareholders. Italy have significant opportunities to Some of the evidence on the importance divert profits to themselves and not share of agency costs is less direct, but perhaps them with nonvoting shareholders. as compelling. In one of the most macabre The evidence on the voting premium in event studies ever performed, Johnson, Israel and Italy suggests that agency costs Magee, Nagarajan, and Newman (1985) may be very large in some countries. But find that sudden executive deaths—in plane how large can they get? Some evidence crashes or from heart attacks—are often from Russia offers a hint. Boycko, Shleifer, accompanied by increases in share prices and Vishny (1993) calculate that, in of the companies these executives man- privatization, manufacturing firms in Russia aged.The price increases are the largest for sold for about $100 per employee, com- some major conglomerates, whose founders pared to market valuations of about built vast empires without returning much $100,000 per employee for Western firms. to investors. A plausible interpretation of The one thousandfold difference cannot this evidence is that the flow of benefits of be explained by a difference in living A SURVEY OF CORPORATE GOVERNANCE 61 standards, which in Russia are about one exchange for their funds, only the hope that tenth of those in the West. Even controlling they will make money in the future. for this difference, the Russian assets sold at Reputation-building is a very common a 99 percent discount. Very similar evidence explanation for why people deliver on their comes from the oil industry, where Russian agreements even if they cannot be forced to companies were valued at under 5 cents per (see, for example, Kreps (1990)). In the barrel of proven reserves, compared to financing context, the argument is that typical $4 to $5 per barrel valuations for managers repay investors because they Western oil firms. An important element of want to come to the capital market and this 99 percent discount is surely the reality raise funds in the future, and hence need to of government expropriation, regulation, and establish a reputation as good risks in taxation. Poor management is probably also order to convince future investors to give a part of the story. But equally important them money.This argument has been made seems to be the ability of managers of initially in the context of sovereign borrow- Russian firms to divert both profits and ing, where legal enforcement of contracts is assets to themselves. The Russian evidence virtually nonexistent (Eaton and Gersovitz suggests that an upper bound on agency (1981), Bulow and Rogoff (1989)). costs in the regime of minimal protection of However, several recent articles have investors is 99 percent of value. presented reputation-building models of private financing. Diamond (1989, 1991) shows how firms establish reputations as II. FINANCING WITHOUT good borrowers by repaying their short GOVERNANCE term loans, and Gomes (1996) shows how dividend payments create reputations that The previous section raised the main enable firms to raise equity. question of corporate governance: why do There surely is much truth to the reputation investors part with their money, and give it models, although they do have problems. As to managers, when both the theory and the pointed out by Bulow and Rogoff (1989), evidence suggests that managers have pure reputational stories run into a back- enormous discretion about what is done ward recursion problem. Suppose that at with that money, often to the point of being some point in the future (or in some future able to expropriate much of it? The question states of the world), the future benefits to is particularly intriguing in the case of the manager of being able to raise outside investors because, unlike highly trained funds are lower than the costs of paying employees and managers, the initial what he promised investors already. In this investors have no special ability to help the case, he rationally defaults on his repay- firm once they have parted with their ments. Of course, if investors expect that money. Their investment is sunk and such a time or state is reached in the nobody—especially the managers—needs future, they would not finance the firm in them. Yet despite all these problems, the first place. Under some plausible outside finance occurs in almost all market circumstances discussed by Bulow and economies, and on an enormous scale in the Rogoff, the problem unravels and there is developed ones. How does this happen? no possibility of external finance. While In this section, we begin to discuss the reputation is surely an important reason various answers to the puzzle of outside why firms are able to raise money, the finance by first focusing on two explana- available research suggests that it is prob- tions that do not rely on governance proper: ably not the whole explanation for external the idea that firms and managers have financing. For example, in Diamond’s reputations and the idea that investors are (1989) model of corporate borrowing, gullible and get taken. Both of these reputation plays a role alongside other approaches have the common element that protections of creditors that prevent managers investors do not get any control rights in from removing assets from the firm. 62 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS An alternative theory of how investors suggesting that the high yield bonds that give their money to companies without were used to finance takeovers in the receiving control rights in exchange United States in the late 1980s were appeals to excessive investor optimism. systematically overvalued by investors. Investors get excited about companies, and Evidence from both the United States and hence finance them without thinking much other countries also indicates that the about getting their money back, simply shares of companies issuing equity in initial counting on short run share appreciation. or secondary offerings are systematically An extreme version of this story is a Ponzi overvalued (Ritter (1991), Loughran, Ritter, scheme, in which promoters raise external and Rydqvist (1994), Pagano, Panetta, and funds sequentially, and use the funds raised Zingales (1995), Teoh, Welch, and Wong from later investors to pay off initial (1995)).This evidence points to concentration investors, thereby creating an illusion of of new issues during times when stock prices high returns. Even without Ponzi schemes, are high, to poor long run performance of if investors are sufficiently optimistic about initial public offerings, to earnings manipu- short term capital gains and are prepared lation prior to the issue, and to deterioration to part with their money without regard for of profitability following the issue. In short, how the firm will ultimately pay investors excessive investor optimism as an explana- back, then external finance can be sus- tion of security issues appears to have at tained without effective governance. least some explanatory power. Delong, Shleifer, Summers, and Waldmann Still, we do not believe that investors as (1989, 1990) provide early models of a general rule are prepared to pay good external finance based on excessive money for securities that are actually investor optimism. worthless because managers can steal Pyramid schemes have been an essential everything. As the evidence on agency element of all major financial markets, theory indicates, managers can expropriate going back at least to the Louisiana and only limited wealth, and therefore the the South Sea Bubbles (Kindleberger securities that investors buy do have some (1978)). Most railroad booms in the world underlying value.To explain why these secu- were financed by investors who had virtu- rities have value, we need theories that go ally no protection, only hope. In the United beyond investor overoptimism. States, such schemes were very common as recently as the 1920s (Galbraith (1955)), and still happen occasionally today. They III. LEGAL PROTECTION also occur in many transition economies, as Russia’s famous pyramid scheme, MMM, in The principal reason that investors provide which millions of people subscribed to external financing to firms is that they shares of a company that used the proceeds receive control rights in exchange. External to advertise on television while running a financing is a contract between the firm as Ponzi scheme, vividly illustrates. Nor is it a legal entity and the financiers, which crazy to assume that enormous volumes of gives the financiers certain rights vis a vis equity financing in the rapidly growing East the assets of the firm (Hart (1995), part Asian economies are based in part on II). If firm managers violate the terms of investor optimism about near-term appre- the contract, then the financiers have the ciation, and overlook the weakness of right to appeal to the courts to enforce mechanisms that can force managers to their rights. Much of the difference in repay investors. corporate governance systems around the In recent years, more systematic statisti- world stems from the differences in the cal evidence has pointed to the importance nature of legal obligations that managers of investor optimism for financing in at have to the financiers, as well as in the least some markets. Kaplan and Stein differences in how courts interpret and (1993), for example, present evidence enforce these obligations. A SURVEY OF CORPORATE GOVERNANCE 63 The most important legal right systematic evidence is mixed. In the United shareholders have is the right to vote on States, boards, especially those dominated important corporate matters, such as by outside directors, sometimes remove top mergers and liquidations, as well as in managers after poor performance elections of boards of directors, which in (Weisbach (1988)). However, a true per- turn have certain rights vis a vis the formance disaster is required before boards management (Manne (1965), Easterbrook actually act (Warner, Watts, and Wruck and Fischel (1983)). (We discuss voting (1988)). The evidence on Japan and rights as the essential characteristic of Germany (Kaplan (1994a,b)) similarly equity in Section VI.) Voting rights, how- indicates that boards are quite passive ever, turn out to be expensive to exercise except in extreme circumstances. Mace and to enforce. In many countries, share- (1971) and Jensen (1993) argue very holders cannot vote by mail and actually strongly that, as a general rule, corporate have to show up at the shareholder meeting boards in the United States are captured by to vote—a requirement that virtually guar- the management. antees nonvoting by small investors. In In many countries, shareholder voting developed countries, courts can be relied on rights are supplemented by an affirmative to ensure that voting takes place, but even duty of loyalty of the managers to share- there managers often interfere in the voting holders. Loosely speaking, managers have a process, and try to jawbone shareholders duty to act in shareholders’ interest. into supporting them, conceal information Although the appropriateness of this duty from their opponents, and so on (Pound is often challenged by those who believe (1988), Grundfest (1990)). In countries that managers also ought to have a duty of with weaker legal systems, shareholder loyalty to employees, communities, credi- voting rights are violated more flagrantly. tors, the state, and so on (see the articles in Russian managers sometimes threaten Hopt and Teubner, Eds. (1985)), the courts employee-shareholders with layoffs unless in Organization for Economic Cooperation these employees vote with the manage- and Development (OECD) countries have ment, fail to notify shareholders about generally accepted the idea of managers’ annual meetings, try to prevent hostile duty of loyalty to shareholders. There is a shareholders from voting based on techni- good reason for this. The investments by calities, and so on. Besides, as Stalin noted, shareholders are largely sunk, and further “it is important not how people vote, but investment in the firm is generally not who counts the votes,” and managers count needed from them. This is much less the shareholders’ votes. Still, even in Russia, case with employees, community members, courts have protected a large shareholder and even creditors. The employees, for when a firm’s management erased his example, get paid almost immediately for name from the register of shareholders. In their efforts, and are generally in a much sum, both the legal extent and the court better position to hold up the firm by protection of shareholder voting rights threatening to quit than the shareholders differ greatly across countries. are. Because their investment is sunk, Even if shareholders elect the board, shareholders have fewer protections from directors need not necessarily represent expropriation than the other stakeholders their interests. The structure of corporate do. To induce them to invest in the first boards varies greatly even across developed place, they need stronger protections, such economies, ranging from two-tier supervi- as the duty of loyalty. sory and management boards in Germany, Perhaps the most commonly accepted to insider-dominated boards in Japan, to element of the duty of loyalty are the legal mixed boards in the United States restrictions on managerial self-dealing, such (Charkham (1994)).The question of board as outright theft from the firm, excessive effectiveness in any of these countries has compensation, or issues of additional securi- proved to be controversial. The available ties (such as equity) to the management and 64 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS its relatives. In some cases, the law explicitly Like shareholders, creditors have a prohibits self-dealing; in other cases, courts variety of legal protections, which also vary enforce corporate charters that prohibit it across countries. (Again, we say more (see Easterbrook and Fischel (1991)). about this in the discussion of debt and Some legal restrictions on managers bankruptcy in Section VI.) These may constrain their actions, by for example include the right to grab assets that serve demanding that managers consult the as collateral for the loans, the right to board of directors before making major liquidate the company when it does not pay decisions, or giving shareholders appraisal its debts, the right to vote in the decision to remedies to stop asset sales at low reorganize the company, and the right to prices. Other restrictions specify that remove managers in reorganization. Legal minority shareholders be treated as well as protection of creditors is often more effec- the insiders (Holderness and Sheehan tive than that of the shareholders, since (1988a)). default is a reasonably straightforward Although the duty of loyalty is accepted violation of a debt contract that a court in principle in most OECD countries, the can verify. On the other hand, when the strictness with which the courts enforce it bankruptcy procedure gives companies the varies greatly. In the United States, courts right of automatic stay of the creditors, would interfere in cases of management managers can keep creditors at bay even theft and asset diversion, and they would after having defaulted. Repossessing assets surely interfere if managers diluted existing in bankruptcy is often very hard even for shareholders through an issue of equity to the secured creditors (White (1993)).With themselves. Courts are less likely to inter- multiple, diverse creditors who have fere in cases of excessive pay, especially if conflicting interests, the difficulties of col- it takes the complex form of option con- lecting are even greater, and bankruptcy tracts, and are very unlikely to second proceedings often take years to complete guess managers’ business decisions, includ- (Baird and Jackson (1985), Gertner and ing the decisions that hurt shareholders. Scharfstein (1991), Weiss (1990)). This, Perhaps most importantly, shareholders in of course, makes debt a less attractive the United States have the right to sue the financing instrument to begin with (Bolton corporation, often using class action suits and Scharfstein (1996)). Still, while costly that get around the free rider problem, if to the creditors, bankruptcy is very tough they believe that the managers have vio- on the debtor firms as well, since their lated the duty of loyalty. managers typically get fired, assets liqui- The United States is generally viewed as dated, and debt kept largely in place (Baird relatively tough on managers in interpret- (1995)). Creditors’ legal rights are thus ing the duty of loyalty, although some, enforced in a costly and inefficient way, but including Bebchuk (1985) and Brudney they are enforced. and Chirelstein (1978), believe it is not Because bankruptcy procedures are so tough enough. For example, in France the complicated, creditors often renegotiate doctrine of corporate opportunities, which outside of formal bankruptcy proceedings prohibits managers from personally profit- both in the United States (Gilson, John, and ing from business opportunities that are Lang (1990), Asquith, Gertner, and offered to the corporation, is not accepted Scharfstein (1994)) and in Europe (OECD by courts (Tunc (1991)). Outside the (1995)). The situation is worse in developing United States and Canada, class action countries, where courts are even less reliable suits are not generally permitted and con- and bankruptcy laws are even less complete. tingent fees are prohibited (Romano The inefficiency of existing bankruptcy pro- (1993a)). Outside the OECD, the duty of cedures has prompted some economists loyalty is a much weaker concept, at least (Bebchuk (1988), Aghion, Hart, and Moore in part because courts have no capability (1992)) to propose new ones, which try to or desire to interfere in business. avoid complicated negotiations by first A SURVEY OF CORPORATE GOVERNANCE 65 converting all the claims of a bankrupt concentrate share holdings. This can mean company into equity, and then allowing the that one or several investors in the firm equity holders to decide what to do with the have substantial minority ownership bankrupt firm. It is possible that in the long stakes, such as 10 or 20 percent. A sub- run, these proposals will reduce the cost of stantial minority shareholder has the enforcing creditor rights. incentive to collect information and moni- In sum, the extent of legal protection of tor the management, thereby avoiding the investors varies enormously around the traditional free rider problem. He also has world. In some countries, such as the United enough voting control to put pressure on States, Japan, and Germany, the law pro- the management in some cases, or perhaps tects the rights of at least some investors even to oust the management through a and the courts are relatively willing to proxy fight or a takeover (Shleifer and enforce these laws. But even in these coun- Vishny (1986b)). In the more extreme tries, the legal system leaves managers with cases, large shareholders have outright considerable discretion. In most of the rest control of the firms and their management of the world, the laws are less protective of with 51 or more percent ownership. Large investors and courts function less well and shareholders thus address the agency stop only the clearest violations of investor problem in that they both have a general rights. As a result, legal protection alone interest in profit maximization, and enough becomes insufficient to ensure that control over the assets of the firm to have investors get their money back. their interests respected. In the United States, large share holdings, and especially majority ownership, are IV. LARGE INVESTORS relatively uncommon—probably because of legal restrictions on high ownership and If legal protection does not give enough exercise of control by banks, mutual funds, control rights to small investors to induce insurance companies, and other institutions them to part with their money, then (Roe (1994)). Even in the United States, perhaps investors can get more effective however, ownership is not completely control rights by being large. When control dispersed, and concentrated holdings by rights are concentrated in the hands of a families and wealthy investors are more small number of investors with a collec- common than is often believed (Eisenberg tively large cash flow stake, concerted (1976), Demsetz (1983), Shleifer and action by investors is much easier than Vishny (1986b)). Holderness and Sheehan when control rights, such as votes, are split (1988a,b) in fact found several hundred among many of them. In particular, this cases of over 51 percent shareholders in concerted action is possible with only public firms in the United States. One other minimal help from the courts. In effect, country where the rule is broadly dispersed concentration of ownership leverages up ownership by diversified shareholders is the legal protection. There are several distinct United Kingdom (Black and Coffee forms that concentration can take, includ- (1994)). ing large shareholders, takeovers, and large In the rest of the world, large share hold- creditors. In this section, we discuss these ings in some form are the norm. In forms of concentrating ownership, and how Germany, large commercial banks through they address the agency problem. In the proxy voting arrangements often control following section, we discuss some costs of over a quarter of the votes in major having large investors. companies, and also have smaller but significant cash flow stakes as direct share- A. Large shareholders holders or creditors (Franks and Mayer (1994), OECD (1995)). In addition, one The most direct way to align cash flow and study estimates that about 80 percent of control rights of outside investors is to the large German companies have an over 66 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 25 percent nonbank large shareholder defeated, management turnover is higher in (Gorton and Schmid (1996)). In smaller poorly performing firms that have block German companies, the norm is family holders. All these findings support the view control through majority ownership or that large shareholders play an active role pyramids, in which the owner controls in corporate governance (Shleifer and 51 percent of a company, which in turn Vishny (1986b)). controls 51 percent of its subsidiaries and Because large shareholders govern by so on (Franks and Mayer (1994)). exercising their voting rights, their power Pyramids enable the ultimate owners to depends on the degree of legal protection control the assets with the least amount of of their votes. Majority ownership only capital (Barca (1995)). In Japan, although works if the voting mechanism works, and ownership is not nearly as concentrated as the majority owner can dictate the deci- in Germany, large cross-holdings as well as sions of the company. This may require share holdings by major banks are the fairly little enforcement by courts, since norm (Prowse (1992), Berglof and Perotti 51 percent ownership is relatively easy to (1994), OECD (1995)). In France, cross- prove, and a vote count is not required once ownership and so-called core investors are the majority shareholder expresses his common (OECD (1995)). In most of the preferences.With large minority sharehold- rest of the world, including most of Europe ers, matters are more complicated, since (e.g., Italy, Finland, and Sweden), as well they need to make alliances with other as Latin America, East Asia, and Africa, investors to exercise control. The power of corporations typically have controlling the managers to interfere in these alliances owners, who are often founders or their off- is greatly enhanced, and the burden on spring. In short, heavily concentrated share courts to protect large shareholder rights is holdings and a predominance of controlling much greater. For this reason, large minor- ownership seems to be the rule around the ity share holdings may be effective only in world. countries with relatively sophisticated legal The evidence on the role of large share- systems, whereas countries where courts holders in exercising corporate governance are really weak are more likely to have is beginning to accumulate. For Germany, outright majority ownership. Franks and Mayer (1994) find that large Again, the most vivid example comes shareholders are associated with higher from Russia. As one Russian investment turnover of directors. Gorton and Schmid banker has pointed out, a Western investor can (1996) show that bank block holders control a Russian company with 75 percent improve the performance of German com- ownership, whereas a Russian investor can panies in their 1974 sample, and that both do so with only 25 percent ownership. This bank and nonbank block holders improve comment is easy to understand once it is performance in a 1985 sample. For Japan, recognized that the management can use a Kaplan and Minton (1994) and Kang and variety of techniques against foreign Shivdasani (1995) show that firms with investors, including declaring some of their large shareholders are more likely to shares illegal, requiring super majorities to replace managers in response to poor per- bring issues on the agenda of shareholder formance than firms without them. Yafeh meetings, losing voting records, and so on. and Yosha (1996) find that large share- While managers can apply these techniques holders reduce discretionary spending, such against domestic investors as well, the as advertising, Research & Development latter have more mechanisms of their own (R&D), and entertainment expenses, by to protect their power, including better Japanese managers. For the United States, access to other shareholders, to courts, as Shivdasani (1993) shows that large out- well as in some cases to physical force.The side shareholders increase the likelihood effectiveness of large shareholders, then, is that a firm is taken over,whereas Denis and intimately tied to their ability to defend Serrano (1996) show that, if a takeover is their rights. A SURVEY OF CORPORATE GOVERNANCE 67 B. Takeovers become more valuable if the takeover succeeds. If minority rights are not fully In Britain and the United States, two of the protected, then the bidder can get a slightly countries where large shareholders are less better deal for himself than the target common, a particular mechanism for con- shareholders get, but still he may have to solidating ownership has emerged, namely surrender much of the gains resulting from the hostile takeover (Jensen and Ruback his acquisition of control. (1983), Franks and Mayer (1990)). In a Second, acquisitions can actually typical hostile takeover, a bidder makes a increase agency costs when bidding tender offer to the dispersed shareholders managements overpay for acquisitions that of the target firm, and if they accept this bring them private benefits of control offer, acquires control of the target firm (Shleifer and Vishny (1988)). A fluid and so can replace, or at least control, the takeover market might enable managers to management.Takeovers can thus be viewed expand their empires more easily, and not as rapid-fire mechanisms for ownership just stop excessive expansion of empires. concentration. Jensen (1993) shows that disciplinary A great deal of theory and evidence hostile takeovers were only a small fraction supports the idea that takeovers address of takeover activity in the 1980s in the governance problems (Manne (1965), United States. Jensen (1988), Scharfstein (1988)). The Third, takeovers require a liquid capital most important point is that takeovers typ- market, which gives bidders access to vast ically increase the combined value of the amounts of capital on short notice. In the target and acquiring firm, indicating that 1980s in the United States, the firm of profits are expected to increase afterwards Drexel, Burnham, Lambert created such a (Jensen and Ruback (1983)). Moreover, market through junk bond financing. The takeover targets are often poorly perform- collapse of this firm may have contributed ing firms (Palepu (1985), Morck, Shleifer, to the end of that takeover wave. and Vishny (1988a, 1989)), and their Last but not least, hostile takeovers are managers are removed once the takeover politically an extremely vulnerable mechanism, succeeds (Martin and McConnell (1991)). since they are opposed by the managerial Jensen (1986, 1988) argues that takeovers lobbies. In the United States, this political can solve the free cash flow problem, since pressure, which manifested itself through they usually lead to distribution of the state anti-takeover legislation, contributed firm’s profits to investors over time. to ending the 1980s takeovers (Jensen Takeovers are widely interpreted as the (1993)). In other countries, the political critical corporate governance mechanism opposition to hostile takeovers in part in the United States, without which mana- explains their general nonexistence in the gerial discretion cannot be effectively first place. The takeover solution practiced controlled (Easterbrook and Fischel in the United States and the United (1991), Jensen (1993)). Kingdom, then, is a very imperfect and There remain some questions about the politically vulnerable method of concen- effectiveness of takeovers as a corporate trating ownership. governance mechanism. First, takeovers are sufficiently expensive that only major performance failures are likely to be C. Large creditors addressed. It is not just the cost of mounting a takeover that makes them expensive. Significant creditors, such as banks, are As Grossman and Hart (1980) point out, also large and potentially active investors. the bidder in takeovers may have to pay the Like the large shareholders, they have expected increase in profits under his man- large investments in the firm, and want to agement to target firm’s shareholders, for see the returns on their investments mate- otherwise they will not tender and simply rialize. Their power comes in part because hold on to their shares, which automatically of a variety of control rights they receive 68 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS when firms default or violate debt are not well established, bank governance is covenants (Smith and Warner (1979)) and likely to be less effective (see Barca (1995) in part because they typically lend short on Italy). term, so borrowers have to come back at The need for at least some legal protec- regular,short intervals for more funds. As a tion is shared by all large investors. Large result of having a whole range of controls, shareholders need courts to enforce their large creditors combine substantial cash voting rights, takeover artists need court- flow rights with the ability to interfere in protected mechanisms for buying shares and the major decisions of the firm. Moreover, changing boards of directors, and creditors in many countries, banks end up holding need courts to enable them to repossess equity as well as debt of the firms they collateral. The principal advantage of large invest in, or alternatively vote the equity of investors (except in takeovers) is that they other investors (OECD (1995)). As a rely on relatively simple legal interventions, result, banks and other large creditors are which are suitable for even poorly informed in many ways similar to the large share- and motivated courts. Large investors put a holders. Diamond (1984) presents one of lighter burden on the legal system than the the first models of monitoring by the large small investors might if they tried to creditors. enforce their rights. For this reason, Although there has been a great deal of perhaps, large investors are so prevalent in theoretical discussion of governance by most countries in the world, where courts large creditors, the empirical evidence of are less equipped to meddle in corporate their role remains scarce. For Japan, affairs than they are in the United States. Kaplan and Minton (1994) and Kang and Shivdasani (1995) document the higher incidence of management turnover in V. THE COSTS OF LARGE INVESTORS response to poor performance in companies that have a principal banking relationship The benefits of large investors are at least relative to companies that do not. For theoretically clear: they have both the Germany, Gorton and Schmid (1996) find interest in getting their money back and the evidence of banks improving company per- power to demand it. But there may be costs formance (to the extent they hold equity) of large investors as well.The most obvious more so than other block holders do in of these costs, which is also the usual 1974, although this is not so in 1985. For argument for the benefits of dispersed the United States, DeLong (1991) points ownership, is that large investors are not to a significant governance role played diversified, and hence bear excessive risk (see, by J. P. Morgan partners in the companies e.g., Demsetz and Lehn (1985)). However, J. P. Morgan invested in in the early 20th the fact that ownership in companies is so century. More recently, U.S. banks play a concentrated almost everywhere in the major governance role in bankruptcies, world suggests that lack of diversification when they change managers and directors is not as great a private cost for large (Gilson 1990). investors to bear as relinquishing control. The effectiveness of large creditors, like A more fundamental problem is that the the effectiveness of large shareholders, large investors represent their own depends on the legal rights they have. In interests, which need not coincide with the Germany and Japan, the powers of the interests of other investors in the firm, or banks vis a vis companies are very signifi- with the interests of employees and man- cant because banks vote significant blocks agers. In the process of using his control of shares, sit on boards of directors, play a rights to maximize his own welfare, the dominant role in lending, and operate in a large investor can therefore redistribute legal environment favorable to creditors. In wealth—in both efficient and inefficient other countries, especially where proce- ways—from others. This cost of concen- dures for turning control over to the banks trated ownership becomes particularly A SURVEY OF CORPORATE GOVERNANCE 69 important when others—such as employees by Zingales (1994) suggests that the or minority investors—have their own expropriation problem is larger in Italy, firm-specific investments to make, which are consistent with a much larger voting distorted because of possible expropriation premium he finds for that country. by the large investors. Using this general Some related evidence on the benefits of framework, we discuss several potential costs control and potential expropriation of of having large investors: straightforward minority shareholders comes from the expropriation of other investors, managers, studies of ownership structure and per- and employees; inefficient expropriation formance. Although Demsetz (1983) and through pursuit of personal (nonprofit- Demsetz and Lehn (1985) argue that there maximizing) objectives; and finally the should be no relationship between ownership incentive effects of expropriation on the structure of a firm and its performance, the other stakeholders. evidence has not borne out their view. To begin, large investors might try to Morck, Shleifer, and Vishny (1988b) pres- treat themselves preferentially at the ent evidence on the relationship between expense of other investors and employees. cash flow ownership of the largest share- Their ability to do so is especially great if holders and profitability of firms, as their control rights are significantly in measured by their Tobin’s Qs. Morck et al. excess of their cash flow rights. This hap- find that profitability rises in the range of pens if they own equity with superior voting ownership between 0 and 5 percent, and rights or if they control the firm through a falls afterwards. One interpretation of this pyramid structure, i.e., if there is a sub- finding is that, consistent with the role of stantial departure from one-share-one-vote incentives in reducing agency costs, (Grossman and Hart (1988), Harris and performance improves with higher manager Raviv (1988)). In this case, large investors and large shareholder ownership at first. have not only a strong preference, but also However, as ownership gets beyond a the ability not to pay out cash flows as pro- certain point, the large owners gain nearly rata distributions to all investors, but full control and are wealthy enough to pre- rather to pay themselves only. They can do fer to use firms to generate private benefits so by paying themselves special dividends of control that are not shared by minority or by exploiting other business relation- shareholders. Thus there are costs associ- ships with the companies they control. ated with high ownership and entrenchment, Greenmail and targeted share repurchases as well as with exceptionally dispersed are examples of special deals for large ownership. Stulz (1988) presents a formal investors (Dann and DeAngelo 1983). model of the roof-shaped relationship A small number of papers focus on between ownership and performance, which measuring the degree of expropriation of has also been corroborated by subsequent minority shareholders. The very fact that empirical work (McConnell and Servaes shares with superior voting rights trade at (1990), Wruck (1989)). a large premium is evidence of significant It has also been argued that German and private benefits of control that may come Japanese banks earn rents from their at the expense of minority shareholders. control over industrial firms, and therefore Interestingly, the two countries where the effectively benefit themselves at the voting premium is the lowest—Sweden and expense of other investors. Rajan (1992) the United States—are the two countries presents a theoretical model explaining for which the studies of expropriation of how banks can extract rents from investors minorities have been made. Not surpris- by using their informational advantage. ingly, Bergström and Rydqvist (1990) for Weinstein and Yafeh (1994) find that, Sweden and Barclay and Holderness controlling for other factors, Japanese firms (1989, 1992) for the United States do not with main banks pay higher average interest find evidence of substantial expropriation. rates on their liabilities than do unaffiliated In contrast, the casual evidence provided firms. Their evidence is consistent with 70 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS rent-extraction by the main banks. Even be large (Bhagat et al. (1990), Rosett more telling is the finding of Hoshi, (1990), Pontiff et al. (1990)). Of course, Kashyap, and Scharfstein (1993) that, the significant protection of investors and when regulatory change enabled Japanese employees in the United States may give an firms to borrow in public capital markets unrepresentative picture of expropriation and not just from the banks, high net worth in other countries. firms jumped at the opportunity. This evi- Expropriation by large investors can be dence suggests that, for these firms, the detrimental to efficiency through adverse costs of bank finance exceeded its benefits. effects on the incentives of managers and Franks and Mayer (1994) present a few employees, who might reduce their firm- cases of German banks resisting takeovers specific human capital investments when of their customer companies, either they are closely monitored by financiers or because they were captured by the man- may be easily dismissed with the conse- agement or because they feared losing quent loss of rents. Schmidt (1996) and profits from the banking relationship. On Cremer (1995) make the general point of the other hand, Gorton and Schmid (1996) how a principal’s high powered incentives find no evidence of rent extraction by the can reduce an agent’s effort. In the case of German banks. large shareholders, a similar point is made The problem of expropriation by large by Burkart, Grom, and Panunzi (1997), in investors becomes potentially more signifi- the case of takeovers by Shleifer and cant when other investors are of a different Summers (1988), and in the case of banks type, i.e., have a different pattern of cash by Rajan (1992). In all these examples, the flow claims in the company. For example, if idea is that a large investor cannot commit the large investor is an equity holder, he himself not to extract rents from the man- may have an incentive to force the firm to ager ex post, and this adversely affects take on too much risk, since he shares in ex ante managerial and employee incentives. the upside while the other investors, who When the targets of expropriation by might be creditors, bear all the costs of large investors are other investors, the failure (Jensen and Meckling (1976)). adverse incentive effect of such expropria- Alternatively, if the large investor is a cred- tion is the decline of external finance. Many itor, he might cause the company to forego countries, for example, do not do much to good investment projects because he bears protect minority investor rights, yet have some of the cost, while the benefits accrue large investors in the form of families or to the shareholders (Myers (1977)). banks.While this governance structure may Finally, large investors might have a control managers, it leaves potential greater incentive to redistribute rents from minority investors unprotected and hence the employees to themselves than the man- unwilling to invest. Perhaps for this reason, agers do (Shleifer and Summers (1988)). countries in Continental Europe, such is The available evidence of redistributions Italy, Germany, and France, have relatively between different types of claim holders in small public equity markets. In this regard, the firm comes largely from corporate con- the existence of a large equity market in trol transactions. Several studies, for example, Japan despite the weak protection of ask whether shareholders expropriate minority investors is puzzling. The puzzle bondholders in leveraged buy outs or may be explained by the predominance of leveraged recapitalizations. Typically, these low powered incentives within large redistributions are relatively small Japanese institutions or in the workings of (Asquith and Wizman (1990)). Another reputations and implicit contracts in group of studies ask whether takeovers lead Japan. to large redistributions of wealth from the The Japanese example brings up a very employees in the form of wage reductions, different view of large investors, namely layoffs, and pension cutbacks. Again, these that they are too soft rather than too redistributions typically do not appear to tough. This can be so for several reasons. A SURVEY OF CORPORATE GOVERNANCE 71 First, large investors, whether shareholders organizational form that, for reasons or creditors, may be soft when they them- discussed in this article, is rarely conducive selves are corporations with their own to efficiency. agency problems. Charkham (1994) shows, for example, that German banks virtually control themselves.“At general meetings in A. The debt versus equity choice recent years, Deutsche Bank held voting Recent years saw a veritable flood of rights for 47.2 percent of its shares, research on the debt contract as a mecha- Dresdner for 59.25 percent, and Commerz- nism for solving agency problems. In this bank for 30.29 percent” (p. 36). Moreover, new work, unlike in the Modigliani-Miller banks have no incentive to discipline man- (1958) framework, where debt is associ- agers, and some incentive to cater to them ated only with a particular pattern of cash to get more business, as long as the firm is flows, the defining feature of debt is the far away from default (Harris and Raviv ability of creditors to exercise control. (1990)). Edwards and Fischer (1994) Specifically, debt is a contract in which a summarize evidence suggesting that borrower gets some funds from the lender, German banks are not nearly as active in and promises to make a prespecified corporate governance as might be expected stream of future payments to the lender. In given their lending power and control over addition, the borrower typically promises equity votes. Second, some recent articles not to violate a range of covenants (Smith show that, even if they don’t suffer from and Warner (1979)), such as maintaining their own agency problems, large investors the value of assets inside the firm. If the such as banks may be too soft because they borrower violates any covenant, and fail to terminate unprofitable projects they especially if he defaults on a payment, the have invested in when continuation is pre- lender gets certain rights, such as the ferred to liquidation (Dewatripont and ability to repossess some of the firm’s Maskin (1995), Gertner, Scharfstein, and assets (collateral) or the opportunity to Stein (1994)). Finally, a large investor may throw the firm into bankruptcy. An essential be rich enough that he prefers to maximize feature of debt, then, is that a failure by the private benefits of control rather than borrower to adhere to the contract triggers wealth. Unless he owns the entire firm, he the transfer of some control rights from will not internalize the cost of these control him to the lender. benefits to the other investors. While these The literature on debt can be usefully arguments suggest a different set of problems divided into that before Grossman-Hart with large investors, they too point to fail- (1986), and that after. Townsend (1978) ures of large investors to force managers to and Gale and Hellwig (1985) consider maximize profits and pay them out. models in which the borrower can abscond with the profits of the firm. However, if the lender is not repaid, he has the right to VI. SPECIFIC GOVERNANCE investigate the books of the firm, and grab ARRANGEMENTS its cash before the borrower can steal it. Thus failure to repay triggers the transfer In the previous sections, we discussed the of control over the assets from the bor- roles of legal protection and concentrated rower to the lender. Gale and Hellwig ownership in assuring that investors can (1985) show that the optimal contract that collect their returns from firms. We have minimizes the expected investigation costs postponed the discussion of specific is a debt contract. Grossman and Hart contractual mechanisms used to address (1982) and Jensen (1986) model the role the agency problem until this section. In of debt in committing the payout of free particular,we now focus on debt and equity cash flows to investors. In Grossman and as instruments of finance. In addition, we Hart (1982), in particular, default enables discuss state ownership—a particular creditors to deprive the manager of the 72 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS benefits of control. A final important early Several other articles model the costs article, which is not cast in the agency con- and benefits of the debt contract. The text, but contains a highly relevant idea, is benefit is usually the reduction in the Myers and Majluf (1984). They show that, agency cost, such as preventing the man- because management has superior informa- ager from investing in negative net present tion, external finance is costly. Moreover, value projects, or forcing him to sell assets they argue that this adverse selection prob- that are worth more in alternative use.The lem is minimized by the issuance of the main costs of debt are that firms may be “safest” security, i.e., the security whose prevented from undertaking good projects pricing is least sensitive to the manager’s because debt covenants keep them from private information. Thus highly rated debt raising additional funds, or else they with a fairly certain payoff stream is issued may be forced by creditors to liquidate before equity, since equity is difficult to when it is not efficient to do so. Stulz price without knowing the precise value of (1990), Diamond (1991), Harris and the firm’s assets in place and future growth Raviv (1990), and Hart and Moore (1995) opportunities. Debt is particularly easy to present some of the main models incorpo- value where there is abundant collateral, so rating these ideas, whereas Lang, Ofek, and that investors need only concern themselves Stulz (1996) present evidence indicating with the value of the collateral and not with that leverage indeed curtails investment by the valuation of the entire firm, as equity firms with poor prospects. Williamson investors would need to. (1988) and Shleifer and Vishny (1992) The next generation of papers adopts the argue that liquidations might be particu- incomplete contracts framework more larly costly when alternative use of the explicitly, and focuses on the transfer of asset is limited or when the potential control from managers to creditors. Aghion buyers of the asset cannot raise funds and Bolton (1992) use incomplete themselves. Dewatripont and Tirole (1994) contract theory to characterize debt as an derive the optimal amount of debt in a instrument whose holders take control of model where the tough negotiating stance the firm in a bad state of the world. They of debt holders after default deters mana- show that if the managerial benefits of gerial shirking ex ante. The model explains control are higher in good states of the how the cash flow structure of debt as world, then it may be efficient for managers senior claimant with little upside potential to have control of assets in good states, and makes debt holders tough on managers for creditors to have it in bad states. Their after a default. This makes it optimal to model does not incorporate the idea that combine the specific form of cash flow control reverts to the creditors in the case rights of debt with contingent control of of default as opposed to some general bad the firm in the bad state. Berglof and von state. Bolton and Scharfstein (1990) present Thadden (1994) similarly show why short a model in which upon default, creditors term debt holders—who are the tough have enough power to exclude the firm financiers in their model—should have from the capital market, and hence stop control in the bad states. Many of these future financing altogether. Hart and articles take advantage of Myers’ (1977) Moore (1989, 1994a) explicitly model insight that debt overhang might be an the idea that debt is a contract that gives effective deterrent to new financing and the creditor the right to repossess investment. collateral in case of default. Fear of such Because the rights of creditors are liquidation keeps money flowing from the clearer, and violations of those rights are debtors to the creditors. Hart and Moore’s easier to verify in courts, the existing liter- models of debt show exactly how the ature has anointed debt as providing better schedule of debt repayments depends on protection to outside investors than equity. what creditors can realize once they gain However,the focus on large investors sheds control. new light on the relative powers of debt and A SURVEY OF CORPORATE GOVERNANCE 73 equity. Specifically, debt and equity ought (1991), Bolton and Scharfstein (1996)). be compared in terms of the combination of In contrast, it may be easier to renegotiate legal protections and ease of ownership with a bank.The difficulty of renegotiation, concentration that each typically provides. and the power of dispersed creditors, might First, does debt promote concentrated explain why public debt is an extremely ownership? By far the dominant form of uncommon financing instrument, used only lending around the world is bank lending. in a few developed countries, and even Banks are usually large investors, who gain there much less than bank debt (Mayer numerous control rights in the firm at the (1990)). time of or even before default. For example, Unlike creditors, individual shareholders the main bank can often take physical are not promised any payments in return control of the firm’s bank account—which for their financial investment in the firm, resides at that very bank—if it misses a although often they receive dividends at the payment, thereby assuring fairly complete discretion of the board of directors. Unlike control of the firm by the bank without creditors, individual shareholders have no much involvement of the courts. This claim to specific assets of the firm, and control is often bolstered by direct equity have no right to pull the collateral (one ownership in the firm, as well as a large commonly studied exception is mutual degree of monopoly power over any future funds, in which individual equity holders credit extended to the firm (OECD can force a liquidation of their pro rata (1995)). In contrast, American, Canadian, share of the assets and a repayment of its and British firms make more extensive use value). Unlike creditors, shareholders do of syndicated bank lending and even of not even have a final date at which the firm is public debt, in which creditors are fairly liquidated and the proceeds are distributed. dispersed (Mayer (1990)). In principle, they may never get anything But even where debt is not very concen- back at all. trated, the effective legal protection In addition to some relatively weak legal afforded creditors is likely to be greater protections, the principal right that equity than that enjoyed by dispersed equity holders typically get is the right to vote for holders.The crucial feature of the creditors’ the board of directors. Even this right is legal rights is that concerted action by not universal, since many countries have multiple creditors is not required to take multiple classes of common stock, and action against a delinquent debtor. The hence equity holders with inferior voting legal obligation of the firm is an obligation rights get proportionately fewer votes than to each and every creditor,and any of these their financial investment in the company. creditors can typically sue the firm for pay- Because concerted action by a large group ment of what is owed or for sale of assets. of shareholders is required to take control Of course, once action is taken by one cred- via the voting mechanism, voting rights are itor, the other creditors and the courts will of limited value unless they are concen- take action to ensure that the first creditor trated. Most small shareholders do not does not grab a disproportionate share for even have an incentive to become informed himself. In fact, this ability to unilaterally on how to vote. Contacting and persuading initiate the grab for assets in a multiple a large group of small shareholders creditor situation lends the theoretical jus- through the proxy mechanism is difficult tification for bankruptcy protection. and expensive, especially when the Unlike equity, debt in a peculiar way may management stands in the way (Dodd and be tougher when it is not concentrated. If a Warner (1983)). In contrast, when votes borrower defaults on debt held by a large are concentrated—either in a large share number of creditors, renegotiating with holding block or through a takeover—they these creditors may be extremely difficult, become extremely valuable, since the party and the borrower might be forced into that controls the concentrated votes can bankruptcy (Gertner and Scharfstein make virtually all corporate decisions. 74 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Concentrated equity in this respect is more This may be especially problematic when powerful than concentrated debt.The value the firm’s value consists primarily of future of individual shares comes from the fact growth opportunities, but the bank’s debt that the votes attached to them are valu- claim and unwillingness to take equity able to those trying to control the firm, and give it little interest in the upside and a the protection of minority shareholders distorted incentive to liquidate (Diamond assures that those who have control must (1991), Hart and Moore (1995), share some of the benefits with the minor- Dewatripont and Tirole (1994)). Rather ity (Grossman and Hart (1988), Harris than give away control to the bank, such and Raviv (1988)). firms often have highly concentrated equity Because the equity holders have voting ownership by the entrepreneur and a power and legal protection of minority venture capitalist. This may pave the way shareholders, they have the ability to for some dispersed outside equity owner- extract some payments from the managers ship as long as minority rights are well in the form of dividends. Easterbrook enough protected. (1984) articulates the agency theory of In fact, we do observe equity financing dividend payments, in which dividends are primarily for young, growing firms, as well for equity what interest is for debt: pay out as for firms in rapidly growing economies, by the managers supported by the control whereas mature economies and mature rights of the financiers, except in the case firms typically use bank finance when they of equity these control rights are the voting rely on external funds at all (see Mayer rights. More recently, Fluck (1995) and (1990), Singh (1995)). In the same spirit, Myers (1995) present agency-theoretic Titman and Wessels (1988) and Rajan and models of dividends, based on the idea that Zingales (1995) show for the United shareholders can threaten to vote to fire States and several OECD economies managers or liquidate the firm, and there- respectively that debt finance is most fore managers pay dividends to hold off the common for firms with tangible assets. shareholders.These models do not explicitly This analysis of equity financing still address the free rider problem between leaves an important question open: how shareholders; namely, how do they manage can firms raise equity finance in countries to organize themselves to pose a threat to with virtually no protection of minority the management when they are small investors, even if these countries are rap- and dispersed? Concentration of equity idly growing? Singh (1995) provides some ownership, or at least the threat of such evidence on the importance of equity concentration, must be important to get financing in LDCs, although some of his companies to pay dividends. data on equity financing might include pri- One of the fundamental questions that vatizations and equity exchanges within the equity contracts raise is how—given industrial groups, both of which often take the weakness of control rights without the form of sales of large blocks and hence concentration—do firms manage to issue need not reflect any minority purchases. equity in any substantial amounts at all? One possible explanation is that, during a Equity is the most suitable financing tool period of rapid economic growth, reputa- when debt contracts are difficult to tional effects and the prospects of coming enforce, i.e., when no specific collateral can back soon to the capital market sustain be used to back credit and when near-term good behavior until the requisite institu- cash flows are insufficient to service debt tions and legal protections are put in place payments. Young firms, and firms with (Gomes (1996)). Investors can thus count intangible assets, may need to be equity on reputation in the short run, and legal financed simply because their assets have protection in the longer run when the firm’s little or no liquidation value. If they are needs for access to capital markets are financed by debt, their managers effectively smaller. Also, in some rapidly growing give full control to the bank from the start. countries, such as Korea, the rates of return A SURVEY OF CORPORATE GOVERNANCE 75 on investment may exceed the rates of there is some evidence that the way in appropriation by the insiders. However, which profits are increased has to do with another possibility is that speculative lower agency costs. Many LBOs are tar- bubbles and investor overoptimism are geted at highly diversified firms, which sell playing an important role in equity financing off many of their noncore divisions shortly in rapidly growing economies.The available after the LBO (Bhagat, Shleifer,and Vishny evidence does not satisfactorily account for (1990)). If the agency problem expresses the puzzle of external equity financing in itself in the form of excessive size and countries with only minimal legal protec- diversification, then the effect of debt over- tion of investors. hang and large shareholders is to reduce agency costs. At the same time, LBOs illustrate the B. LBOs potential costs of heavily concentrated A remarkable recent phenomenon in the ownership. Jensen (1989a) conjectures United States that illustrates both the ben- that because LBOs are so efficient, they efits and the costs of having large investors would become a predominant organizational is leveraged buy outs. In these transactions, form in the United States. Rappaport shareholders of a publicly owned company (1990) in contrast argues that the heavy are bought out by a new group of investors, oversight from investors might prevent that usually includes old managers, a spe- future investment and growth, and hence be cialized buyout firm, banks and public debt unattractive to the management. Bhagat, holders (Jensen (1989a, 1989b)). With Shleifer, and Vishny (1990) argue that the fewer constraints on compensation principal purpose of LBOs in the 1980’s arrangements than when the firm was was to serve as a temporary financing tool public, managers typically sharply increase for implementation of drastic short-run their percentage ownership of the new improvements, such as divestitures. Kaplan company, even though they take out some (1991) looks empirically at the question of of their money invested in the firm (Kaplan whether LBOs are permanent organizations, and Stein (1993)). The buyout firm typi- or whether, alternatively, they eventually cally buys enough equity to control the return to the public equity market. His firm. Most of the financing, however,comes evidence suggests that, while LBOs are not from banks and from buyers of subordi- very short lived organizations, the median nated public debt, which in the 1980s firm sells equity to the public within five to became known as junk bonds. In some six years. Although this suggests that LBOs cases, the decisions of the dispersed holders are not permanent organizations, Kaplan of junk debt were coordinated by its under- also finds that even those firms issuing writers. In short, LBOs had concentrated equity to the public retain a very heavy equity ownership by managers and LBO concentration of both debt and equity funds, as well as debt ownership by banks, ownership. Large investors remain even and, in effect, the holders of public debt. when the original financing structure is too Consistent with the idea that large tough to be permanent. investors reduce agency problems, the available evidence indicates that LBOs are C. Cooperatives and state ownership efficient organizations. First, like other takeovers, LBOs usually buy out the old We have suggested that, in some situations, shareholders at a substantial premium, concentrated ownership may not be optimal meaning at least prima facie that they were because nonshareholder constituencies going to increase profits (DeAngelo, such as managers, employees, and con- DeAngelo, and Rice (1984)). Second, there sumers are left with too few rents, and too is direct evidence from the sample of LBOs little incentive to make relationship-specific that subsequently went public that they do investments. In these situations, coopera- increase profits (Kaplan (1989)). Third, tives might be a more efficient ownership 76 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS structure (Hansmann (1988), Hart and The view of corporate governance taken Moore (1994b)). For example, private firms in this article helps explain the principal with large investors might under-provide elements of the behavior of state firms. quality or otherwise shortchange the firm’s While in theory these firms are controlled stakeholders because of their single- by the public, the de facto control rights minded focus on profits.This logic has been belong to the bureaucrats. These bureau- used to explain why health care, child care, crats can be thought of as having and even retailing are sometimes best pro- extremely concentrated control rights, but vided by cooperatives, including consumer no significant cash flow rights because the cooperatives. By voting on prices and qual- cash flow ownership of state firms is effec- ity, stakeholders achieve a better outcome tively dispersed amongst the taxpayers of than would a profit-maximizing owner. the country. Moreover, the bureaucrats A similar argument has been used to typically have goals that are very different justify state ownership of firms. Where from social welfare, and are dictated by monopoly power, externalities, or distribu- their political interests (Shapiro and tional issues raise concerns, private profit- Willig (1990), Boycko et al. (1996), maximizing firms may fail to address these Shleifer and Vishny (1994)). For example, concerns. A publicly spirited politician can they often cater to special interest groups then improve efficiency by controlling the that help them win elections, such as decisions of firms. Such social welfare public employee trade unions, which not arguments underlie the traditional case for surprisingly typically strongly support state ownership of railroads, electricity, state ownership (Lopez-de-Silanes, prisons, schools, health care, and many Shleifer, and Vishny (1997)). In sum, the other activities (Laffont and Tirole (1993), bureaucrats controlling state firms have Sappington and Stiglitz (1987)). Versions at best only an indirect concern about of this argument are used to justify state profits (because profits flow into the ownership of industrial firms as well. government budget), and have objectives With a few exceptions of activities where that are very different from the social the argument for state ownership carries interest. Nonetheless, they have virtually the day, such as police and prisons (Hart, complete power over these firms, and can Shleifer, and Vishny (1997)), the reality of direct them to pursue any political state ownership is broadly inconsistent objective. State ownership is then an with this efficiency argument. First, state example of concentrated control with no firms do not appear to serve the public cash flow rights and socially harmful interest better than private firms do. For objectives. Viewed from this perspective, example, in many countries state enterprises the inefficiency of state firms is not at all are much worse polluters than private surprising. firms. Indeed, the pollution problems are The recognition of enormous inefficiency most severe in the former communist coun- of state firms, and the pressures on public tries that were dominated by state firms budgets, have created a common response (Grossman and Krueger (1993)). Second, around the world in the last few years, contrary to the theory, state firms are typi- namely privatization. In most cases, cally extremely inefficient, and their losses privatization replaces political control with result in huge drains on their countries’ private control by outside investors. At the treasuries (Kikeri, Nellis, and Shirley same time, privatization in most countries (1992) and Boycko, Shleifer, and Vishny creates concentrated private cash flow (1995) survey the relevant evidence). In ownership to go along with control. The their frequent disregard of social objectives, result of the switch to these relatively more as well as in their extreme inefficiency, the efficient ownership structures is typically a behavior of state firms is inconsistent significant improvement in performance of with the efficiency justification for their privatized firms (Megginson et al. (1994), existence. Lopez-de-Silanes (1994)). A SURVEY OF CORPORATE GOVERNANCE 77 The cases where privatization does not VII. WHICH SYSTEM IS THE BEST? work as well as intended can also be understood from the corporate governance Corporate governance mechanisms vary a perspective. For example, when firms are great deal around the world. Firms in the privatized without the creation of large United States and the United Kingdom investors, agency costs of managerial substantially rely on legal protection of control may rise even when the costs of investors. Large investors are less preva- political control fall. In the United lent, except that ownership is concentrated Kingdom, managers of privatized firms sporadically in the takeover process. In such as water utilities receive large wage much of Continental Europe as well as in increases (Wolfram (1995)).This outcome Japan, there is less reliance on elaborate is not surprising, given that the controlling legal protections, and more reliance on outside shareholders no longer exist in large investors and banks. Finally, in the these firms, leaving managers with more rest of the world, ownership is typically discretion. At the same time, we doubt that heavily concentrated in families, with a few the problems of managerial discretion in large outside investors and banks. Legal these companies are nearly as serious as the protection of investors is considerably prior problems of political control. weaker than in Japan and Germany, let Another example of postprivatization alone in Britain and the United States.This difficulties with corporate governance is diversity of systems raises the obvious Russia (Boycko et al. (1995)). For political question: what arrangement is the best reasons, the Russian privatization has led from the viewpoint of attracting external to controlling ownership by the manage- funds to firms? In this section, we attempt ment of many companies.The management to deal with this question. has almost complete control and substantial cash flow rights, which can in principle A. Legal protection and large lead to dramatically improved incentives. investors However, there are two problems—both of which could have been predicted from the Our analysis leads us to conclude that both theory. First, the virtual absence of protec- the legal protection of investors and some tion of minority shareholders makes it form of concentrated ownership are attractive for managers to divert resources essential elements of a good corporate gov- from the firms despite their large personal ernance system. Large investors appear to cash flow stakes, since in this way they do be necessary to force managers to distrib- not need to share with outside investors at ute profits. These investors require at least all. Second, managers in many cases are some basic legal rights, such as the voting not competent to restructure the privatized rights or the power to pull collateral, to firms, yet in virtue of their control rights exercise their power over the management. remain on the job and “consume” the ben- If small investors are to be attracted to the efits of control. In fact, some of the most business of financing companies, they as successful privatizations in Russia have well require some legal protection against been the ones where outside investors have expropriation by both the managers and accumulated enough shares to either the large investors. Legal protection and replace or otherwise control the manage- large investors are complementary in an ment. Such outside investors have typically effective corporate governance system. been less capable of diverting the profits Indeed, the successful corporate gover- for themselves than the managers, as well nance systems, such as those of the United as better capable of maximizing these profits. States, Germany, and Japan, rely on some The example of the Russian privatization combination of concentrated ownership vividly illustrates both the benefits and the and legal protection of investors. In the costs of concentrated ownership without United States; both small and large share- legal protection of minority investors. holders are protected through an extensive 78 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS system of rules that protects minority went public in Italy, compared to several rights, allows for easy transfer of shares, thousand in the United States. Barca keeps elections of directors relatively (1995) suggests that bank finance is also uninhibited by managers, and gives share- difficult to obtain. Although Mayer (1990) holders extensive powers to sue directors reports a significant amount of bank for violations of fiduciary duty, including financing in Italy, most of it comes from through class-action suits. Because of state bank financing of state firms. In Italy, extensive bankruptcy protection of compa- most large firms not supported by the nies, however,creditors in the United States government are family controlled and have relatively fewer rights than do credi- internally financed. tors in Germany and Japan. These legal Although there is little systematic evi- rules support a system of active public dence available, most of the world appears participation in the stock market, concen- to be more like Italy than like the United tration of ownership through takeovers, but States, Germany, or Japan. A recent study little governance by banks. of India, for example, shows that large In Germany, creditors have stronger firms tend to be family controlled, and to rights than they do in the United States, rely almost entirely on internal financing but shareholder rights are weaker.Germany except when they get money from the then has a system of governance by both government (Khanna and Palepu (1996)). permanent large shareholders, for whom Latin American firms also face little the existing legal rules suffice to exercise external corporate governance, and financing their power, and by banks, but has virtually tends to be either internal or from no participation by small investors in the government-controlled banks.The conclusion market. Japan falls between the United we draw is simple: corporate governance States and Germany in the degree of systems of the United States, Germany, and protection of both shareholder and creditor Japan have more in common than is rights, and as a result has powerful banks typically thought, namely a combination of and powerful long term shareholders, large investors and a legal system that although neither is evidently as powerful as protects investor rights. Corporate governance they are in Germany. In addition, the systems elsewhere are less effective because Japanese governance system has succeeded they lack the necessary legal protections. in attracting small investors into the stock market. Because both Germany and Japan B. Evolution of governance systems have a system of permanent large investors, hostile takeovers are rare in both countries. The above discussion does not address the Although we compare the merits of the three question that has interested many people, systems below, it is essential to remember namely which of the developed corporate that all of them have effective legal protec- governance systems works the best? One tion of at least some types of investors. could argue that, since all these systems In much of the rest of the world, legal survived and the economies prospered, the protection of investors is less substantial, governance systems of the United States, either because laws are bad or because Japan, and Germany must be about equally courts do not enforce these laws. As a con- good. However, recent research has shown sequence, firms remain family-controlled that, historically, political pressures are as and, even in some of the richest countries, important in the evolution of corporate have difficulty raising outside funds, and governance systems as the economic ones. finance most of their investment internally In a much-discussed recent book, Roe (Mayer 1990). Pagano, Panetta, and (1994) argues that politics rather than Zingales (1995) report the extraordinary economic efficiency shaped American difficulties that firms face raising outside corporate law, at least at the Federal level. funds in Italy. Over an 11-year period Roe provides a detailed account on how the between 1982 and 1992, only 123 firms American political system systematically A SURVEY OF CORPORATE GOVERNANCE 79 discouraged large investors. Banks, insurance have focused on is: what type of large companies, mutual funds, and pension investors are best? How do U.S.-style funds were all prevented from becoming takeovers compare to more permanent influential in corporate affairs. The hostile large shareholders and creditors in West political response to the 1980s takeovers Germany and Japan? We do not believe can be viewed as a continuation of the that the available research provides a firm promanagement and antilarge-shareholder answer to this question. policies (Grundfest (1990), Jensen (1993)). Not surprisingly, the most enthusiastic Roe does not explain whether the extremely assessments of American corporate gover- fine development of the legal protection of nance system come from those who put small shareholders in the United States is greater emphasis on the role of legal pro- in part a response to the suppression of tection than on that of large investors large investors, but this conclusion is (Easterbrook and Fischel (1991), Romano actually suggested by some other work (1993a)). Romano (1993a) argues that (e.g., Douglas (1940), Coffee (1991), competition between U.S. states has Bhide (1993)). Roe’s conclusion is caused the State of Delaware, where many nonetheless that the American system is far large companies are incorporated, to adopt from efficient because of its discouragement corporate laws that effectively serve the of the large investors. interests of shareholders, and thus secure The trouble is, the argument that the effective corporate governance. Romano political process accommodates the power- (1993a) even argues that Delaware ful interests in the economy rather than adopted the most benign antitakeover maximizing social welfare applies to legislation of all the states, thereby not pre- Germany and Japan as well. Both countries cluding a future role for hostile takeovers. have shaped their systems of powerful Easterbrook and Fischel (1991) do not banks at the end of the 19th century, dur- discuss the role of large shareholders at all. ing the period of rapid economic growth, Romano (1993b) believes that the fre- and with strong support from the state quently mentioned hopes that institutional (Gerschenkron (1962)). In both countries, investors in the United States become the United States attempted to destroy the active, value-maximizing shareholders powerful financial institutions during the (e.g., Black (1990)) are exaggerated. She occupation after World War II (Adler is also skeptical about the potential gover- (1949)), and in both countries it failed. nance role of banks. In short, the bet Moreover, once German banks became suf- among these scholars is on the legal ficiently powerful, they discouraged the protection of investors. To the extent that introduction of disclosure rules, prohibitions takeovers complement this legal protection, on insider trading, and other protections of they are viewed as sufficient. minority shareholders—thus making sure In contrast, advocates of the German that these investors never became a signif- and Japanese corporate governance system icant economic or political force to protect point to the benefits of permanent long their rights. Through this political channel, term investors relative to those of the legal system has developed to accom- takeovers. Hoshi, Kashyap, and Scharfstein modate the prevailing economic power, (1990, 1991) show that firms with a main which happened to be the banks. banking relationship in Japan go through Evolutionary arguments evidently do not financial distress with less economic dis- adjudicate the question of which system is tress and better access to financing. In more efficient. addition, a large theoretical and anecdotal literature argues that the American corpo- C. What kind of large investors? rate governance system, particularly takeovers, imposes short horizons on the The question that many of the comparisons behavior of corporate managers, and hence of the United States, Japan, and Germany reduces the efficiency of investment 80 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS (Stein (1988, 1989), Shleifer and Vishny small investors from participating in financial (1990)). The theories and the arguments markets. In sum, despite a great deal of (Porter (1992)) in this area are remarkably controversy, we do not believe that either short of any empirical support (see the theory or the evidence tells us which of Poterba and Summers (1995)). Still, the the three principal corporate governance superior performance of the Japanese and systems is the best. In this regard, we are German economies, at least until the not surprised to see political and economic 1990s, has caused many to prefer their pressures for the three systems to move governance systems to the American one toward each other, as exemplified by the (see Aoki (1990), Roe (1993), and growing popularity of large shareholders in Charkham (1994)). the United States, the emergence of public We do not think that these debates have debt markets in Japan, and the increasing been conclusive. True, American takeovers bank-bashing in Germany. are a crude governance mechanism. But the At the same time, in thinking about the U.S. economy has produced mechanisms of evolution of governance in transition this kind repeatedly during the 20th economies, it is difficult to believe that century, including mergers, proxy fights, either significant legal protection of LBOs, and more recently vulture funds. investors or takeovers are likely to play a Although many of these mechanisms run key role. In all likelihood, then, unless into political trouble, new ones keep being Eastern Europe is stuck with insider invented. The end of 1980s hostile domination and no private external finance takeovers probably does not spell the end of at all (a risk in Russia), it will move toward active large investors. Moreover,partly as a governance by banks and large shareholders. result of takeovers, the American economy The early evidence from the Czech in the 1980s went through a more radical, Republic (van Wijnbergen and Mancini and possibly effective, restructuring than (1995)) and Russia (Blasi and Shleifer the economies of Japan and Western (1996)) indeed suggests that large share- Europe. Finally, because of extensive legal holders, which in the Czech Republic are protection of small investors, young often bank-controlled mutual funds, play a American firms are able to raise capital in central role in corporate governance. It the stock market better than firms else- would be extremely fortunate if transition where in the world. It is difficult to dismiss economies managed to approach the the U.S. corporate governance system in corporate governance systems of light of these basic facts. Germany and Japan, particularly in the On the other hand, permanent large dimension of the legal protection of shareholders and banks, such as those dom- investors. But this does not imply that the inating corporate governance in Japan and United States should move in the same Germany, obviously have some advantages, direction as well. such as the ability to influence corporate management by patient, informed investors. These investors may be better able to help VIII. CONCLUSION distressed firms as well. Still, there are serious questions about the effectiveness of In the course of surveying the research on these investors, largely because their corporate governance, we try to convey a toughness is in doubt. As Charkham particular structure of this field. Corporate (1994) has shown, German banks are large governance deals with the agency problem: public institutions that effectively control the separation of management and themselves. There is little evidence from finance. The fundamental question of either Japan or Germany that banks are corporate governance is how to assure very tough in corporate governance. financiers that they get a return on their Finally, at least in Germany, large-investor- financial investment. We begin this survey oriented governance system discourages by showing that the agency problem is A SURVEY OF CORPORATE GOVERNANCE 81 serious: the opportunities for managers to by optimal design of incentives, by fear of abscond with financiers’ funds, or to self-dealing, or by distributive politics? squander them on pet projects, are plenti- Second, what is the nature of legal ful and well-documented. protection of investors that underlies We then describe several broad corporate governance systems in various approaches to corporate governance. We countries? How do corporate laws differ, begin by considering the possibility of and how does enforcement of these laws financing based on reputations of managers, vary across countries? Although a lot has or on excessively optimistic expectations of been written about law and corporate investors about the likelihood of getting governance in the United States, much less their money back. We argue that such is written (in English) about the rest of the financing without governance is unlikely to world, including other wealthy economies. be the whole story. We then discuss legal Yet legal rules appear to play a key role in protection of investors and concentration corporate governance. of ownership as complementary approaches Third, are the costs and benefits of to governance. We argue that legal concentrated ownership significant? In protection of investor rights is one essen- particular, do large investors effectively tial element of corporate governance. expropriate other investors and stakeholders? Concentrated ownership—through large Are they tough enough toward managers? share holdings, takeovers, and bank Resistance to large investors has driven finance—is also a nearly universal method the evolution of corporate governance in the of control that helps investors to get their United States, yet they dominate corporate money back. Although large investors can governance in other countries. We need be very effective in solving the agency to know a great deal more about these problem, they may also inefficiently questions to objectively compare the redistribute wealth from other investors to successful corporate governance systems. themselves. Fourth, do companies in developing Successful corporate governance countries actually raise substantial equity systems, such as those of the United States, finance? Who are the buyers of this equity? Germany, and Japan, combine significant If they are dispersed shareholders, why are legal protection of at least some investors they buying the equity despite the apparent with an important role for large investors. absence of minority protections? What are This combination separates them from the real protections of shareholders in most governance systems in most other coun- countries anyway? We were surprised to tries, which provide extremely limited legal find very little information on equity protection of investors, and are stuck with finance outside the United States. family and insider-dominated firms receiving Finally, and perhaps most generally, little external financing. At the same time, what are the political dynamics of we do not believe that the available corporate governance? Do political and evidence tells us which one of the successful economic forces move corporate governance governance systems is the best. toward greater efficiency or, alternatively, In writing this survey, we face a variety of do powerful interest groups, such as the still open questions. In conclusion, we simply managers in the United States or the raise some of them. While the literature in banks in Germany, preserve inefficient some cases expresses opinions about these governance systems? How effective is the questions, we are skeptical that at the political and economic marketplace in moment persuasive answers are available. delivering efficient governance? While our First, given the large impact of executives’ survey has described some evidence in this actions on values of firms, why aren’t very area from the United States, our high powered incentive contracts used understanding of the politics of corporate more often in the United States and governance around the world remains elsewhere in the world? Is their use limited extremely limited. 82 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS NOTE Barca, Fabrizio, 1995, On corporate governance in Italy: Issues, facts, and agency, manuscript, * Shleifer is from Harvard University.Vishny is Bank of Italy, Rome. from the University of Chicago. Prepared for Barclay, Michael, and Clifford Holderness, the Nobel Symposium on Law and Finance, 1989, Private benefits from control of public Stockholm, August 1995. We are grateful to corporations, Journal of Financial Oliver D. Hart for many conversations, to Doug Economics 25, 371–395. Diamond, Frank Easterbrook, Milton Harris, Barclay, Michael, and Clifford Holderness, Martin Hellwig, James Hines,Tor Jonsson, Steve 1992, The law and large-block trades, The Kaplan, Rafael La Porta, Florencio Lopez-de- Silanes, Raghu Rajan, David Scharfstein, René Journal of Law and Economics 35, 265–294. Stulz, and Luigi Zingales for comments, and to Baumol, William, 1959, Business Behavior, the NSF for financial support. Value and Growth (Macmillan, New York). Bebchuk, Lucian, 1985, Toward undistorted choice and equal treatment in corporate REFERENCES takeovers, Harvard Law Review 98, 1693–1808. Adler, Hans, 1949, The post-war reorganization Bebchuk, Lucian, 1988, A new method for of the German banking system, Quarterly corporate reorganization, Harvard Law Journal of Economics 63, 322–341. Review 101, 775–804. Aghion, Philippe, and Patrick Bolton, 1992, An Benston, George, 1985, The self-serving man- “incomplete contracts” approach to financial agement hypothesis: Some evidence, Journal contracting, Review of Economic Studies 59, of Accounting & Economics 7, 67–83. 473–494. Berglof, Erik, and Enrico Perotti, 1994, The Aghion, Philippe, Oliver Hart, and John Moore, governance structure of the Japanese finan- 1992, The economics of bankruptcy reform, cial keiretsu, Journal of Financial Economics Journal of Law, Economics, and 36, 259–284. Organization 8, 523–546. Berglof, Erik, and Ernst-Ludwig von Thadden, Alchian, Armen, 1950, Uncertainty, evolution, 1994, Short-term versus long-term interests: and economic theory, Journal of Political Capital structure with multiple investors, Economy 58, 211–221. Quarterly Journal of Economics 109, Allen, Franklin, and Douglas Gale, 1994, 1055–1084. Financial Innovation and Risk Sharing (MIT Bergström, Clas, and Kristian Rydqvist, 1990, Press, Cambridge, Mass.). Ownership of equity in dual-class firms, Aoki, Masahiko, 1990, Towards an economic Journal of Banking and Finance 14, 255–269. model of the Japanese firm, Journal of Berle, Adolf, and Gardiner Means, 1932, The Economic Literature 28, 1–27. Modern Corporation and Private Property Asquith, Paul, Robert Gertner, and David (Macmillan, New York). Scharfstein, 1994, Anatomy of financial Bhide, Amar, 1993, The hidden costs of stock distress: An examination of junk bond issuers, market liquidity, Journal of Financial Quarterly Journal of Economics 109, Economics 34, 31–51. 625–658. Bhagat, Sanjay, Andrei Shleifer, and Robert Asquith, Paul, and Thierry Wizman, 1990, Event Vishny, 1990, Hostile takeovers in the 1980s: risk, covenants, and bondholders’ returns in The return to corporate specialization, leveraged buyouts, Journal of Financial Brookings Papers on Economic Activity: Economics 27, 195–214. Microeconomics, Special Issue, 1–72. Baird, Douglas, 1995, The hidden virtues of Black, Bernard, 1990, Shareholder passivity chapter 11: An overview of the law and reexamined, Michigan Law Review 89, economics of financially distressed firms, 520–591. manuscript, University of Chicago Law School. Black, Bernard, and John Coffee, 1994, Hail Baird, Douglas, and Thomas Jackson, 1985, Britannia?: Institutional investor behavior Cases, Problems, and Materials on Bankruptcy under limited regulation, Michigan Law (Little, Brown and Co., Boston). Review 92, 1997–2087. A SURVEY OF CORPORATE GOVERNANCE 83 Blasi, Joseph, and Andrei Shleifer, 1996, Comment, Robert, and Gregg Jarrell, 1995, Corporate governance in Russia: An initial Corporate focus and stock returns, Journal of look, in Roman Frydman, Cheryl W. Gray, and Financial Economics 37, 67–87. Andrzej Rapaczynski, Eds.: Corporate Comment, Robert, and G. William Schwert, Governance in Central Europe and Russia: 1995, Poison or placebo? Evidence on the Vol. 2 Insiders and the State (Central deterrent and wealth effects of modern anti- European University Press, Budapest). takeover measures, Journal of Financial Bolton, Patrick, and David Scharfstein, 1990, Economics 39, 3–44. A theory of predation based on agency Coughlan, Anne, and Ronald Schmidt, 1985, problems in financial contracting, American Executive compensation, management Economic Review 80, 94–106. turnover, and firm performance: An empirical Bolton, Patrick, and David Scharfstein, 1996, investigation, Journal of Accounting and Optimal debt structure and the number of cred- Economics 7, 43–66. itors, Journal of Political Economy 104, 1–25. Cremer, Jacques, 1995, Arm’s length relation- Boycko, Maxim, Andrei Shleifer, and Robert W. ships, Quarterly Journal of Economics 110, Vishny, 1993, Privatizing Russia, Brookings 275–296. Papers on Economic Activity, 139–192. Dann, Larry, and Harry DeAngelo, 1983, Boycko, Maxim, Andrei Shleifer, and Robert W. Standstill agreements, privately negotiated Vishny, 1995, Privatizing Russia (M.I.T. stock repurchases, and the market for corpo- Press, Cambridge, Mass.). rate control, Journal of Financial Economics Boycko, Maxim, Andrei Shleifer, and Robert W. 11, 275–300. Vishny, 1996, A theory of privatization, Paish DeAngelo, Harry, and Linda DeAngelo, 1985, lecture, Economic Journal 106, 309–319. Managerial ownership of voting rights, Brudney, Victor, and Marvin A. Chirelstein, Journal of Financial Economics 14, 33–69. 1978, A restatement of corporate freeze- DeAngelo, Harry, Linda DeAngelo, and Edward outs, Yale Law Journal 87, 1354–1375. Rice, 1984, Going private: Minority freeze- Bulow, Jeremy, and Kenneth Rogoff, 1989, outs and stockholder wealth, Journal of Law A constant recontracting model of sovereign and Economics 27, 367–401. debt, Journal of Political Economy 97, DeAngelo, Harry, and Edward Rice, 1983, 155–178. Antitakeover amendments and stockholder Burkart, M., Denis Gromb, and Fausto Panunzi, wealth, Journal of Financial Economics 11, 1997, Large shareholders, monitoring, and 329–360. fiduciary duty, Quarterly Journal of De Long, J. Bradford, 1991, Did Morgan’s Men Economics 112. Add Value? An Economist’s Perspective on Burrough, Bryan, and John Helyar, 1990, Financial Capitalism, in Peter Temin, Ed.: Barbarians at the Gate: The Fall of RJR Inside the Business Enterprise: Historical Nabisco, (Harper & Row, New York). Perspectives on the Use of Information Charkham, Jonathan, 1994, Keeping Good (University Press, Chicago). Company: A Study of Corporate Governance De Long, J. Bradford, Andrei Shleifer,Lawrence in Five Countries (Clarendon Press, Oxford). Summers, and Robert Waldmann, 1989, The Clark, Robert, 1985, Agency costs versus fiduci- size and incidence of the losses from noise ary duties, in John Pratt and Richard trading, Journal of Finance 44, 681–696. Zeckhauser,Eds.: Principals and Agents: The De Long, J. Bradford, Andrei Shleifer, Structure of Business (Harvard Business Lawrence Summers, and Robert Waldmann, School Press, Cambridge, Mass.). 1990, Noise trader risk in financial mar- Coase, Ronald, 1937, The nature of the firm, kets, Journal of Political Economy 98, Economica 4, 386–405. 703–738. Coase, Ronald, 1960, The problem of social Demsetz, Harold, 1983,The structure of owner- cost, Journal of Law and Economics 3, 1–44. ship and the theory of the firm, Journal of Coffee, John, 1991, Liquidity versus control:The Law and Economics 26, 375–390. institutional investor as corporate monitor, Demsetz, Harold, and Kenneth Lehn, 1985, The Columbia Law Review 91, 1277–1368. structure of corporate ownership: Causes and 84 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS consequences, Journal of Political Economy Eisenberg, Melvin, 1976, The Structure of the 93, 1155–1177. Corporation: A Legal Analysis (Little, Brown Denis, David, and Jan Serrano, 1996, Active and Co., Boston). investors and management turnover following Fama, Eugene, 1980, Agency problems and the unsuccessful control contests, Journal of theory of the firm, Journal of Political Financial Economics 40, 239–266. Economy 88, 288–307. Dewatripont, Mathias, and Eric Maskin, 1995, Fama, Eugene, Lawrence Fisher, Michael Credit and efficiency in centralized and Jensen, and Richard Roll, 1969, The adjust- decentralized economies, Review of ment of stock prices to new information, Economic Studies 62, 541–556. International Economic Review 10, 1–21. Dewatripont, Mathias, and Jean Tirole, 1994, Fama, Eugene, and Michael Jensen, 1983a, A theory of debt and equity: Diversity of Separation of ownership and control, Journal securities and manager-shareholder congru- of Law and Economics 26, 301–325. ence, Quarterly Journal of Economics 109, Fama, Eugene, and Michael Jensen, 1983b, 1027–1054. Agency problems and residuals claims, Dewatripont, Mathias, and Jean Tirole, 1995, Journal of Law and Economics 26, 327–349. The Prudential Regulation of Banks (MIT Fluck, Zsussanna, 1995, The optimality of debt Press, Cambridge). versus outside equity, manuscript, New York Diamond, Douglas, 1984, Financial intermedia- University. tion and delegated monitoring, Review of Franks, Julian, and Colin Mayer, 1990, Economic Studies 51, 393–414. Takeovers: Capital markets and corporate Diamond, Douglas, 1989, Reputation acquisi- control: A study of France, Germany, and the tion in debt markets, Journal of Political UK, Economic Policy: A European Forum Economy 97, 828–862. 10, 189–231. Diamond, Douglas, 1991, Debt maturity struc- Franks, Julian, and Colin Mayer,1994,The own- ture and liquidity risk, Quarterly Journal of ership and control of German corporations, Economics 106, 1027–1054. manuscript, London Business School. Dodd, Peter, and Jerold Warner, 1983, On cor- Galbraith, John Kenneth, 1955, The Great porate governance: A study of proxy contests, Crash, 1929 (Houghton Mifflin, Boston). Journal of Financial Economics 11, Gale, Douglas, and Martin Hellwig, 1985, 401–439. Incentive-compatible debt contracts:The one- Douglas, William O., 1940, Democracy and period problem, Review of Economic Studies Finance (Yale University Press, New Haven). 52, 647–663. Downs, Anthony, 1957, An Economic Theory of Gerschenkron, Alexander, 1962, Economic Democracy (Harper, New York). Backwardness in Historical Perspective Easterbrook, Frank, 1984, Two agency-cost (Harvard University Press, Cambridge MA). explanations of dividends, American Gertner, Robert, and David Scharfstein, 1991, Economic Review 74, 650–659. A theory of workouts and the effects of Easterbrook, Frank, and Daniel Fischel, 1983, reorganization law, Journal of Finance 46, Voting in corporate law, Journal of Law and 1189–1222. Economics 26, 395–427. Gertner, Robert, David Scharfstein, and Jeremy Easterbrook, Frank, and Daniel Fischel, 1991, Stein, 1994, Internal versus external capital The Economic Structure of Corporate Law markets, Quarterly Journal of Economics (Harvard University Press, Cambridge, Mass.). 109, 1211–1230. Eaton, Jonathan, and Mark Gersovitz, 1981, Gilson, Stuart, 1990, Bankruptcy, boards, Debt with potential repudiation: Theoretical banks, and block holders, Journal of and empirical analysis, Review of Economic Financial Economics 27, 355–387. Studies 48, 289–309. Gilson, Stuart, Kose John, and Larry Lang, Edwards, Jeremy, and Klaus Fischer, 1994, 1990,Troubled debt restructurings: an empir- Banks, Finance and Investment in West ical study of private reorganization of firms in Germany Since 1970 (Cambridge University default, Journal of Financial Economics 27, Press, Cambridge). 315–355. A SURVEY OF CORPORATE GOVERNANCE 85 Gomes, Armando, 1996, Dynamics of stock Hart, Oliver,and John Moore, 1989, Default and prices, manager ownership, and private bene- renegotiation: A dynamic model of debt, man- fits of control, manuscript, Harvard University. uscript, Harvard University. Gorton, Gary, and Frank Schmid, 1996, Hart, Oliver, and John Moore, 1990, Property Universal banking and the performance of rights and the nature of the firm, Journal of German firms, Working Paper 5453, National Political Economy 98, 1119–1158. Bureau of Economic Research, Cambridge, MA. Hart, Oliver, and John Moore, 1994a, A theory Grossman, Sanford, and Oliver Hart, 1980, of debt based on the inalienability of human Takeover bids, the free-rider problem, and the capital, Quarterly Journal of Economics 109, theory of the corporation, Bell Journal of 841–879. Economics 11, 42–64. Hart, Oliver,and John Moore, 1994b,The gover- Grossman, Sanford, and Oliver Hart, 1982, nance of exchanges: Members’ cooperatives Corporate financial structure and managerial versus outside ownership, manuscript, incentives, in J. J. McCall Ed.: The Economics Harvard University. of Information and Uncertainty (University of Hart, Oliver, and John Moore, 1995, Debt and Chicago Press, Chicago), 123–155. seniority: An analysis of the role of hard Grossman, Sanford, and Oliver Hart, 1986, The claims in constraining management, costs and benefits of ownership: A theory of American Economic Review 85, 567–585. vertical and lateral integration, Journal of Hart, Oliver, Andrei Shleifer, and Robert Vishny, Political Economy 94, 691–719. 1997, The proper scope of government: Grossman, Sanford, and Oliver Hart, 1988, One Theory and an application to prisons, share-one vote and the market for corporate Quarterly Journal of Economics 112. control, Journal of Financial Economics 20, Haubrich, Joseph, 1994, Risk aversion, perform- 175–202. ance pay, and the principal-agent problem, Grossman, Gene, and Alan Krueger, 1993, Journal of Political Economy 102, 258–276. Environmental impact of a North-American Hellwig, Martin, 1994, Banking and finance at the free trade agreement, in The U.S.-Mexico end of the twentieth century, Wirtschaftswis- Free Trade Agreement, P. Garber, Ed.: (MIT senschaftliches Zentrum Discussion Paper No. Press, Cambridge, Mass.). 9426, University of Basel. Grundfest, Joseph, 1990, Subordination of Holderness, Clifford, and Dennis Sheehan, American capital, Journal of Financial 1988a, What constrains managers who own Economics 27, 89–114. large blocks of stock? Managerial Economics Hansmann, Henry, 1988, Ownership of the firm, Research Center Working Paper 88–07, Journal of Law, Economics, and University of Rochester. Organization 4, 267–304. Holderness, Clifford, and Dennis Sheehan, Hansmann, Henry, 1996, The Ownership of 1988b, The role of majority shareholders in Enterprise (Harvard University Press, publicly held corporations: An exploratory Cambridge, MA). analysis, Journal of Financial Economics 20, Harris, Milton, and Artur Raviv, 1988, 317–346. Corporate governance: Voting rights and Holmstrom, Bengt, 1979, Moral hazard and majority rules, Journal of Financial observability, Bell Journal of Economics 10, Economics 20, 203–235. 4–29. Harris, Milton, and Artur Raviv, 1990, Capital Holmstrom, Bengt, 1982, Managerial incentive structure and the informational role of debt, problems: A dynamic perspective, in Essays in Journal of Finance 45, 321–350. Economics and Management in Honor of Hart, Oliver, 1995, Firms, Contracts, and Lars Wahlbeck (Swedish School of Financial Structure (Oxford University Economics, Helsinki), 210–235. Press, London). Holmstrom, Bengt, and Jean Tirole, 1989, The Hart, Oliver, and Bengt Holmstrom, 1987, The theory of the firm, in R. Schmalensee and Theory of Contracts, in T. Bewley, Ed.: R. D. Willig, Eds.: Handbook of Industrial Advances in Economic Theory (University Organization, Vol. 1, (North-Holland, Press, Cambridge, U.K.), 294–351. Amsterdam), 61–133. 86 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Hopt, Klaus, and Gunther Teubner, Eds., 1985, agency costs, and ownership structure, Corporate Governance and Directors’ Journal of Financial Economics 3, 305–360. Liabilities, (de Gruyter, Berlin). Jensen, Michael, and Kevin Murphy, 1990, Horner, M., 1988, The value of the corporate Performance pay and top management incen- voting right, Journal of Banking and Finance tives, Journal of Political Economy 98, 12, 69–83. 225–263. Hoshi, Takeo, Anil Kashyap, and David Jensen, Michael, and Richard Ruback, 1983, Scharfstein, 1990,The role of banks in reduc- The market for corporate control: The scien- ing the costs of financial distress in Japan, tific evidence, Journal of Financial Journal of Financial Economics 27, 67–88. Economics 11, 5–50. Hoshi, Takeo, Anil Kashyap, and David Johnson, W. Bruce, Robert Magee, Nandu Scharfstein, 1991, Corporate structure, liq- Nagarajan, and Harry Newman, 1985, An uidity, and investment: Evidence from analysis of the stock price reaction to sudden Japanese industrial groups, Quarterly Journal executive deaths: Implications for the man- of Economics 106, 33–60. agement labor market, Journal of Accounting Hoshi, Takeo, Anil Kashyap, and David & Economics 7, 151–174. Scharfstein, 1993,The choice between public Kang, Jun-Koo, and Anil Shivdasani, 1995, Firm and private debt: An analysis of post- performance, corporate governance, and top deregulation corporate finance in Japan, executive turnover in Japan, Journal of National Bureau of Economic Research Financial Economics 38, 29–58. Working Paper 4421, Cambridge, Mass. Kaplan, Steven, 1989, The effects of manage- Hunt, Bishop Carleton, 1936, The Development ment buyouts on operating performance and of the Business Corporation in England, value, Journal of Financial Economics 24, 1800–1867, (Harvard University Press, 217–254. Cambridge, Mass.). Kaplan, Steven, 1991, The staying power of Jarrell, Gregg, and Annette Poulsen, 1988a, leveraged buyouts, Journal of Financial Shark repellents and stock prices:The effects Economics 29, 287–313. of antitakeover amendments since 1980, Kaplan, Steven, 1994a,Top executives, turnover, Journal of Financial Economics 19, and firm performance in Germany, Journal of 127–168. Law, Economics, and Organization 10, Jarrell, Gregg, and Annette Poulsen, 1988b, 142–159. Dual-class recapitalizations as antitakeover Kaplan, Steven, 1994b, Top executive rewards mechanisms: The recent evidence, Journal of and firm performance: A comparison of Financial Economics 20, 129–152. Japan and the United States, Journal of Jensen, Michael, 1986, Agency costs of free Political Economy 102, 510–546. cash flow, corporate finance, and takeovers, Kaplan, Steven, and Bernadette Minton, 1994, American Economic Review 76, 323–329. Appointments of outsiders to Japanese Jensen, Michael, 1988, Takeovers: Their causes boards: Determinants and implications for and consequences, Journal of Economic managers, Journal of Financial Economics Perspectives 2, 21–48. 36, 225–257. Jensen, Michael, 1989a, Eclipse of the public Kaplan, Steven, and Jeremy Stein, 1993, The corporation, Harvard Business Review 67, Evolution of Buy out Pricing and Financial 60–70. Structure in the 1980s, Quarterly Journal of Jensen, Michael, 1989b, Active investors, LBOs Economics 108, 313–357. and the privatization of bankruptcy, Journal Kaplan, Steven, and Michael Weisbach, 1992, of Applied Corporate Finance 2, 35–44. The success of acquisitions: Evidence from Jensen, Michael, 1993, The modern industrial divestitures, Journal of Finance 47, 107–138. revolution, exit, and the failure of internal Khanna, Tarun, and Krishna Palepu, 1996, control systems, Journal of Finance 48, Corporate scope and severe market imperfec- 831–880. tions: An empirical analysis of diversified Jensen, Michael, and William Meckling, 1976, business groups in India, manuscript, Harvard Theory of the firm: Managerial behavior, University. A SURVEY OF CORPORATE GOVERNANCE 87 Kikeri, Sunita, John Nellis, and Mary Shirley, Mace, Myles, 1971, Directors, Myth, and 1992, Privatization: The Lessons of Experience, Reality (Harvard Business School Press, (The World Bank, Washington, D.C.). Boston). Kindelberger, Charles P., 1978, Manias, Panics, Malatesta, Paul, and Ralph Walkling, 1988, and Crashes, (Basic Books, New York). Poison Pill Securities: Stockholder Wealth, Kreps, David, 1990, Corporate Culture and Profitability, and Ownership Structure, Journal Economic Theory, in J. Alt and K. Shepsle, of Financial Economics 20, 347–376. Eds.: Perspectives on Positive Political Manne, Henry, 1965, Mergers and the market Economy (University Press, Cambridge). for corporate control, Journal of Political Laffont, Jean-Jacques, and Jean Tirole, 1993, Economy 75, 110–126. A Theory of Incentives in Regulation and Marris, Robin, 1964, The Economic Theory of Procurement (MIT Press, Cambridge, Mass.). Managerial Capitalism (Free Press of Lambert, Richard, and David Larcker, 1985, Glencoe, Illinois). Golden parachutes, executive decision- Martin, Kenneth, and John McConnell, 1991, making, and shareholder wealth, Journal of Corporate performance, corporate takeovers, Accounting & Economics 7, 179–203. and management turnover, Journal of Lang, Larry, and René Stulz, 1994, Tobin’s Q, Finance 46, 671–688. corporate diversification, and firm perform- Mayer, Colin, 1990, Financial systems, corpo- ance, Journal of Political Economy 102, rate finance, and economic development, in 1248–1280. R. G. Hubbard, Ed.: Asymmetric Information, Lang, Larry, Eli Ofek, and René Stulz, 1996, Corporate Finance, and Investment Leverage, investment, and firm growth, (University of Chicago Press, Chicago). Journal of Financial Economics 40, 3–30. McConnell, John, and Chris Muscarella, 1986, Lang, Larry, René Stulz, and Ralph Walkling, Corporate capital expenditure decisions and 1991, A test of the free cash flow hypothesis: the market value of the firm, Journal of The case of bidder returns, Journal of Financial Economics 14, 399–422. Financial Economics 29, 315–336. McConnell, John, and Henri Servaes, 1990, Loughran, Timothy, Jay Ritter, and Kristian Additional evidence on equity ownership and Rydqvist, 1994, Initial public offerings: corporate value, Journal of Financial International insights, Pacific-Basin Finance Economics 27, 595–612. Journal 2, 165–199. Megginson,William, Robert Nash, and Matthias Lease, Ronald, John McConnell, and Wayne van Randenborgh, 1994, The financial and Mikkelson, 1983,The market value of control operating performance of newly privatized in publicly traded corporations, Journal of firms, Journal of Finance 49, 403–452. Financial Economics 11, 439–471. Mirrlees, James, 1976,The optimal structure of Lease, Ronald, John McConnell, and Wayne incentives and authority within an organiza- Mikkelson, 1984, The market value of differ- tion, Bell Journal of Economics 7, 105–131. ential voting rights in closely held corpora- Modigliani, Franco, and Merton Miller, 1958, tions, Journal of Business 57, 443–467. The cost of capital, corporation finance, and Levy, Haim, 1982, Economic valuation of voting the theory of investment, American Economic power of common stock, Journal of Finance Review 48, 261–297. 38, 79–93. Morck, Randall, Andrei Shleifer, and Robert Lewellen, Wilbur, Claudio Loderer, and Ahron Vishny, 1988a, Characteristics of targets of Rosenfeld, 1985, Merger decisions and execu- hostile and friendly takeovers, in A. Auerbach, tive stock ownership in acquiring firms, Journal Ed.: Corporate Takeovers: Causes and of Accounting & Economics 7, 209–231. Consequences (University of Chicago Press, Lopez-de-Silanes, Florencio, 1994, Chicago). Determinants of privatization prices, manu- Morck, Randall, Andrei Shleifer, and Robert script, Harvard University. Vishny, 1988b, Management ownership and Lopez-de-Silanes, Florencio, Andrei Shleifer, and market valuation: An empirical analysis, Robert W. Vishny, 1997, Privatization in the Journal of Financial Economics 20, United States, Rand Journal of Economics 28. 293–315. 88 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Morck, Randall, Andrei Shleifer, and Robert Pound, John, 1988, Proxy contests and the Vishny, 1989, Alternative Mechanisms of efficiency of shareholder oversight, Journal of Corporate Control, American Economic Financial Economics 20, 237–265. Review 79, 842–852. Prowse, Stephen, 1990, Institutional investment Morck, Randall, Andrei Shleifer, and Robert patterns and corporate financial behavior in Vishny, 1990, Do managerial objectives drive the United States and Japan, Journal of bad acquisitions? Journal of Finance 45, Financial Economics 27, 43–66. 31–48. Rajan, Raghuram, 1992, Insiders and Outsiders: Murphy, Kevin J., 1985, Corporate performance The choice between relationship and arms- and managerial remuneration: An empirical length debt, Journal of Finance 47, analysis, Journal of Accounting & Economics 1367–1400. 7, 11–41. Rajan, Raghuram, and Luigi Zingales, 1995, Myers, Stuart, 1977, Determinants of corporate What do we know about capital structure? borrowing, Journal of Financial Economics Some evidence from international data, 5, 147–175. Journal of Finance 50, 1421–1460. Myers, Stuart, 1995, Inside and outside equity Rappaport, Alfred, 1990, The staying power of financing, manuscript, Massachusetts the public corporation, Harvard Business Institute of Technology. Review 1, 96–104. Myers, Stuart, and Nicholas Majluf, 1984, Ritter, Jay, 1991, The long run performance of Corporate financing and investment decisions IPOs, Journal of Finance 46, 3–27. when firms have information that investors do Roe, Mark, 1993, Some differences in corporate not have, Journal of Financial Economics 13, structure in Germany, Japan, and the United 187–221. States, The Yale Law Review 102, New York Times, 1986, Tax trial only one of 1927–2003. Posner problems, June 23, Section D, 3. Roe, Mark, 1994, Strong Managers Weak Organization for Economic Cooperation and Owners: The Political Roots of American Development, 1995, Corporate governance Corporate Finance (University Press, environments in OECD countries, February 1. Princeton, N.J.). Owen, Thomas, 1991, The Corporation under Roll, Richard, 1986, The hubris hypothesis of Russian Law, 1800–1917: A Study in corporate takeovers, Journal of Business 59, Tsarist Economic Policy (Cambridge 197–216. University Press, New York). Ross, Stephen, 1973, The economic theory of Pagano, Marco, F. Panetta, and Luigi agency: The principal’s problem, American Zingales, 1995, Why do companies go Economic Review 63, 134–139. public? An empirical analysis, manuscript, Rossett, Joshua, G., 1990, Do union wealth Graduate School of Business, University of concessions explain takeover premiums? The Chicago. evidence on contract wages, Journal of Palepu, Krishna, 1986, Predicting takeover Financial Economics 27, 263–282. targets: A methodological and empirical Romano, Roberta, 1993a, The Genius of analysis, Journal of Accounting and American Corporate Law (American Economics 8, 3–35. Enterprise Institute Press, Washington, D.C.). Pontiff, Jeffrey, Andrei Shleifer, and Michael Romano, Roberta, 1993c, Public pension fund Weisbach, 1990, Reversions of excess pension activism in corporate governance reconsid- assets after takeovers, Rand Journal of ered, Columbia Law Review 93, 795–853. Economics 21, 600–613. Rydqvist, Kristian, 1987, Empirical investiga- Porter, Michael, 1992, Capital disadvantage: tion of the voting premium, Working Paper America’s failing capital investment system, 35, Northwestern University. Harvard Business Review, Sept.–Oct., 65–83. Ryngaert, Michael, 1988, The effect of poison Poterba, James M., and Lawrence H. Summers, pill securities on shareholder wealth, Journal 1995, A survey of U.S. companies, time hori- of Financial Economics 20, 377–417. zons, and hurdle rates, Sloan Management Sappington, David, and Joseph Stiglitz, 1987, Review 37, 43–53. Privatization, information, and incentives, A SURVEY OF CORPORATE GOVERNANCE 89 Journal of Policy Analysis and Management Singh, Ajit, 1995, Corporate financial patterns 6, 567–582. in industrializing economies: A comparative Scharfstein, David, 1988, The disciplinary role international study, Technical Paper 2, of takeovers, Review of Economic Studies 55, April, Washington, DC: World Bank and 185–199. International Finance Corporation. Schmidt, Klaus, 1996,The costs and benefits of Smith, Clifford, and Jerold Warner, 1979, On privatization, The Journal of Law, Economics financial contracting: An analysis of bond & Organization 12, 1–24. covenants, Journal of Financial Economics 7, Shapiro, Carl, and Robert D. Willig, 1990, 117–161. Economic rationales for the scope of privati- Stein, Jeremy, 1988,Takeover threats and man- zation, in E. N. Suleiman and J. Waterbury, agerial myopia, Journal of Political Economy Eds.: The Political Economy of Public Sector 96, 61–80. Reform and Privatization (Westview Press, Stein, Jeremy, 1989, Efficient capital markets, London), 55–87. inefficient firms: A model of myopic corpo- Shivdasani, Anil, 1993, Board composition, rate behavior, Quarterly Journal of ownership structure, and hostile takeovers, Economics 104, 655–669. Journal of Accounting and Economics 16, Stigler, George, 1958, The economies of scale, 167–198. Journal of Law and Economics 1, 54–71. Shleifer,Andrei, and Lawrence Summers, 1988, Stiglitz, Joseph, 1975, Incentives, risk, and Breach of trust in hostile takeovers, in A. J. information: Notes toward a theory of hierar- Auerbach, Ed.: Corporate Takeovers: Causes chy, Bell Journal of Economics 6, 552–579. and Consequence (University of Chicago Stulz, René, 1988, Managerial control of voting Press, Chicago), 65–88. rights, Journal of Financial Economics 20, Shleifer, Andrei, and Robert Vishny, 1986a, 25–59. Greenmail, white knights, and shareholder’s Stulz, René, 1990, Managerial discretion and interest, Rand Journal of Economics 17, optimal financing policies, Journal of 293–309. Financial Economics 26, 3–27. Shleifer, Andrei, and Robert Vishny, 1986b, The Economist, June 1995, A Survey of Korea. Large shareholders and corporate control, Teoh, S. H., Ivo Welch, and T. J. Wong, 1995, Journal of Political Economy 94, 461–488. Earnings management and post-issue Shleifer, Andrei, and Robert Vishny, 1988, under performance of seasoned equity Value maximization and the acquisition offerings, manuscript, University of process, Journal of Economic Perspectives California. 2, 7–20. Tirole, Jean, 1994, Incomplete contracts:Where Shleifer, Andrei, and Robert Vishny, 1989, do we stand? manuscript, IDEI, Toulouse, Management entrenchment: The case of France. manager-specific investments, Journal of Titman, Sheridan, and Roberto Wessels, 1988, Financial Economics 25, 123–140. The determinants of capital structure choice, Shleifer, Andrei, and Robert Vishny, 1990, Journal of Finance 43, 1–19. Equilibrium short horizons of investors and Townsend, Robert, 1978, Optimal contracts and firms, American Economic Review Papers competitive markets with costly state verifi- and Proceedings 80, 148–153. cation, Journal of Economic Theory 21, Shleifer, Andrei, and Robert Vishny, 1992, 265–293. Liquidation values and debt capacity: A mar- Tunc, Andre, 1991, Corporate Law, in R. ket equilibrium approach, Journal of Finance Buxbaum, G. Hertig, A. Hirsch, and K. Hopt, 47, 1343–1366. Eds.: European Business Law: Legal and Shleifer, Andrei, and Robert Vishny, 1993, Economic Analysis of Integration and Corruption, Quarterly Journal of Economics Harmonization (de Gruyter, Berlin). 108, 599–618. Walkling, Ralph, and Michael Long, 1984, Shleifer, Andrei, and Robert Vishny, 1994, Agency theory, managerial welfare, and Politicians and firms, Quarterly Journal of takeover bid resistance, Rand Journal of Economics 109, 995–1025. Economics 15, 54–68. 90 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Warner, Jerold, Ron Watts, and Karen Wruck, Williamson, Oliver, 1985, The Economic 1988, Stock prices and top management Institutions of Capitalism (Free Press, New changes, Journal of Financial Economics 20, York). 461–492. Williamson, Oliver,1988, Corporate finance and Weinstein, David E., and Yishay Yafeh, 1994, On corporate governance, Journal of Finance 43, the costs of a bank-centered financial system: 567–592. Evidence from the changing main bank rela- Wolfram, Catherine, 1995, Increases in tions in Japan, manuscript, Harvard University. executive pay following privatization, manu- Weisbach, Michael, 1988, Outside directors and script, MIT. CEO turnover, Journal of Financial Wruck, Karen, 1989, Equity ownership concen- Economics 20, 431–460. tration and firm value, Journal of Financial Weiss, Lawrence, 1990, Bankruptcy resolution: Economics 23, 3–28. Direct costs and violation of priority of Yafeh, Yishay, and Oved Yosha, 1996, Large claims, Journal of Financial Economics 27, shareholders and banks: Who monitors and 285–314. how? manuscript, Hebrew University, White, Michelle, 1993, The costs of corporate Jerusalem, Israel. bankruptcy: a U.S.-European comparison, Yermack, David, 1997, Good timing: CEO stock manuscript, University of Michigan. option awards and company news announce- van Wijnbergen, Sweder, and Anton Mancini, ments, Journal of Finance 52, 449–476. 1995,Voucher privatization, corporate control, Zingales, Luigi, 1994, The value of the voting and the cost of capital: An analysis of the right: A study of the Milan stock exchange Czech privatization program, manuscript, LSE. experience, The Review of Financial Studies Williamson, Oliver, 1964, The Economics of 7, 125–148. Discretionary Behavior: Managerial Zingales, Luigi, 1995, What determines the Objectives in a Theory of the Firm (Prentice value of corporate votes? Quarterly Journal Hall, Englewood Cliffs, N.J.). of Economics 110, 1075–1110. Chapter 3

Rafael La Porta, Florencio Lopez-de-Silanes, Andrei Shleifer† and Robert Vishny INVESTOR PROTECTION AND CORPORATE GOVERNANCE*

Source: Journal of Financial Economics, 58(1–2)(2000): 3–27.

ABSTRACT Recent research has documented large differences among countries in ownership concentration in publicly traded firms, in the breadth and depth of capital markets, in dividend policies, and in the access of firms to external finance. A common element to the explanations of these differences is how well investors, both shareholders and creditors, are protected by law from expropriation by the managers and controlling shareholders of firms.We describe the differences in laws and the effectiveness of their enforcement across countries, discuss the possible origins of these differences, summarize their consequences, and assess potential strategies of corporate governance reform. We argue that the legal approach is a more fruitful way to understand corporate governance and its reform than the conventional distinction between bank-centered and market-centered financial systems.

1. INTRODUCTION of minority shareholders and creditors by the controlling shareholders is extensive. RECENT RESEARCH ON CORPORATE GOVERNANCE When outside investors finance firms, they around the world has established a number face a risk, and sometimes near certainty, of empirical regularities. Such diverse that the returns on their investments will elements of countries’ financial systems as never materialize because the controlling the breadth and depth of their capital shareholders or managers expropriate markets, the pace of new security issues, them. (We refer to both managers and con- corporate ownership structures, dividend trolling shareholders as “the insiders”.) policies, and the efficiency of investment Corporate governance is, to a large extent, allocation appear to be explained both a set of mechanisms through which outside conceptually and empirically by how well investors protect themselves against expro- the laws in these countries protect outside priation by the insiders. investors. According to this research, the Expropriation can take a variety of protection of shareholders and creditors by forms. In some instances, the insiders the legal system is central to understanding simply steal the profits. In other the patterns of corporate finance in instances, the insiders sell the output, the different countries. assets, or the additional securities in the Investor protection turns out to be crucial firm they control to another firm they own because, in many countries, expropriation at below market prices. Such transfer 92 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS pricing, asset stripping, and investor may use these resources for their own dilution, though often legal, have largely benefit. Jensen and Meckling view financial the same effect as theft. In still other claims as contracts that give outside instances, expropriation takes the form of investors, such as shareholders and credi- diversion of corporate opportunities from tors, claims to the cash flows. In their the firm, installing possibly unqualified model, the limitation on expropriation is family members in managerial positions, the residual equity ownership by entrepre- or overpaying executives. In general, neurs that enhances their interest in expropriation is related to the agency dividends relative to perquisites. problem described by Jensen and Research by Grossman, Hart, and Meckling (1976), who focus on the Moore, summarized in Hart (1995), makes consumption of “perquisites” by man- a further key advance by focusing squarely agers and other types of empire building. on investor power vis a vis the insiders, and It means that the insiders use the profits distinguishing between the contractual and of the firm to benefit themselves rather residual control rights that investors have. than return the money to the outside Economists have used this idea to model investors. financial instruments not in terms of their If extensive expropriation undermines cash flows, but in terms of the rights they the functioning of a financial system, how allocate to their holders. In this frame- can it be limited? The legal approach to work, investors get cash only because they corporate governance holds that the have power. This can be the power to key mechanism is the protection of change directors, to force dividend outside investors – whether shareholders or payments, to stop a project or a scheme creditors – through the legal system, meaning that benefits the insiders at the expense of both laws and their enforcement. Although outside investors, to sue directors and get reputations and bubbles can help raise compensation, or to liquidate the firm and funds, variations in law and its enforcement receive the proceeds. Unlike in the are central to understanding why firms Modigliani-Miller world, changing the raise more funds in some countries than in capital structure of the firm changes others. To a large extent, potential share- the allocation of power between the insiders holders and creditors finance firms because and the outside investors, and thus their rights are protected by the law.These almost surely changes the firm’s investment outside investors are more vulnerable to policy. expropriation, and more dependent on the In both the contractual framework of law, than either the employees or the Jensen and Meckling and the residual con- suppliers, who remain continually useful to trol rights framework of Grossman, Hart, the firm and are thus at a lesser risk of and Moore, the rights of the investors are being mistreated. protected and sometimes even specified by The legal approach to corporate gover- the legal system. For example, contract law nance is a natural continuation of the field deals with privately negotiated arrange- as it has developed over the last 40 years. ments, whereas company, bankruptcy, and Modigliani and Miller (1958) think of securities laws specifically describe some firms as collections of investment projects of the rights of corporate insiders and out- and the cash flows these projects create, side investors. These laws, and the quality and hence naturally interpret securities of their enforcement by the regulators and such as debt and equity as claims to these courts, are essential elements of corporate cash flows. They do not explain why the governance and finance (La Porta et al., managers would return the cash flows to 1997, 1998). When investor rights such as investors. Jensen and Meckling (1976) the voting rights of the shareholders and point out that the return of the cash flows the reorganization and liquidation rights from projects to investors cannot be taken of the creditors are extensive and well for granted, and that the insiders of firms enforced by regulators or courts, investors INVESTOR PROTECTION AND CORPORATE GOVERNANCE 93 are willing to finance firms. In contrast, law also shapes the opportunities for when the legal system does not protect out- external finance. side investors, corporate governance and The legal approach to corporate gover- external finance do not work well. nance has emerged as a fruitful way to Jensen and Meckling (1976) recognize think about a number of questions in the role of the legal system when they finance. In Section 2, we discuss the differ- write: ences in legal investor protection among countries and the possible judicial, This view of the firm points up the political, and historical origins of these important role which the legal system differences. In Section 3, we summarize the and the law play in social organizations, research on the economic consequences of especially, the organization of economic investor protection. In Section 4, we compare activity. Statutory law sets bounds on the the legal approach to corporate governance kinds of contracts into which individuals to the more standard focus on the relative and organizations may enter without importance of banks and stock markets risking criminal prosecution. The police as ways to explain country differences. In powers of the state are available and Section 5, we discuss both the difficulties used to enforce performance of contracts and the opportunities for corporate gover- or to enforce the collection of damages nance reform. Section 6 concludes. for non-performance. The courts adjudicate contracts between contracting parties and establish precedents which 2. INVESTOR PROTECTION form the body of common law. All of these government activities affect both When investors finance firms, they typically the kinds of contracts executed and the obtain certain rights or powers that are extent to which contracting is relied generally protected through the enforce- upon (p. 311). ment of regulations and laws. Some of these rights include disclosure and One way to think about legal protection accounting rules, which provide investors of outside investors is that it makes the with the information they need to exercise expropriation technology less efficient. At other rights. Protected shareholder rights the extreme of no investor protection, the include those to receive dividends on insiders can steal a firm’s profits perfectly pro-rata terms, to vote for directors, to efficiently. Without a strong reputation, no participate in shareholders’ meetings, to rational outsider would finance such a firm. subscribe to new issues of securities on the As investor protection improves, the insiders same terms as the insiders, to sue directors must engage in more distorted and waste- or the majority for suspected expropriation, ful diversion practices such as setting up to call extraordinary shareholders’ meetings, intermediary companies into which they etc. Laws protecting creditors largely deal channel profits. Yet these mechanisms are with bankruptcy and reorganization still efficient enough for the insiders to procedures, and include measures that choose to divert extensively. When investor enable creditors to repossess collateral, to protection is very good, the most the insiders protect their seniority, and to make it can do is overpay themselves, put relatives harder for firms to seek court protection in in management, and undertake some reorganization. wasteful projects. After a point, it may be In different jurisdictions, rules protect- better just to pay dividends. As the diver- ing investors come from different sources, sion technology becomes less efficient, the including company, security, bankruptcy, insiders expropriate less, and their private takeover, and competition laws, but also benefits of control diminish. Firms then from stock exchange regulations and obtain outside finance on better terms. By accounting standards. Enforcement of laws shaping the expropriation technology, the is as crucial as their contents. In most 94 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS countries, laws and regulations are Whether contracts, court-enforced legal enforced in part by market regulators, in rules, or government-enforced regulations part by courts, and in part by market are the most efficient form of protecting participants themselves. All outside financial arrangements is largely an empir- investors, be they large or small, share- ical question. As the next section shows, the holders or creditors, need to have their evidence rejects the hypothesis that private rights protected. Absent effectively contracting is sufficient. Even among coun- enforced rights, the insiders would not have tries with well functioning judiciaries, those much of a reason to repay the creditors or with laws and regulations more protective to distribute profits to shareholders, and of investors have better developed capital external financing mechanisms would tend markets. to break down. La Porta et al. (1998) discuss a set of The emphasis on legal rules and regulations key legal rules protecting shareholders and protecting outside investors stands in creditors and document the prevalence of sharp contrast to the traditional “law and these rules in 49 countries around the economics” perspective on financial con- world. They also aggregate these rules into tracting. According to that perspective, shareholder (antidirector) and creditor most regulations of financial markets are rights indices for each country, and con- unnecessary because financial contracts sider several measures of enforcement take place between sophisticated issuers quality, such as the efficiency of the judicial and sophisticated investors. On average, system and a measure of the quality of investors recognize a risk of expropriation, accounting standards. La Porta, Lopez-de- penalizing firms that fail to contractually Silanes, Shleifer, and Vishny use these disclose information about themselves and variables as proxies for the stance of the to contractually bind themselves to treat law toward investor protection to examine investors well. Because entrepreneurs bear the variation of legal rules and enforcement these costs when they issue securities, they quality across countries and across legal have an incentive to bind themselves families. through contracts with investors to limit Legal scholars such as David and expropriation (Jensen and Meckling, Brierley (1985) show that commercial 1976). As long as these contracts are legal systems of most countries derive from enforced, financial markets do not require relatively few legal “families,” including regulation (Stigler, 1964; Easterbrook and the English (common law), the French, and Fischel, 1991). the German, the latter two derived from the This point of view, originating in the Roman Law. In the 19th century, these sys- Coase (1961) theorem, crucially relies on tems spread throughout the world through courts enforcing elaborate contracts. In conquest, colonization, and voluntary adop- many countries, such enforcement cannot tion. England and its former colonies, be taken for granted. Indeed, courts are including the U.S., Canada, Australia, New often unable or unwilling to invest the Zealand, and many countries in Africa and resources necessary to ascertain the facts South East Asia, have ended up with the pertaining to complicated contracts. They common law system. France and many are also slow, subject to political pressures, countries Napoleon conquered are part and at times corrupt. When the enforce- of the French civil law tradition. This ment of private contracts through the court legal family also extends to the former system is costly enough, other forms of pro- French, Dutch, Belgian, and Spanish tecting property rights, such as judicially- colonies, including Latin America. enforced laws or even government-enforced Germany, Germanic countries in Europe, regulations, may be more efficient. It may and a number of countries in East Asia are be better to have contracts restricted by part of the German civil law tradition. The laws and regulations that are enforced than Scandinavian countries form their own unrestricted contracts that are not. tradition.1 INVESTOR PROTECTION AND CORPORATE GOVERNANCE 95 Table 3.1 presents the percentage of distinguish between two broad kinds of countries in each legal family that give answers: the “judicial” explanations that investors the rights discussed by La Porta, account for the differences in the legal Lopez-de-Silanes, Shleifer, and Vishny, as philosophies using the organization of the well as the mean for that family antidirector legal system, and the “political” explana- and creditor rights scores. How well legal tions that account for these differences rules protect outside investors varies using political history. systematically across legal origins. The “judicial” explanation of why com- Common law countries have the strongest mon law protects investors better than civil protection of outside investors – both law has been most recently articulated by shareholders and creditors – whereas Coffee (2000) and Johnson et al. (2000b). French civil law countries have the weakest Legal rules in the common law system are protection. German civil law and usually made by judges, based on prece- Scandinavian countries fall in between, dents and inspired by general principles although comparatively speaking they have such as fiduciary duty or fairness. Judges stronger protection of creditors, especially are expected to rule on new situations by secured creditors. In general, differences applying these general principles even when among legal origins are best described by specific conduct has not yet been described the proposition that some countries protect or prohibited in the statutes. In the area of all outside investors better than others, and investor expropriation, also known as self- not by the proposition that some countries dealing, the judges apply what Coffee calls protect shareholders while other countries a “smell test,” and try to sniff out whether protect creditors. even unprecedented conduct by the insiders Table 3.1 also points to significant is unfair to outside investors.The expansion differences among countries in the quality of legal precedents to additional violations of law enforcement as measured by the of fiduciary duty, and the fear of such efficiency of the judiciary, (lack of) corrup- expansion, limit the expropriation by the tion, and the quality of accounting stan- insiders in common law countries. In con- dards. Unlike legal rules, which do not trast, laws in civil law systems are made by appear to depend on the level of economic legislatures, and judges are not supposed to development, the quality of enforcement is go beyond the statutes and apply “smell higher in richer countries. In particular,the tests” or fairness opinions. As a conse- generally richer Scandinavian and German quence, a corporate insider who finds a way legal origin countries receive the best not explicitly forbidden by the statutes to scores on the efficiency of the judicial sys- expropriate outside investors can proceed tem.The French legal origin countries have without fear of an adverse judicial ruling. the worst quality of law enforcement of the Moreover, in civil law countries, courts do four legal traditions, even controlling for not intervene in self-dealing transactions as per capita income. long as these have a plausible business Because legal origins are highly corre- purpose.The vague fiduciary duty principles lated with the content of the law, and of the common law are more protective of because legal families originated before investors than the bright line rules of the financial markets had developed, it is civil law, which can often be circumvented unlikely that laws were written primarily in by sufficiently imaginative insiders. response to market pressures. Rather, the The judicial perspective on the differ- legal families appear to shape the legal ences is fascinating and possibly correct, rules, which in turn influence financial but it is incomplete. It requires a further markets. But what is special about legal assumption that the judges have an inclina- families? Why, in particular,is common law tion to protect the outside investors rather more protective of investors than civil law? than the insiders. In principle, it is easy to These questions do not have accepted imagine that the judges would use their dis- answers. However, it may be useful here to cretion in common law countries to narrow 96 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS ual reports on their shareholder protec- shareholder rights that can only rights poses restrictions, such “creditor rights index” rights “creditor es a shareholder to call es a shareholder els). “Corruption” is an ation of minorities in the in of minorities ation x” by the law. is provided which each component of which g the average of investors’ g the average tatement, stan- accounting te to the firm; shareholders result from the disposition result from organization.The rest of the data is La Porta et al. data is La Porta (1998). Legal origin law (18 law (21 civil law civil law average Common French civil German Scandinavian World Panel A: Measures of shareholder protection ends to file for reorganization; for once the reorganiza- of their security ends to file possession to gain are able secured creditors divid ten representing the average of investors’ the average ten representing each country between 1982 and 1995 in government in assessments of corruption Legal origin and investors rights and investors origin Legal inclusion or omission of 90 items falling in the categories of general information, of general statements, income the categories in of 90 items falling or omission balance sheets,inclusion funds flow s (lower scores represent higher corruption). companies’ standards” scores represent higher and rating (lower “Accounting 1990 ann created by examining is an index index ranging from zero to assessments of conditions of the judicial system in each country between 1980–1983 (lower scores represent lower efficiency lev efficiency scores represent lower each country system between 1980–1983 (lower in of the judicial assessments of conditions Panel C shows measures of legal enforcement.“Efficiency of the judicial system” is an index ranging from zero to ten representin system” zero enforcement.“Efficiency from of the judicial measures of legal C shows ranging Panel is an index rows in Panel B show the percentage of countries within each legal origin for which each component of the “creditor rights inde rights “creditor each component of the which for origin each legal within of countries the percentage B show Panel in rows tion petition has been approved (no automatic stay); has been approved petition tion that of the proceeds the distribution in first are ranked secured creditors of the re the resolution firm;of the assets of a bankrupt pending of its property the administration the debtor does not retain as creditors’ consent or minimum the index” rights “antidirector origins. by the law. legal The across protection is provided the measures of creditor B shows Panel is a summary measure of creditor protection.This index ranges from zero to four and is formed by adding one when: by adding and is formed to four zero from the country im ranges protection.Thisis a summary index measure of creditor be waved by a shareholders’, vote.The rest of the rows in Panel A show the percentage of countries within each legal origin for for be waved by a shareholders’, origin each legal within of countries the percentage A show vote.The Panel in rest of the rows for an Extraordinary Shareholders’. an Extraordinary for (the sample median); is less than or equal to 10 percent Meeting have preemptive shareholders board of directors is allowed; place; is in mechanism of share capital that entitl an oppressed minorities percentage the minimum are not required to deposit their shares prior to the General Shareholders’. to the General are not required to deposit their shares prior Meeting; represent or proportional voting cumulative Proxy by mailProxy 39% 5% 0% 25% 18% tion.This index ranges from zero to six and is formed by adding one when: by adding vo to six and is formed their proxy zero to mail from shareholders the country ranges allows index tion.This This table presents data on measures of investor protection for 49 countries classified by their legal origin.The source of the source origin.The by their legal classified 49 countries for protection presents data on measures of investor table This Panel A shows the measures of shareholder protection across legal origins. legal The across protection the measures of shareholder A shows index”Panel rights “antidirector is a summary measure of Antidirector rights index rights Antidirector 4.00 2.33 2.33 3.00 3.00 Table 3.1 dards, stock data, items. and special Variables countries) countries) (6 countries) (4 countries) (49 countries) INVESTOR PROTECTION AND CORPORATE GOVERNANCE 97 Panel C: Measures of enforcement Panel B: Measures of creditor protection 10% 94% 52% 0% 0% 78% nto reorganization 72% 42% 33% 75% 55% going i Corruption 7.06 5.84 8.03 10.00 6.90 Efficiency of of the judicial system of the judicial Efficiency 8.15 6.56 8.54 10.00 7.67 Management does not stay in reorganizationManagement does not stay in 78% 26% 33% 0% 45% Paid restrictions for for restrictions Paid Secured creditors firstSecured creditors 89% 65% 100% 100% 81% No automatic stay on secured assets 72% 26% 67% 25% 49% Creditor rights index rights Creditor 3.11 1.58 2.33 2.00 2.30 Preemptive right to new issues right Preemptive % Share of capital to call and ESM 44% 62% 33% 75% 53% Oppressed minority 94% 29% 50% 0% 53% Cumulative voting/proportional represent’n voting/proportional Cumulative 28% 29% 33% 0% 27% Shares not blocked before meeting before Shares not blocked 100% 57% 17% standardsAccounting 100% 71% 69.92 51.17 62.67 74.00 60.93 98 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS the interpretation of fiduciary duty and to Recent research supports the proposition sanction expropriation rather than prohibit that civil law is associated with greater it. Common law judges could also in princi- government intervention in economic ple use their discretion to serve political activity and weaker protection of private interests, especially when the outside property than common law. La Porta et al. investors obstruct the government’s goals. (1999a) examine the determinants of To explain investor protection, it is not government performance in a large number enough to focus on judicial power; a political of countries. To measure government and historical analysis of judicial objectives interventionism, they consider proxies for is required. From this perspective, impor- the amount and quality of regulation, the tant political and historical differences prevalence of corruption and of red tape, between mother countries shape their laws. and bureaucratic delays. As a general rule, This is not to say that laws never change they find that civil law countries, particu- (in Section 5 we focus specifically on legal larly French civil law countries, are more reform) but rather to suggest that history interventionist than common law countries. has persistent effects. The inferior protection of the rights of out- La Porta et al. (1999a) argue that an side investors in civil law countries may be important historical factor shaping laws is one manifestation of this general phenome- that the state has a relatively greater role non. This evidence provides some support in regulating business in civil law countries for interpreting the differences in legal than in common law ones. One element of families based on political history.3 this view, suggested by Finer (1997) and other historians, points to the differences in the relative power of the king and the prop- 3. CONSEQUENCES OF INVESTOR erty owners across European states. In PROTECTION England from the seventeenth century on, the crown partially lost control of the Three broad areas in which investor courts, which came under the influence of protection has been shown to matter are the parliament and the property owners the ownership of firms, the development of who dominated it. As a consequence, financial markets, and the allocation of common law evolved to protect private real resources. property against the crown. Over time, courts extended this protection of property owners to investors. In France and 3.1. Patterns of ownership and Germany, by contrast, parliamentary power control was weaker. Commercial Codes were adopted only in the nineteenth century by The focus on expropriation of investors and the two great state builders, Napoleon and its prevention has a number of implications Bismarck, to enable the state to better for the ownership structures of firms. regulate economic activity. Over time, the Consider first the concentration of control state maintained political control over rights in firms (as opposed to the dividend firms and resisted the surrender of that or cash flow rights). At the most basic power to financiers.2 Perhaps as impor- level, when investor rights are poorly tantly, the state in civil law countries did protected and expropriation is feasible on a not surrender its power over economic substantial scale, control acquires enormous decisions to courts, and hence maintained value because it gives the insiders the the statutory approach to commercial opportunity to expropriate efficiently. laws. As we noted above, however, fairness When the insiders actually do expropriate, assessments of self-dealing transactions, the so called private benefits of control for which judicial power and discretion become a substantial share of the firm’s are essential, may be central to limiting value. This observation raises a question: expropriation. will control in such an environment be INVESTOR PROTECTION AND CORPORATE GOVERNANCE 99 concentrated in the hands of an entrepreneur outsiders and still retain control by holding or dispersed among many investors? on to the shares with superior voting rights. The research in this area originates in He can also use a pyramidal structure, in the work of Grossman and Hart (1988) which a holding company he controls sells and Harris and Raviv (1988), who examine shares in a subsidiary that it itself controls. the optimal allocation of voting and cash Wolfenzon (1999) shows that an entrepre- flow rights in a firm. The specific question neur can then control the subsidiary of how control is likely to be allocated has without owning a substantial fraction of its not received a clear answer. For several cash flow rights, and that such schemes are reasons, entrepreneurs may wish to keep more attractive when the protection of control of their firms when investor protec- outside investors is weaker.An entrepreneur tion is poor. La Porta et al. (1999) note can also keep control through cross- that if expropriation of investors requires shareholdings among firms, which make it secrecy, sharing control may restrain the harder for outsiders to gain control of one entrepreneur beyond his wishes. Zingales group firm without buying all of them. (1995), La Porta et al. (1999), and What about the distribution of cash flow Bebchuk (1999) argue that if entrepreneurs rights between investors as opposed to disperse control between many investors, control? If an entrepreneur retains control they give up the “private benefits” premium of a firm, how can he raise any external in a takeover. In Bebchuk’s (1999) model, funds from outside investors – for financing diffuse control structures are unstable or for diversification – who expect to be when investors can concentrate control expropriated? Jensen and Meckling (1976) without fully paying for it. Finally, an entre- would suggest that cash flow ownership by preneur or his family may need to retain an entrepreneur reduces incentives for control of the firm because the family’s expropriation and raises incentives to pay reputation is needed to raise external funds out dividends. La Porta et al. (1999b) when the legal protection of outside show that this need for higher cash flow investors is poor. For all these reasons, ownership as a commitment to limit expro- firms in countries with poor investor priation is higher in countries with inferior protection may need concentrated control. shareholder protection. Bennedsen and Wolfenzon (2000) make The available evidence on corporate a countervailing argument. When investor ownership patterns around the world protection is poor, dissipating control supports the importance of investor protec- among several large investors – none of tion. This evidence was obtained for a whom can control the decisions of the firm number of individual countries, including without agreeing with the others – may Germany (Edwards and Fischer, 1994; serve as a commitment to limit expropria- Gorton and Schmid, 2000), Italy (Barca, tion. When there is no single controlling 1995), and seven Organization for shareholder, and the agreement of several Economic Cooperation and Development large investors (the board) is needed for countries (European Corporate Governance major corporate actions, these investors Network, 1997). La Porta et al. (1998) might together hold enough cash flow describe ownership concentration in their rights to choose to limit expropriation of sample of 49 countries, while La Porta the remaining shareholders and pay the et al. (1999) examine patterns of control in profits out as efficient dividends. When the largest firms from each of 27 wealthy the dissipation of control reduces ineffi- economies. The data show that countries cient expropriation, it may emerge as an with poor investor protection typically optimal policy for a wealth-maximizing exhibit more concentrated control of firms entrepreneur. than do countries with good investor pro- An entrepreneur has a number of ways tection. In the former, even the largest to retain control of a firm. He can firms are usually controlled either by the sell shares with limited voting rights to the state or by the families that founded or 100 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS acquired these firms. In the latter countries, Creditor rights encourage the development the Berle and Means corporation – with of lending, and the exact structure of these dispersed shareholders and professional rights may alternatively favor bank lending managers in control – is more common.4 or market lending. Shareholder rights Claessens et al. (2000) examine a encourage the development of equity sample of nearly 3,000 firms from 9 East markets, as measured by the valuation of Asian economies. Except in Japan, which firms, the number of listed firms (market has fairly good shareholder protection, they breadth), and the rate at which firms go find a predominance of family control and public. For both shareholders and credi- family management of the corporations in tors, protection includes not only the rights their sample, with some state control as written into the laws and regulations but well.They also present remarkable evidence also the effectiveness of their enforcement. of “crony capitalism” in Asia: outside Consistent with these predictions, La Porta Japan, the top 10 families in each of the et al. (1997) show that countries that remaining 8 countries studied control protect shareholders have more valuable between 18 and 58 percent of the aggre- stock markets, larger numbers of listed gate value of listed equities. securities per capita, and a higher rate of In sum, the evidence has proved to be IPO (initial public offering) activity than broadly consistent with the proposition do the unprotective countries. Countries that the legal environment shapes the value that protect creditors better have larger of the private benefits of control and credit markets. thereby determines the equilibrium owner- Several recent studies have also estab- ship structures. Perhaps the main implica- lished a link between investor protection, tions of this evidence for the study of insider ownership of cash flows, and corpo- corporate governance are the relative rate valuation.5 Gorton and Schmid irrelevance of the Berle and Means corpo- (2000) show that higher ownership by the ration in most countries in the world and large shareholders is associated with the centrality of family control. Indeed, higher valuation of corporate assets in La Porta et al. (1999) and Claessens et al. Germany. Claessens et al. (1999) use a (2000) find that family-controlled firms sample of East Asian firms to show that are typically managed by family members greater insider cash flow ownership is asso- so that the managers appear to be kept on ciated with higher valuation of corporate a tighter leash than what Berle and Means assets, whereas greater insider control of describe. As Shleifer and Vishny (1997) voting rights is associated with lower valu- have argued, in large corporations of most ation of corporate assets. Using a sample countries, the fundamental agency problem of firms from 27 wealthy economies, is not the Berle and Means conflict La Porta et al. (1999b) find that firms in between outside investors and managers, countries with better shareholder protection but rather that between outside investors have higher Tobin’s Q than do firms in and controlling shareholders who have countries with inferior protection.They also nearly full control over the managers. find that higher insider cash flow ownership is (weakly) associated with higher corporate 3.2. Financial markets valuation, and that this effect is greater in countries with inferior shareholder The most basic prediction of the legal protection. These results support the roles approach is that investor protection of investor protection and cash flow encourages the development of financial ownership by the insiders in limiting markets. When investors are protected expropriation. from expropriation, they pay more for Johnson et al. (2000a) draw an ingenious securities, making it more attractive for connection between investor protection and entrepreneurs to issue these securities.This financial crises. In countries with poor applies to both creditors and shareholders. protection, the insiders might treat outside INVESTOR PROTECTION AND CORPORATE GOVERNANCE 101 investors well as long as future prospects papers show that an exogenous component are bright and they are interested in con- of financial market development, obtained tinued external financing. When future by using legal origin as an instrument, prospects deteriorate, however, the insiders predicts economic growth. step up expropriation, and the outside More recent research distinguishes the investors, whether shareholders or credi- three channels through which finance can tors, are unable to do anything about it. contribute to growth: saving, factor accu- This escalation of expropriation renders mulation, and efficiency improvements. security price declines especially deep in Beck et al. (2000) find that banking sector countries with poor investor protection. To development exerts a large impact on total test this hypothesis, Johnson et al. (2000a) factor productivity growth and a less examine the depreciation of currencies and obvious impact on private savings and capital the decline of the stock markets in 25 accumulation. Moreover, this influence countries during the Asian crisis of continues to hold when an exogenous 1997–1998. They find that governance component of banking sector development, variables, such as investor protection obtained by using legal origin as an instru- indices and the quality of law enforcement, ment, is taken as a predictor. Wurgler are powerful predictors of the extent of (2000) finds that financially developed market declines during the crisis. These countries allocate investment across indus- variables explain the cross-section of tries more in line with the variation in declines better than do the macroeconomic growth opportunities than do financially variables that have been the focus of the underdeveloped countries. Morck et al. initial policy debate. (2000) find that stock markets in more developed countries incorporate firm- specific information better, helping to 3.3. Real consequences allocate investment more effectively. This Through its effect on financial markets, research suggests that financial development investor protection influences the real improves resource allocation. Through this economy. According to Beck et al. (2000), channel, investor protection may benefit financial development can accelerate eco- the growth of productivity and output. nomic growth in three ways. First, it can enhance savings. Second, it can channel these savings into real investment and 4. BANK AND MARKET CENTERED thereby foster capital accumulation. Third, GOVERNANCE to the extent that the financiers exercise some control over the investment decisions Traditional comparisons of corporate of the entrepreneurs, financial development governance systems focus on the institu- allows capital to flow toward the more pro- tions financing firms rather than on the ductive uses, and thus improves the effi- legal protection of investors. Bank-centered ciency of resource allocation. All three corporate governance systems, such as channels can in principle have large effects those of Germany and Japan, are on economic growth. compared to market-centered systems, A large body of research links financial such as those of the United States and the development to economic growth. King and United Kingdom (see, e.g., Allen and Gale, Levine (1993) initiate the modern incarna- 2000). Relatedly, relationship-based tion of this literature by showing that coun- corporate governance, in which a main tries with larger initial capital markets bank provides a significant share of grow faster in the future. Demirguc-Kunt finance and governance to each firm, is and Maksimovic (1998), Levine and contrasted with market-based governance, Zervos (1998), Rajan and Zingales in which finance is provided by large num- (1998), and Carlin and Mayer (1999) bers of investors and in which takeovers extend these findings. Several of these play a key governance role. 102 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS These institutional distinctions have of listed securities? Or what about the been central to the evaluation of alternative French civil law based financial systems, in corporate governance regimes and to policy which neither credit markets nor stock proposals for improvement. In the 1980s, markets are especially well developed? when the Japanese economy could do no Sapienza (1999), for example, finds that in wrong, bank-centered governance was Italy the stock market is extremely under- widely regarded as superior because, as developed, but so is the banking system, Aoki and Patrick (1993) and Porter with a typical firm raising a small amount (1992) argue, far-sighted banks enable of money from each of a dozen banks. firms to focus on long term investment More generally, La Porta et al. (1997) decisions. According to Hoshi et al. (1991), show that, on average, countries with big- banks also deliver capital to firms facing ger stock markets also have higher ratios of liquidity shortfalls, thereby avoiding costly private debt to gross domestic product financial distress. Finally, banks replace the (GDP), contrary to the view that debt and expensive and disruptive takeovers with equity finance are substitutes for each more surgical bank intervention when the other.The prevalent financing modes gener- management of the borrowing firm under- ally do not help with the classification. performed. Another way to classify financial systems In the 1990s, as the Japanese economy is based on the existence of Glass-Steagall collapsed, the pendulum swung the other regulations restricting bank ownership of way.6 Kang and Stulz (1998) show that, corporate equity. This approach is again far from being the promoters of rational useful for distinguishing the United States investment, Japanese banks perpetrate soft from Germany, which does not have such budget constraints, over-lending to declining regulations. On the other hand, most firms that require radical reorganization. countries in the world do not have these And according to Weinstein and Yafeh regulations. Some of them, like the United (1998) and Morck and Nakamura (1999), Kingdom, have a highly developed stock Japanese banks, instead of facilitating gov- market and few equity holdings by banks, ernance, collude with enterprise managers even though banks are not prevented by to deter external threats to their control law from holding equity. Other countries and to collect rents on bank loans. In the have neither a developed banking system recent assessments by Edwards and nor a developed stock market. Glass- Fischer (1994) and Hellwig (1999), Steagall regulations in themselves do not German banks are likewise downgraded to assure a development of a market system ineffective providers of governance. by interfering with corporate governance by Market-based systems, in contrast, rode the banks. Consistent with our skepticism American stock market bubble of the about the usefulness of such regulations for 1990s into the stratosphere of wide classifying financial systems, La Porta support and adulation. et al. (1999) show that Glass-Steagall Unfortunately, the classification of finan- regulations have no predictive power for cial systems into bank and market centered ownership concentration across countries. is neither straightforward nor particularly Perhaps most important, the reliance on fruitful. One way to do this is by looking at either the outcomes or the Glass-Steagall the actual outcomes. It is easy to classify regulations to classify corporate governance Germany as bank-centered because its regimes misses the crucial importance of banks influence firms through both debt investor rights. All financiers depend on and equity holdings and its stock market is legal protection to function. A method of underdeveloped.7 But what about Japan, financing develops when it is protected by which boasts both powerful banks with the law that gives financiers the power to influence over firms and a highly developed get their money back. Germany and some and widely-held equity market (second or other German civil law countries have third in the world by size) with thousands developed banking systems because they INVESTOR PROTECTION AND CORPORATE GOVERNANCE 103 have strong legal protection of creditors, (1931) and others have argued that particularly of secured creditors. Without bubbles play an important and positive role such rights German banks would have in stimulating investment. much less power.The United Kingdom also To summarize, bank-versus market has a large banking and public debt sector, centeredness is not an especially useful way again because creditors have extensive to distinguish financial systems. Investor rights, as well as a large equity market. rights work better to explain differences Italy and Belgium, by contrast, have among countries, and in fact are often developed neither debt nor equity markets necessary for financial intermediaries to because no outside investors are protected develop. Moreover, even if some countries there.8 The point here is simple: all outside go through monopoly banking in their investors, be they large or small, creditors development process, this stage has little to or shareholders, need rights to get their recommend it other than as a stepping money back. Investor rights are a stone toward more developed markets. And more primitive determinant of financial to get to more developed markets, it is development than is the size of particular essential to improve the rights of outside institutions. investors. Despite the difficulty of classifying financial systems into bank- and market centered, economists at least since 5. POSSIBILITIES FOR REFORM Gerschenkron (1962) have engaged in a lively debate as to which one is superior, In the last decade, the reform of corporate focusing on the hypothesis that bank- governance has attracted interest in centered systems are particularly suitable for Western and Eastern Europe, Latin developing economies.This is not a place to America, and Asia. The discussions have review this debate. Rather, our concern is intensified since the Asian financial crisis, that the interest in monopoly bank lending and took on the flavor of reforming “the distracts attention from the important role global financial architecture”. To discuss that stock markets play in external finance. any reform, it is important to start with its Equity financing is essential for the expan- goals. Our analysis suggests that one objec- sion of new firms whose main asset are the tive of corporate governance reform is to growth opportunities. In principle, firms protect the rights of outside investors, could utilize private equity financing, but it including both shareholders and creditors. has many of the same problems of exces- As the evidence described in Section 3 sive investor power suppressing entrepre- shows, the benefits of such reform would be neurial initiative as does monopoly banking to expand financial markets, to facilitate (see, e.g., Myers, 1977; Burkart et al., external financing of new firms, to move 1997). Public equity financing, for which a away from concentrated ownership, to developed stock market is needed, has improve the efficiency of investment alloca- other advantages over private equity tion, and to facilitate private restructuring financing. It allows the buyers of equity to of financial claims in a crisis. diversify. It offers the initial equity holders, So what, if anything, can be done to such as venture capitalists, an attractive achieve these goals, and what are the exit option through the public equity mar- obstacles? To organize this discussion, we kets. Last but not least, it allows firms to follow Coffee (1999) and Gilson (2000) in time their equity issues to take advantage drawing a distinction between legal and of favorable investor sentiment toward functional convergence. Legal convergence their industry, or toward the market as a refers to the changes in rules and enforce- whole. Such sentiment may play a benefi- ment mechanisms toward some successful cial role when shareholders are skeptical standard. To converge to effective investor about the likelihood of getting back a protection in this way, most countries return on their money. Indeed, Keynes require extensive legal, regulatory, and 104 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS judicial reform. Alternatively, functional reformers see as protection of investors, convergence refers to more decentralized, the founding families call “expropriation of market-based changes, which do not entrepreneurs”. No wonder,then, that in all require legal reform per se, but still bring countries – from Latin America to Asia to more firms and assets under the umbrella Europe – the families have opposed legal of effective legal protection of investors. reform. We discuss these paths of reform in turn. There is a further reason why the insid- For most countries, the improvement of ers in major firms oppose corporate gover- investor protection requires radical nance reform and the expansion of capital changes in the legal system. Securities, markets. As Mayer (1988) shows, existing company, and bankruptcy laws generally large firms typically finance their own need to be amended. The particular list of investment projects internally or through legal protections of investors studied by captive or closely connected banks. In fact, La Porta et al. (1998) is neither necessary La Porta et al. (1997) show that the lion’s nor sufficient for such reforms. There may share of credit in countries with poor cred- be significant complementarities between itor protection goes to the few largest various laws in protecting minority share- firms. These firms obtain the finance they holders: securities laws, for example, can need, the political influence that comes mandate disclosure of material information with the access to such finance, and the while company laws enable minority share- protection from competition that would holders to act on it. Moreover,the regulatory come if smaller firms could also raise and judicial mechanisms of enforcing external capital. When new entrepreneurs shareholders and creditor rights would have good projects, they often have to come need to be radically improved. In fact, the to the existing firms for capital. Poor cor- evidence on the importance of the histori- porate governance delivers the insiders cally determined legal origin in shaping secure finance, secure politics and secure investor rights – which could be thought of markets. They have an interest in keeping as a proxy for the law’s general stance the system as is. toward outside investors – suggests at least Consistent with the dominance of inter- tentatively that many rules need to be est group politics, successful reforms have changed simultaneously to bring a country occurred only when the special interests with poor investor protection up to best could be destroyed or appeased. In this practice. respect, corporate governance reform is no The political opposition to such change different from most other reforms in devel- has proved intense. Governments are often oping or industrialized countries (see, e.g., reluctant to introduce laws that surrender Hirschman, 1963; Shleifer and Treisman, to the financiers the regulatory control they 2000). But examples of significant legal currently have over large corporations. reform of corporate governance do exist. Important objections to reform also come Ramseyer and Nakazato (1999) describe from the families that control large corpo- legal reform in Japan after World War II, rations. From the point of view of these when General McArthur, assisted by attor- families, an improvement in the rights of neys from Chicago and an occupying army, outside investors is first and foremost a introduced an Illinois-based company law. reduction in the value of control due to the Another example is securities markets regu- deterioration of expropriation opportuni- lation in the United States in 1933–1934, ties. The total value of these firms may introduced in the middle of the Great increase as a result of legal reform, as Depression, which substantially increased expropriation declines and investors corporate disclosure. A third example is finance new projects on more attractive some streamlining of bankruptcy procedures terms; still, the first order effect is a tax on in East Asia following the crisis of 1997. the insiders for the benefit of minority Although such opportunities for corpo- shareholders and creditors. What the rate governance reform do arise, they often INVESTOR PROTECTION AND CORPORATE GOVERNANCE 105 have been wasted, in part because of a lack intermediaries. Thus the accounting of appreciation of the need to protect profession, once it recognized the increased investors. Recent research points to some demand for its services, became an crucial principles of investor protection independent private force in assuring the that reforms need to focus on. compliance with disclosure regulations. As The first such principle is that legal rules a consequence, a small Commission was do matter. It is not just the stance of the able to regulate a huge market with law or the political sentiment of the day relatively few resources. The principle of that shapes financial markets. One illustra- recruiting private intermediaries into the tion of this principle, described by Johnson enforcement of securities regulations has (1999), is the Neuer Markt in Germany, a since been followed by a number of segment of the Frankfurt Stock Exchange countries, including Germany and Poland. created especially for listing new firms. A third and related principle of Because the Neuer Markt operates in successful reform, stressed by Glaeser et al. Germany, the corporate law, the securities (2001), is that government regulation of law, and other basic laws and regulations financial markets may be useful when court that are applied to the companies listing enforcement of private contracts or laws there are the general German rules. The cannot be relied upon. An example of how politics are German as well. As part of a regulation can work when judicial enforce- private contract with firms wishing to list ment is limited comes from the securities on the Neuer Markt, the Deutsche Bourse – law reform in Poland and the Czech which operates the Frankfurt Stock Republic, two transition economies whose Exchange – has mandated that these firms judiciaries in the early 1990s were gener- must comply with international accounting ally viewed as ineffective. At that time the standards and agree to greater disclosure Polish government introduced a tough than that required of already listed firms. securities law focused on shareholder The new listing venue, with its greater protection. Like the U.S. securities law, the restrictions on the entrepreneurs, has Polish regulations focused on significant sharply accelerated the pace of initial pub- disclosure by new issuers and already listed lic offerings in Germany. At the same time, firms, as well as on licensing and close the captains of German industry have administrative oversight of intermediaries. accepted it because their firms were not The law also provided for a creation of a directly affected.This points to one possible powerful SEC with significant enforcement strategy of overcoming political opposition powers that did not require reliance on to reform. courts. This reform was followed by a A second principle is that good legal remarkable development of the Polish rules are the ones that a country can stock market, with both new and already enforce. The strategy for reform is not to listed companies raising equity in the create an ideal set of rules and then see market. how well they can be enforced, but rather to By contrast, the Czech government chose enact the rules that can be enforced within neither to introduce tough securities laws the existing structure. One example of the nor to create a powerful market regulator success of such a policy is the U.S. securi- at the time of privatization. Perhaps as a ties legislation of 1933–1934, described by consequence, the Czech markets have been Landis (1938) and McCraw (1984). This plagued by massive expropriation of minority legislation placed much of the responsibil- shareholders – the so-called “tunneling” of ity for accurate corporate accounting and assets from both firms and mutual funds. disclosure on intermediaries, and focused In contrast to the Polish market, the Czech the regulatory oversight by the Securities market stagnated, with hundreds of compa- and Exchange Commission (SEC) on these nies getting delisted and virtually no public relatively few intermediaries. The SEC equity financing by firms (see Coffee, also emphasized self-regulation by the 1999; Pistor, 1999; Glaeser et al., 2001). 106 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS The comparison of Poland to the Czech A related and increasingly important Republic is especially instructive because mechanism of opting into a more protective the two countries share roughly similar legal regime is being acquired by a firm incomes, economic policies, and quality of already operating in such a regime. When a judiciaries. Under these circumstances, reg- British firm fully acquires a Swedish firm, ulation of the stock market and listed firms the possibilities for legal expropriation of in Poland, with its focus on investor pro- investors diminish. Because the controlling tection, appeared to play a beneficial role. shareholders of the Swedish company are The successful regulations of the U.S. compensated in such a friendly deal for securities markets, the Polish financial the lost private benefits of control, they are markets, and the Neuer Markt in Germany more likely to go along. By replacing the share a common element: the extensive and wasteful expropriation with publicly shared mandatory disclosure of financial informa- profits and dividends, such acquisitions tion by the issuers, the accuracy of which is enhance efficiency. enforced by tightly regulated financial inter- It is important to recognize the limita- mediaries. Although such disclosure is not tions of functional convergence, particularly sufficient by itself without the right of the in the area of creditor rights. Assets shareholders to act on it, it does appear to located in particular countries generally be a key element of shareholder protection. remain under the jurisdiction of these With the legal reform slow and halting in countries’ laws. Without bankruptcy most countries, “functional convergence” reform, opt-in mechanisms are unlikely to may play a role in improving investor address the legal problems faced by credi- protection. The liberalization of capital tors. Thus, despite the benefits of opting markets in many countries has increased into the more protective legal regime for not only the flow of foreign investment into external finance, this mechanism is unlikely them, as Henry (2000) and Stulz (1999) to fully replace bona fide legal reform. document, but also the economic and political pressure to create financial instruments acceptable to foreign investors. 6. CONCLUSION These pressures give rise to several forms of functional convergence. When contracts This paper describes the legal protection of are enforced well, companies in unprotec- investors as a potentially useful way of tive legal regimes can offer their investors thinking about corporate governance. customized contracts such as corporate Strong investor protection may be a charters with greater rights than the law particularly important manifestation of the generally provides. This strategy relies on greater security of property rights against perhaps a greater enforcement capacity of political interference in some countries. courts than is warranted, and also ignores Empirically, strong investor protection is the public good benefit of standard rules. associated with effective corporate A more promising approach is for compa- governance, as reflected in valuable and nies to opt into the more investor friendly broad financial markets, dispersed owner- legal regimes. One way of doing this is to ship of shares, and efficient allocation of list a company’s securities on an exchange capital across firms. Using investor protection that protects minority shareholders as the starting point appears to be a more through disclosure or other means. In fruitful way to describe differences in fact, this is done by many companies that corporate governance regimes across coun- list their shares as American Depositary tries than some of the more customary Receipts (ADRs) in the U.S. But such list- classifications such as bank- or marketing imposes only limited constraints on the centeredness. insiders: although it improves disclosure, it An important implication of this typically does not give minority shareholders approach is that leaving financial markets many effective rights. alone is not a good way to encourage them. INVESTOR PROTECTION AND CORPORATE GOVERNANCE 107 Financial markets need some protection of the law remains a crucial channel through which outside investors, whether by courts, gov- politics affects corporate governance. ernment agencies, or market participants 4 The evidence also reveals that control is themselves. Improving such protection is a valued, and specifically that voting premiums difficult task. In part, the nature of investor increase as shareholder protection deteriorates (see, for example, Modigliani and Perotti, 1998; protection, and more generally of regula- Nenova, 1999; Zingales, 1994). tion of financial markets, is deeply rooted 5 In addition, La Porta et al. (2000) show in the legal structure of each country and that better minority shareholder protection is in the origin of its laws. Marginal reform associated with higher dividend pay-outs in a may not successfully achieve the reformer’s cross-section of firms from around the world. goals. In part, the existing corporate 6 Jensen (1989) expresses some early skepti- governance arrangements benefit both the cism about the Japanese financial system. politicians and the entrenched economic 7 Hellwig (1999) doubts that banks are so interests, including the families that man- powerful, even in the case of Germany. age the largest firms in most countries in 8 Levine et al. (2000) find that the La Porta et al. (1998) measure of creditor rights is the world. Corporate governance reform correlated with measures of financial intermedi- must circumvent the opposition by these aries development across countries, while their interests. Despite these difficulties, reform measure of shareholder rights is correlated with of investor protection is politically feasible stock market development. in some circumstances, and can bring significant benefits. It can take the form of opting into more protective legal regimes REFERENCES or introducing more radical changes in the legal structure. The integration of world Allen, F., Gale, D., 2000. Comparing Financial capital markets makes such reforms Systems. MIT Press, Cambridge, MA. more likely today than they have been in Aoki, M., Patrick, H., 1993.The Japanese Main decades. Bank System: Its Relevance for Developing and Transforming Economies. Oxford University Press, New York. NOTES Barca, F., 1995. On corporate governance in Italy: issues, facts, and agency. Unpublished * We are grateful to Nicholas Barberis, working paper. Bank of Italy, Rome. Simeon Djankov, Oliver Hart, Michael Jensen, Bebchuk, L., 1999.The rent protection theory of Simon Johnson, Ross Levine, and Daniel corporate ownership and control. Wolfenzon for helpful comments, and also to the Unpublished working paper. Harvard Law NSF for financial support of this research. School, Cambridge, MA. † Corresponding author. Tel.: 617–495–5046; Beck, T., Levine, R., Loayza, N., 2000. Finance fax: 617–496–1708. E-mail address: ashleifer@ harvard.edu (A. Shleifer). and the sources of growth. Journal of 1 Socialist countries had a legal tradition Financial Economics 58, 261–300. based on Soviet law, but because the laws of Bennedsen, M., Wolfenzon, D., 2000. The bal- these countries are changing rapidly during the ance of power in closely held corporations. transition out of socialism, La Porta et al. Journal of Financial Economics 58, 113–139. (1998) do not consider them. Berglof, E., von Thadden, L., 1999. The changing 2 According to Cameron (1961), France had corporate governance paradigm: implications a lively stock market in the nineteenth century. for transition and developing countries. Nearly all firms listed on it, however, benefitted Unpublished working paper. Stockholm from government concessions, investment, own- Institute of Transition Economics, ership, subsidies, protection, and often outright guarantees to investors. Stockholm, Sweden. 3 Berglof and von Thadden (1999) and Rajan Burkart, M., Gromb, D., Panunzi, F., 1997. and Zingales (1999) argue that political factors Large Shareholders, monitoring, and fiduci- affect corporate governance through channels ary duty. Quarterly Journal of Economics other than the law itself. This may be true, but 112, 693–728. 108 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Cameron, R., 1961. France and the Economic Unpublished working paper. Stanford Development of Europe. Princeton University University, Stanford, CA. Press, Princeton, NJ. Glaeser, E., Johnson, S., Shleifer, A., 2001. Carlin, W., Mayer,C., 1999. Finance, investment Coase versus the Coasians. Quarterly Journal and growth. Unpublished working paper. of Economics, forthcoming. University College, London. Gorton, G., Schmid, F.,2000. Universal banking Claessens, S., Djankov, S., Lang, L., 2000. The and the performance of German firms. separation of ownership and control in East Journal of Financial Economics 58, 29–80. Asian corporations. Journal of Financial Grossman, S., Hart, O., 1988. One-share-one- Economics 58, 81–112. vote and the market for corporate control. Claessens, S., Djankov, S., Fan, J., Lang, L., Journal of Financial Economics 20, 1999. Expropriation of minority shareholders 175–202. in East Asia. Unpublished working paper.The Harris, M., Raviv, A., 1988. Corporate gover- World Bank, Washington, DC. nance: voting rights and majority rules. Coase, R., 1961. The problem of social cost. Journal of Financial Economics 20, Journal of Law and Economics 3, 1–44. 203–236. Coffee, J., 1999. The future as history: the Hart, O., 1995. Firms, Contracts, and Financial prospects for global convergence in corporate Structure. Oxford University Press, London. governance and its implications. Hellwig, M., 1999. On the economics and Northwestern Law Review 93, 631–707. politics of corporate finance and corporate Coffee, J., 2000. Privatization and corporate control. Unpublished working paper. governance: the lessons from securities mar- University of Mannheim, Mannheim. ket failure. Working paper no. 158. Columbia Henry, P., 2000. Do stock market liberalizations Law School, New York. cause investment booms? Journal of David, R., Brierley, J., 1985. Major Legal Financial Economics 58, 301–334. Systems in the World Today. Stevens and Hirschman, A., 1963. Journeys Toward Sons, London. Progress. The Twentieth Century Fund, New Demirguc-Kunt, A., Maksimovic, V., 1998. Law, York. finance, and firm growth. Journal of Finance Hoshi, T., Kashyap, A., Scharfstein, D., 1991. 53, 2107–2139. Corporate structure, liquidity, and invest- Easterbrook, F., Fischel, D., 1991. The ment: evidence from Japanese industrial Economic Structure of Corporate Law. groups. Quarterly Journal of Economics 106, Harvard University Press, Cambridge, MA. 33–60. Edwards, J., Fischer, K., 1994. Banks, Finance Jensen, M., 1989. Eclipse of the modern and Investment in West Germany Since 1970. corporation. Harvard Business Review 67, Cambridge University Press, Cambridge, UK. 61–74. European Corporate Governance Network Jensen, M., Meckling, W., 1976. Theory of the (ECGN). 1997. The Separation of Ownership firm: managerial behavior, agency costs, and and Control: A Survey of 7 European ownership structure. Journal of Financial Countries Preliminary Report to the Economics 3, 305–360. European Commission. Volumes 1–4. Johnson, S., 1999. Does investor protection Brussels, European Corporate Governance matter? Evidence from Germany’s Neuer Network. Markt. Unpublished working paper. MIT Finer, S., 1997. The History of Government, Press, Cambridge, MA. Vol. III. Cambridge University Press, Johnson, S., Boone, P.,Breach, A., Friedman, E., Cambridge, UK. 2000a. Corporate governance in the Asian Gerschenkron, A., 1962. Economic financial crisis, Journal of Financial Backwardness in Historical Perspective. Economics 58, 141–186. Belknap Press of Harvard University Press, Johnson, S., La Porta, R., Lopez-de-Silanes, F., Cambridge, MA. Shleifer, A., 2000b. Tunneling. American Gilson, R., 2000. Globalizing corporate Economic Review Papers and Proceedings governance: convergence of form or function. 90, 22–27. INVESTOR PROTECTION AND CORPORATE GOVERNANCE 109 Kang, J., Stulz, R., 1998. Do banking shocks Modigliani, F., Perotti, E., 1998. Security versus affect borrowing firm performance? An bank finance: the importance of a proper analysis of the Japanese experience. enforcement of legal rules. Unpublished Unpublished working paper. Ohio State working paper. MIT Press, Cambridge, MA. University, Columbus, OH. Morck, R., Nakamura, M., 1999. Banks and Keynes, J., 1931. An economic analysis of corporate control in Japan. Journal of unemployment. In Collected Writings,Vol. XII. Finance 54, 319–340. Macmillan, London. Morck, R., Yeung, B., Yu, W., 2000. The King, R., Levine, R., 1993. Finance and growth: information content of stock markets: why do Schumpeter might be right. Quarterly Journal emerging markets have synchronous price of Economics 108, 717–738. movements? Journal of Financial Economics Landis, J., 1938. The Administrative Process. 58, 215–260. Yale University Press, New Haven, CT. Myers, S., 1977. Determinants of corporate La Porta, R., Lopez-de-Silanes, F., Shleifer, A., borrowing. Journal of Financial Economics 5, 1999. Corporate ownership around the world. 147–175. Journal of Finance 54, 471–517. Nenova, T., 1999. The value of a corporate vote La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and private benefits: a cross-country analysis. Vishny, R., 1997. Legal determinants of exter- Unpublished working paper. Harvard nal finance. Journal of Finance 52, University, Cambridge, MA. 1131–1150. Pistor,K., 1999. Law as a determinant of equity La Porta, R., Lopez-de-Silanes, F., Shleifer, A., market development. Unpublished working Vishny, R., 1998. Law and finance. Journal of paper. Harvard University, Cambridge, MA. Political Economy 106, 1113–1155. Porter, M., 1992. Capital disadvantage: La Porta, R., Lopez-de-Silanes, F., Shleifer, A., America’s falling capital investment system. Vishny, R., 1999a.The quality of government. Harvard Business Review 46, 65–72. Journal of Law, Economics and Organization Rajan, R., Zingales, L., 1998. Financial depend- 15, 222–279. ence and growth. American Economic Review La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 88, 559–586. Vishny, R., 1999b. Investor protection and Rajan, R., Zingales, L., 1999. The politics of corporate valuation. NBER Working Paper financial development. Unpublished working 7403. National Bureau of Economic paper. University of Chicago, Chicago, IL. Research, Cambridge, MA. Ramseyer, M., Nakazato, M., 1999. Japanese La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Law: an Economic Approach. University of Vishny, R., 2000. Agency problems and divi- Chicago Press, Chicago, IL. dend policies around the world. Journal of Sapienza, P., 1999.The effects of bank mergers Finance 55, 1–33. on loan contracts. Unpublished working Levine, R., Zervos, S., 1998. Stock markets, paper. Northwestern University, Chicago, IL. banks and economic growth. American Shleifer,A.,Treisman, D., 2000.Without a Map: Economic Review 88, 537–558. Political Tactics and Economic Reform in Levine, R., Loayza, N., Beck,T., 2000. Financial Russia. MIT Press, Cambridge, MA. intermediary development and economic Shleifer, A., Vishny, R., 1997. A survey of cor- growth: causality and causes. Journal of porate governance. Journal of Finance 52, Monetary Economics, 46, 31–77. 737–783. Mayer, C., 1988. New issues in corporate Stigler,G., 1964. Public regulation of the securi- finance. European Economic Review 32, ties market. Journal of Business 37, 117–142. 1167–1188. Stulz, R., 1999. International Portfolio Flows McCraw, T.K., 1984. Prophets of Regulation. and Securities Markets.Working Paper 99–3, The Belknap Press of Harvard University Fisher College of Business. The Ohio State Press, Cambridge, MA. University, Columbus, OH. Modigliani, F., Miller,M., 1958.The cost of cap- Weinstein, D., Yafeh,Y., 1998. On the costs of a ital, corporation finance, and the theory of bank-centered financial system: evidence investment. American Economic Review 48, from the main bank relations in Japan. 261–297. Journal of Finance 53, 635–672. 110 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Wolfenzon, D., 1999. A theory of pyramidal Zingales, L., 1994. The value of the voting structures. Unpublished working paper. right: a study of the Milan stock exchange. Harvard University, Cambridge, MA. The Review of Financial Studies 7, 125–148. Wurgler, J., 2000. Financial markets and the Zingales, L., 1995. Inside ownership and the allocation of capital. Journal of Financial decision to go public. Review of Economic Economics 58, this issue. Studies 62, 425–448. Chapter 4

James S. Ang, Rebel Cole, and James Wuh Lin* AGENCY COSTS AND OWNERSHIP STRUCTURE

Source: Journal of Finance, 55(1) (2000): 81–106.

ABSTRACT We provide measures of absolute and relative equity agency costs for corporations under different ownership and management structures. Our base case is Jensen and Meckling’s (1976) zero agency-cost firm, where the manager is the firm’s sole shareholder.We utilize a sample of 1,708 small corporations from the FRB/NSSBF database and find that agency costs (i) are significantly higher when an outsider rather than an insider manages the firm; (ii) are inversely related to the manager’s ownership share; (iii) increase with the number of nonmanager shareholders, and (iv) to a lesser extent, are lower with greater monitoring by banks.

THE SOCIAL AND PRIVATE COSTS OF AN and Meckling agency theory, the zero AGENT’S ACTIONS due to incomplete align- agency-cost base case is, by definition, the ment of the agent’s and owner’s interests firm owned solely by a single owner-manager. were brought to attention by the seminal When management owns less than 100 contributions of Jensen and Meckling percent of the firm’s equity, shareholders (1976) on agency costs. Agency theory has incur agency costs resulting from manage- also brought the roles of managerial ment’s shirking and perquisite consumption. decision rights and various external and Because of limitations imposed by personal internal monitoring and bonding mechanisms wealth constraints, exchange regulations to the forefront of theoretical discussions on the minimum numbers of shareholders, and empirical research. Great strides have and other considerations, no publicly traded been made in demonstrating empirically firm is entirely owned by management. the role of agency costs in financial Thus, Jensen and Meckling’s zero agency decisions, such as in explaining the choices cost base case cannot be found among the of capital structure, maturity structure, usual sample of publicly traded firms for dividend policy, and executive compensation. which information is readily available. The However, the actual measurement of the absence of information about sole owner- principal variable of interest, agency costs, manager firms explains why agency costs in both absolute and relative terms, has are often inferred but not directly meas- lagged behind. ured in the empirical finance literature. To measure absolute agency costs, a zero No-agency-cost base case firms, how- agency-cost base case must be observed to ever, can be found among non–publicly serve as the reference point of comparison traded firms. Until recently, data on for all other cases of ownership and man- non–publicly traded firms, which tend to be agement structures. In the original Jensen much smaller than their publicly traded 112 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS counterparts, have been sparse. In 1997, specific operational knowledge on the part the Federal Reserve Board released its of nonmanaging shareholders, and the lack National Survey of Small Business of an external market for shares, however, Finances (NSSBF), which collected data may offset the presence of dominant from a nationally representative sample of shareholders. Additionally, heavy reliance small businesses. Data from the NSSBF of the non–publicly traded firms on bank enable us to analyze the relationship financing could give banks a special role in between agency costs and ownership struc- delegated monitoring on behalf of other ture because the survey provides financial shareholders. Thus, it would seem that data on a group of firms whose manage- determination of the size of agency costs ment owns 100 percent of equity. These for these firms is an empirical issue. firms enable us to estimate the expected Our results provide direct confirmation expense for the no-outside-equity agency-cost of the predictions made by Jensen and base case. Furthermore, the database Meckling (1976). Agency costs are indeed includes firms with a wide range of ownership higher among firms that are not 100 percent and manager/owner structures, including owned by their managers, and these costs firms owned by two individuals as well as increase as the equity share of the owner- firms managed by outsiders with no equity manager declines. Hence, agency costs stake. As a consequence, small firms increase with a reduction in managerial appear well suited for a study of equity- ownership, as predicted by Jensen and related agency costs. Meckling.These results hold true after con- We use two alternative measures of trolling for differences across industries, agency costs. The first is direct agency the effects of economies of scale, and dif- costs, calculated as the difference in dollar ferences in capital structure. We also find expenses between a firm with a certain some evidence that delegated monitoring of ownership and management structure and small firms by banks reduces agency costs. the no-agency-cost base case firm. This The paper is organized as follows. In measure captures excessive expenses Section I, we discuss the nature of equity including perk consumption. To facilitate agency costs in various ownership struc- cross-sectional comparisons, we standard- tures and explain the broad outline of our ize expenses by annual sales. Our second empirical model. In Section II, we provide measure of agency costs is a proxy for the a description of the data.We present results loss in revenues attributable to inefficient and analysis in Section III, followed by a asset utilization, which can result from summary and conclusions in Section IV. poor investment decisions (e.g., investing in negative net-present-value assets) or from management’s shirking (e.g., exerting too I. AGENCY COSTS AMONG SMALL little effort to help generate revenue). This BUSINESSES second measure of agency costs is calculated as the ratio of annual sales to total When compared to publicly traded firms, assets, an efficiency ratio. We can then small businesses come closest to the type measure agency costs as the difference in of firms depicted in the stylized theoretical the efficiency ratio, or,equivalently, the dol- model of agency costs developed by Jensen lar revenues lost, between a firm whose and Meckling (1976). At one extreme of manager is the sole equity owner and a ownership and management structures are firm whose manager owns less than 100 firms whose managers own 100 percent of percent of equity. the firm. These firms, by their definition, Monitoring of managers’ expenditures on have no agency costs. At the other extreme perquisites and other personal consumption are firms whose managers are paid relies on the vigilance of the nonmanaging employees with no equity in the firm. In shareholders and/or related third parties, between are firms where the managers own such as the company’s bankers.The lack of some, but not all, of their firm’s equity. AGENCY COSTS AND OWNERSHIP STRUCTURE 113 Agency costs arise when the interests of perks, but only percent of each dollar in the firm’s managers are not aligned with firm profit, the manager who owns less those of the firm’s owner(s), and take the than 100 percent of the firm has the form of preference for on-the-job perks, incentive to consume perks rather than to shirking, and making self-interested and maximize the value of the firm to all share- entrenched decisions that reduce shareholders. At the extreme is the manager with holder wealth.The magnitude of these costs zero ownership ( 0), who gains 100 is limited by how well the owners and dele- percent of perquisite consumption, but zero gated third parties, such as banks, monitor percent of firm profits (in the case when the actions of the outside managers. salary is independent of firm performance). To illustrate, consider those firms where Aggregate expenditure on monitoring by a single owner controls 100 percent of the the nonmanaging shareholders decreases stock but hires an outsider to manage the as their individual ownership shares business. On the one hand, agency costs decline. This is due to the well-known free- may be small because the sole owner can rider problem in spending for quasi-public internalize all monitoring costs and has the goods, such as monitoring effort. Each right to hire and fire the manager. More monitoring shareholder, with ownership i specifically, such an owner incurs 100 percent must incur 100 percent of the monitoring of the monitoring costs and receives 100 costs, but realizes only i percent of the percent of the resulting benefits. On the monitoring benefits (in the form of reduced other hand, the sole owner may not be able agency costs). A nonmonitoring share- to monitor perfectly for the same reasons holder,however,enjoys the full benefits of a that he or she hired an outside manager, monitoring shareholder’s activity without such as lack of time or ability. Owners incurring any monitoring cost. Thus, as the of small firms typically lack financial number of non-manager shareholders sophistication, and may not be capable of increases, aggregate expenditure on performing random audits or fully under- monitoring declines, and the magnitude of standing the operating or financial results. owner-manager agency-cost problems Consequently, these firms incur residual increases. Offsetting this relationship are agency costs. If these costs are significant, concerns among shareholders about an they must reflect a failure of the owner’s increase in the probability that the firm will monitoring activities. Potential explana- be unable to pay off bank debt or secure tions for this failure are lax monitoring by future financing from the same or new the owners and the lack of an adequate investors, which may produce some monitoring technology available for the restraint in agency behavior. However, as owners. In this case, the separation of the noted by Williams (1987), these counter- management function (initiation and vailing forces to agency behavior are implementation) versus the control func- expected to decline in effectiveness when tion by nonmanaging owners/shareholders the firm is not in imminent danger of (ratification and monitoring), as suggested insolvency. by Fama and Jensen (1983a, 1983b), To summarize, against the null hypothe- may not be complete or effective. Thus, sis that agency costs are independent of the residual agency costs are still expected in a ownership and control structure,1 we sole owner firm when the manager is an postulate the following hypotheses derived outsider. from agency theory when compared to the Agency costs attributable to the diver- base case: (i) agency costs are higher at gence of interests vary inversely with the firms whose managers own none of the manager’s ownership stake. As the number firm’s equity, (ii) agency costs are an of shareholders increases from one, the inverse function of the managers’ owner- ownership of the owner/manager falls to , ship stake, and (iii) agency costs are an where 0 1. Because the manager increasing function of the number of gains 100 percent of each dollar spent on nonmanager shareholders. 114 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS II. DATA financial statements of proprietorships, which typically commingle personal Our empirical approach utilizes two and business funds. We eliminate partner- fundamental assumptions about agency ships and S-corporations because, unlike costs: (1) A firm managed by a 100 percent C-corporations, they are not subject to owner incurs zero agency costs and, (2) corporate taxation, and this may lead owner- agency costs can be measured as the managers to take compensation in the form difference in the efficiency of an imper- of partner distributions or dividends rather fectly aligned firm and the efficiency of a than salary expense because there is no perfectly aligned firm. To operationalize double taxation of such earnings at the firm this approach for measuring agency costs, level. By focusing solely on C-corporations, we need certain data inputs: (i) data on we avoid the complications of comparing firm efficiency measures; (ii) data on firm operating expenses across organizational ownership structure, including a set of forms.This restriction on the NSSBF database firms that are 100 percent owned by man- yields an analysis sample of 1,708 firms.3 agers; and (iii) data on control variables, including firm size, characteristics, and A. Agency costs monitoring technology. Of these data requirements, the most To measure agency costs of the firm, we demanding in terms of availability is item use two alternative efficiency ratios that (ii) because sole-ownership firms typically frequently appear in the accounting and are not publicly listed, and because financial economics literature: the expense financial information on U.S. private firms ratio, which is operating expense scaled by usually is not available to the public. The annual sales,4 and the asset utilization Federal Reserve Board’s National Survey ratio, which is annual sales divided by total of Small Business Finances (NSSBF), for- assets. The first ratio is a measure of how tunately, does provide financial information effectively the firm’s management controls about privately held firms, including their operating costs, including excessive ownership structure, and does include a set perquisite consumption, and other direct of firms entirely owned by managers. agency costs. More precisely, the difference Consequently, we use data from the in the ratios of a firm with a certain own- NSSBF to measure agency costs.2 ership and management structure and the The NSSBF is a survey conducted by the no-agency-cost base case firm, multiplied Federal Reserve Board to gather information by the assets of the former,gives the excess about small businesses, which have largely agency cost related expense in dollars. been ignored in the academic literature The second ratio is a measure of how because of the limited availability of data. effectively the firm’s management deploys The survey collected detailed information its assets. In contrast to the expense ratio, from a sample of 4,637 firms that is agency costs are inversely related to the broadly representative of approximately sales-to-asset ratio. A firm whose sales-to- 5 million small nonfarm, nonfinancial asset ratio is lower than the base case firm businesses operating in the United States experiences positive agency cost. These as of year-end 1992. Cole and Wolken costs arise because the manager acts in (1995) provide detailed information about some or all of the following ways: makes the data available from NSSBF. poor investment decisions, exerts insuffi- For this study, we limit our analysis to cient effort, resulting in lower revenues; small C-corporations, collecting informa- consumes executive perquisites, so that the tion on the governance structure, manage- firm purchases unproductive assets, such as ment alignment, extent of shareholder and excessively fancy office space, office fur- external monitoring, size, and financial nishing, automobiles, and resort properties. information. We focus on corporations to These efficiency ratios are not measured minimize problems associated with the without error. Sources of measurement AGENCY COSTS AND OWNERSHIP STRUCTURE 115 error include differences in the accounting family controls the firm, the controlling methods chosen with respect to the recog- family also fulfills the monitoring role that nition and timing of revenues and costs, large blockholders perform at publicly poor record-keeping typical of small busi- traded corporations. Due to more diffused nesses, and the tendency of small-business ownership among older businesses with owners to exercise flexibility with respect to larger families, however, monitoring by certain cost items. For example, owners family members whose interests may not may raise/lower expenses, including their always be aligned should be less effective own pay, when profits are high/low. than monitoring by a sole owner. Fortunately, these items are sources of Agency costs should increase with the random measurement errors that may be number of nonmanager shareholders. As reduced with a larger sample across firms the number of shareholders increases, the in different industries and age. free-rider problem reduces the incentives for limited-liability shareholders to monitor. With less monitoring, agency costs B. Ownership structure increase. Hence, we hypothesize that the The corporate form of organization, with expense and asset-utilization ratios the limited-liability provision that makes it should be positively and negatively related more efficient for risk-sharing than propri- to the natural logarithm of one plus the etorships or partnerships, allows the firm number of nonmanaging shareholders, 7 to expand and raise funds from a large respectively. number of investors.5 Thus, it has a richer Finally, agency costs should be higher at set of ownership and management struc- firms managed by an outsider. This rela- tures. The NSSBF provides four variables tionship follows directly from the agency that we use to capture various aspects of the theory of Jensen and Meckling (1976). As ownership structure of small-business noted above, this is the extreme case where corporations: (i) the ownership share of the the manager gains 100 percent of primary owner, (ii) an indicator for firms perquisite consumption, but little of the where a single family controls more than firm’s profits. 50 percent of the firm’s shares, (iii) the 6 number of nonmanager shareholders, and, C. External monitoring by banks (iv) an indicator for firms managed by a shareholder rather than an outsider. Banks play a pivotal role in small business According to theory, agency costs should financing because they are the major be inversely related to the ownership share source of external funds for such firms. of the primary owner. For a primary owner Cole,Wolken, and Woodburn (1996) report who is also the firm’s manager, the incen- that more than 60 percent of the dollar tive to consume perquisites declines as his amount of small business credit outstanding ownership share rises, because his share of takes the form of bank loans. Petersen and the firm’s profits rises with ownership while Rajan (1994), Berger and Udell (1995), his benefits from perquisite consumption and Cole (1998) argue and present are constant. For a primary owner who evidence that firm-creditor relationships employs an outside manager, the gains generate valuable information about from monitoring in the form of reduced borrower quality. agency costs increase with his ownership Because banks generally require a firm’s stake. Here, the primary owner fulfills the managers to report results honestly and to monitoring role that large blockholders run the business efficiently with profit, perform at publicly traded corporations. bank monitoring complements shareholder Agency costs should be lower at firms monitoring of managers, indirectly reduc- where a single family controls more than ing owner-manager agency costs.That is, by 50 percent of the firm’s equity. At a small, incurring monitoring costs to safeguard closely held corporation where a single their loans, banks lead firms to operate 116 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS

Professional Services

Business Services

Insurance and Real Estate

Retail

Wholesale

Transportation

Other Manufacturing

Primary Manufacturing

Construction

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Operating Expense-to-Sales Ratio

Figure 4.1 Operating expense-to-sales ratio by one-digit SIC for a sample of 1,708 small corporations. more efficiently by better utilizing assets greater perceived likelihood of the firm and moderating perquisite consumption in switching its banking business between order to improve the firm’s reported finan- banks. cial performance to the bank. Thus, lower The bank’s ability to monitor is proxied priority claimants, such as outside share- by the length of a firm’s relationship with holders, should realize a positive externality its primary bank. A longer relationship from bank monitoring, in the form of lower enables the bank to generate information agency costs. Additionally, local bankers’ about the firm that is useful in deciding its ability to acquire knowledge concerning the creditworthiness (Diamond (1984)). Both firms from various local sources, such as Petersen and Rajan (1994) and Cole churches, social gatherings, and interactions (1998) find that longer relationships with the firm’s customers and suppliers, improve the availability of credit to small makes them especially good monitors. We firms while Berger and Udell (1995) find use two variables to represent bankers’ that longer relationships improve the terms incentive, cost, and ability to monitor: the of credit available to small firms. number of banks used by the firm and The bank’s incentive to monitor is prox- the length of the firm’s longest banking ied by the firm’s debt-to-asset ratio. relationship.8 Because our sample consists entirely of The bank’s cost of monitoring is proxied small businesses, virtually all of the firm’s by the number of banks from which the debt is private rather than public, and the firm obtains financial services. The incen- majority of this debt is in the form of bank tive for each bank to monitor may decrease loans. As leverage increases, so does the as the number of banks with which the firm risk of default by the firm, hence the incen- deals increases (Diamond (1984)). Part of tive for the lender to monitor the firm.While the reduced incentive to monitor is due to a the primary purpose of this monitoring is to form of lenders’ free-rider problems, and prevent risk-shifting by shareholders to part is due to the shorter expected length debtholders, increased monitoring should of banking relationships when there is a also inhibit excessive perquisite consumption AGENCY COSTS AND OWNERSHIP STRUCTURE 117

Professional Services

Business Services

Insurance and Real Estate

Retail

Wholesale

Transportation

Other Manufacturing

Primary Manufacturing

Construction

01234567 Sales-to-Assets Ratio

Figure 4.2 Sales-to-asset ratio by one-digit SIC for a sample of 1,708 small corporations. by managers. (Most of the sample firms’ of 0.65 for finance and real estate and nonbank debt is in the form of loans from professional services. Figure 4.2 shows the finance companies and other nonbank ratio of annual sales to total assets by one- private lenders, who also have greater digit SIC. This efficiency ratio ranges from incentive to monitor the firm as leverage 3.6 for manufacturing to 6.2 for professional increases.) services. Hence, these figures underscore the importance of controlling for differences D. Control variables across industries in our analysis of agency costs. We do this by including a set of 35 We realize that the length of banking dummy variables, one for each two-digit relationship variable may be correlated SIC that accounts for more than one with firm age, which in turn could be percent of our sample of firms. related to a firm’s efficiency. Due to the Small firms such as those surveyed by effects of learning curve and survival bias, the NSSBF seem likely to realize scale older firms are likely to be more efficient economies in operating expenses (e.g., than younger ones and, especially, than overhead items). Thus, there is a need to start-up firms. Hence, we include firm control for firm size. This adjustment is age as a control variable in all our tests especially important for comparisons of involving the variable measuring the operating expenses across firms where the length of the firm’s relationship with its difference in average size is of several primary bank. orders of magnitude, as it is with the small Both of our efficiency ratios vary widely businesses in our sample. Figure 4.3 across industries because of the varying confirms this, showing that the operating- importance of inventory and fixed assets. expense-to-sales ratio declines monotonically Figure 4.1 shows the ratio of operating by sales quartile, decreasing from 0.56 for expenses to sales by one-digit SIC. These the smallest quartile to 0.38 for the largest ratios vary from a low of 0.39 for quartile. If we regress the expense-to-sales construction and manufacturing to a high ratio against annual sales, we find a 118 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS

0.6 0.5 0.4 0.3 0.2 0.1 0 $0.00–$0.35 $0.35–$1.50 $1.50–$5.20 > $5.20 Sales Quartile ($Millions)

Figure 4.3 Operating expense-to-sales ratio by sales quartile for a sample of 1,708 small corporations. negative relationship that is statistically types of managers: owners versus outsiders. significant at better than the 1 percent level Panel A shows results when agency costs (t 6.9). are measured by the ratio of operating It is not clear,however,that efficiency in expenses to annual sales; Panel B shows scale economies is realized as measured by results when agency costs are measured by the ratio of sales to assets, where both the the ratio of annual sales to total assets. It numerator and denominator are popular is important to note here and in all subse- alternative measures of size. Indeed, quent analyses that the expected signs for Figure 4.4 shows that the ratio of sales to the expense ratio and the asset utilization assets is higher for the two middle sales ratio are opposite to each other. Higher quartiles than for either the largest or sales-to-assets ratios are associated with smallest quartile, suggesting, if any, a greater efficiency and lower agency costs, quadratic relationship. When we regress whereas higher expense-to-sales ratios are the sales-to-asset ratio against sales we associated with less efficiency and higher find a positive but statistically insignifi- agency costs. cant relationship (t 0.18). Similar results are obtained when the sales-to- asset ratio is regressed against the natural A.1. Agency costs as measured by logarithm of sales. the ratio of operating expenses to annual sales III. RESULTS AND ANALYSIS In Panel A of Table 4.1, columns 2 and 3 show the number of observations and the A. Some preliminary results mean (median) ratios of operating expenses regarding the separation of (which does not include salary to managers), ownership and control to sales for firms whose manager is an owner. Columns 4 and 5 show the same We first examine how agency costs vary information for firms whose manager is an with the separation of ownership and outsider. Consistent with our prior expecta- control—that is, whether the firm’s tions, most small businesses are managed manager is a shareholder or an outsider by shareholders rather than by outsiders with no ownership stake. This analysis may (1,249, or 73 percent of the 1,708 sample offer some insights into the effects of firms). However,there is not an insignificant managerial alignment with owners on number of firms that hire outside managers equity agency costs.9 Table 4.1 compares (459, or 27 percent of the sample). Thus, the agency costs of firms under two there appear to be a sufficient number of AGENCY COSTS AND OWNERSHIP STRUCTURE 119

0 $0.00–$0.35 $0.35–$1.50 $1.50–$5.20 > $5.20 Sales Quartile ($Millions)

Figure 4.4 Sales-to-asset ratio by sales quartile for a sample of 1,708 small corporations.

firms in these two groups for making the firm’s equity. At 368 of these 515 meaningful statistical comparisons of their firms, the owner also serves as manager; at operating expense ratios. 147 firms, the owner employs an outsider We find that both the median and as manager. The former group fits the average ratios of operating expenses to definition of our no-agency-cost base case, annual sales are considerably higher for where the manager owns 100 percent of firms managed by outsiders (column 5) the firm and the interests of manager and than for firms owned by shareholders owner are completely aligned. For the (column 3). For the full sample (line 1 of latter group, the interests of owner and Panel A), the average ratios of operating manager are completely unaligned. Thus, expenses to assets at insider-managed these groups are of interest because they firms and outsider-managed firms are represent the two ends of the Jensen and 46.9 percent and 51.9 percent, respectively; Meckling’s spectrum of ownership and the 5.0 percentage-point difference in these managerial structures. Line 2 of Panel A in means is statistically significant at the Table 4.1 shows that the ratio of operating 1 percent level. expenses to sales for the no-agency-cost Our data enable us to provide a rough base case firm, where the manager owns estimate of the agency costs per year 100 percent of the firm’s equity, is 46.4 attributable to the nonalignment of outside percent, as compared with 49.8 percent for managers and shareholders. A back-of-the- firms whose owners hold all of the firms’ envelope calculation shows that, in equity but hire an outside manager. For absolute dollars, a five-percentage-point these two groups of firms, the difference in difference implies that the operating operating expense ratios is 3.4 percentage expenses at a firm with median annual points. Although this univariate difference sales of $1.3 million are $65,000 per year in means is not statistically significant, a higher when an outsider rather than a multiple regression model that corrects for shareholder manages the firm. The present size and industry effect, shown in Table 4.3 values of these residual equity agency costs later, indicates that firms hiring outside are of course several times higher.10 managers have operating expenses that are Included in the full sample are 515 firms 5.4 percent greater than those at firms in which the primary owner controls all of managed by a shareholder. 120 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS Also included in the full sample are interest in the firm (line 3), where a single 1,001 firms in which the primary owner holds family holds a controlling interest in the firm a controlling interest of more than half of the (line 4), and where no individual or family firm’s equity. As shown in Table 4.1, Panel A, owns more than half of the firm (line 5). line 3, the ratio of operating expenses to For the full sample, displayed in line 1, sales for these firms is 2.8 percentage the average sales-to-asset ratio at insider- points lower when the owner manages the managed firms is almost 10 percent higher firm than when the owner hires an outside than at outsider-managed firms at 4.76 manager. However, this difference is not and 4.35, respectively. The 0.41 difference statistically significant. in these means is statistically significant at There are also 1,249 firms in which a the 10 percent level.This difference implies single family holds a controlling interest of that the revenues of a median-size firm, more than half of the firm’s equity. As which has $438,000 in total assets, are shown in line 4 of Panel A, the average $180,000 per year higher when a share- ratio of operating expenses to sales for holder rather than an outsider manages the these firms is 3.9 percentage points higher firm. In each of the remaining four when the firm is managed by an outsider comparisons (lines 2–5 of Panel B), the than when the firm is managed by a average ratio of annual sales to total assets shareholder. This difference is statistically also is greater when the firm is managed by significant at the 5 percent level. a shareholder than when the firm is One final group of interest is composed managed by an outsider. However, this of 336 firms in which no person or family difference is statistically significant at holds a controlling interest of more than least at the 10 percent level only when the 50 percent of the firm’s equity. As predicted, primary owner holds a controlling interest because of the more diffuse ownership of in the firm (line 3). these firms, the average ratio of operating Overall, the results displayed in Table 4.1 expenses to sales is indeed much higher: suggest that both the ratio of operating 7.2 percentage points more at firms man- expenses to annual sales and the ratio of aged by outsiders than at firms managed by annual sales to total assets are adequate shareholders. This difference is statistically proxies for small corporations’ agency costs. significant at the 5 percent level.To confirm Each provides results consistent with the that our finding is robust with respect to predictions of agency theory for a wide range sample distributions, we also perform non- of potentially high to low agency cost parametric tests on the difference between organizational and management structures. the medians, and find similar results. A.3. Determinants of high- and A.2. Agency costs as measured by the low-agency cost firms ratio of annual sales to total assets Table 4.2 presents descriptive statistics for In Panel B of Table 4.1 we present results the variables hypothesized to explain from a similar analysis of agency costs, but agency costs. Statistics are presented both here we measure agency costs by the ratio for the entire sample and for two groups of of annual sales to total assets rather than firms constructed by dividing the entire the ratio of operating expenses to annual sample in half, based on the sample’s sales.11 As predicted, the results show that median ratios of agency costs. For the the sales-to-asset ratios are higher in all entire sample (Panel A), ownership and categories of shareholder-managed firms control is highly concentrated. On average, versus outsider-managed firms. This is true a shareholder manages the firm 73 percent for the full sample of 1,249 firms (line 1) of the time, the primary owner controls and for the subsamples where the primary 65 percent of the firm’s equity, and a single owner holds all of the firm’s equity (line 2), family owns a controlling interest in the where the primary owner holds a controlling firm 73 percent of the time. The average AGENCY COSTS AND OWNERSHIP STRUCTURE 121 Table 4.1 Agency Costs, Ownership Structure, and Managerial Alignment with Shareholders Agency costs are presented for a sample of 1,708 small corporations divided into two groups of firms: those managed by owners (aligned with shareholders) and those managed by an outsider (not aligned with shareholders). Agency costs are proxied alternatively by the ratio of operating expenses to annual sales and the ratio of annual sales to total assets. Separate analyses are presented for each agency cost proxy and for subgroups where the primary owner owns 100 percent of the firm, where the primary owner owns more than half of the firm, where a single family owns more than half of the firm, and where no owner or family owns more than half of the firm. The last column shows the difference between the mean (median) ratios of the outsider-managed firms and the insider-managed firms. Statistical significance of the differences in the mean ratios is based on the t-statistic from a parametric test (based on the assumption of unequal variances) of whether the difference in the mean ratios of the two groups of firms is significantly different from zero. Statistical significance of the differences in the median ratios is based on a chi-square statistic from a nonparametric test of whether the two groups are from populations with the same median (Mood (1950)). Data are taken from the Federal Reserve Board’s National Survey of Small Business Finances. Type of Manager

Owner-Manager Outsider-Manager Difference Number of Ratio Mean Number of Ratio Mean in Means (in Firms (Median) Firms (Median) Median)

Panel A: Operating Expense-to-Annual Sales Ratio

All firms 1,249 46.9 459 51.9 5.0*** (42.0) (52.2) (10.2)*** Primary owner owns 368 46.4 147 49.8 3.4 100 percent of the firm (41.7) (47.6) (5.9)** Primary owner owns 743 46.8 258 49.6 2.8 50 percent of the firm (41.5) (47.7) (6.2)** A single family owns 943 46.2 306 50.1 3.9** 50 percent of the firm (41.7) (49.0) (7.3)*** No owner or family owns 220 48.1 116 55.3 7.2** 50 percent of the firm (42.7) (55.6) (12.9)***

Panel B: Annual Sales-to-Total Assets Ratio

All firms 1,249 4.76 459 4.35 0.41* (3.18) (2.88) (0.30)* Primary owner owns 368 5.35 147 4.78 0.57 100 percent of the firm (3.54) (3.33) (0.21) Primary owner owns 743 5.08 258 4.49 0.59* 50 percent of the firm (3.33) (3.13) (0.20) A single family owns 943 4.74 306 4.41 0.33 50 percent of the firm (3.19) (3.07) (0.12) No owner or family owns 220 4.63 116 3.89 0.74 50 percent of the firm (3.14) (2.49) (0.65)**

*, **,*** indicate statistical significance at the 10, 5, and 1 percent levels, respectively. 122 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 0.09*** 2.33*** 0.06*** 0.02 0.04* le into two equal-size le into ns and medians appear ns and medians tio). Column 6 (column 9) Below Above Below Above Operating Expense-to-Assets Sales-to-Assets Ratio Groups Ratio Groups -test for significant differences in the means of the low- and high-ratio groups of groups and high-ratio the means of low- in differences significant -test for t Panel A Panel B Panel C Mean Median Median Median Difference Median Median Difference 50 percent of the firm50 percent 0.73 1 0.75 0.71 0.04* 0.72 0.74 Descriptive Statistics forDescriptive Agency Costs Used to Analyze Variables (years) Annual sales ($millions) 5.9 1.3 8.2 3.6 4.6*** 6.1 5.7 0.4 Debt-to-asset ratio 0.60 0.52 0.61 0.58 0.03 0.55 0.64 Number of banking relationships of banking Number 1.65 1 1.75 1.56 0.19*** 1.73 1.58 0.15*** Length of the longest banking relationship of the longest banking Length 10.6 8 11.3 10.0 1.3*** 11.5 9.7 1.8*** Number of nonmanager shareholders of nonmanager Number 3.51 1 3.89 3.14 0.75** 2.35 4.68 Ownership share of primary owner share of primary Ownership 0.65 0.54 0.66 0.64 0.02 0.62 0.68 One family owns One family Firm manager is a shareholder manager Firm 0.73 1 0.78 0.69 0.09*** 0.71 0.75 Firm age (years)Firm 17.6 14 18.4 16.8 1.6** 19.0 16.2 2.8*** Control variables Control External monitoring variables monitoring External firms. Finances. Survey of Small Business Reserve Board’s National the Federal Data are from shows the difference in the two groups’ in the difference shows means, a and the results from groups of 854 firms based on the entire sample’s median operating expense-to-annual sales ratio (annual sales-to-total asset ra sales ratio expense-to-annual operating based on the entire sample’s median of 854 firms groups in columns 2 and 3.in the samp constructed by splitting of firms two groups In columns 4 and 5 (columns 7 8) are the means for Ownership variables Ownership Selected variables used to study agency costs at a sample of 1,708 small corporations are identified in column 1, in are identified used to study agency costs at a sample of 1,708 small corporations variables Selected the sample mea Table 4.2 (1)*, **, at the 10, statistical significance *** indicate 5, (2) levels, and 1 percent respectively. (3) (4) (5) (6) (7) (8) (9) AGENCY COSTS AND OWNERSHIP STRUCTURE 123 number of nonmanager shareholders is B.1. Agency costs as measured by 3.51, but this statistic is strongly influ- the ratio of operating expenses to enced by extreme values, as the median annual sales number of nonmanager shareholders is one. The average firm’s longest banking Table 4.3 presents the results from relationship is 10.6 years.The average firm multivariate regressions analyzing agency maintains relationships with 1.65 banks, costs as measured by the ratio of operating reports $5.9 million in annual sales, is 17.6 expenses to annual sales. Column 1 identifies years old, and has a debt-to-asset ratio the explanatory variable and columns 2 of 0.60. through 9 display parameter estimates for When we split the sample into low- eight different model specifications. In expense and high-expense ratio groups columns 2 through 8 we analyze each of (Panel B), we observe strong differences in the seven ownership structures, external the two groups. Based on t-tests for signif- monitoring, and capital structure variables icant differences in the means of the two independently. In column 9 we test whether groups, the high-expense firms are less the independent results stand up when all likely to be managed by a shareholder, are seven variables are included in a single less likely to be controlled by one family, regression. Because of the importance of have fewer nonmanaging shareholders, industry structure and economies of scale, have shorter and fewer banking relation- as established in Section II, we include in ships, report lower sales, and are younger each regression variables to control for than the low-expense firms. Similar results firm size and industry effects. Our measure are obtained when the top third and bottom of size is the logarithm of annual sales, and third of the sample are compared. our controls for industry are 35 two-digit When we split the sample into low and SIC indicator variables, one for each two- high asset-utilization groups (Panel C), we digit standard industrial classification that also find strong differences in the two accounts for more than one percent of our groups. Low-efficiency firms are less likely sample of firms. to be managed by a shareholder,have lower In column 2 of Table 4.3 we find that a percentage ownership by the primary owner, firm managed by a shareholder has agency have fewer nonmanaging shareholders, have costs that are 5.4 percentage points lower longer and more numerous banking rela- than those at firms managed by an tionships, have lower debt-to-asset ratios, outsider. This is very close to the 5.0 per- and are older than high-efficiency firms. centage point difference reported for all firms in Panel A of Table 4.1. For a firm B. Multivariate regression results with the $1.3 million median annual sales, explaining agency costs the coefficient in column 2 of Table 4.3 implies agency costs of approximately Tables 4.3 and 4.4 present the results $70,000. obtained from estimating multivariate In column 3 of Table 4.3 we find that a regressions to explain the determinants of firm in which one family owns a controlling our two proxies for agency costs, the ratio interest has agency costs that are 3.0 per- of operating expenses to annual sales and centage points lower than other firms. For of annual sales to total assets. Each proxy the median-size firm, this implies agency is regressed against the ownership, external costs of approximately $39,000. In column 4, monitoring, and control variables intro- we find that agency costs decline by 0.082 duced and discussed in Section II. These percentage points for each percentage point regressions compare the relative, as well as increase in the ownership share of the the absolute, agency costs of various firm’s primary owner. This implies that a ownership structures vis-à-vis the no- median-size firm where the primary owner agency-cost base case—the 100 percent has a 100 percent share has agency manager-owned firm. costs that are approximately $105,000, or 124 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 7.9) 0.7) 1.7) 3.4) 2.8) 0.3) 4.3) 2.9*** 0.7 1.0* 0.3*** 8.6*** 0.4 5.7*** ( ( 8.3) ( 0.6) ( size is 1,708. size Each 2.8*** 0.6 ( sample of firms. Data are ependent variables: common 7.4) ( 1.9) ( 2.6*** 1.1* ( 7.8) ( 3.2) ( 2.6*** 0.22*** 0.03(0.6) 0.02 (0.4) ( 8.8) ( 3.1*** (2.8) (0.1) 9.0) ( 3.9) ( 3.1*** 8.2*** ( 8.6) ( 2.2) 3.1*** 3.0** ( 8.8) ( 4.1) 2.9*** 5.4*** ( ( 94.7*** 91.7*** 74.9*** 91.3*** 88.3*** 87.4*** 88.7*** 104.5*** 0.246 0.241 0.245 0.242 0.244 0.240 0.239 0.259 50 percent of the firm 50 percent rminants of Agency Costs at Small Corporations rminants 2 R Dete -statistic 16.04*** 15.63*** 16.00*** 15.73*** 15.50*** 15.57*** 15.46*** 14.57*** Adjusted Firm age Firm Log of annual sales Two-digit SIC dummiesTwo-digit Yes Yes Yes Yes Yes Yes Yes Yes Debt-to-asset ratio Number of banking relationships of banking Number Length of the longest banking relationship of the longest banking Length Log of the number of nonmanager stockholders of nonmanager Log of the number 1.9*** 0.01 Ownership share of primary owner share of primary Ownership One family owns One family Manager is a shareholder Manager F Regression summary statisticsRegression Control variables Control External monitoring variables monitoring External from the Federal Reserve Board’s National Survey of Small Business Finances. Survey of Small Business Reserve Board’s National the Federal from specification includes a set of 35 dummy variables indicating each two-digit SIC that accounts for more than one percent of the more than one percent SIC that accounts for each two-digit indicating a set of 35 dummy variables includes specification ownership/managerial alignment variables, variables, alignment monitoring ownership/managerial external capital structure variables, variables. and control Sample The dependent variable proxying for agency costs is the ratio of operating expense to annual sales.There are four groups of ind groups to annual sales.There are four expense of operating agency costs is the ratio for proxying dependent variable The Ownership variables Ownership Table 4.3 (1)Intercept (17.4) (2) (16.9) (3) (17.1) (17.2) (4) (16.9) (5) (16.6)*, **, at the 10, statistical significance *** indicate 5, levels, and 1 percent respectively. (16.9) (6) (17.3) (7) (8) (9) AGENCY COSTS AND OWNERSHIP STRUCTURE 125 8.1 percentage points, lower than those at finding conflicts with our hypothesis in a firm where the primary owner has only a which multiple banking relationships one percent share. Each of the variables reduce each bank’s incentive to monitor, analyzed in columns 2 through 4 is statis- and, therefore, increase agency costs. One tically significant at least at the 5 percent possible explanation reconciling the two level. seemingly contradictory results is that the In column 5 of Table 4.3 we analyze (the number of banking relationships may proxy natural logarithm of one plus) the number for factors other than the banks’ incentive of nonmanager shareholders. We expect a to monitor the firm. The most prominent positive relationship between agency costs explanations are the increasing financial and this variable, as the returns to monitor- sophistication and maturity of the firms ing decrease and free-rider problems and their managers, and regulatory limita- increase with the number of nonmanager tions on loans to a single borrower, which shareholders. We use the natural logarithm may constrain a small bank’s ability to rather than the level of this variable supply funds to a larger firm. because we expect that the relationship is In column 8 of Table 4.3 we analyze the stronger at smaller values of the variable. complex relation between capital structure The estimated coefficient is positive and and ownership on agency costs. As discussed significant at better than the one percent in Section II, we expect an inverse rela- level, confirming our expectations. For a tionship between agency costs and the firm with 30 non-manager shareholders, debt-to-asset ratio. We do, indeed, find a the maximum value imposed by our cap at negative relationship, but the coefficient is the 95th percentile, the estimated coeffi- not significantly different from zero. cient of 1.9 implies that agency costs are In each of the seven specifications 6.5 percentage points higher, or $85,000 displayed in columns 2 through 8 of Table greater, than at a firm with zero nonman- 4.3, observe that our size variable, the nat- ager shareholders. ural logarithm of annual sales, is negative In columns 6 and 7 of Table 4.3 we and statistically significant at better than analyze the two bank monitoring variables: the 1 percent level, which is strong evidence the length of the firm’s longest banking of economies of scale. Not shown in Table relationship and the firm’s number of bank- 4.3 are statistics indicating that at least 20 ing relationships. As discussed in Section II, of the 35 two-digit SIC indicator variables we expect agency costs to vary inversely included in each specification are statisti- with the length of the longest banking cally significant at least at the 5 percent relationship and directly with the number level.These findings underscore the critical of banking relationships. To distinguish importance of controlling for differences between the private information generated across industries when examining the oper- by bank monitoring and the public informa- ating expense-to-sales ratio. The adjusted tion generated by a firm’s durability, we R2 for each of the seven specifications in also include firm age in the specification columns 2 through 8 indicates that the analyzing the length of the firm’s longest models explain approximately one-quarter banking relationship. of the variability in the ratio of operating As shown in column 6 of Table 4.3, expenses to annual sales. agency costs are reduced by a statistically Our final specification appears in column significant 0.22 percent for each additional 9 of Table 4.3, where we include each of year in the length of the firm’s longest four of the ownership variables, the two banking relationship. The coefficient on bank-monitoring variables, and the capital firm age is not significantly different from structure variable, along with the control zero. In column 7, however, a related variables for firm size, age, and industrial variable, the number of banking relation- classification. We find that each of the four ships, is negative and statistically significant ownership variables has the predicted sign. at better than the 10 percent level. This However,only two of the four—the indicator 126 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS variable for shareholder-managed firm and include in each regression a series of 35 the variable for the ownership share of the two-digit SIC indicator variables, one for primary owner—are statistically signifi- each two-digit standard industrial classifi- cant, but each is significant at better than cation that accounts for more than one the 1 percent level.The statistical insignifi- percent of our sample of firms. We include cance of the other two ownership variables the natural logarithm of annual sales as a may be attributable to the high correlation measure of size, because Figure 4.4 among the ownership variables.The signifi- suggests a possible quadratic relationship cant coefficients indicate that agency costs between sales and the ratio of sales-to- at a firm managed by a shareholder are 5.7 asset ratio. percentage points lower than those at a Column 2 of Table 4.4 shows that a firm firm managed by an outsider, and that managed by a shareholder has a sales-to- agency costs are reduced by 0.086 per- asset ratio that is 0.51 greater than that of centage points for each percentage point a firm managed by an outsider, and this increase in the primary owner’s ownership coefficient is statistically significant at better share. This latter result supports the than the 5 percent level. This evidence hypothesis that large shareholders make supports the hypothesis that agency costs more effective monitors (Shleifer and are higher when an outsider manages the Vishny (1986) and Zeckhauser and Pound firm. In column 3, we see that the variable (1990)).12 indicating those firms in which one family Both of the external monitoring variables owns a controlling interest has a coefficient are negative and significant, just as they are that is positive but not significantly differ- in columns 6 and 7. The debt-to-asset ratio ent from zero. Column 4 shows that the is negative but not significantly different coefficient on the ownership share of the from zero, just as it is in column 8. Overall, primary owner is positive and significant at the results displayed in column 9 generally better than the 1 percent level. The coeffi- confirm the findings when the analysis cient indicates that the sales-to-asset ratio variables are examined independently in increases by 0.012 for each percentage columns 2 through 8. point increase in the ownership share of the firm’s primary owner.This finding supports B.2. Agency costs as measured by the the hypothesis that agency costs decrease ratio of annual sales to total assets as the ownership becomes more concentrated. Column 5 shows that the coefficient Table 4.4 displays the results from multi- on (the natural logarithm of) the number variate regressions analyzing agency costs of nonmanager stockholders is negative as measured by the ratio of annual sales to and statistically significant at better than the total assets. In interpreting these results, it 1 percent level, supporting the hypothesis is important to remember that the sales-to- that agency costs increase as the free-rider asset ratio varies inversely with agency problem worsens. costs. As in Table 4.3, column 1 identifies In columns 6 and 7 of Table 4.4 we the explanatory variables and columns 2 analyze the two bank monitoring variables: through 9 display parameter estimates for the length of the firm’s longest banking different specifications of the regression relationship and the number of the firm’s model. In columns 2 through 8, we analyze banking relationships. Once again, to distin- each of seven ownership structure, external guish between the private information monitoring, and capital structure variables generated by bank monitoring and the independently. In column 9, we test public information generated by a firm’s whether the independent results stand up durability, we also include firm age in the when all seven are included in a single specification analyzing the length of the regression. firm’s longest banking relationship. As Because of the importance of industry shown in column 6, the length of the firm’s structure, established in Section II, we longest banking relationship variable is AGENCY COSTS AND OWNERSHIP STRUCTURE 127 1.0) 0.8) 0.01 4.3) 0.50*** 0.4) 0.01 5.4) 1.05*** 0.22 0.8) iness Finances. iness size is 1,708. size Each ent variables: common 0.8) ( 0.01 4.3) ( 0.50*** 0.4) ( 0.01 5.4) ( 1.05*** 1.8) (3.1) 0.011* 0.012*** 0.34 1.2) ( 0.30 0.46* (1.2) (1.8) sample of firms. In column 1.05*** 1.11*** 1.11*** (5.3) (5.5) (5.5) 4.7) ( 1.09*** ( 1.2) ( 0.012 1.7) ( 0.025* ( 6.2) ( 0.82*** (3.0) ( (0.3) ( 0.032 0.030 0.035 0.051 0.035 0.042 0.045 0.080 0.080 (0.7) (0.5) (1.1) (2.7) (1.1) (1.7) (0.3) (3.9) (3.9) (2.0) 2.01* 2.51** 1.27 1.27 2.49** 1.95* 3.26*** 1.34 1.17 50% of the firm 0.087 rminants of Agency Costs at Small Corporations rminants 2 R Dete -statistic 2.51*** 2.40*** 2.65*** 3.49*** 2.63*** 3.02*** 3.20*** 4.37*** 4.37*** relationship ( stockholders ( Adjusted Firm age Firm Log of annual sales 0.05 0.03 0.07 0.18*** 0.07 0.11* 0.02 0.28*** 0.28*** Two-digit SIC dummiesTwo-digit Yes Yes Yes Yes Yes Yes Yes Yes Yes Debt-to-asset ratio Number of banking relationships of banking Number Length of the longest banking Length Log of the number of nonmanager Log of the number Ownership share of primary owner share of primary Ownership 0.012*** One family owns One family Manager is a shareholderManager 0.51** F Regression summary statisticsRegression Control variables Control External monitoring variables monitoring External 10, have been orthogonalized. variables Survey of Small Bus Reserve Board’s National ownership the four the Federal Data are from specification includes a set of 35 dummy variables indicating each two-digit SIC that accounts for more than one percent of the includes more than one percent SIC that accounts for specification each two-digit indicating a set of 35 dummy variables ownership/managerial alignment variables, variables, alignment monitoring ownership/managerial external capital structure variables, variables. and control Sample Ownership variables Ownership The dependent variable proxying for agency costs is the ratio of annual sales to total assets. agency costs is the ratio for of independ groups are four proxying dependent There variable The Table 4.4 (1)Intercept (1.9) (2) (2.4) (3) (1.2) (4) (1.2) (5) (2.5) (1.9) (6) (3.2) (7)*, **, at the 10, statistical significance *** indicate 5, levels, and 1 percent respectively. (1.2) (8) ( (9) (10) 128 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS inversely related to the sales-to-asset ratio, from positive and significant at better than and is statistically significant at better than the 1 percent level in column 4 to negative the 10 percent level. This runs counter to and significant at better than the 10 per- our hypothesis that agency costs are lower cent level in column 9.This counterintuitive when a firm’s bank has had more time to finding may be attributable to the high develop valuable private information about correlation of the ownership share variable the firm. However, this variable is not sig- with the log of number of nonmanager nificantly different from zero in the full shareholders ( 0.75). specification shown in column 9. In column Both of the external monitoring variables 7 we see that a related variable, number of are inversely related to the sales-to-asset banking relationships, is negative and sta- ratio, as they are in columns 6 and 7, when tistically significant at better than the 1 these variables are analyzed independ- percent level. This latter finding supports ently. However, the length of the longest the hypothesis that the values of a bank’s banking relationship variable no longer monitoring effort and private information even approaches statistical significance about a firm are dissipated when the firm (t 0.3), but the number of banking obtains financial services from multiple relationships is significant at better than sources, but, on the whole, the results the 1 percent level. The debt-to-asset ratio regarding the bank monitoring variables remains positive and significant at better are ambiguous. than the 1 percent level, as it is in column In column 8 we analyze the effect of 8. Overall, the results displayed in column 9 capital structure on the sales-to-asset tend to confirm the findings previously ratio. The results indicate that firms with discussed when the analysis variables higher debt ratios have higher sales-to- are examined one-by-one in columns 2 asset ratios, and that this relationship is through 8. statistically significant at better than the 1 To test whether the correlations among percent level.This finding is supportive of a the ownership variables are responsible for version of the theory put forth by Williams this finding, we orthogonalize the four own- (1987) that additional debt decreases ership variables and then reestimate the agency costs. model specification appearing in column 9. Not shown in Table 4.4 are the results The results of this reestimation, which concerning the industry indicator variables. appear in column 10, confirm our suspi- In each specification, at least 20 of the 35 cions.Three of the four ownership variables two-digit SIC indicator variables are signif- are statistically significant at least at the icant at the 5 percent level. Once again, this 10 percent level.13 The log of the number of finding underscores the critical importance shareholders remains negative and signifi- of controlling for differences across indus- cant at better than the 1 percent level, but tries when comparing agency costs. the primary owner’s ownership share Our final specification appears in column switches back from negative to positive and 9 of Table 4.4, where we include each of the also is significant at better than the 1 per- four ownership variables, the two monitoring cent level. The dummy indicating that the variables, and the capital structure vari- firm is managed by a shareholder also is able, along with the control variables for positive, but is significant at only the 10 firm size, age, and industrial classification. percent level. The dummy indicating that a We find that only two of the four correlated family controls the firm is not significantly ownership variables are statistically signif- different from zero. In sum, the results for icant at better than the 10 percent level. the four ownership variables are not quali- The natural logarithm of the number of tatively different from those reported in nonmanager shareholders varies inversely columns 2 through 5, when each of these with the sales-to-asset ratio and is signifi- variables is examined independently, and cant at better than the 1 percent level. The provide strong support for the agency-cost primary owner’s ownership share switches theory of Jensen and Meckling (1976).14 AGENCY COSTS AND OWNERSHIP STRUCTURE 129 IV. SUMMARY AND CONCLUSIONS pay to the managers as well as perks and firm level monitoring cost, is the appropriate measure In this article, we use data on small to test the hypothesis. businesses to examine how agency costs 2 Data from the NSSBF yield significant and interesting results that appear in several recent vary with a firm’s ownership structure. published papers. See the studies on banking Because the managerial ownership of small relationships and credit markets by Petersen firms is highly variable, with a range from and Rajan (1994, 1995, 1997), Berger and zero to 100 percent, we are able to Udell (1995), and Cole (1998). estimate a firm’s agency costs across a 3 The staff at the Federal Reserve Board wide variety of management and ownership partially edited the financial statement items structures.15 By comparing the efficiency for violations of accounting rules, such as when of firms that are managed by shareholders gross profit is not equal to sales less cost of with the efficiency of firms managed by goods sold, and some improbable events such as outsiders, we can calculate the agency when accounts receivable are greater than sales, or cost of goods sold equals inventory. costs attributable to the separation of 4 Operating expenses are defined as total ownership and control. expenses less cost of goods sold, interest We also examine the determinants of expense, and managerial compensation. agency costs in a multivariate regression Excessive expense on perks and other nonessen- framework and find that our results tials should be reflected in the operating support predictions put forth by the theo- expenses. Strictly speaking, agency costs that ries of Jensen and Meckling (1976) and are measured by this ratio are those incurred at Fama and Jensen (1983a) about owner- the firm level (i.e., shirking and perquisite ship structure, organizational form, and the consumption by the managers).This may under- alignment of managers’ and shareholders’ estimate total agency costs since this ratio does not fully measure firm-level indirect agency costs, interests. First, we find that agency costs such as the distortion of operating decisions due are higher when an outsider manages the to agency problems. (See Mello and Parsons firm. Second, we find that agency costs (1992) for an attempt to measure such costs in vary inversely with the manager’s owner- the presence of debt.) Nor does it measure ship share.Third, we find that agency costs off-income-statement agency costs, such as the increase with the number of nonmanager private monitoring costs by the nonmanagement shareholders. Fourth, we also find that, to a shareholders or the private costs of bonding lesser extent, external monitoring by banks incurred by the manager. produces a positive externality in the form 5 Manne (1967) and Alchian and Demsetz of lower agency costs. (1972) agree that limited liability is an attractive feature of the corporate form of organization. Jensen and Meckling (1976) point out that although unlimited liability gives more incentive NOTES for each shareholder to monitor, in the aggregate it leads to excessive monitoring. Thus, it * Ang is from Florida State University; Cole is may be more economical to offer a single high from The University of Auckland, New Zealand; premium to creditors to bear risk of nonpay- and Lin is from Montana State University. We ment and, thus, monitoring in exchange for appreciate the comments of David Mauer, limited liability. Michael Long, René Stulz (the editor), and an 6 Technically, the survey does not provide a anonymous referee. variable for the number of nonmanager share- 1 Theoretical support for the null hypothesis holders. Rather, we define this variable as the is due to Demsetz (1983), who suggests that the number of shareholders for firms that have an sum of amenities for on-the-job consumption outside manager and as the number of share- and take-home pay for similar quality managers holders less one for firms that have an insider is the same for both high-cost and low-cost manager. monitoring organizations. The proportion paid 7 This formulation recognizes the unequal and to the managers, however, differs according to diminishing role of additional shareholders, and the cost of monitoring. Here, it would seem that the problem of undefined zero when there is no total operating expense, which include direct other shareholder. 130 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 8 These are also the same governance 14 For the sake of completeness, we also variables used by Berger and Udell (1995) and perform the same procedure on the regression Cole (1998) in their studies of banking rela- equation in column 9 of Table 4.3. We find that tionship. However,none of their variables, except orthogonalizing the ownership variables does for the corporate form dummy, are found to not qualitatively affect the results. Only the significantly affect either the loan term or the same two ownership variables are statistically use of collateral. significant. Overall, although both measures of 9 In Fama and Jensen (1983a), the delegation agency costs provide qualitatively similar of decision control management and residual results, the expense ratio regression yields owner (i.e., hiring of outsiders as managers) is greater explained variations. related to the decision skill and the accompanying 15 There are few empirical studies in related specialized knowledge that are needed to run areas of corporate finance that analyze ownership, the firm. Shareholders, however, still have to organizational, and management structures in bear the costs of monitoring. detail. For example, see a study of executive 10 As a way of comparison, Dong and Dow compensation in Israel by Ang, Hauser, and (1993) estimate that 10 to 20 percent of total Lauterbach (1997). labor hours are attributed to supervision or monitoring in the Chinese collective farms. Dobson (1992) finds that X-inefficiency REFERENCES measures 0.2 percent of sales among large U.S. manufacturing firms. Alchian, Armen A., and Harold Demsetz, 1992, 11 The distribution of the sales-to-asset ratio is Production, information costs, and economic highly skewed by the presence of large outliers. organization, American Economic Review Consequently, the ratio is capped at the value 62, 777–795. found at the 95th percentile, a ratio of 19.0. Ang, James S., Samuel Hauser, and Beni 12 Fama and Jensen (1983) realize that for Lauterbach, 1997, Top executive compensa- shareholders to monitor the firm’s management tion under alternate ownership and they must hold sufficient ownership; however, governance structure, in Mark Hirschey and the cost of large ownership shares is suboptimal M. Wayne Marr, eds.: Advances in Financial risk-taking, and, possibly, underinvestment. Also, Demsetz (1983) suggests that firms with concen- Economics (JAI Press Inc., Greenwich, trated ownership have lower monitoring costs. Conn.). 13 The sales-to-asset ratio is likely subject to Arnold, R. J., 1985, Agency costs in banking additional biases that render it much noisier firms: An analysis of expense preference than the operating expense-to-sales ratio. Note behavior, Journal of Economics and Business, that the adjusted-R2 statistics appearing at the 103–112. bottoms of Tables 4.3 and 4.4 indicate that we Barclay, Michael J., and Clifford W. Smith Jr., are able to explain about 26 percent of the 1995, The maturity structure of corporate variability in the latter ratio but only about eight debt, Journal of Finance 50, 609–632. percent of the variability in the former ratio. Berger, Allen N., and Gregory F. Udell, 1995, This led us to investigate additional control variables in the sales-to-asset regression, including Relationship lending and lines of credit in the ratio of operating expenses to sales. Results small firm finance, Journal of Business 68, from specifications including the operating 351–382. expense-to-sales ratio as an additional regressor Cole, Rebel A., 1998, The importance of rela- are not qualitatively different from those in tionships to the availability of credit, Journal Table 4.4. In no case was the operating expense- of Banking and Finance, forthcoming. to-sales ratio statistically significant at even the Cole, Rebel A., and Hamid Mehran, 1998, CEO 20 percent level. Because we view the operating compensation, ownership structure, and taxa- expense-to-sales ratio as an endogenous vari- tion: Evidence from small businesses, mimeo, able, we also tested a specification that included Northwestern University. a predicted value of this ratio rather than the actual value. The predicted value was obtained Cole, Rebel A., and John D. Wolken, 1995, using the model appearing in column 9 of Sources and uses of financial services by Table 4.3. Again, the results from this robustness small businesses: Evidence from the 1993 check are not qualitatively different from those National Survey of Small Business Finances, appearing in Table 4.4 Federal Reserve Bulletin 81, 629–667. AGENCY COSTS AND OWNERSHIP STRUCTURE 131 Cole, Rebel A., John D. Wolken, and Louise agency costs and capital structure, Journal of Woodburn, 1996, Bank and nonbank compe- Financial Economics 3, 305–360. tition for small business credit: Evidence Kim, Wi S., and Eric H. Sorensen, 1986, from the 1987 and 1993 National Surveys of Evidence on the impact of the agency costs of Small Business Finances, Federal Reserve debt in corporate debt policy, Journal of Bulletin 82, 983–995. Financial and Quantitative Analysis 21, Demsetz, Harold, 1983,The structure of owner- 131–144. ship and theory of the firm, Journal of Law Knight, Frank H., 1921, Risk, Uncertainty and and Economics 26, 375–393. Profit (University of Chicago Press, Demsetz, Harold, and Kenneth Lehn, 1985, The Chicago). structure of corporate ownership: Causes and Manne, Henry G., 1967, Our two corporate sys- consequences, Journal of Political Economy tems: Law and economics, Virginia Law 93, 1155–1177. Review 53, 259–284. Denning, Karen C., and Kuldeep Shastri, 1993, Mello, Antonio S., and John E. Parsons, 1992, Changes in organizational structure and Measuring the agency cost of debt, Journal of shareholder wealth, Journal of Financial and Finance 47, 1887–1904. Quantitative Analysis 28, 553–564. Milgrom, Paul, and John Roberts, 1995, Diamond, Douglas W., 1984, Financial interme- Economics, Organization, and Management diation and delegated monitoring, Review of (Prentice Hall, Englewood Cliffs, N.J.). Economic Studies 51, 393–414. Petersen, Mitchell A., and Raghuram G. Rajan, Dobson, John, 1992, Agency costs in U.S. man- 1994, The benefits of lending relationships: ufacturing: An empirical measure using Evidence from small business data, Journal X-efficiency, Journal of Economics and of Finance 49, 3–38. Finance 16, 1–10. Petersen, Mitchell A., and Raghuram G. Rajan, Dong, Xiao Y., and Gregory K. Dow, 1993, 1995,The effect of credit market competition Monitoring costs in Chinese agricultural on lending relationships, Quarterly Journal of teams, Journal of Political Economy 101, Economics 110, 407–442. 539–553. Petersen, Mitchell A., and Raghuram G. Rajan, Fama, Eugene F., and Michael C. Jensen, 1997, Trade credit: Theories and evidence, 1983a, Separation of ownership and control, Review of Financial Studies 10, 661–691. Journal of Law and Economics 26, 301–325. Shleifer, Andrei, and Robert Vishny, 1986, Fama, Eugene F., and Michael C. Jensen, Large shareholders and corporate control, 1983b, Agency problems and residual claims, Journal of Political Economy 94, 461–488. Journal of Law and Economics 26, 327–349. Williams, Joseph, 1987, Perquisites, risk, and Fama, Eugene F., and Michael C. Jensen, 1985, capital structure, Journal of Finance 42, Organizational forms and investment deci- 29–48. sions, Journal of Financial Economics 14, Zeckhauser, Richard, and John Pound, 1990, 101–119. Are large shareholders effective monitors? Jensen, Michael C., 1986, Agency costs of free An investigation of share ownership and cash flow, corporate finance and takeovers, corporate performance, in R. G. Hubbard, ed.: American Economic Review 76, 323–329. Asymmetric Information, Corporate Finance, Jensen, Michael C., and William H. Meckling, and Investment (University of Chicago Press, 1976,Theory of the firm: Managerial behavior, Chicago).

Part 2 Equity ownership structure and control

INTRODUCTION

S SEEN IN PART 1, EFFICACY OF A CORPORATE governance system requires Aeffective monitoring of managers by the shareholders. Shleifer and Vishny (Ch.2) have argued that large shareholders can perform this monitoring function. They also acknowledge that this role imposes costs such as free riding by non-block shareholders who may avoid the monitoring effort but reap the benefits of the block shareholder’s monitoring. Large shareholders in this event may only be willing to undertake monitoring if they can expect to receive compensation for their efforts including an appropriate share of the added value that monitoring generates.They may also look to design the securities they hold in the firms with features that guarantee such reward. In the first paper of this part, Hiroshi Osano (Ch.5) addresses the problem of security design that will satisfy a large investor and motivate her to undertake monitoring.The author shows that the optimal security is a debt-like security such as standard debt with a positive probability of default, or debt with call options. Such a design also leads to the financial market equilibrium being Pareto optimal. The remaining three papers by Becht and Röell (Ch.6), Faccio and Lang (Ch.7) and Goergen and Renneboog (Ch.8) present empirical evidence on the ownership structure of European corporations with particular focus on block shareholdings and control exercised through pyramidal structures. Pyramidal structures allow block holders to control vast groups of companies with relatively small investment capital. Becht and Röell (Ch.6) find the most salient finding is the extraordinarily high degree of concentration of shareholder voting power in Continental Europe relative to the USA and the UK.Thus the relationship between large controlling shareholders and weak minority shareholders is at least as important to understand as the more commonly studied interface between management and dispersed shareholders. Faccio and Lang (Ch.7) report similar findings of high concentration in Western Europe excluding the UK but they also provide a rich analysis of the ownership structure of Western European corporations including pyramids, cross-shareholdings, multiple control chains, etc. They employ a mechanism to differentiate control (i.e. voting rights from cash flow rights and report the relative importance of each). Faccio and Lang find that both dispersed shareholdings and family controlled shareholdings are prevalent with the former more common in the UK and Ireland and the latter in Continental Europe.This pattern also differs between financial and non-financial firms. This paper is important in order to understand the ownership pattern in Europe and how it may influence corporate financial behaviour. Goergen and Renneboog (Ch.8) make an interesting comparison between the UK and Germany in terms of their ownership structure and how such a structure leads to different 134 EQUITY OWNERSHIP STRUCTURE AND CONTROL outcomes after initial public offerings in terms of takeover incidence, the ownership structure of bidders, and the differential characteristics of acquired firms that attract concentrated and dispersed acquirers. Goergen and Renneboog relate these differences to the differences in shareholder rights and laws protecting minority shareholders in Germany and the UK. Although in the UK concentrated ownership of the kind observed in Germany is uncommon, as noted in the previous paper,large block shareholdings in the form of institutional shareholding are quite widespread in the UK. Institutional shareholders – insurance companies, pension funds, mutual funds, investment trusts – collectively hold a substantial majority of shares in UK listed companies. Given their dominance, do institutional large block shareholders monitor their investee companies effectively? This question is addressed by Faccio and Lasfer (Ch.9) with particular reference to occupational pension funds. By comparing companies in which these funds hold large stakes with a control group of companies listed on the London Stock Exchange, they show that occupational pension funds hold large stakes over a long time period mainly in small companies. However, the value added by these funds is negligible and their holdings do not lead companies to comply with the Code of Best Practice or outperform their industry counterparts. Overall, their results suggest that occupational pension funds are not effective monitors. Chapter 5

Hiroshi Osano* SECURITY DESIGN, INSIDER MONITORING, AND FINANCIAL MARKET EQUILIBRIUM

Source: European Finance Review, 2(3) (1999): 273–302.

ABSTRACT This paper considers a problem of security design in the presence of monitoring done by a large investor to discipline the management of a firm. Since the large investor enjoys only part of the benefits generated by her monitoring activities but incurs all the associated costs, the design and amount of security need to be structured so as to motivate her to maintain an efficient level of monitoring, if no other mechanism exists to make her commit to specific levels of monitoring in advance. By assuming that the large investor takes account of the effect of the issued amount of security on the revenues received, we show that the optimal security is a debt-like security such as standard debt with a positive probability of default, or debt with call options. We also verify that the financial market equilibrium is constrained Pareto optimal.

1. INTRODUCTION be organized so as to best enhance the role of the large investor as a delegated monitor LARGE INVESTORS – INSTITUTIONAL INVESTORS who disciplines the performance of a such as pension funds, mutual funds, company’s management. commercial banks, insurance companies, and To attain our goal, we consider the finance companies, as well as private problem of security design under asymmetric investors such as initial owners – monitor information when investors can freely and control the management of companies trade in the market. In particular, we on behalf of the other investors.1 However, develop a model in which a large investor in recent years, the securitizing of financial affects the return structure of a risky assets and a shift away from bank borrowing project by having access to costly monitor- toward bond and equity financing have ing technology but cannot be made to commit grown significantly.2 If this tendency to monitoring the management of the firm increases the number of shares owned by at any particular level of intensity because small and passive investors, it may reduce her monitoring level is hidden from outside the incentive for large investors to monitor investors. In this framework, one natural the performance of the management of a question is whether standard debt or some given company. This possibility arises from other simple security design would be a free-rider problem: small and passive suitable under reasonable assumptions. investors realize part of the benefits of Another question is how the interaction monitoring performed by large investors between the design of a security, amount of but they incur none of the costs of the the security implemented, and the inten- monitoring.This paper investigates how the sity of the large investor’s monitoring is design and issued amount of security can structured. 136 EQUITY OWNERSHIP STRUCTURE AND CONTROL The main thesis of this paper depends on expectation implies that in this situation a a mechanism in which the design and lower security price will prevail. Such an issued amount of security influences the indirect effect of the design and issued intensity of the large investor’s monitoring, amount of security causes a liquidity cost not only through a direct effect on her that is borne by the large investor, who ex ante payoff but also through an indirect issues the security. Thus, the large investor effect via the market security price on her needs to consider not only the direct effect ex ante payoff. To clarify the basic idea, of the design and issued amount of security, consider first an entrepreneur (a large including a retention cost associated with a investor) who holds a project that generates high unsecuritized or unsold portion of the risky returns in future periods. She may cash flows, but also the indirect effect of keep some of the future returns for herself the design and issued amount of security and securitize the other portion of the through a change in the security price, future returns to raise capital and maintain including a liquidity cost associated with a liquidity. Once the large investor issues a low rate of monitoring. security, she turns effective control of the The above arguments show that there is firm over to a manager. However, the large a potential conflict between the need for investor is assumed to have access to costly liquidity preference and the efficiency of technology for monitoring the management monitoring. Furthermore, these arguments of the firm and thus can affect the return indicate that the design of the security and structure of the project.3 However, if the the amount issued are incentive devices large investor carries out such monitoring, independent of one another for reconciling small and passive investors can free-ride on the friction between these two aims. part of the benefits of the control function Nevertheless, the interaction between the of the large investor, although these design of the security, the amount issued, investors incur none of the costs of and the large investor’s monitoring monitoring. This externality reduces the remains imperfectly understood in the likelihood of the large investor’s carrying literature of corporate governance or out high intensity monitoring unless she finance. Indeed, the literature reveals can be made to commit to prior monitoring several difficulties.The first problem is that levels before she designs and issues the only a limited menu of debt and equity has security. been studied. The second problem is that In this model, the design and issued most of the arguments in this field depend amount of security affects the large on the bilateral contract framework but not investor’s monitoring because of both the on the market equilibrium framework.4 direct effect on her ex ante payoff and the To avoid these difficulties and formalize indirect effect on her ex ante payoff the trade-off between the retention cost through a change in the security price. The and the liquidity cost, we incorporate four logic of the indirect effect is explained as key issues into our model. First, to motivate follows. If small and passive investors are the entrepreneur (large investor) to issue a aware that a change in the design and security backed by future risky returns, we issued amount of security affects the large assume that she must sell her issued investor’s monitoring, the security price at security for liquidity reasons, as studied in which the trade takes place will reflect this Allen and Gale (1988, 1991), Gale awareness. (1992), and DeMarzo and Duffie (1999). More specifically, small and passive Thus, the retention cost arises from retaining investors rationally anticipate a low inten- a large amount of unsecuritized returns or sity of monitoring if the large investor unsold securities. Second, we assume that designs and sells to small and passive the large investor can choose a level of investors a security that is more sensitive monitoring to discipline the management of to the variations of future returns or if she the firm and affect the structure of risky sells a security more aggressively. This returns in future periods, while she cannot SECURITY DESIGN AND INSIDER MONITORING 137 be committed to any specific levels of some extent by monitoring considerations, monitoring until she designs and issues a the optimal security becomes a debt-like security. As a result, the free-rider problem security such as standard debt with a associated with monitoring in the presence positive probability of default or debt with of small and passive investors naturally call options. occurs. This causes a liquidity cost to the We also compare the allocation of the large investor because her weaker incentive financial market equilibrium with three to monitor makes the security price lower. benchmark allocations.The first benchmark Third, we do not take the forms of financial case is concerned with the competitive claims as given, such as debt and equity, but equilibrium allocation, in which the large rather try to derive these instruments as an investor does not take account of the effect optimal security design. By designing secu- of the amount of security issued on her rities, the large investor can balance the sales revenues.The second benchmark case retention cost against the liquidity cost. is the “passive” equilibrium allocation, in Finally, instead of a bilateral contract which the large investor becomes a passive model, we set up a simple market (general) investor, that is, she chooses the minimum equilibrium model in which the large monitoring level. The final benchmark case investor sells her issued security to small deals with the constrained social surplus and passive investors, taking account of the maximizing allocation, in which the social effect of the amount sold on the revenues planner maximizes the social surplus subject received. This framework enables us to to the constraints: (i) that the large investigate the price effect of the design investor initially owns the risky project,5 and issued amount of security. This frame- (ii) that her monitoring level is unobserv- work also allows us to consider how the able to outside investors, and (iii) that she recent trend in the financing patterns of cannot be committed to any specific levels nonfinancial firms away from (indirect) of monitoring before she designs and issues bank borrowing and toward (direct) bond the security. In the first two benchmark financing in Germany and Japan affects cases, we show that the competitive equi- their corporate governance systems. librium allocation is identical with the passive By considering these four issues, we show one. More specifically, under these two that the optimal security is a debt-like equilibrium allocations, either the optimal security such as standard debt with a posi- security is equity or the financial market tive probability of default, or debt with call breaks down, and the monitoring level of options in the presence of the monitoring the large investor is minimized. Compared problem.This result suggests that it cannot to the final benchmark allocation, the allo- be necessarily optimal to issue risk free cation of the financial market equilibrium, debt with zero default probability.The intu- in which the large investor considers the itive reason for this result is that in the effect of the amount sold on the revenues present model, the large investor wants to received, is proved to be constrained Pareto sell some portion of the security and optimal: it attains the constrained social receive the revenues from the sale of the surplus maximizing allocation. security in the initial period to maintain Our research is related to two strands of liquidity; as a result, to promote sales the literature. One is concerned with security revenues, the large investor is motivated to design under the bilateral contract or the set security claims at each possible contin- general equilibrium model, while the other gency to increase by the maximum amount is involved with insider or speculator mon- that the firm can pay by taking account of itoring under the general equilibrium the adverse effect on monitoring. If the model. In the analysis of security design security claims were not constrained by under the bilateral contract model, Hart monitoring considerations, then the optimal and Moore (1989) and Bolton and security would become equity. In fact, since Scharfstein (1990) show that standard the security claims must be constrained to debt is an optimal contract when managers 138 EQUITY OWNERSHIP STRUCTURE AND CONTROL can appropriate to themselves income not financial market equilibrium. Section 4 paid out. Berglöf and von Thadden (1994), compares the allocation at financial mar- incorporating partial liquidation and multi- ket equilibrium with the three benchmark ple investors into the Hart and Moore allocations. Section 5 summarizes our model, discuss whether short-term debt, results and discusses some directions for long-term debt, or equity are the best secu- future study. rities the firm can issue. However, these papers do not investigate the monitoring activity of investors. Townsend (1979), 2. THE MODEL Diamond (1984), Gale and Hellwig (1985), and von Thadden (1995) do con- We consider a model in which there exist sider the monitoring activity of investors one large investor (the issuer) and a set of under the bilateral contract model, and outside investors.The large investor owns a prove that standard debt is an optimal risky project that generates future cash contract. Although these papers make flows, and also has access to some moni- important contributions, they shed little toring activities by which she can affect the light on the effect of security design on the structure of the returns of the risky project. market security price because their frame- We assume that all investors are risk work is restricted to bilateral contracts. neutral and normalize the market interest The problem of optimal security design rate to zero. in the general equilibrium model is ana- The model has three periods, indexed lyzed with symmetric information by Allen t 0, 1, 2. In the initial period, the large and Gale (1988, 1991), and with adverse investor designs a security whose return selection by Boot and Thakor (1993), represents a claim backed by some part of Nachman and Noe (1994), Demange and the risky project returns of the firm, with Laroque (1995), Ohashi (1995), Rahi the total supply normalized to one. The (1995, 1996), and DeMarzo and Duffie large investor also offers the security for (1999).This line of research, however,does sale on the security market. The large not examine the implications of monitoring investor then decides to sell a fraction done by investors under asymmetric qL ∈ [0, 1] of the security to the market at information. a price p. The large investor herself keeps 7 In work most closely related to ours, the fraction (1 – qL) of the security. On the Holmström and Tirole (1993) and Admati, other side of the market, outside investors Pfleiderer, and Zechner (1994) explore the establish their share of security allocations, implications of monitoring done by qO ∈ [0, 1]. Once the large investor issues investors under the general equilibrium the security, she turns effective control of model with asymmetric information.6 the risky project (firm) over to the man- Holmström and Tirole develop a model in agement. In period 1, the large investor has which the equity ownership structure of access to costly technology of monitoring firms affects the value of market monitoring the management, and can affect the return through its effect on market liquidity. structure of the risky project. Given the Admati, Pfleiderer, and Zechner study the costs and benefits of monitoring, the large effect of monitoring done by large share- investor chooses the level of monitoring. holders on security market equilibrium. Our Note that the large investor makes the analysis goes beyond these two interesting monitoring decision by taking into account papers by considering the problem of the share of security allocations established security design endogenously instead of in period 0.There is no further trading from focusing on particular financial claims period 1 to period 2. In period 2, the such as equity. returns of the risky project are realized and The paper is organized as follows. the claims of the security are paid off to Section 2 describes the basic model. security holders. The large investor also Section 3 characterizes the allocation at receives the residual claims that remain SECURITY DESIGN AND INSIDER MONITORING 139 after the security claims are paid to cm, where c is the unit monitoring cost. Let security holders including the large m be the minimum monitoring level investor herself. required to sustain the monitoring To formalize the model, we require process.12 specific assumptions about the liquidity All the project returns at each state, (Xg, motive of the large investor, the preference Xb), and the fraction of the security sold by of outside investors, and the monitoring the large investor to the market, qL, can be technology of the large investor. verified with no cost. On the other hand, the To provide the motivation for liquidation, monitoring level chosen by the large we assume that the large investor evaluates investor, m, cannot be observed by any the period 2 cash flows at a rate ∈ (0, 1). outside agents. Thus, the large investor This assumption implies that the large cannot be committed to any specific levels investor is indifferent as to whether she of monitoring until period 1. The functions keeps future project returns or sells them (g(m), b(m)) and the parameters for cents in the dollar in cash by securi- (Xg, Xb, c, ) are assumed to be common tizing them. This also means that, if the knowledge. large investor retains a portion of the secu- The security is represented by a claim, rity, the private value of one dollar’s worth that is, any promise to make a future pay- of security returns to the large investor is . ment, contingent on the state of nature Similar assumptions are made in Allen and (that is, the project returns). Let F˜ be a Gale (1988, 1991), Gale (1992), and real-valued random variable that generates 8 DeMarzo and Duffie (1999). Since we a security claim Fg at the good state, and assume that high transaction costs deter a security claim Fb at the bad state, the large investor and outside investors respectively. We focus our attention on the from building a bilateral contract relation, set of admissible securities. These are the large investor has an incentive to defined by a set of functions satisfying the securitize some portion of the future risky following limited liability and monotonicity project returns and sell the security to out- conditions: side investors. In contrast, outside investors are 0 Fs Xs, s g, b, (1) assumed to have no liquidity motives. Since F F . (2) outside investors do not care when asset b g returns are generated, they have no concern In the subsequent discussion, we denote the about the timing of payments. set of admissible securities as . In the The large investor has access to costly limited liability condition (1), the restric- technology of monitoring the manage- tion of 0 Fs for s g, b expresses limited ment.The technology affects the structure liability for security holders.The restriction of the returns of the risky project as of Fs Xs for s g, b implies limited lia- follows:9,10 bility for the large investor as a residual claimant. The monotonicity condition (2) X , with probability (m), X˜ g g shows that the security claim is nonde- Xb , with probability b (m), creasing in the available firm cash flows.13 These assumptions are quite common in where Xg Xb 0, (g(m), b(m)) ∈ the finance literature, and are justified in {(g(m), b(m)) | g(m) b(m) 1, Innes (1990), Nachman and Noe (1994), g(m) 0, b(m) 0}, and m is the inten- Hart and Moore (1994), and DeMarzo and sity of monitoring.11 One part of the Duffie (1999). efficiency of monitoring is captured by the Now, the security represented by the function g(m), where g’(m) 0 and claim (Fg, Fb) pays an amount min (Fs, Xs) ”g(m) 0.The other part of the efficiency for s g, b when the returns of the project of monitoring is characterized by the are realized in period 2.14 The large cost of monitoring at level m, expressed by investor then receives the cash revenues from 140 EQUITY OWNERSHIP STRUCTURE AND CONTROL the sale of the fraction qL of the security in Since outside investors do not behave period 1, and the residual cash flows of strategically, they take the security price p max (Xs Fs, 0) for s g, b and the as given. claims for the unsold fraction (1 qL) of A financial market equilibrium is now the security in period 2. characterized by (m, F˜, qL, qO, p) that The trading process is described as a satisfies the following conditions (see Walrasian process in which the large Figure 5.1): investor is strategic. The large investor FINANCIAL MARKET EQUILIBRIUM: chooses the fraction of the security sold to (i) Sequential Optimality of the Large the market, q , by taking into account the L Investor. In period 1, the large investor effect of the issued amount on the security chooses a monitoring level m such that price, p. We assume that the large investor perceives an inverse demand correspon- p ∈ P q F˜ ˜ dence ( L; ) for the issued security L1(F˜, qL) max RL(m, F, qL). (5) mm F˜, where P : [0, 1] →ℜ. In the next ˜ section, we specify how P(qL;)F is determined. On the other hand, outside In period 0, given the knowledge of the investors behave as representative, risk - price correspondence P(qL;F˜ ), the large neutral, price-taking investors. Thus, out- investor chooses a design F˜ and a fraction side investors do not receive expected net qL of the security sold to the market gains from buying any security at market such that equilibrium. ˜ The expected payoff of the large investor L0 max L1(F, qL), (6) ˜ʦ ʦ in period 0 is then given by F ,qL [0,1] where is the set of admissible securities ˜ RL(m, F˜, qL) E max (X F˜, 0) defined by (1) and (2). (ii) Optimality of Outsider Investors.In (1 q )E min (F˜, X˜) L period 1, outside investors purchase a fraction qLP(qL; F˜) cm, (3) qO of the security supplied in the market, relative to an observed price p, such that where E is the expectation operator. The right-hand side of (3) consists of the (p) max R ( q , p). (7) expected residual claims retained by the 0 o o qoʦ[0,1] large shareholder, E max (X˜ F˜, 0) the expected claims for the unsold fraction (iii) Market Equilibrium.The security price (1 qL) of the security, (1 qL)E min p and the sold and purchased fractions ˜ ˜ (F, X ), the cash revenues raised from the (qL, qO) must satisfy sale of the security, qLP(qL;F˜ ), and the ˜ monitoring cost, cm.The first two terms p ʦ P(qL; F ), (8) are discounted by the rate because the and returns of these two claims accrue only in qL qO. (9) period 2. The monitoring cost is not discounted, since it must be paid in period 1. Note that the large investor selects a Note that the first two terms clearly security design F˜ from the set of admissible depend on the monitoring level m because securities , a sold fraction qL from [0, 1], the realized probability of the state is and a monitoring level m from (m, ∞). affected by the choice of m. Outside investors select a purchased fraction The expected payoff of outsiders in qO from [0, 1]. Since the large investor can- period 0 is similarly represented by not be committed in advance to any specific levels of monitoring, she must solve the ˜ ˜ Ro(qo, p) qo [E min (F, X) p]. (4) sequential problem made up of (5) and (6). SECURITY DESIGN AND INSIDER MONITORING 141

period 0

large investor risky project (insider)

security design F = (Fg , Fb)

security trade

unsold fraction sold fraction (1-qL) (qL)

security market outside investors

qL = qO (qO) security price (p)

period 1 large investors

monitoring (m) management monitoring cost (cm)

period 2 stochastic project returns

Rg R = g Rb b

Figure 5.1 Sequence of the events.

3. FINANCIAL MARKET 0if E min (F˜, X˜ )< p, EQUILIBRIUM q0 ʦ [0,1] if E min (F˜, X˜ ) p, 0 if E min (F˜, X˜ )< p. In this section, we first derive the inverse demand correspondence P(qL;F˜ ) from the (10) condition of the optimality of outside investors and the condition of market Combining (9) and (10), we have equilibrium by taking as given a security design chosen by the large investor,F˜ . We next solve the sequential optimization 0if E min (F˜, X˜ ) < p, problem of the large investor in two steps, qL ʦ [0,1] if E min (F˜, X˜ ) p, and determine the remaining variables 0 if E min (F˜, X˜ ) > p. endogenously. (11) 3.1. Inverse demand correspondence The relation between qL and p determined Solving the optimization problem of out- by (11) gives the inverse demand corre- side investors, (7), with (4), we see spondence p ∈ P(qL;).F˜ 142 EQUITY OWNERSHIP STRUCTURE AND CONTROL 3.2. The monitoring problem faced We next examine the case of E min ˜ ˜ by the large investor (,F X ) p, where qL ʦ [0, 1] from (11). Rearranging (3) with the definitions of X˜ We now proceed to the sequential and F˜ , we explicitly describe the interim optimization problem of the large investor. maximization problem (5) by We discuss the optimality conditions for this problem by moving backward from ˜ period 1 to period 0. max RL(m, F, qL) m We begin with solving the monitoring m problem (5) faced in period 1 by the large max max(Xi Fi, 0) i(m) investor, who designs a security F˜ and mm ig,b holds a fraction 1 q of the security L (1 qL) min(Fi Xi) after trading. Since the large investor takes ig,b the price correspondence pPʦ (qL; F˜) as given because qL and F˜ are determined in i(m) pqL cm. (15) period 0, it follows from (11) that the following three cases are distinguished: E min (F˜ ,X˜ ) p, E min (F˜ ,X˜ ) p, or We should again keep in mind that the E min (F˜ ,X˜ ) p. security price p is determined in period 0. We first discuss the case of E min Thus, the large investor in this stage takes (,F˜ X˜ ) p, where q 0 from (11). Given the security price p as given. L In the subsequent analysis, we assume (1), (3) and qL 0, the objective function of (5) is reduced to that the maximization problem (15) has an interior solution that satisfies the second- ˜ ˜ order sufficient condition for m to be a RL(m, F,0) EX cm. (12) 15 local maximum of RL (m, F˜, qL). Then, the first-order condition for the problem Rewriting (12) with the definition of X˜ , we (15) implies a local optimality condition, specify the interim maximization problem of and is expressed by the large investor as follows: max R (m, F˜,0) mm L max(Xi Fi, 0) i(m) ig,b max X m cm F X i i( ) (13) (1 qL) min( i i) m m ig,b ig,b Let us assume that the maximization i(m) c 0. (16) problem (13) has an interior solution. Since the objective function (13) is Note that although the large investor pays globally concave with respect to m under the full cost of monitoring, she enjoys only a Xg Xb, the first-order condition for this part of the benefits of the monitoring problem becomes a global optimality because the monitoring also affects the condition. Solving (13) with g(m) claims on the outsiders’ holdings of the b(m) 1, we see the following first-order security. Thus, the monitoring level is condition: always less than ex post efficient, which is achieved if the large investor is the sole ’ (Xg Xb)g(m) c 0. (14) owner of the security. In fact, (16) is not easily tractable. To In this case, the monitoring level is ex post obtain a more manageable form, let us efficient because the large investor holds distinguish the following four cases by all the issued security. However, she cannot exploiting the monotonicity condition (2): satisfy her liquidity needs. (A) Fb Fg Xb Xg or Fb Xb Fg Xg, SECURITY DESIGN AND INSIDER MONITORING 143

Fb Fg Xb Xg

∆ (A)

Fb Xb Fg Xg

Fb Xb Fg = Xg

∆ (B)

Fb = Xb Fg Xg

∆ (C)

Fb =Xb Fg =Xg

∆ (D)

Figure 5.2 Four possible cases of F˜ .

(B) Fb Xb Xg Fg, (C) Xb Fb Fg and Section 4, the optimality condition will be Xg, and (D) Xb Fb Xg Fg or Xb Xg fully exploited. Fb Fg. Furthermore, given the limited Finally, we investigate the case of E min ˜ ˜ liability condition (1), the four possible (,F X ) p, where qL must be equal to 1 cases of F˜ introduced above are reduced to from (11). Then, it is immediately seen the following four possible sets F˜ (see from (3) and qL 1 that the objective Figure 5.2): (A) ≡ {(Fg, Fb) | Fb Fg function of (5) is described by Xg and 0 Fb Xb}, (B) ≡ {(Fg, Fb) | 0 ˜ ˜ ˜ Fb Xb Xg Fg}, (C) ≡ {(Fg, Fb) | RL(m,,1)F E max (X F, 0) Xb Fb Fg Xg}, and (D) ≡ {(Fg, Fb) | P(1;F˜ ) cm. Xb Fb Xg Fg}. Note that (A) is ˜ identical to the set F , {(Fg, Fb) | Fb Fg Because of E min (F˜ ,X˜ ) p for pPʦ (1;F˜ ) Xb Xg or Fb Xb Fg Xg}, under con- from (11), we have E min (F˜ ,X˜ ) sup P(1;F˜ ). ditions (1) and (2). Similarly, (D) is iden- Thus, using the definitions of F˜ and F˜ , we find ˜ tical with the set F , {(Fg, Fb) | Xb Fb X F or X X F F }, under con- ˜ g g b g b g max RL(m, F,1) ditions (1) and (2). From now on, we divide mm the space F˜ into (A), (B), (C), and (D). For each division of the space of F˜ , max max(Xi Fi,0)i(m) mm the first-order condition for an interior ig,b solution to the monitoring problem is specified in the Appendix. In Subsection 3.3 P(1; F˜) cm 144 EQUITY OWNERSHIP STRUCTURE AND CONTROL problem (6) is provided by (A5)–(A8) in the max max(Xi Fi,0)i(m) Appendix. mm i g,b The most notable feature of these optimality conditions is that changes in min(Fi Xi)i(m) cm ig,b the design of the security, (Fg, Fb), or in the issued amount of security, q has both lim max R (m,F˜, q ). (17) L, L L a direct effect on the monitoring level and qL→1 m m an indirect effect through a change in the Here, the inequality in (17) is derived security price. If the large investor did not ˜ ˜ ˜ from E min (F ,X ) sup P(1;F ), and the take into account the effect of a change in final equality in (17) is obtained from the (Fg, Fb) or qL on sales revenues, the opti- limit of the right-hand side of (15) mal strategy would be given by F X ˜ ˜ g g, with E min (F ,X ) p, as qL converges Fb Xb, and qL 1 in all possible cases, to unity. unless the security market breaks down. The arguments of the three cases exam- However, this implies that the large ined above suggest the following. First, we investor would securitize all the project ˜ ˜ can rule out the case of E min (F ,X ) p. returns and sell all of the security. Thus, In this case, we cannot examine the problem the monitoring level would be minimized, of the security issue because the large which might cause a collapse of the secu- investor holds all the issued security rity market: outside investors would (qL 0). Second, we can substitute the rationally anticipate the strategy of the ˜ ˜ case of E min (F ,X ) p for the case of large investor and participate in the ˜ ˜ E min (F ,X ) p because the latter is security market only if the security price is included in the former as qL converges to low enough. unity.We will, therefore, focus on the case of On the basis of the optimality conditions ˜ ˜ both E min (F ,X ) p and qL 0 in the (A5)–(A8) given in the Appendix, we now subsequent analysis. have the following proposition on the optimal strategy of the large investor for ˜ 3.3. The problem faced by the large each possible set of F .The detailed proof is investor of issuing the security. presented in the Appendix. We are now in a position to discuss the PROPOSITION 1. (i) If F˜ is restricted optimization problem of issuing the secu- to (A) or (C), the optimal solution rity, (6), faced by the large investor in to the constrained maximization period 0. To do so, we first solve the problem, (m*, F*g, F*b, q*L), is character- problem (6) with the restriction in which ized by ˜ F (Fg, Fb) belongs to one of the four sets defined in the preceding subsection: X X q* F* X ’ m* c (A) F˜ ʦ (A), (B) F˜ ʦ (B), (C) F˜ ʦ (C), [ g b L( g b)] g( ) , and (D) F˜ ʦ (D). Then, we compute the (18a) optimal value for the problem (6) under * * each of the four possible restrictions, and Fg min Xb (1 ) g(m ) choose the maximal value among the opti- 1 mal values of the four possible cases. The m* g(m*) , Xg, (18b) corresponding solution gives an optimal Fg* ˜ security design F(Fg, Fb) and an optimal level of the sold fraction q to the problem * L Fg Xb, (18c) (6). In contrast to the analysis of the monitoring problem, the large investor must * * * consider the effect of a change in her (1 )[Fgg(m ) Xbb(m )] strategy on the security price. A detailed * * ’ m* qL(Fg Xb)g(m*) 0. (18d) analysis of the optimizing conditions to the q*L SECURITY DESIGN AND INSIDER MONITORING 145 The large investor’s payoff is then begin with (18a). Given the security price determined in period 0, an increase in the A c * * L0 L0 [Xgg(m ) Xbb(m )] monitoring effort in period 1 affects the ex ante payoff of the large investor through two routes. One effect increases * * * qL(1 )[Fg g(m ) the expected project returns, thereby further increasing not only the discounted * * Xbb(m )] cm . (18e) residual returns to be received by the large investor,that is,(Xg Fg* Xb Fb*)g’(m*), but also the discounted security returns for (ii) If F˜ is restricted to (B) or (D), the unsold portion to be received by the the optimal solution to the constrained large investor, that is, (1 q*) (F* F*) maximization problem, (m**, F**, F**, q**), L g b g b L (m*).The combined effects increase the is represented by ’g ex ante payoff of the large investor, ** ’ ** (1 qL )(Xg Xb)g(m ) c, (19a) expressed by the left-hand side of (18a). The other effect increases the total moni- ** Fg Xg, (19b) toring cost, c, thus decreasing the ex ante payoff of the large investor.This cost effect ** Fb Xb, (19c) is represented by the right-hand side of ** ** (1 )[Xgg(m ) Xbb(m )] (18a). The monitoring level is then chosen to balance the former benefit against the m** q**(X X )’ (m**) 0. latter cost. L g b g ** qL Before proceeding to (18b) and (18c), we discuss the implications of (18d) (19d) because it is then more straightforward to The large investor’s payoff is then understand (18b) and (18c). An increase in the sold fraction has two potentially B D conflicting effects on the ex ante payoff of ** Lo Lo [ qL (1 )] the large investor. One effect causes the large investor to receive higher revenues ** ** ** [Xgg(m ) Xbb(m )] cm . from the sale of the security. This effect enables the large investor to satisfy any (19e) desire to obtain as much revenue as possible at date 0 rather than at date 1 thus Proposition 1 shows that, if F˜ is restricted directly increasing her ex ante payoff, to (A) or (C), the optimal security can expressed by (1 )p* (1 ) be a debt-like security with Fg* Xg and [F*g g(m*) Xb b(m*)]. The other effect F*b Xb. One interpretation of this kind of of an increase in qL reduces the correlation security is standard debt with default such between the ex post project returns and the as that analyzed in DeMarzo and Duffle ex post payoff of the large investor because (1999) and von Thadden (1995): Fg*can be she obtains the more deterministic sales viewed as the face value, and F*b as the revenues at period 0. Since this effect leads default payment of debt. Another interpre- to a lower monitoring level, outside tation is debt with call options (convertible investors rationally anticipate a lower bond or debt with warrant): Xg Fg* can intensity of monitoring. This expectation then be interpreted as the option premium. reduces the security price, thus indirectly Proposition 1 also indicates that if F˜ is decreasing the large investor’s revenues restricted to (B) or (D), the optimal from the sale of the security, represented by security is equity whose claim fully reflects q*L(F*g Xb)g’(m*)[∂m*/∂q*L].The former the project returns. direct benefit cancels out the latter indi- It is instructive to explore the implica- rect cost at the optimal level of the sold tions of equations (18) and (19). Let us fraction. 146 EQUITY OWNERSHIP STRUCTURE AND CONTROL

* * * * We can now examine the implications q L(1 )b(m ) qL(Fg Xb) * * * of (18b) and (18c), and show how the ’g(m )∂m /∂Fb becomes positive. The design of the security affects the ex ante reason for F*b Xb mainly depends on payoff of the large investor. If F*g Xg, these facts: (i) that an increase in Fb q we multiply both sides of (18b) by *L and further motivates the large investor to attain rewrite it: the good state because its direct effects decrease her discounted bad state returns * * * * qLg(m ) qLg(m ) mainly through a decline in the discounted m* bad state residual returns to be received q*(F* X )’ (m*) 0. (20) L g b g * by the large investor, and (ii) that its posi- Fg tive indirect effects dominate its negative Given (20), an increase in Fg brings two direct effects. direct effects to the ex ante payoff of the In a similar manner,we can investigate the large investor. The first increases the implications of (19) by substituting Xg for Fg. discounted good state security returns to be Using Proposition 1, an optimal solution received by the large investor, expressed by to the sequential optimization problem δ (1 q*L) g(m*).The second direct effect faced by the large investor,(m, Fg, F b, qL), in turn decreases the discounted good state is represented by residual returns to be obtained by the large (m, Fg, F b, qL) investor, represented by δ g(m*). The combined two direct effects are then sum- A B , > LO marized by q*L δg(m*), which is negative (m*, F*g ,F*b ,q*L)if LO (m**,F**,F**,q**)ifA B and is shown by the first term in the left- g b L LO LO . F hand side of (20). An increase in g also (21) leads to two indirect effects on the ex ante payoff of the large investor due to a change A B in the security price. Due to an increase in Here, L0 and L0 are determined by the security returns at the good state, one (18e) and (19e), respectively. indirect effect raises the security price. This A B Comparing L0 with L0, we now obtain effect results in an increase in the large the following proposition that characterizes investor’s revenues from the sale of the the optimal security design. The detailed q m security, represented by *L g( *), which is proof is provided in the Appendix. positive and is indicated by the middle term in the left-hand side of (20). The other indi- PROPOSITION 2. (i) In the presence of rect effect of an increase in Fg reduces the the monitoring problem, the optimal secu- incentive for the large investor to attain the rity is standard debt with default or debt good state because the combined two direct with call options (F*g Xg and F*b Xb). effects represented by the first term in the (ii) In the absence of the monitoring prob- left-hand side of (20) decrease her dis- lem, the optimal security is equity. counted good state returns.This indirect effect causes the large investor to choose a lower One crucial implication of this proposition is monitoring level. Since outside investors that it is not optimal to issue risk-free debt anticipate this, the security price and the large such as Fg* F*b because risk-free debt pre- investor’s revenues from the sale of the secu- vents the large investor from satisfying the rity decrease. This effect is captured by the desire to obtain the revenues as much as pos- final term in the left-hand side of (20). The sible at date 0 rather than at date 1. Another claim at the good state, Fg, is thus determined important implication of this proposition is by considering all of these effects. that in the presence of the monitoring prob- In contrast, the claim at the bad state, lem, it is not optimal to issue equity. Fb, is found to be on the boundary of Fb Xb, A natural intuition is that risk-free debt since the total effect of an increase in Fb on would be an optimal security if we neg- the ex ante payoff of the large investor, lected the large investor’s desire to receive SECURITY DESIGN AND INSIDER MONITORING 147 the revenues as much as possible at date 0 possible outcomes and so on. Second, rather than at date 1. This is because the Admati, Pfleiderer, and Zechner (1994) large investor would receive the full discuss the monitoring activity of the large margin benefit from monitoring as well as investor when only equity can be issued. bearing the full marginal cost of monitoring. However, Proposition 2 implies that the However, in the present case, the large optimal security becomes debt-like in the investor can sell some portion of the secu- presence of the monitoring problem, rity and receive the revenues from the sale although our framework is restricted. This of the security. Since the discount factor suggests that the security design problem is less than one, the sales revenues received needs to be analyzed endogenously in the in period 0 are higher than the residual study of the monitoring of the large investor. claims received in period 2. Hence, unless the monitoring level decreases through a change in security payments, the large 4. NORMATIVE ANALYSIS OF THE investor is motivated to increase the sales FINANCIAL MARKET EQUILIBRIUM revenues received in period 0. This implies that, in the absence of monitoring consid- In this section, we compare the financial erations, the large investor is motivated to market equilibrium in the preceding section set security claims at each state to with several benchmark cases: the compet- increase by the maximum amount that the itive equilibrium allocation, the passive firm can pay. If the security claims need equilibrium allocation, and the constrained not be constrained by monitoring consider- social surplus maximizing allocation. ations, then the optimal security becomes The competitive equilibrium allocation equity because F*g Xg and F*b Xb.On deals with the situation in which the large the other hand, if the security claims are investor does not take account of the effect constrained by monitoring considerations of the issued amount of security on her to some extent, the large investor needs to sales revenues. The passive equilibrium enhance her monitoring activity by allocation arises from the situation in increasing the residual revenues received which the large investor chooses the mini- at the good state in period 2. The large mum monitoring level. The constrained investor would therefore face the possibil- social surplus maximizing allocation corre- ity that the optimal security becomes stan- sponds to the constrained Pareto optimal dard debt with default because she must allocation such that the social planner set F*g less than Xg while keeping F*b Xb. maximizes the social surplus subject to sev- More specifically, in the presence of eral information constraints. In the analysis monitoring considerations, the large that follows, we show that the financial investor needs to consider the trade-off market equilibrium in the preceding section between a decline in Fg from Xg and a rise is constrained Pareto optimal, whereas in qL. In other words, she needs to balance the competitive and passive equilibrium the costs of securitizing a smaller portion allocations are not. With regard to the of the project returns against the costs of competitive and passive equilibrium selling a smaller portion of the issued secu- allocations, we suggest that they are identical. rity.16 Since the total effects of a decrease in Fg from Xg and an increase in qL on the ex PROPOSITION 3. If the large investor does ante payoff of the large investor per issued not take account of the effect of the issued amount are positive, the optimal security amount of security on her sales revenues, the involves debt-like types such as standard optimal allocation, (mˆ , Fˆg, Fˆb, qˆL), would be debt with default or debt with call options. given such that the monitoring level of the Several remarks about this proposition large investor,mˆ , is minimized: more specif- are in order.First, we should notice that the ically,mˆ = m, Fˆg =Xg, Fˆb = Xb, and qˆL 1. result of this proposition depends on Proof. If the large investor does not take the simplifying assumptions such as two into account the effect of the issued 148 EQUITY OWNERSHIP STRUCTURE AND CONTROL amount of security on her sales revenues, Let S(m, F˜, q) be the social surplus the optimality conditions for the problem defined by (6) provided by (A5)–(A8) in the Appendix S (m, F˜, q) EX˜ q(1 ) show that her optimal strategy is given by E min (F˜ ,X˜ ) cm, (23) mˆ m, Fˆg Xg, Fˆb Xb, and qˆL 1 in all possible cases, provided the security market where the first term represents the project does not collapse. returns, the second term expresses the liquidity benefits of the large investor, and the third The intuitive explanation for Proposition 3 term indicates the monitoring cost. is clarified as follows. If the large investor Then, the constrained social surplus does not consider the effect of the issued maximizing allocation is formally charac- amount of security on her sales revenues, it terized by the following sequential is optimal for the large investor to securi- maximization problem: tize all the project returns and sell all of the security, because she evaluates the sales Constrained social surplus maximizing revenues received in period 0 more than the allocation. In period 1, the social planner residual claims received in period 2. Since chooses a monitoring level m such that the large investor need not take account of the monitoring problem, she can minimize S1(F˜, q) max S(m, F˜, q). (24) the monitoring level. However, this might mm cause the collapse of the security market In period 0, the social planner chooses a because outside investors would rationally design F˜ and a fraction q of security sold to anticipate the large investor’s strategy and outside investors such that participate in the security market only if the security price is low enough. S0 max S1(F˜, q). (25) Proposition 2 indicates that in the F˜ʦ,qʦ[0,1] presence of the monitoring problem, the optimal security is a debt-like security if Given (A1) and (A3) with E min (F˜ ,X˜ ) p the large investor takes account of the in the Appendix, it can be seen immediately effect of the amount sold on the revenues from (23) that the constrained social received. On the other hand, Proposition 3 surplus maximization problem is equivalent suggests that even in the presence of the to the sequential optimization problem of monitoring problem, the optimal security is the large investor at financial market equi- equity if the large investor does not take librium. Thus, we obtain the following account of the effect of the amount of proposition: security issued on her sales revenues, that is, if she chooses the minimum monitoring PROPOSITION 4. If the large investor level. The result of Proposition 3 depends takes account of the effect of the amount on the assumption that the large investor sold on the revenues received, the financial cannot be committed to any prior levels of market equilibrium characterized in the pre- monitoring until she designs and issues the ceding section is constrained Pareto optimal security. in the sense that it attains the constrained We next examine the relation between social surplus maximizing allocation. the financial market equilibrium and constrained social surplus maximizing In view of Propositions 2–4, we see that allocations, defined such that the social the competitive equilibrium allocation is planner maximizes the social surplus subject not constrained Pareto optimal. Thus, the to the constraints: (i) that the large investor large investor must take into account the initially owns the risky project,17 (ii) that her effect of the issued amount of security on monitoring level is unobservable to outside her sales revenues to attain the constrained investors, and (iii) that she cannot be com- social surplus maximizing allocation. mitted to any specific levels of monitoring This finding is reminiscent of the results of before she designs and issues the security. Allen and Gale (1991) in a different context. SECURITY DESIGN AND INSIDER MONITORING 149 The intuition behind this proposition is constrained Pareto optimal or the market that the social surplus can be transformed collapses. Furthermore, unless the market into the surplus of the large investor as a breaks down, the optimal security in this result of the price adjustment at financial case is always equity; and the monitoring market equilibrium. This finding mainly level of the large investor is minimized. depends on both the price-making behavior of As suggested in DeMarzo and Duffie the large investor and the risk neutrality (1999), the model rather closely fits into of the large investor and outside investors. the context of the design of corporated securities that are sold by an informed underwriter who retains some fraction of the security of issuing firms. In Germany 5. CONCLUSION and Japan, the recent trend in the financing patterns of nonfinancial firms away from This paper has developed a framework for (indirect) bank borrowing and toward analyzing a problem of security design in (direct) bond financing has been affecting the presence of monitoring done by a their corporate governance systems. large investor to discipline the manage- However, since German house banks and ment of a firm. Since the large investor Japanese main banks usually retain the enjoys only a part of the benefits brought security of their borrowing firms, our result about by her monitoring activities but indicates that these intermediaries may still incurs all the associated costs, she needs discipline the management of firms by to structure the problem of security designing a debt-like security for their design to motivate herself to make an borrowing firms and retaining some portion efficient level of monitoring if she cannot of the security. be committed to any prior levels of moni- Within our framework, many issues toring before she designs and issues the remain to be investigated. Among other security. In fact, it is costly to design and things, the problem of multi-security design issue the security to attain an efficient is an important question that we hope to level of monitoring because of both the pursue in future research. Furthermore, the direct effect on the ex ante payoff of the spanning issue caused by the risk averting large investor and the indirect effect on actions of investors should also be exam- her ex ante payoff through a change in the ined. Finally, as has been recently studied security price. under the incomplete contract model, the By assuming that the large investor bankruptcy and liquidation problem is an takes account of the effect on her sales rev- important topic of corporate governance enues of the amount of security issued, we (see Hart and Moore (1989), Bolton and have shown that the optimal security is a Scharfstein (1990), Aghion and Bolton debt-like security such as standard debt (1992), Berglöf and von Thadden (1994), with a positive probability of default, or and Dewatripont and Tirole (1994)). It will debt with call options in the presence of the be very interesting to explore how the design monitoring problem. This result suggests of securities can resolve the bankruptcy and that it is not necessarily optimal to issue a liquidation problem at financial market security with zero default probability such equilibrium. as risk-free debt, and that the monitoring model of Admati, Pfleiderer, and Zechner (1994) is restricted in the sense that only APPENDIX equity is permitted to be issued. We have also proved that the financial market equi- 5.1. First-order condition (16) for librium is constrained Pareto optimal. In each divition of the space F˜ : contrast, if the large investor does not take into account the effect of the amount of For each possible set of F˜ , we obtain the security issued on her sales revenues, the corresponding value of the objective competitive equilibrium allocation is not function of (5) and the corresponding 150 EQUITY OWNERSHIP STRUCTURE AND CONTROL expression for the first-order condition 5.2. Optimizing conditions for the (16): Objective function of (5) in the case problem (6): of E min (F˜ ,X˜ ) p: For this analysis, we need to respecify the (A) F˜ ʦ (A): ˜ objective function L1 (F, qL) of the problem ˜ (6) for each possible case generated from RL(m, F, qL) [(Xg Fg)g(m) the restriction of F˜ . Substituting E min (Xb Fb)b(m)] (F˜ ,X˜ ) for p in (A1A)–(A1D) yields: (1 qL)[Fgg(m) F (m) pq cm, b b L ˜ (A1A) (A) F ʦ (A): A ˜ (B) F˜ ʦ (B): L1 (F,qL) [Xgg(mA) Xbb (mA)] ˜ RL(m, F, qL) (Xb Fb)b(m) qL(1 )[Fg g(mA) (1 qL)[Xgg(m) Fbb(mA)] cmA, Fbb(m)] pqL cm, (A1B) (A3A) (C) F˜ ʦ (C): (B) F˜ ʦ (B): ˜ B ˜ RL(m, F, qL) (Xg Fg)g(m) L1 (F,qL) [Xg g(mB) Xb b (mB)] (1 qL)[Fgg(m) qL(1 )[Xg g(mB) Xbb(m)] pqL cm, (A1C) Fb b(mB)] cmB, (D) F˜ ʦ(D): (A3B)

RL(m, F˜, qL) (1 qL)[Xgg(m) ˜ ʦ Xbb(m)] pqL cm. (C) F (C): (A1D) C L1 (F˜,qL) [Xgg(mC) Xbb (mC)] First-order condition (16): qL(1 )[Fg g(mC) ˜ ʦ (A) F (A): Xbb(mC)] cmC, ’ [Xg Xb qL)(Fg Fb)]g(m) c, (A3C) (A2A) ˜ (B) Fʦ(B): (D) F˜ ʦ(D): [X X q )(X F )]’ (m) c, D g b L g b g L1 (F˜,qL) [ qL(1 )] (A2B) [Xgg(mD) Xbb(mD)] ˜ ʦ (C) F (C): cmD, (A3D) ’ [Xg Xb qL)(Fg Xb)]g(m) c, (A2C) where mA, mB, mC, and mD denote the monitoring levels that satisfy the optimality (D) F˜ ʦ (D): conditions (A2A), (A2B), (A2C), and (A2D), respectively. ( q X X ’ m c 1 L)( g b) g( ) . (A2D) Now, for each possible set of F˜ , we solve the following constrained maximization We should again notice that the large investor problem: takes the security price as given when solving the monitoring problem in period 1. i max i (F˜, q ), In Subsection 3.3, we will use (A1) and L0 L1 L F˜ʦ(i),qLʦ[0,1] (A2) to solve the problem (6) faced by the large investor in period 0. i A, B, C, D. (A4) SECURITY DESIGN AND INSIDER MONITORING 151

Using the envelope theorem and (C) F˜ʦ (C): (A2A)–(A2D), the first-order conditions ’ for each constrained maximization problem qL(1 )g(mC) (Fg Xb)g are described as follows:

(A) F˜ ʦ (A): mC (mC) g0 g1 0, (A7a) Fg qL(1 )g(mA) (Fg Fb) (1 )[Fgg(mC) Xbb(mC)] m m q (F X ) ’ (m ) C 0, ’ (m ) A 0, (A5a) L g b g C q g A F g0 g1 L g (A7b) qL(1 ) b(mA) (Fg Fb) where g0, g1, and ς are the nonnegative multipliers associated with Fg Xb, Xg m Fg and 1 qL, respectively. The partial ’ (m ) A 0, derivatives ∂m /∂F and ∂m /∂q are g A F g0 b0 b1 C g C L b constructed by totally differentiating (A5b) (A2C) with respect to mC, Fg, and qL. Note (1 )[Fgg(mA) Fbb(mA)] that Fb is equal to Xb in this case. ’ mA qL(Fg Fb) g (mA) 0, (D) F˜ʦ (D): qL (A5c) (1 )[Xgg(mD) Xbb(mD)] m ς q (X X ) ’ (m ) D 0, where g0, g1, b0, b1, and are the non- L g b g D q negative multipliers associated with F L g (A8) Fb, Xg Fg, Fb 0, Xb Fb, and 1 qL, respectively. The partial derivatives where ς is the nonnegative multiplier asso- ∂mA/∂Fg, ∂mA/∂Fb, and ∂mA/∂qL are derived ciated with 1 qL. The partial derivative by totally differentiating (A2A) with ∂mD /∂qL is derived by totally differentiating respect to mA, Fg, Fb, and qL. (A2D) with respect to mD and qL. Note that F X and F X in this case. (B) F˜ ʦ (B): g g b b Proof of Proposition 1. We begin by q (1 ) (m ) (X F )’ evaluating the first-order conditions for L b B g b g ˜ ˜ F and qL, assuming F ʦ(A), that is, (A5a)–(A5c). To do so, we first need to mB specify ∂mA/∂Fg, ∂mA/∂Fb, and ∂mA/∂qL, (mB) b0 b1 0, (A6a) Fb which are obtained from (A2A). Totally differentiating (A2A) with respect to mA, (1 )[Xgg(mB) Fbb(mB)] Fg, Fb, and qL yields mB q (X F ) ’ (m ) 0, ’ L g b g B q m qLg(mA) L A <0, (A9) (A6b) Fg A q ’ (m ) where b0, b1, and ς are the nonnegative mA L g A >0, (A10) multipliers associated with Fb 0, Xb Fb A Fb, and 1 qL, respectively. The partial ’ mA (Fg Fb) g(mA) derivatives ∂mB/∂Fb and ∂mB/∂qL are 0, (A11) obtained by totally differentiating (A2B) qL A with respect to mB, Fb, and qL. Note Λ that Fg is equal to Xg in this parameter where A [Xg Xb qL(Fg Fb)] ”g constellation. (mA) 0. Note that ΛA is negative 152 EQUITY OWNERSHIP STRUCTURE AND CONTROL because the second-order condition must be (A7b), we totally differentiate (A2C) with satisfied.The signs of (A9)–(A11) then fol- respect to mC, Fg, and qL: low from (2), 0 qL 1, and g’(m) 0. Since , , and are the nonnegative m q ’ (m ) g0 b0 b1 C L g C multipliers associated with F F F < 0, (A14) g b, b Fg ΛC 0, and X F (A5b) and (A10) show that b b, ’ mC (Fg Xb) g(mC) g0 0 or b1 0 or both, that is, Fg Fb 0, (A15) or Xb Fb or Fg Fb Xb. If g0 0 so qL ΛC that Fg Fb, it is seen from (A5a) that g1 0, which means Xg Fg Fb. This where Λ [X X q (F X )] ˜ ʦ C g b L g b contradicts Fb Xb Xg for F (A). ” (m ) 0, which is negative from the g C Thus, we must have b1 0 and Fb Xb. second-order condition. Suppose that Furthermore, if Fg Fb Xb, it again X F F .Then, it is found from (A7a) b b g follows from (A5a) that g1 0, which that 0, that is, X F . Since this con- g1 g g implies Xg Fg. However, this contra- tradicts F X , we must have X F F g b b b g dicts Fg Fb Xb Xg. Therefore, we for F˜ ʦ (C). Given ∂m /∂F 0 from (A14) C g must have Xb Fb Fg and g1 0. and ∂m /∂q 0 from (A15) with F X , ∂ ∂ C L g b Given mA/ Fg 0 from (A9) and we can assume that F and q are determined ∂ ∂ g L mA/ qL 0 from (A11) with Xb Fb as an interior solution by (A7a) and (A7b). Fg, we can assume that Fg and qL are Thus, rearranging (A2C), (A7a), and (A7b) determined as an interior solution by with Xb Fb and g0 g1 ς 0, we (A5a) and (A5c).Thus, rearranging (A2A), have (18a)–(18e) if F˜ ʦ (C). (A5a), and (A5c) with Xb Fb and Finally, we investigate the first-order ς g0 g1 0, we obtain (18a)–(18e) condition for q when F˜ ʦ (D), that is, ˜ L if F ʦ (A). (A8). We totally differentiate (A2D) with We next discuss the first-order condi- respect to mD and qL, and see tions for F˜ and qL, assuming F˜ ʦ (B), that is, (A6a) and (A6b). To specify ∂m /∂F and ∂m /∂q , we totally differen- m (X X )’ (m ) B b B L D g b g D tiate (A2B) with respect to m , F , <0. (A16) B b qL ΛD and qL: Here, Λ (1 q ) (X X )” (m ) m q ’ (m ) D L g b g D B L g B 0, which is negative due to the second- > 0, (A12) Fb ΛB order condition. Given ∂mD/∂qL 0 from (A16), we can assume that qL is deter- m (X F )’ (m ) mined as an interior solution by (A8).Thus, B g b g B 0, (A13) rearranging (A2D) and (A8) with Xb Fb, qL B Xg Fg, and ς 0, we obtain (19a)–(19e) if F˜ ʦ (D). Proof of Proposition 2. Since (F˜**, q**) is where ΛB [Xg Xb qL(Xg Fb)] ”g L feasible in the problem (6) even if F˜ is (mB) 0, which is negative from the second-order condition. It then follows restricted to the set (A) ʜ (C), we should notice that (F˜*, q*) (F˜**, q **). from (A6a) and (A12) that b1 0, that L1 L L1 L Thus, the remaining problem is to check is, Xb Fb. Given ∂mB/∂qL 0 from whether or not a solution with F* X is (A13) and Fb Xb Xg, we can assume g g really optimal. that qL is determined as an interior solution by (A6b). Thus, rearranging (A2B) and In the absence of the monitoring prob- (A6b) with ς 0 and (F , F ) (X , X ) lem, g(m) is independent of m so that b g b g from F X and F˜ ʦ (B), we have ’g(m) 0. Thus, the problem (6) has no b b ˜ (19a)–(19e) if F˜ ʦ (B). interior solution with respect to Fg if F is To examine the first-order conditions for F˜ restricted to the set (A) ʜ (C). More ˜ specifically, it follows from (A5a) and and qL when F ʦ (C), that is, (A7a) and SECURITY DESIGN AND INSIDER MONITORING 153

(A7a) that g1 0 in this case. Thus, we Combining (A18) and (A19) yields * ˜ see Fg Xg even if F is restricted by the set , ** (A) ʜ (C). This finding implies that the { [Xg Xb qL (1 ) optimal security is equity. ᭛ ᭛ (X X )] ’ (m ) c}(m** m ) We now return to the situation in which g b g q**(1 )(X X ) g’(m) 0. Let us take the solution (Fg, Fb, L g b ** ** ** ** ᭛ ** qL (Fg, Fb, q L ) (Xg, Xb, q L ) as a [g (m ) g(m )] starting point.Then, consider a permutation q** X ᭛ ᭛ ᭛ ** ** ** L (1 )[( g ) (Fg, Fb, qL) from the solution (Fg, Fb, q L ); ** ** ᭛ ᭛ ᭛ ** g(m ) Xbb(m )] that is, (Fg, Fb, qL) (Xg – ,Xb, q ( 1 + )), L ** ** ** ᭛ where Xg – Xb 0 and 0. Choose qL (1 ) g(m ), for m m . m᭛ that satisfies Since m᭛ satisfies (A17), this inequality ᭛ ᭛ ’ ᭛ [Xg Xb qL(Fg Xb)]g(m ) c, reduces to , ** which means Γ qL {(1 )(Xg X m᭛ m** [Xg Xb qL** (1 )(Xg b)[ g( ) g( )] ’ ᭛ ** Xb)] g(m ) c. (A17) (1 )[(Xg )g(m ) ** ** ᭛ ᭛ Xb b(m )] (1 )g(m )}, Since (Fg, Fb) ʦ (A) ʝ (C), the large investor’s payoff is represented by for m** m᭛. (A20)

A ᭛ ᭛ ᭛ ᭛ L1(F˜ , qL) [Xgg(m ) Xbg(m )] Now, suppose that we can take a pair (, ) ʦ {( ,) | Xg – Xb 0 and ** qL (1 )(1 )[(Xg ) 0} which satisfies ᭛ ᭛ ᭛ g(m ) Xbb(m )] cm . ** Xg Xb qL (1 ) B (X X ) X X Now, for L0 defined in Proposition 1, g b g b ** qL (Xg Xb), (A21) construct ** [(Xg )g(m ) Xbb , A ˜᭛ ᭛ B ** ** Γ L1(F , qL) L0 (m )] g(m ). (A22)

᭛ ** Condition (A21) with (19a), (A17), and (Xg Xb)[ g(m ) g(m )] ᭛ ** ”g 0 leads to m m , thereby ** qL (1 )(1 )[(Xg Xb) ensuring that the first term in the large ᭛ ᭛ bracket of the right-hand side of (A20) [ (m ) (m**)] c(m m**) g g is nonnegative. Condition (A22) also ** ** qL (1 )[(Xg )g(m ) means that the sum of the remaining two terms in the large bracket of the right- X m** q** m** b b ( )] L (1 ) g( ). hand side of (A20) are nonnegative. (A18) Thus, if (A21) and (A22) hold, the right- Because of ”g 0, we find hand side of (A20) is nonnegative. Combining (A21) and (A22) under ( ,) ᭛ ** ʦ (Xg Xb)[g(m ) g(m )] {( , ) | Xg – Xb 0 and 0}, q**(1 )(X X )[ (m᭛) we now have the following sufficient L g b g condition for Γ, 0: ** g(m )] [Xg ** ** g(m ) Xb qL (1 ) Ξ 1 ** ** ’ ᭛ ** ᭛ (Xg )g(m ) Xbb(m ) (Xg Xb)] g(m )(m m ), Ξ ** ᭛ 2, (A23) for m m . (A19) Xg Xb 154 EQUITY OWNERSHIP STRUCTURE AND CONTROL

** ** where m is determined with qL from and Mayer (1994), OECD (1995), and Gorton and Schmid (1996)). Furthermore, in Japan, (19a) and (19d). Note that Xg – Xb is also fulfilled for any pair of ( , ) 0 that large crossholdings also strengthen the role of satisfies (A23). Thus, if these exists a pair large investors (Prowse (1992), Berglöf and Perotti (1994), and OECD (1995)). In France, (, ) 0 that satisfies (A23), then cross-ownership and core industries are the Γ, ʦ 0 for ( , ){(,)| Xg –Xb norm (OECD (1995)). Finally, in most of the 0 and 0}.This finding shows rest of the world except the United Kingdom, firms are controlled by large owners such as the families of founders or the State. For the litera- A A ˜ ᭛ ᭛ B L0 L1(F , qL) L0. ture and empirical evidence on this issue, see Shleifer and Vishny (1997) and La Porta, Lopez-de-Silanes, and Shleifer (1999). The remaining problem is to prove that a 2 For the decline in the importance of bank pair ( , ) 0 that satisfies (A23) always financing in Japan, see Hoshi, Kashyap, and ** ** exists.Given b(m ) 1 g(m ), we see Scharfstein (1990) and Campbell and Hamao (1994). For the German evidence, see Baums (1994). ** g(m ) 3 The monitoring level of the large investor ** ** may also be interpreted as her managerial effort (Xg )g(m ) Xbb(m ) or her investment level in the production or investment processes. If we follow this interpre- X X X b tation, our research enables us to investigate the g b ** underinvestment problem such as Williamson g(m ) (1985), Klein, Crawford, and Alchian (1978), < , and Grout (1984) within the financial market Xg Xb equilibrium where the design and issued amount of security are endogenously determined. for any such that Xg Xb 0. This 4 Holmström and Tirole (1993) and Admati, completes the proof of Proposition 2. Pfleiderer, and Zechner (1994) are the excep- tional ones. 5 Unless we assume that the large investor initially owns the risky project, we cannot NOTE incorporate her liquidity benefits into the social surplus. * I am most grateful for helpful comments by 6 The recent work of Bolton and von Thadden Noriyuki Yanagawa and seminar participants at (1998) and Maug (1998) compares the liquid- the ISER Seminar (Osaka University), the ity benefits obtained through dispersed corpo- Finance Forum Seminar (Kansai Economic rate ownership with the benefits from efficient Research Center), and the Tezukayama University management control achieved by some degree Workshop. I would also like to thank an anony- of ownership concentration. Kahn and Winton mous referee for detailed and valuable comments. (1998) also demonstrate that market liquidity 1 In the United States, large shareholders are can undermine effective control by a large more common than is often believed (see shareholder because it gives the large share- Demsetz (1983), Shleifer and Vishny (1988), holder excessive incentives to speculate rather and Holderness and Sheehan (1988)). than monitor. Furthermore, large investors such as pension 7 The fraction qL sold to the market does not funds and mutual funds have increasingly played necessarily equal to 1 because the large investor active roles as delegated monitors in recent cannot sell all the holding fraction of the secu- years (see Admati, Pfleiderer, and Zechner rity even though she adjusts the design of the (1994)). In other countries, large investors are security. In other words, a change in the fraction more prevalent. In Germany and Japan, com- of the security sold affects the revenues of the mercial banks have significant powers because large investor in a different way from a change they vote significant blocks of shares, sit on in the design of the security (see equation (3)). boards of directors, play a dominant role in Thus, the fraction of the security sold is not a lending, and operate in a legal environment perfect substitute for the design of the security favorable to creditors (see Aoki (1990), Franks in our model. SECURITY DESIGN AND INSIDER MONITORING 155

8 More specifically, 0 in Allen and Gale REFERENCES (1988, 1991) and Gale (1992), while ʦ (0, 1) in De-Marzo and Duffie (1999). Another line of Admati, A. R., Pfleiderer, P., and Zechner, J. literature assumes the presence of noise or liq- (1994) Large shareholder activism, risk uidity traders, instead of assuming the liquidity sharing, and financial market equilibrium, motive for the issuer. See Boot and Thakor Journal of Political Economy 102, (1993), Holmström and Tirole (1993), and Demange and Laroque (1995). As pointed out 1097–1130. in DeMarzo and Duffie (1999), this kind of Aghion, P.and Bolton, P.(1992) An ‘Incomplete assumption (1) is justified if the issuer faces contracts’ approach to bankruptcy and the strict credit constraints or binding minimal cap- financial structure of the firm, Review of ital requirements. It is also plausible to impose Economic Studies 59, 473–494. this assumption if firms need to raise capital to Allen, F. and Gale, D. (1988) Optimal security fund other valuable investment opportunities design, Review of Financial Studies 1, under various forms of market imperfections. 229–263. 9 To simplify the analysis, we do not specify Allen, F. and Gale, D. (1991) Arbitrage, short the action of managers nor the structure of sales, and financial innovation, Econometrica managerial contracts. Holmström and Tirole (1993) throw light on this problem in their 59, 1041–1068. model of market monitoring. Aoki, M. (1990) Towards an economic model of 10 In the subsequent discussion, “tildes” is the Japanese firm, Journal of Economic used to denote random variables. Literature 28, 1–27. 11 As mentioned in note 3, the monitoring Baums, T. (1994) The German banking system level of the large investor, m, may also be inter- and its impact on corporate finance and preted as her managerial effort or her invest- governance, in M. Aoki and H. Patrick (eds.), ment level. The Japanese Main Bank System: Its 12 The minimum level of monitoring m may Relevance for Developing and Transforming be set equal to zero. Economies, Oxford University Press, Oxford, 13 As indicated in Innes (1990) and Nachman and Noe (1994), the monotonicity pp. 409–449. assumption is important in establishing the opti- Berglöf, E. and Perotti, E. (1994) The gover- mality of the debt-like security because this nance structure of the Japanese financial assumption can exclude the type of “do or die” keiretsu, Journal of Financial Economics 36, contract under which the firm pays everything 259–284. to investors in the bad state and nothing in the Berglöf, E. and von Thadden, E.-L. (1994) good state. Nachman and Noe (1994) addition- Short-term versus long-term investors: ally assume another monotonicity condition that Capital structure with multiple investors, X F the residual claims s s are nondecreasing in Quarterly Journal of Economics 109, the available project returns, X . Even when this s 1055–1084. assumption is additionally imposed, all of our results in this paper still remain valid. Bolton, P.and Scharfstein, D. S. (1990) A theory 14 If the security belongs to the set of of predation based on agency problems in admissible securities, , then we always have financial contracting, American Economic min (Xs, Fs) Fs. Review 80, 93–106. 15 The second-order sufficient condition for m Bolton, P. and von Thadden, E.-L. (1998) ˜ to be a local maximum of RL(m, F, qL) is auto- Blocks, liquidity, and corporate control, matically satisfied if we also assume another Journal of Finance 53, 1–25. monotonicity condition, that the residual claims Boot, A.W. A. and Thakor,A.V. (1993) Security X F X s s are nondecreasing in s. design, Journal of Finance 48, 1349–1378. 16 We should keep in mind that securitizing a Campbell, J. Y. and Hamao, Y. (1994) Changing larger portion of the project returns is not a perfect substitute for selling a larger portion of patterns of corporate financing and the main the issued security. This mainly depends on the bank system in Japan, in M. Aoki and assumption that a change in Fg can affect the H. Patrick (eds.), The Japanese Main Bank payoff of the large investor in a different way System: Its Relevance for Developing and from a change in qL. See note 7. Transforming Economies, Oxford University 17 See note 5. Press, Oxford, pp. 325–349. 156 EQUITY OWNERSHIP STRUCTURE AND CONTROL Demange, G. and Laroque, G. (1995) Private Innes, R. (1990) Limited liability and incentive information and the design of securities, contracting with ex-ante action choices, Journal of Economic Theory 65, 233–257. Journal of Economic Theory 52, 45–67. DeMarzo, P. and Duffie, D. (1999) A liquidity- Kahn, C. and Winton, A. (1998) Ownership based model of security design, Econometrica structure, speculation, and shareholder inter- 67, 65–99. vention, Journal of Finance 53, 99–129. Demsetz, H. (1983) The structure of ownership Klein, B., Crawford, R., and Alchian, A. (1978) and the theory of the firm, Journal of Law Vertical integration, appropriable rents, and and Economics 26, pp. 375–390. the competitive contracting process, Journal Dewatripont, M. and Tirole, J. (1994) A theory of Law and Economics 21, 297–326. of debt and equity: Diversity of securities and La Porta, R., Lopez-de-Silanes, F., and manager-shareholder congruence, Quarterly Shleifer, A. (1999) Corporate ownership Journal of Economics 109, 1027–1054. around the world, Journal of Finance 54, Diamond, D. W. (1984) Financial intermedia- 471–517. tion and delegated monitoring, Review of Maug, E. (1998) Large shareholders as Economic Studies 51, 393–414. monitors: Is there a trade-off between liquid- Franks, J. and Mayer, C. (1994) The ownership ity and control?, Journal of Finance 53, and control of German corporations, mimeo. 65–98. Gale, D. (1992) Standard securities, Review of Nachman, D. C. and Noe,T. H. (1994) Optimal Economic Studies 59, 731–755. design of securities under asymmetric Gale, D. and Hellwig, M. (1985) Incentive- information, Review of Financial Studies compatible debt contracts: The one-period 7, 1–44. problem, Review of Economic Studies 52, Organization for Economic Cooperation and 647–664. Development, Corporate Governance Gorton, G. and Schmid, F. (1996) Universal Environments in OECD countries (1995). banking and the performance of German Ohashi, K. (1995) Endogenous determination of firms, NBER Working Paper No. 5453, the degree of market-incompleteness in National Bureau of Economic Research, future innovation, Journal of Economic Cambridge, Mass. Theory 65, 198–217. Grout, P. (1984) Investment and wages in the Prowse, S. (1992) The structure of corporate absence of binding contracts: A Nash ownership in Japan, Journal of Finance 47, bargaining approach, Econometrica 52, 1121–1140. 449–460. Rahi, R. (1995) Optimal incomplete markets Hart, O. and Moore, J. (1989) Default and rene- with asymmetric information, Journal of gotiation: A dynamic model of debt, mimeo. Economic Theory 65, 171–197. Hart, O. and Moore, J. (1994) A theory of debt Rahi, R. (1996) Adverse selection and security based on the inalienability of human capital, design, Review of Economic Studies 63, Quarterly Journal of Economics 109, 287–300. 841–879. Shleifer, A. and Vishny, R. (1986) Large share- Holderness, C. and Sheehan, D. (1988) The role holders and corporate control, Journal of of majority shareholders in publicly held Political Economy 94, 461–488. corporations: An exploratory analysis, Shleifer, A. and Vishny, R. (1997) A survey of Journal of Financial Economics 20, corporate governance, Journal of Finance 52, 317–346. 737–783. Holmström, B. and Tirole, J. (1993) Market von Thadden, E.-L. (1995) Short-term versus liquidity and performance monitoring, long-term investors: Capital structure with Journal of Political Economy 101, 678–709. multiple investors, Review of Economic Hoshi,T., Kashyap, A., and Scharfstein, D. (1990) Studies 62, 557–575. Bank monitoring and investment: Evidence Townsend, R. (1979) Optimal contracts and from the changing structure of Japanese competitive markets with costly state verifi- corporate banking relationships, in G. Hubbard cation, Journal of Economic Theory 21, (ed.), Asymmetric Information, Corporate 265–293. Finance, and Investment, University of Williamson, O. (1985) The Economic Chicago Press, Chicago, pp. 105–126. Institutions of Capitalism, Free Press. Chapter 6

Marco Becht* and Ailisa Röell BLOCKHOLDINGS IN EUROPE: AN INTERNATIONAL COMPARISON1

Source: European Economic Review, 43(4–6) (1999): 1049–1056.

ABSTRACT We preview empirical work by the European Corporate Governance Network on the size of block shareholdings in Europe. The most salient finding is the extraordinarily high degree of concentration of shareholder voting power in Continental Europe relative to the U.S.A. and the U.K. Thus the relationship between large controlling shareholders and weak minority shareholders is at least as important to understand as the more commonly studied interface between management and dispersed shareholders.

1. INTRODUCTION European Union. The impetus for this development was given by the European THE PURPOSE OF THIS PRESENTATION IS TO Commission’s 1988 Transparency DRAW attention to and briefly preview the Directive (more formally, Large Holdings work on large shareholding in Europe carried Directive, 88/627/EEC). It requires mem- out in the context of an ongoing international ber states to enact laws directing share- research initiative, the European Corporate holders of companies listed on a member Governance Network (ECGN). The data and state exchange to notify the relevant tables on which this article are based were authorities and the company itself within collected for each country by participating seven days whenever their voting rights ‘country teams’ (listed at the end of this cross the thresholds of 10%, 20%, 1/3 (or paper), and are currently being prepared for 25%), 50% and 2/3 (or 75%). The direc- publication in book form. tive is quite strict as to the attribution of The structure of the paper is as follows. voting rights. Any voting rights controlled Section 2 discusses the quality and by a person or entity must be included, such availability of the data, and describes the as voting rights held indirectly through con- recent developments in Europe that have trolled companies, those obtained through made the collection of reasonably compara- written voting agreements, or those on ble data on large blockholdings possible. shares lodged for safekeeping and exercised Section 3 presents some comparative cross- at discretion in the absence of instructions country data and briefly discusses the findings from the holders. As a result of member and their relationship to differences among state legislation carrying out this directive, countries’ legal and other institutions. data on large shareholders’ voting stakes has become available, starting in the early to mid-1990s, throughout the European 2. THE DATA Union. There are several reasons why the data Data on large shareholders’ stakes in listed on large share stakes thus made available companies has only recently become widely are imperfect in terms of quality and and publicly available throughout the international comparability. The directive 158 EQUITY OWNERSHIP STRUCTURE AND CONTROL left countries with considerable leeway 3. EVIDENCE ON BLOCKHOLDINGS regarding how it was carried out; as a result, the information available varies The most striking fact about blockholdings greatly in quality and detail. For example, in Europe is that they are so much higher most countries do not require the notifica- than in the U.S.A. Table 6.1 presents the tion of large ownership stakes (we will use median size of the largest ultimate voting the term ‘ownership’ to refer to cash-flow block in listed industrial companies for var- rather than voting rights) alongside voting ious countries. In several countries, the stakes; a notable exception is the median largest voting stake in listed com- Netherlands.There is also some variation in panies is over 50%, suggesting that voting the reporting thresholds used and in the control by a large blockholder is the rule accuracy of the snapshot of a company’s rather than the exception; and in no control structure that can be obtained at European country studied is the median any moment in time (as a result of report- largest shareholder small enough to fall ing errors, deliberate omissions, changes in below the 5% disclosure threshold. Table the total number of shares outstanding, 6.2 provides a more detailed breakdown etc.). More fundamentally, different coun- into ranges for the largest ultimate voting tries have different legal devices available block. Note that, whereas in the U.S.A. as a means of separating ownership from over 50% of companies have a largest voting rights. Common examples are the shareholder who holds less than 5% of the use of dual class stock, the issue of shares, in Austria and Germany there are essentially non-voting share certificates via virtually no such companies.These findings trust offices, restrictions on voting rights of were summarised at the first ECGN con- large share positions (voting caps), the ference in March 1997 under the heading issue of stock with contingent voting rights ‘Strong Blockholders–Weak Minority triggered by control disputes, and the use of Owners’ (see Becht, 1997a, b): the separa- pyramidal groups. Such intricacies are not tion of ownership and control manifests always easy to reconstruct from the raw itself in a fundamentally different way in numbers reported in the context of the Europe than in the U.S.A. While in the transparency legislation. The individual U.S.A. the main agency problems seem to country teams of ECGN have thus to some stem from conflicts of interest between degree found themselves working within managers and dispersed, insufficiently the idiosyncrasies of their own country’s interventionist shareholders, in much of situation; this has led to considerable continental Europe there are generally differences in the focus, coverge and large blockholders present who can and do degree of detail in the different country exercise control over management. Instead, reports. the main potential conflict of interest lies Figure 6.1 describes the concept of between controlling shareholders and pow- ‘ultimate’ voting power that must be erless minority shareholders. Recently notified pursuant to the transparency LaPorta et al. (1998) have come to a sim- legislation. Company 4 is a listed company ilar conclusion focusing on a sample of 691 whose control is at issue. While the largest firms in a broader set of 27 countries.2 direct stake in the company is 50%, in fact Within Europe, the level of concentration bank 2 controls 72% of the votes in company of voting power is by no means uniform; 4: 50% directly through its majority-held and these differences are rooted in differ- holding 3, plus the 20% owned by 1 via a ences in customs and the legal environment voting pact, plus the 2% owned by individual in different countries. To illustrate the 5 who has deposited the shares with 2, who striking differences in concentration of vot- controls the associated votes. Thus the ing power among countries, consider the ultimate voting block notified by 2 is 72%. size of the largest voting block. In the U.K., The data presented in this paper concerns the sample of 250 listed companies reports such ultimate voting blocks. a very modest median value of 9.9%; while BLOCKHOLDINGS IN EUROPE 159

voting pact deposited Company (1) Bank (2) Individual (5)

80% 20% Holding Company (3) 2%

20% 50%

Listed Company (4)

Figure 6.1 Direct stakes and voting blocks.

Table 6.1 Median size (%) of largest ultimate outside voting block for listed industrial companies

Number of Country companies Median largest voting block

Austria 50 52.0 Belgium 121 50.6 – BEL20 20 45.1 Germany 374 52.1 – DAX30 30 11.0 Spain 193 34.2 France – CAC40 40 20.0 Italy 216 54.53 The Netherlands 137 43.5 United Kingdoma 250 9.9 United States – NYSE 1309 0b – NASDAQ 2831 0b

Sources: Austria: Gugler et al. (1999); Belgium: Becht et al. (1999); France: Bloch and Kremp (1999); Germany: Becht and Böhmer (1999); Italy: Bianchi et al. (1999); Spain: Crespí-Cladera and Garcia-Cestona (1999);The Netherlands: De Jong et al. (1999); UK: Goergen and Renneboog (1999); USA: Becht (1999a). a Random sample of 250 companies. b More precisely, below the 5% disclosure threshold. at the other extreme, Germany, Austria and voting power concentration is inversely Italy all exceed 50%. The figures for the related to company size: blocks held in German DAX30 are somewhat atypical smaller companies tend to be much larger because there is a minimum turnover on average, see some of the country reports requirement for membership of the forthcoming in Barca and Becht (1999). DAX30, leading these companies to widen The numbers mask strong differences in their shareholder base in order to ensure the way the large voting stakes are built up, their continued inclusion in the index. which are discussed more fully in the various Similarly, the French CAC40 is an unrepre- country reports. For example, in Italy, sentative group of the largest and most liq- France and Belgium, pyramidal holding uid companies on the exchange. In general, company groups are a preferred method of 160 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 6.2 Voting power concentration, percentage of companies for which largest voting power stake lies in (in and below) various ranges

Range Austria Germany Netherlands USA (NYSE) USA (NASDAQ)

[x, y] (%) Cum (%) Cum (%) Cum (%) Cum (%) Cum (%) (%) (%) (%) (%)

0–5% 0 0 1.1 1.1 10.2 10.2 52.8 52.8 54.4 54.4 5–10% 0 0 1.9 3.0 11.7 21.9 21.1 73.9 17.4 71.9 10–25% 14 14 14.5 17.5 13.9 35.8 20.9 94.8 20.6 92.4 25–50% 18 32 18.3 35.8 24.8 60.6 3.5 98.3 5.5 98.0 50–75% 54 86 25.5 61.3 19.7 80.3 1.5 99.8 1.5 99.4 75–90% 8 94 17.5 78.8 6.6 86.9 0.2 100 0.5 99.9 90–95% 6 100 5.7 84.4 5.1` 92.0 0 100 0.0 99.9 95–100% 0 100 15.6 100 8.0 100 0 100 0.1 100

Sources: See Table 6.1. amassing voting power without concentrating and of course the 50% majority threshold. (cash flow) ownership. In the Netherlands, Such thresholds differ from country to many companies use a device called an country and are reflected in differences in Administratiekantoor, a foundation controlled clustering of voting blocks. by company insiders that owns all or most of the voting stock and distributes non-voting share certificates to ordinary shareholders; 4. CONCLUSION as well as a defensive arsenal of preference or priority shares, with large blocks of An immediate very clear conclusion from potential voting power triggered by impor- the EGCN’s work on blockholdings in tant control-relevant events. In Germany, Europe is that voting power is highly voting pacts are a commonly used device for concentrated in Continental Europe. Our concentrating voting power, together with survey, which was purely descriptive, raises voting caps and the default voting power of more questions than it answers. Why is banks where shares are placed in custody (a there this strong preference for voting source of voting power excluded from the power, and to what extent is it matched by figures in Tables 6.1 and 6.2). concentration in cash flow ownership Figures 6.2 and 6.3, taken from the stakes (which, unfortunately, need not be ECGN U.K. and Germany Country Reports, disclosed under the EU transparency respectively, illustrate how countries’ legal directive)? Becht (1999b) provides an environment affects the pattern of voting analysis of the trade-offs involved. Is the power concentration. Thus is the U.K. the current level of disclosure of voting and Takeover Panel’s mandatory bid rule that ownership stakes sufficient or excessive? prescribes a general cash offer to all share- Can EU policy on these matters do better holders whenever a blockholder crosses the in ensuring the smooth functioning of capi- 30% threshold, has effectively ensured that tal markets: in the words of U.S. Supreme the growth of large blocks stops short of Court justice Louis Brandeis, is trans- 30%. For the 250 companies examined, all parency the best disinfectant? And Bianco blocks over 30% were held by founders and and Casavola (1999) make progress their families. In Germany, important towards tackling the most fundamental thresholds induce clusters of holdings: the question: what are the effects of corporate 10% anti-director threshold at which share- governance variables such as (pyramidal) holders can take to court board members, ownership concentration on company per- the 25% blocking minority and 75% super formance?3 There is enormous scope for majority for changes to company statutes, further empirical work on these issues. BLOCKHOLDINGS IN EUROPE 161

100

25 Maximum Outside Voting Block % Block Voting Outside Maximum

0 0 .25 .5 .75 1 Fraction of the data

Figure 6.2 U.K.: Percentile plot of largest ultimate voting block. Source: Goergen and Renneboog (1999).

100

25 Maximum Outside Voting Block % Block Voting Outside Maximum

0 0 .25 .5 .75 1 Fraction of the data

Figure 6.3 Germany: Percentile plot of largest ultimate voting block. Source: Becht and Böhmer (1999).

NOTES Network. The ECGN was founded in 1996 as a vehicle for encouraging comparative empirical research on corporate governance in Europe. * Corresponding author. 1 The authors wish to stress that this paper Lack of hard data has been a major impediment reports on a collective research effort by the to such research, and the ECGN brings together members of the European Corporate Governance local ‘country teams’ (listed in reference 162 EQUITY OWNERSHIP STRUCTURE AND CONTROL section) familiar with the language and Becht, M., Chapelle, A., Renneboog, L., 1999. corporate culture of their own country to inves- Shareholding cascades: The separation of tigate issues of common interest. The data col- ownership and control in Belgium. In: lection efforts focus on varibales for which there Barca, F., Becht, M. (Eds.), Ownership are common (and high) disclosure standards, and Control: A European Perspective, such as those mandated by the European Union forthcoming. in various directives relating to company law and securities market; the current results Becht, M., Mayer,C. 1999. Ownership and voting regarding blockholdings are an example. The power in Europe. In: Barca, F., Becht, M. readers should consult the ECGN website, (Eds.), Ownership and Control: A European http://www.ecgn.org, for more detailed analysis Perspective, forthcoming. and continuing updates of country reports and Bianchi, M., Bianco, M., Enriques, L., 1999. other work. During 1996–1998 the Network Pyramidal groupe and the separation received financial support from the Directorate between ownership and control in Italy. General for Industry of European Commission, In: Barca, F., Becht, M. (Eds.), Ownership Fondazione Eni Enrico Mattei and the and Control: A European Perspective, Politecnico di Milano. forthcoming. 2 They also document that most companies in their sample are family or state controlled; see Bianco, M., Casavola, P., 1999. Italian the ECGN country reports (Barca and Becht, corporate governance: Effects of financial 1999) for detailed information on this point for structure and firm performance. European European listed companies. Economic Reviews 43, this issue. 3 See Gugler (1999) for a survey of work on Bloch, L., Kremp, E., 1999. Ownership and voting this issue, including appendices by the various power in France. In: Barca, F., Becht, M. (Eds.), ECGN country teams on local research. Ownership and Control: A European Perspective, forthcoming. Crespí-Cladera, R., Garcia-Cestona, M.A., REFERENCES 1999. Ownership and control: A Spanish survey. In: Barca, F., Becht, M. (Eds.), Barca, F., Becht, M. (Eds.), 1999. Ownership and Ownership and Control: A European Control: A European Perspective, forthcoming. Perspective, forthcoming. Becht, M., 1997a. Strong blockholders, weak De Jong, A., Kabir, R., Marra, T., Röell, A., owners and the need for European mandatory 1999. Ownership and control in The disclosure. In: European Corporate Netherlands. In: Barca, F., Becht, M. (Eds.), Governance Network, The Separation of Ownership and Control: A European Ownership and Control: A Survey of 7 Perspective, forthcoming. European Countries. Preliminary Report to Goergen, M., Renneboog, L., 1999. Strong man- the European Commission, Brussels. agers and passive institutional investors in Becht, M., 1997b. Strong blockholders – Weak the U.K. In: Barca, F., Becht, M. (Eds.), minority owners. Fondazione Eni Enrico Ownership and Control: A European Mattei Newsletter, May. Perspective, forthcoming. Becht, M., 1999a. Beneficial ownership in the Gugler, K., 1999. Corporate governance and United States. In: Barca, F., Becht, M. (Eds.), economic performance. Preliminary Report Ownership and Control: A European to the OECD, Paris. Perspective, forthcoming. Gugler, K., Kalss, S., Stomper, A., Zechner, J., Becht, M., 1999b. European corporate 1999.The separation of ownership and control: governance:Trading off liquidity against control. An Austrian perspective. In: Barca, F., Becht, M. European Economic Reviews 43, this issue. (Eds.), Ownership and Control: A European Becht, M., Böhmer,E., 1999. Ownership and vot- Perspective, forthcoming. ing power in Germany. In: Barca, F., Becht, M. LaPorta, R., Lopez de Silanes, F., Shleifer, A., (Eds.), Ownership and Control: A European 1998. Corporate ownership around the world. Perspective, forthcoming. Journal of Finance, forthcoming. Chapter 7

Mara Faccio* and Larry H. P. Lang THE ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS†

Source: Journal of Financial Economics, 65(3) (2002): 365–395.

ABSTRACT We analyze the ultimate ownership and control of 5,232 corporations in 13 Western European countries.Typically firms are widely held (36.93%) or family controlled (44.29%).Widely held firms are more important in the UK and Ireland, family controlled firms in continental Europe. Financial and large firms are more likely widely held, while non-financial and small firms are more likely family controlled. State control is important for larger firms in certain countries. Dual class shares and pyramids enhance the control of the largest shareholders, but overall there are significant discrepancies between ownership and control in only a few countries.

1. INTRODUCTION between ultimate ownership and control, allowing a small number of families to con- RECENT STUDIES SUCH AS SHLEIFER AND trol firms representing a large percentage VISHNY (1997), Claessens et al. (2000), and of stock market capitalization. Holderness et al. (1999) suggest that Berle This paper answers two questions. What and Means’ (1932) model of widely dis- is the structure of the ultimate ownership persed corporate ownership is not com- of Western European firms? What are the mon, even in developed countries. In fact, means by which owners gain control rights large shareholders control a significant in excess of ownership rights? To answer number of firms in many countries, includ- these questions, we collect ultimate owner- ing developed ones. To examine ownership ship data for a sample of 5,232 listed firms and control by large shareholders, La Porta in Austria, Belgium, Finland, France, et al. (1999) traced the control chains of a Germany, Ireland, Italy, Norway, Portugal, sample of 30 firms in each of 27 countries. Spain, Sweden, Switzerland, and the UK. They documented the ultimate controlling We include a large number of medium- and owners and how they achieved control small-sized corporations, and we include rights in excess of their ownership rights both non financial and financial companies. through deviations from the one-share-one- We measure ownership and control in vote rule, pyramiding, and cross-holdings. terms of cash-flow and voting rights. For Claessens et al. (2000) carried out a similar example, if a family owns 25% of Firm X task for 2,980 listed firms in nine East that owns 20% of Firm Y, then this family Asian countries including Hong Kong, owns 5% of the cash-flow rights of Firm Y Indonesia, Japan, South Korea, Malaysia, (the product of the ownership stakes the Philippines, Singapore, Taiwan, and along the chain) and controls 20% of Firm Y Thailand.They found significant discrepancies (the weakest link along the control chain). 164 EQUITY OWNERSHIP STRUCTURE AND CONTROL Western European firms are most likely (1999), in that we find fewer State-controlled to be widely held (36.93%) or family firms and more widely held firms, fewer controlled (44.29%).Widely held firms are pyramids, and more dual class shares. especially important in the UK and Ireland, Compared to the findings of Claessens et al. while family control is more important in (2000) for East Asia, we find that families continental Europe. Widely held firms are control a higher proportion of firms; each more important for financial and large family controls fewer firms on average; top firms, while families are more important families control a lower proportion of total for non financial and small firms. In certain stock market capitalization;1 a higher countries, widely held financial institutions proportion of family controlled companies also control a significant proportion of have family members in top management; firms, especially financial firms. In some and the largest shareholder is less often countries of continental Europe, the State alone, but averages much higher cash-flow also controls a significant proportion of rights, control rights, and ratio of cash-flow to firms, especially the largest. Widely held voting rights.These differences may be due corporations control few firms. to weaker law enforcement in Asia that We report the use of multiple classes of allows controlling owners to achieve shares, pyramidal structures, holdings effective control of a large number of firms through multiple control chains, and cross- by controlling and owning a smaller part of holdings, which are devices that give the each firm. controlling shareholders control rights in Section 2 describes the data. Section 3 excess of their cash-flow rights. Pyramiding discusses ultimate ownership patterns. occurs when the controlling shareholder Section 4 discusses the means whereby owns one corporation through another owners gain control rights in excess of own- which he does not totally own. Firm Y is ership rights and Section 5 measures the held through “multiple control chains” if it extent to which this has been achieved. has an ultimate owner who controls it via a Section 6 presents conclusions. multitude of control chains, each of which includes at least 5% of the voting rights at each link. “Cross-holdings” means com- 2. DATA pany Y directly or indirectly controls its own stocks. Dual class shares are used by Some of the previous studies of corporate few firms in Belgium, Portugal, and Spain, ownership and control, such as Lins and but by 66.07%, 51.17%, and 41.35% of Servaes (1999a, b), rely primarily on firms in Sweden, Switzerland, and Italy. Worldscope. However, we find its coverage Pyramids and holdings through multiple inadequate. For example, Worldscope control chains are used to control only includes only 176 of 632 Spanish listed 19.13% and 5.52% of listed firms firms at the end of 1997. In this case, we respectively, being less important for family instead rely upon the Spanish Stock controlled firms and more important for Exchange regulatory authority’s files firms controlled by the State and by widely (Comision Nacional del Mercado de held financial institutions. 53.99% of Valores, 1998) which provides quarterly European firms have only one controlling information on all shareholders with at owner. More than two-thirds of the family least 5% of control rights, as well as direc- controlled firms have top managers from tors’ ownership for all listed firms. the controlling family. Overall, we find a Moreover, ownership data is sometimes substantial discrepancy between ownership missing: in this case, Worldscope reports and control in Belgium, Italy, Norway, zero ownership stakes. To ensure accuracy, Sweden, and Switzerland, but much less we include only countries for which we can elsewhere. obtain alternative sources (especially pri- Our results for the 20 largest firms differ mary or official) to permit cross-checking. slightly from those of La Porta et al. We do not rely on Worldscope if we have ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 165 an official data source (i.e., the Stock stake above 2%: Compart with a stake of Exchange ownership files). When official 33.45% and Mediobanca with a stake of data sources are not available, we collect 3.77%. Compart is indicated in the Italian data from alternative sources. We use supervisory authority’s files as the “ultimate” Worldscope for ownership data only when owner of Montedison. However, we found information for a specific firm can not that Compart has three shareholders with otherwise be identified. Cross-checking stakes above 10%: Credit (11.01%), excludes Luxembourg, Greece,2 Denmark3 Cassa di Risparmio di Roma (10.14%), and Holland4 leaving comprehensive, and Mediobanca (15.26%). According to reliable ownership data for 13 Western our definition, Compart is the ultimate con- European countries: Austria, Belgium, trolling shareholder of Montedison at the Finland, France, Germany, Ireland, Italy, 20% threshold. However, at the 10% Norway, Portugal, Spain, Sweden, threshold, Mediobanca would be the largest Switzerland, and the UK. For these ultimate owner of Montedison, with a countries, Table 7.1 lists the data sources 15.26% 3.77% 19.03% control for ownership and multiple classes of stake. Furthermore, mechanisms used to shares. Data limitations confined our sam- secure control rights in excess of ownership ple to the period from 1996 to the end of rights are not systematically analyzed. 1999.5 This is not a significant restriction Over the sample period, a total of 5,547 since ownership structures tend to be rather firms were listed in these 13 countries. Of stable, as noted in La Porta et al. (1999). these, we excluded 167 firms whose owner- The European Corporate Governance ship was not recorded, 61 that use nominee Network (ECGN), too, has sponsored sev- accounts (mostly in the UK and Ireland, eral studies on ownership structures within where firms are not required to disclose the the European Union.6 However,compliance identity of their “true” owners) and 87 affil- with the European Union directive on large iates of foreign firms (i.e., a foreign firm shareholdings (88/627/EEC) restricts controls at least 50% of their voting rights) meaningful cross-country analysis with where we could not follow their ownership non-European Union countries. In particu- chain. We retain the affiliates of foreign lar, the ownership measures documented in firms in our database whenever the holding those studies represent neither ultimate firm is included in our sample. This screen- ownership nor ultimate control stakes. ing left 5,232 firms comprising 94.32% of Specifically, they do not consider the use of the listed firms in the 13 countries.7 For multiple voting shares, and simply add up these firms, our database records all owners direct and indirect control stakes without who control at least 5% of voting rights. In tracing them to the ultimate owners. The France, Germany, and Spain, such owners controlling owner is defined as the one who must disclose their identity. The disclosure controls an absolute majority (i.e., over threshold is 2% in Italy and 3% in the UK. 50%) of voting rights, or holds enough vot- The difficulty of organizing dispersed ing rights to have de facto control. For shareholders means that if the largest example, French regulation defines de facto shareholder holds a substantial block control as occurring when a person or a of shares, then that shareholder has effec- legal entity owns, directly or indirectly, tive control. In line with earlier studies, we more than 40% of voting rights and no shall assume that 20% of the voting shares other partners or shareholders own a suffices to ensure control. We shall also higher percentage directly or indirectly discuss some cases that assume a control (Bloch and Kremp, 1998). This corre- threshold of 10%. If no shareholder sponds to a control threshold of 40%. To exceeds a given control threshold, then the illustrate the bias that this definition firm is said to be widely held at that thresh- introduces, consider the ultimate control old.Table 7.2 sets out the screening process structure of Montedison (Italy). and lists the samples of firms analyzed in Montedison has two shareholders with a later tables. 166 EQUITY OWNERSHIP STRUCTURE AND CONTROL d in the company), d in Datastream Die Welt (1996), Welt Datastream Die Dual class shares “Wer gehört zu wem” Worldscope (1998) Worldscope Worldscope (1998)Worldscope (1998) Becht and Boehmer (http://www.commerzbank.com/navigate/date_frm.htm) at: available http://www.welt.de/ Financial TimesFinancial (1997), “Extel Financial” Times Financial (1997), “Extel Financial” http://www.bourse-de-paris.fr/fr/market8/fsg830.htm Worldscope (1998)Worldscope Muus (1998) Financial Times Financial (1997), Financial” “Extel Les Echos (1996) SFB-Paris BourseSFB-Paris Financial (1997), Times Financial” “Extel (http://www.bluebook.fi/en/tuotteet/haku/majorfinnishfirms.html)Hugin, Annual Report CD (1998) (http://www.huginonline.com) Blue Book, Media Internet Helsinki Firms “Major Finnish Database” Worldscope (1998)Worldscope Stock Exchange Brussels Worldscope (1998) Worldscope List of country specific data sources used to collect data on direct ownership (i.e., used to collect data sources ownership data on direct List of country specific names of large shareholders, hel and stakes and dual class share structures Ireland http://www.hemscott.com/equities/firm/ Datastream Worldscope (1998) Germany (1997), Commerzbank France Herald (1997), The Handbook 1997,” Tribune Firm “French Datastream Table 7.1 Countries Direct ownership data Austria Börse (2000), Wiener 1999” “Yearbook Finland Blue Book, Database” Media Internet Helsinki Firms “Major Finnish Datastream Börse (2000), Wiener 1999” “Yearbook Belgium (www.bxs.be) Stock Exchange Brussels Datastream ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 167 “Sociedades Cotadas 1997” “Sociedades Yearbook” Times Financial (1996) Worldscope (1998) Worldscope Financial Times Financial Zurich 96/97,” “Swiss Stock Guide Zurich Worldscope (1998)Worldscope Hugin, Annual Report CD (1998) (http://www.cnmv.es/english/cnmve.htm) (1998) and Garcia-Cestona Crespi-Cladera “Participaciones significativas en sociedades cotizadas” en sociedades significativas “Participaciones Times Financial (1997), Financial” “Extel Worldscope (1998)Worldscope Hugin, Annual Report CD (1998) (http://www.consob.it/trasparenza_soc_quot/trasp_soc_quot.htm) Il Sole 24 ore (1997), dell’azionista” “Il taccuino mercato ristretto al 31 dicembre 1996” al 31 dicembre ristretto mercato azionaria delle società quotate in borsa o ammesse alle negoziazioni nel borsa o ammesse alle negoziazioni quotate in società delle azionaria Il Sole 24 ore (1997), dell’azionista” “Il taccuino http://www.hemscott.com/equities/firm/ Kingdom (1997), London Stock Exchange London Stock Exchange “The Financial (1997), Times Financial” “Extel United Financial Times (1997), “Extel Financial” FinancialUnited Times (1997), “Extel Datastream Switzerland (1998), Bank of Switzerland Union 96/97,” “Swiss Stock Guide (1998), Bank of Switzerland Union Sweden Hugin, Annual Report CD (1998) Datastream Spain de Mercado del Nacional Comision (1998), Valores ABC (1996) Portugal Bolsa de de Lisboa, Valores Cotadas 1997” “Sociedades (1998) Bolsa de de Lisboa, Valores Norway Hugin, Annual Report CD (1998) Datastream Italy CONSOB (1997), “Bollettino n. speciale – edizione 4/97 – Compagine Datastream 168 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 7.2 Screenings used to identify and remove corporations without reliable ownership data, and sub-samples of firms used in the different sections of the paper. Panel A describes the procedure used to eliminate companies with insufficient or unreliable data on ownership, due to unavailable data, or related to the impossibility to trace ownership back to the ultimate owner (i.e., when the direct controlling shareholder is from a country for which we do not have ownership data, or when the direct owner is a nominee). Panel B reports the total number of firms remaining after this screening, as well as the number of firms with an ultimate owner at the 5% and 20% cutoffs. It also lists the different subsets of firms used in the various sections of the paper Panel A: Selection criteria Number Percentage

All listed firms 5,547 100.00 Firms with no ownership data (167) (3.01) Firms majority-controlled by foreign investors (87) (1.57) Firms controlled by “nominees” (61) (1.10) Firms in our sample 5,232 94.32 Panel B: Sample of firms analyzed in the various tables Tables All listed firms 5,232 100.00 7.3, 7.4, 7.5, 7.6 Firms where at least one shareholder holds 4,806 91.86 7.9 at least 5% of voting rights Firms where at least one shareholder holds 3,300 63.07 7.7, 7.8 at least 20% of voting rights Selected samples of firms 7.10; Figures 7.1 and 7.2

The exclusion of firms that use nominee by each owner at the 10% threshold, but accounts may overstate the proportion of family controlled at the 20% threshold. widely held firms in our sample. However, nominee accounts are the largest share- 2.1. Calculation of cash-flow rights holders in only a small proportion of firms and control rights (less than 5%), so any bias is likely to be marginal. Ireland and the UK have the Corporate ownership is measured by highest proportion of nominee accounts so cash-flow rights, and control is measured this bias would be strongest there. These by voting rights. Ownership and control countries also have the highest proportions rights can differ because corporations can of widely held firms, so the exclusion of issue different classes of shares that provide firms using nominee accounts is unlikely to different voting rights for given cash-flow distort our cross-country comparisons. rights. Ownership and control rights can A shareholder of a corporation is said to also differ because of pyramiding and hold- be an ultimate owner at a given threshold if ings through multiple control chains. he controls it via a control chain whose links Firm Y is said to be controlled through all exceed that threshold. If a firm has two pyramiding if it has an ultimate owner, who owners with 12% of control rights each, then controls Y indirectly through another corpo- we say that the firm is half controlled by ration that it does not wholly control. For each owner at the 10% threshold, but that example, if a family owns 15% of Firm the firm is widely held at the 20% threshold. X, that owns 20% of Firm Y, then Y is In the case of a firm with two owners (a fam- controlled through a pyramid at the 10% ily with 20% of control rights and a widely threshold. However, at the 20% threshold, held corporation with 19% of control rights) we would say that Firm Y is directly con- we would say that this firm is half controlled trolled by Firm X (which is widely held at the ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 169 20% threshold) and no pyramiding would be Widely held financial institution: recorded. If Firm X holds 100% of Firm Y, A financial firm (SIC 6000-6999) that is then again there is no pyramid. Pyramiding widely held at the control threshold. implies a discrepancy between the ultimate State: A national government (domestic owner’s ownership and control rights. In the or foreign), local authority (county, munici- above example, the family owns 3% of the pality, etc.), or government agency. cash-flow rights of Firm Y (the product of its Widely held corporation: A non financial ownership stakes along the control chain) firm, widely held at the control threshold. but its control rights are measured by the Cross-holdings: The firm Y is controlled weakest link in its control chain, i.e., 15%. by another firm, that is controlled by Y, or Firm Y is said to be controlled through a directly controls at least 20% of its own multiple control chain if it has an ultimate stocks. owner who controls it via a multitude of Miscellaneous: Charities, voting trusts, control chains, each of which includes at employees, cooperatives, or minority for- least 5% of the voting rights at each link. eign investors. In the previous example, suppose that the Where the ultimate owner of a corpora- family also owns 7% of Firm Y directly. tion is an unlisted firm, we tried to trace its Then the family owns 10% of the cash-flow owners using all available data sources. We rights of Firm Y (0.15 * 0.20 0.07) and had incomplete success because most of controls 22% of its voting rights (min our sample countries do not require (0.15, 0.20) 0.07). Claessens et al. unlisted firms to disclose their owners. One (2000) defined “holdings through multiple exception is the UK, where the 3% disclo- control chain” as “cross-holdings.” sure rule also applies to unlisted firms. A firm can be controlled by holdings However,we were still not able to find own- through multiple control chains, even ership data for all unlisted firms. If we though it is not controlled by pyramiding. failed to identify the owners of an unlisted For example, suppose that Firm A controls firm, then we classified them as a family. 10% of B and 100% of C, which controls This approach is close to that of Claessens 15% of B. Since C is fully controlled by A et al. (2000), who regard as a family any in the control chain A-C-B, there is no controlling shareholder that was an pyramiding. However, Firm A controls unlisted firm in a business group. Below, we Firm B directly and indirectly through offer both a general justification for this Firm C, with control rights of 25%.We con- convention and statistical support for it in clude that Firm A controls Firm B through the largest European economies. multiple control chains because: (1) Firm B In the case of the Fiat group, for exam- has a controlling owner at the 20% level; ple, we traced the ownership of La (2) B is controlled via multiple control Rinascente back to the Agnelli family from chains; and (3) all links in each chain Il Taccuino dell’ Azionista, although we involve at least 5% of the control rights. find two unlisted firms in its chain of Firm Y is controlled by a cross-holding at control (namely, Carfin and Eufin). In fact, the 20% threshold if Firm X holds a stake in La Rinascente is controlled by Eufin (with Firm Y of at least 20%, and Y holds a stake a 32.8% ownership and a 40.51% control in Firm X of at least 20%, or if firm Y holds stake), which is wholly controlled by Ifil. directly at least 20% of its own stocks. Wer gehört zu wem is particularly useful in a number of cases in Germany. However, it helps us identify the owners for only 20% 2.2. Types of ownership of unlisted German firms. For example, we This section discusses our classification of find that TCHIBO Holding AG is the ultimate owners into the following six types: largest owner of Beiersdorf AG (a listed Family: A family (including an individ- firm) with a 25.87% O&C stake. TCHIBO ual) or a firm that is unlisted on any stock Holding, however, is an unlisted firm. We exchange. identified its ultimate owner: the Herz 170 EQUITY OWNERSHIP STRUCTURE AND CONTROL family, which has a 100% O&C stake. At the 20% level, financial firms control Another case is Heidelberger Zement AG, 4.9% of unlisted firms, the State 2.67%, whose largest owner (with a 19.07% O and cross-holdings 1.83%. Thus, both stake and a 21.8% C stake) is Schwenk State and widely held financial firms are Gmbh, again an unlisted firm. We find that insignificant as ultimate owners of unlisted Schwenk is 90% owned and controlled by firms. Furthermore, since the State and the the Schwenk family, and 10% owned by the largest financial institutions are included in Babette family. In some cases, the owner- the Wer gehört zu wem database, we were ship structure of unlisted firms is more able to identify them and trace their complex. For example, the direct owner of ownership chain. This further reduces the Thyssen AG is Thyssen Beteiligunsverw possibility of bias in our sample. Gmbh, which is unlisted. We find that this For Italy, Bianchi et al. (1998) consid- firm is 49.999% owned and controlled by ered a sample of 1,000 manufacturing Commerzbank (a listed, widely held finan- firms surveyed by the Bank of Italy. They cial firm), and 49.981% owned and con- reported that the largest immediate share- trolled by Allianz. A number of complex holder of these firms held, on average, a cases of ownership of unlisted firms were direct stake of 67.69%. Among all types of also reported by La Porta et al. (1999). immediate shareholders, individuals held Our database records the unlisted sub- 48% of equity, non financial firms 36.9%, sidiaries of widely held corporation or financial firms 0.17%, the State 4.6%, financial institution, so any listed firm con- and foreign investors 8%. We secured a trolled by an unlisted firm that is controlled wider sample of 3,800 unlisted Italian by a widely held corporation or financial firms with ultimate owners from the AIDA institution is recorded as controlled by the database.8 We found that the State is the latter in our database. Thus, an unlisted largest owner for 0.4% of firms, financial firm that we identify as the ultimate con- institutions for 0.2%, and families (domes- troller of a listed firm is unlikely to be, tic and foreign) for 99.4%. The average in fact, controlled by a widely held corpo- ultimate control stake of the largest ration or financial institution. The unlisted controlling shareholder was 70.71%. firm isn’t likely controlled by the State For France, Bloch and Kremp (1998) because State-controlled firms tend to be summarize the ownership structure of a listed. In any case, State ownership has sample of 282,322 (mostly unlisted) decreased dramatically in Europe after the firms.9 For firms with more than 500 privatization wave of the 1990s. The low employees, the largest owner held, on aver- likelihood that an unlisted firm is, in fact, age, 88% of the capital. For 56% of the controlled by a widely held corporation, unlisted firms, the largest owner was a widely held financial institution, or the family; for the remaining 44% it was State leaves families as the most likely another corporation, usually an unlisted controller of an unlisted firm. firm. For the UK, Goergen and Renneboog For Germany, we collected a sample of (1998) reported that 78% of unlisted firms 500 unlisted firms with ultimate owners are fully controlled by one shareholder,while from Wer gehört zu wem. We considered the remaining 22% have a shareholder who firms in alphabetical order,including a firm (directly) holds a majority stake.10 in our sample if we could trace its ultimate owners. We stopped when our sample 2.3. Examples numbered 500. We found that the average control stake is 89.44%. In 68% of Figure 7.1 illustrates dual class shares and the cases, the largest owner is the sole pyramiding in the Nordström family group owner; in the remainder, the largest owner of Sweden. All holdings of more than 5% of holds an average stake of 67%. Families, a firm’s shares are listed. All firms shown in both domestic and foreign, are the largest Figure 7.1 have dual class shares: A-shares ultimate owners of 90.6% of these firms. carry one voting right while B-shares ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 171

Panel A: Group Chart Blockfield Nordström family Properties

(12.8%O; (12.8%O; (7.6%O; (8%O; 34.9%C) 35.1%C) 39.3%C) 14.1%C)

Humlegården Columna Sweco AB Fastigheter AB Fastigheter AB (16.2%O; (26.1%O; (35%O; 23.7%C) 17.7%C) 48.1%C) (11.3%O; Latour 5.3%C) (2%O; Realia (2%O; 9.5%C) (2.66%O; 39.3%C) 9.5%C) Robur Minne family (13.7%O; (1.5%O; Kapitalf. 9%C) 8.7%C)

Lingfield Investm. Eriksson family

Panel B: Calculating ownership and voting rights in Realia’s capital structure Number of Number of Percent of share Type shares Voting rights votes capital Percent of votes

A-shares 2.641 million 1 2.641 million 5.8% 38.09% B-shares 42.922 million 1/10 4.292 million 94.2% 61.91% All 45.563 million 6.933 million 100% 100%

Shareholders A-shares B-shares Ownership Control Columna Fastigheter 1.933 million 14.007 million 35.0% 48.1% Lingfield Investments — 6.259 million 13.7% 9.0% The Eriksson family 0.591 million 0.101 million 1.5% 8.7%

Figure 7.1 The Nordström family group (Sweden).This figure describes the major listed firms controlled by the Nordström family. All control stakes of at least 5% are reported. All the firms have dual class shares with A-shares each carrying one voting right and B-shares each carrying a tenth of a vote. The procedure used to compute direct ownership and control rights is illustrated in Panel B. Ownership stakes are denoted by “O” and voting stakes by “C.” Direct ownership and control stakes are shown alongside each of the arrows, while the resulting ultimate ownership and control stakes for Realia (the only case of pyramiding) by the Nordström family are reported in the firm’s box. carrying one-tenth of a voting right. Realia carry 4.292 million votes ( one-tenth of has two classes of shares: 2.641 million 42.922 million). Thus, A-shares carry A-shares and 42.922 million B-shares. 38.09% of the votes ( 2.641 million/ A- and B-shares have the same face value, (2.641 million 4.292 million)), while so the total stock capital is 45.563 million B-shares carry 61.91% of votes. shares. A-shares constitute 5.8% of stock Realia has three direct shareholders: capital ( 2.641 million/45.563 million) Columna Fastigheter, Lingfield Investment, while B-shares constitute 94.2%. A-shares and the Eriksson family. Columna owns carry 2.641 million votes, while B-shares 1.933 million A-shares and 14.007 million 172 EQUITY OWNERSHIP STRUCTURE AND CONTROL B-shares. Thus, Columna owns 35% chains (Ifi-Ifil and Ifi-Unicem, denoted by ( (1.933 million 14.007 million)/ dotted lines in Figure 7.2). 45.563 million) of cash-flow rights, and To compute the Agnelli family’s ultimate controls 48.1% ( (1.933 million ownership of Unicem, we form the 1.401 million)/ (2.641 million 4.292 product of its ownership stakes along million)) of votes in Realia. Realia’s second the pyramidal chain of Agnelli-Ifi- largest direct shareholder, Lingfield Carfin-Ifil-Unicem, namely 100% * Investments, owns no A-shares and only 41.23% * 100% * 20.55% * 8.76% 6.259 million B-shares, which comprise 0.7422%. We then add the stake from the 13.7% (6.259 million/ 45.563 million) of two further (i.e., multiple) control chains the cash-flow rights, and 9% (0.6259/(2.641 (the dotted lines).The first chain is from Ifi million 4.292 million)) of the control into Ifil (the left side of Figure 7.2) which rights. Finally, the Eriksson family owns gives the Agnelli family a stake of 0.591 million A-shares and 0.101 million 100% * 41.23% * 7.97% * 8.76% B-shares, comprising 1.5% ( (0.591 0.2878%. The second chain is from Ifi to million 0.101 million)/ 45.563 million) of Unicem (the right side of Figure 7.2) which the ownership rights and 8.7% ( (0.591 gives the Agnelli family a stake of million 0.0101 million)/ (2.641 million 100% * 41.23% * 19.42% 8.0069%. 4.292 million)) of the control rights.Through The family’s overall ownership stake in Columna Fastigheter, the Nordström family Unicem is the sum of these three has sole control of Realia at the 20% stakes: 0.7422% 0.2878% 8.0069% threshold. However, at the 10% threshold, 9.0369%. Realia has a second large shareholder, The Agnelli family’s ultimate control Blockfield Properties. In Relia’s control stake is the sum of the weakest links along structure there are two pyramiding chains: each control chain. The weakest link in the Nordström family/Columna/Realia and pyramidal chain is min(100%; 82.45%; Blockfield/Columna/Realia. 100%; 37.64%; 14.81%) 14.81%. In Figure 7.2 illustrates the control of the control chain from Ifi to Ifil, the weak- Unicem by pyramiding, holdings through est link is min(100%; 82.45%; 14.16%; multiple control chains, and dual class 0%) 0%. Note that Ifil controls 14.81% shares within the Agnelli family group, the of Unicem; this 14.81% stake has already largest Italian group. The methodology been considered in the pyramidal chain, so presented in Figure 7.1 is used to compute we use 0%. In the control chain from Ifi to cash-flow rights (O) and control rights (C), Unicem, the weakest link is min(100%; taking account of dual class shares. 82.45%; 32.83%) 32.83%.The Agnelli Unicem is directly controlled by two major family’s control stake in Unicem is the sum shareholders: Ifi and Ifil. Ifil is controlled of these three weakest links, namely by Ifi with a direct stake (O 7.97%; 14.81% 0% 32.83% 47.64%. C 14.6%) and an indirect stake The ratio of cash-flow to voting rights is (O 20.55%; C 37.64%) through 0.1897 (9.04%/47.64%). Carfin, a wholly owned, non financial We conclude that Unicem is controlled unlisted firm. Since Carfin is wholly con- by the Agnelli family at both the 10% and trolled by Ifi, we consider Ifi’s stake in Ifil the 20% thresholds.The company has only as a direct holding rather than a pyramid, one controlling shareholder, and its control although we say there is holdings through structure includes pyramiding, multiple multiple control chain (see section 2.1). Ifi control chains, and multiple class shares. is controlled by a single major shareholder, Giovanni Agnelli & C. S.p.A. (the Agnelli family). The Agnelli family’s control of 3. ULTIMATE OWNERSHIP PATTERNS Unicem is thus exercised through pyramiding (Ifi-Carfin-Ifil-Unicem), non voting Table 7.3 analyzes the ultimate controlling shares (within Ifi, Ifil, and Unicem), and owners of Western European corporations two holdings through multiple control at the 20% threshold. In all countries, ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 173

Agnelli Family

(100% O&C)

Giovanni Agnelli & C. (100% O&C )

(41.23%O; 82.45%C)

IFI S.p.A. (41.23%O; 82.45%C)

(100% O&C)

Carfin S.p.A. (41.23%O; 82.45%C)

(20.55%O; 37.64%C)

(7.97%O; 14.6%C) Ifil S.p.A. (3.77%O The Public Instit. (11.76%O; 52.24%C) 6.9%C) for Social Security (8.76%O; 14.81%C)

Unicem S.p.A. (19.42%O; 32.83%C) (9.04%O; 47.46%C)

Figure 7.2 Unicem (Italy). This figure describes the principal shareholders of Unicem, a complex case of pyramiding, holdings through multiple control chains, and the use of dual class shares within the Agnelli family group, the largest Italian group. All control stakes of at least 5% are reported.The company has only one ultimate owner,the Agnelli family, which controls 47.64% of the voting rights and owns 9.04% of the cash-flow rights in Unicem. Hard lines indicate pyramiding while dotted lines indicate holdings through further (i.e., multiple) control chains. Ownership stakes are denoted by “O” and voting stakes by “C.” Direct ownership and control stakes are shown alongside each arrow, while the Agnelli family’s ultimate ownership and control stakes (when different from the direct stakes) are shown in each firm’s box. widely held and family controlled firms are percentages are in the UK (23.68%) and the most important category. 36.93% of Ireland (24.63%); in continental Europe, the firms in our sample are widely held and the lowest percentages are in Norway 44.29% are family controlled. However, (38.55%), Sweden (46.94%), there is a sharp cleavage between Switzerland (48.13) and Finland ownership patterns in continental Europe (48.84%). In every other Western and in the UK and Ireland. Widely held European country, family controlled firms firms comprise 63.08% of UK firms and are in the majority. 62.32% of Irish firms; in continental The UK and Ireland also stand apart in Europe the highest percentages of widely the low percentages of corporations that held firms are all in Scandinavia but are are State controlled (0.08% and 1.45%). substantially lower (Sweden 39.18%, In continental Europe, the State controls Norway 36.77%, Finland 28.68%). The more than 10% of the listed firms in lowest percentages of widely held firms are Austria, Finland, Italy, and Norway. in Germany (10.37%), Austria (11.11%), 9.03% of Western European firms are and Italy (12.98%).The picture for family controlled by widely held financial institu- control is reversed. Here, the lowest tions, this being especially significant in 174 EQUITY OWNERSHIP STRUCTURE AND CONTROL county, f votes are f votes -holdings:The harities, voting into six types. into The table presents table The 0.08 0.76 8.94 3.46 0.00 7.32 1.09 9.35 6.31 0.23 4.90 0.00 2.86 5.71 0.41 4.11 1.64 11.51 0.47 0.05 5.75 0.57 4.60 6.90 0.00 1.45 2.17 4.35 5.07 0.00 6.30 3.65 9.07 3.37 2.62 5.11 3.79 11.37 0.91 0.00 2.31 0.77 12.69 12.69 0.00 4.14 1.68 9.03 3.43 0.51 13.09 0.32 4.46 4.54 2.27 10.34 2.88 12.26 1.20 0.72 15.76 1.55 0.65 4.52 0.00 15.32 0.00 8.59 11.11 1.01 6.25 49.54 5.17 55.17 7.31 44.23 12.22 11.46 22.66 25.47 22.65 24.29 10.59 27.96 39.50 20.11 13.04 11.59 27.03 37.59 26.11 38.71 16.28 32.56 16.93 27.36 12.12 40.74 Family of which: ividual) or a firm that is unlisted on any stock exchange. Widely held financial institution: on any stock that is unlisted exchange. financial or a firm held ividual) firm Widely A financial Number Widely Identified Unlisted Widely held Widely held trusts, employees, cooperatives, investors. foreign at least 20% o or minority controlling that do not have a shareholder Companies firm firm, by another Y is controlled Y, stocks. at least 20% of its own by controls that is controlled or directly Miscellaneous: C municipality, etc.), corporation: held firm, agency.Widely threshold. at the control A non financial held or government widely Cross (SIC 6000-6999) that is widely held at the control threshold. at the control held (SIC 6000-6999) that is widely State: (domestic or foreign), government A national local authority ( Family: A family (including an ind Family: (including family A the percentage of firms controlled by different controlling owners at the 20% threshold. owners controlling are classified shareholders by different controlled Controlling of firms the percentage Ultimate control of publicly traded firms. Data relating to 5,232 publicly traded corporations are used to construct table. this firms. corporations traded traded of publicly to 5,232 publicly control Ultimate Data relating classified as widely held as widely classified UK 1,953 63.08 23.68 Switzerland 214 27.57 48.13 Sweden 245 39.18 46.94 Spain 632 26.42 55.79 Portugal 87 21.84 60.34 Norway 155 36.77 38.55 Italy 208 12.98 59.61 Ireland 69 62.32 24.63 Germany 704 10.37 64.62 France 607 14.00 64.82 Finland 129 28.68 48.84 Belgium 130 20.00 51.54 Table 7.3 Country of Firm held Family families firms StateTotal Corporation Financial 5,232 Miscellaneous Cross-holdings 36.93 44.29 Austria 99 11.11 52.86 ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 175 Belgium (12.69%) and Italy (12.26%) firms will affect its results. Financial firms but insignificant in Finland (0.65%). constitute only 10.1% of sample firms in Control by widely held corporations is Finland and 14.5% in Germany, but trivial (1.68%) in all sample countries. 30.3% in Austria and 39.2% in Belgium. Cross-holdings are only marginally significant in Norway, Germany, and Austria (accounting 3.2. Size effects for 2.27%, 2.62%, and 1.01% of cases respectively) but are insignificant elsewhere. To explore the relationship between owner- If we lower the threshold to 10%, then ship patterns and firm size, we identify in the proportion of widely held firms falls each country the 20 largest, the middle 50, to 13.72%; family control increases from and the 50 smallest firms by market capi- 44.29% to 55.87%; control by financial talization. For Austria, Ireland, and institutions also increases from 8.73% to Portugal which had less than 120 listed 18.34%.This increase is especially signif- firms, we defined the 20 largest firms as icant in the UK and Ireland. By contrast, “Large”, then divided the rest equally into lowering the control threshold to 10% those that were “Medium” and “Small.” has little effect on the percentages of Table 7.5 shows that family ownership is State-controlled firms, firms controlled less likely for larger firms, being particularly by widely held firms, and firms with cross- weak among the largest firms in the UK holdings. and Sweden. In Austria, Finland, Italy, Norway, and Portugal, State control is quite important for the largest firms. Large 3.1. Financial versus non financial firms are also more likely to be widely held firms than small firms in all our sample countries Table 7.4 compares the ownership struc- except Austria, Norway, and Portugal. In ture of financial firms and non-financial the UK, 90% of large firms are widely held firms. Financial firms are slightly more at the 20% threshold, but only 14% of likely to be widely held than non-financial small firms are widely held at the 10% firms (39.92% versus 36.39%), are much threshold. Cross-country differences more likely to be controlled by a widely become less significant among small firms. held financial institution (21.48% versus 5.96%), and are much less likely to be family controlled (26.54% versus 4. MEANS OF ENHANCING CONTROL 48.15%). For other types of controlling owners, the differences between financial This section reports the major mechanisms and non financial firms are insignificant. used to enhance control. We neglect less In both the UK and Ireland, the proportion significant mechanisms such as firm- of widely held financial firms is about the specific voting caps,11 golden shares,12 same as that of non financial firms. informal alliances (i.e., voting blocks), or Financial firms are more likely than transfer restrictions on shares. non-financial firms to be widely held for continental countries except Austria, Finland, and Portugal. Non-financial firms 4.1. Dual class share structures are more likely to be family controlled in all countries except Finland and Portugal. Consistent with the hypothesis that control Differences in patterns of ownership and provides large private benefits (see, control across countries can be explained for example, Barclay and Holderness, by differences in the regulations on finan- 1989; Claessens et al., 2002; Jensen and cial firms (Barth et al., 2000). Since the Meckling, 1976; Johnson et al., 2000; ownership structure of financial firms dif- La Porta et al., 1997, 2002) in Western fers from that of non financial firms, a Europe, several studies report that voting country’s ratio of financial to non financial shares trade at a premium over non voting 176 EQUITY OWNERSHIP STRUCTURE AND CONTROL entage of lity, etc.), or onstruct this controlled by controlled mily: A family d as widely held d as widely employees, coop- 9.43 72.33 11.01 1.57 4.09 1.57 0.00 9.97 69.20 6.37 4.26 5.82 3.61 0.76 8.86 63.92 2.53 0.00 12.03 12.66 0.00 39.92 26.54 5.05 1.31 21.48 4.00 1.71 63.76 24.97 0.09 0.84 6.90 3.44 0.00 60.11 18.01 0.00 0.42 17.94 3.53 0.00 25.93 55.86 4.12 1.44 7.10 5.25 0.31 32.69 24.04 17.31 0.00 16.35 9.62 0.00 37.43 47.49 5.03 0.00 3.35 6.70 0.00 50.00 40.48 4.76 0.00 2.38 2.38 0.00 25.85 62.03 3.49 1.95 6.34 0.28 0.06 29.41 23.37 7.35 0.00 38.40 1.47 0.00 23.44 55.47 7.81 0.78 4.69 7.81 0.00 19.05 71.43 0.00 0.00 4.76 4.76 0.00 33.08 43.65 10.99 0.38 4.55 4.64 2.71 56.00 12.00 24.00 0.00 4.00 4.00 0.00 24.49 18.37 8.16 7.14 38.78 0.00 3.06 62.07 26.72 0.00 1.72 5.17 4.31 0.00 63.64 13.64 9.09 4.55 0.00 9.09 0.00 12.75 37.58 5.88 0.00 28.27 1.96 13.56 12.38 70.92 4.81 4.52 6.58 0.79 0.00 22.45 33.16 6.63 0.00 36.22 1.53 0.00 31.71 44.11 17.07 1.22 0.61 5.28 0.00 23.08 64.10 10.26 0.00 2.56 0.00 0.00 37.25 32.35 1.96 1.96 13.73 12.75 0.00 11.59 59.18 16.18 0.00 4.35 8.70 0.00 10.00 38.33 13.33 0.00 18.33 16.67 3.33 36.39 48.15 3.85 2.13 5.96 3.30 0.21 0.183 0.815 0.185 0.817 75.7 24.3 82.9 17.1 83.9 16.1 75.9 24.1 83.9 16.1 76.4 23.6 84.1 15.9 85.5 14.5 83.9 16.1 89.9 10.1 60.8 39.2 69.7 30.3 % of Widely held Widely held ividual) or a firm that is unlisted on any stock exchange. Widely held financial institution: on any stock financial that is unlisted exchange. (SIC 6000-6999) held or a firm firm ividual) Widely A financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial Non-financial government agency.Widely held corporation: held firm, agency.Widely threshold. at the control A non financial government held widely firm Cross-holdings:The Y is that is widely held at the control threshold. at the control held that is widely State: (domestic or foreign), government A national (county, local authority municipa firm,another Y, at least stocks. by controls 20% of its own that is controlled or directly Miscellaneous: Charities, trusts, voting (including an ind (including firms controlled by different controlling owners at the 20% threshold. owners six types. into controlling are classified by different shareholders controlled firms Controlling Fa table, which groups corporations into financial firms (SIC:table, firms financial into corporations corporations. which groups presents the perc 6000-6999) and non financial table The Concentration of control:Concentration firms. are used to c non financial versus corporations traded financial to 5,232 publicly Data relating eratives, investors. foreign at least are classifie 20% of votes or minority controlling that do not have a shareholder Companies All Financial UK Financial Switzerland Financial Sweden Financial Spain Financial Portugal Financial Norway Financial Italy Financial Ireland Financial Germany Financial France Financial Finland Financial Belgium Financial Table 7.4 CountryAustria Type of firm Financial firms Widely held Family State corporation financial Miscellaneous Cross-holdings ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 177 shares. For example, Zingales (1994) as of May 1998. Prior to this, German reports that in Italy this premium is firms could be authorized to issue shares 81.5%, Megginson (1990) reports a with multiple voting rights by State authorities. 13.3% premium in the UK, Muus and For example, at the end of 1996, Rwe AG Tyrell (1999) find a 29% premium in had multiple voting stocks with a 20 Germany, and Muus (1998) documents a voting right, and Siemens AG had multiple 51.35% premium in France. In addition, voting stocks with a 6 voting right. Horner (1988) documents a 27% pre- Multiple voting shares were issued in mium in Switzerland and Rydqvist (1992) Germany by Bewag, Frankisches finds a 6.5% premium in Sweden. In the Uberlandwerk, Hamburger Hochbahn, US, Lease et al. (1983) find a voting pre- Hamburgische Electricitats Werke, Lech mium of 5.4%. Nenova (2000) analyzes a Elektrizitatswerke, and Uberlandwerk sample of 661 dual class share firms in Unterfranken. Multiple voting shares are 18 countries and reports that control benefits legal in France and most firms grant two constitute 50% of firm value in Mexico, votes for each ordinary share, as long as around 0% of firm value in Denmark, and they have been held for at least two con- between one-quarter and one-half of firm secutive years; for publicly traded firms this value for Brazil, Chile, France, Italy, and minimum holding period can be extended South Korea. Nenova adds that a large up to four years. These multiple voting fraction (i.e., more than 70%) of the stocks do not represent a special category differences in benefits can be explained by of stocks in France, so we treat them as the legal systems, in particular the quality ordinary shares. of general investor protection, minority Table 7.6 ranks countries by this last rights in the case of control transfer, and statistic. Some ironies emerge when we standards of law enforcement. compare this ranking with the ranking of Panel A of Table 7.6 displays the legal countries by the severity of legal restric- restrictions in each country on dual class tions on dual class shares, which are of shares, the proportion of firms issuing dual four types: (1) the one-share-one-vote class shares for each country, and the rule; (2) a cap on the proportion of non average minimum percentage of the stock voting stocks; (3) a rule placing a mini- capital required to control 20% of the mum on the votes accruing to any type of voting rights (denoted by Own 20% share; (4) no restrictions on the type of Con). For example, Realia (Figure 1) has stock issued. 2.641 million A-shares with one vote per The European countries with the lowest share and 42.922 million B-shares with average minimum percentages of shares one-tenth of a vote per share. A-shares required to ensure 20% control are represent 5.8% of stock capital (2.641 Sweden (9.83%), Switzerland (15.26%), million/45.563 million) and 38.09% (i.e., and Finland (15.42%). Thus, Sweden and 2.641 million/(2.641 million 4.292 Finland, the only countries that impose a million)) of votes, while B-shares represent lower limit on the voting rights of shares, 61.91% of votes. Hence, it takes 3.05% exhibit a greater discrepancy between own- ( 0.2 * (5.8%/38.09%)) of A-share ership and control than countries that capital to acquire 20% of the votes place no limits on voting rights, such as (Own 20% Con is 0.0305). Austria and Ireland. Therefore, the explicit Minimum percentages are calculated for floor on the voting rights of a share pro- each firm from its capital structure, then vides a clear indication of the unfairness in averaged over all sample firms to construct voting rights that regulators will tolerate, the table. These figures need not account as well as a defense against public criticism. for the use of multiple voting shares Sweden also has the highest percentage of (except in Scandinavian countries). The firms issuing dual class shares (66.07%), issue of multiple voting shares was out- followed by Switzerland (51.17%), lawed in Italy, Spain, the UK, and Germany, Italy (41.35%), and Finland (37.60%). 178 EQUITY OWNERSHIP STRUCTURE AND CONTROL nd then ch groups edium-size threshold. not have a irm, widely l government ls at least 20% ividual) or a firm that is unlisted on any stock that is unlisted exchange. or a firm ividual) Average market cap Widely Widely held Widely held the rest into two equal groups.The table presents the percentage of firms controlled by different controlling owners at the 20% owners controlling by different controlled of firms the rest two equal groups.The presents the percentage into table Smallest 50 12.0 14.00 81.00 3.00 0.00 2.00 0.00 0.00 Middle 50Middle 158.3 10.00 75.00 2.00 7.00 4.00 1.00 1.00 Smallest 50 6.8 8.00 77.00 2.00 4.00 9.00 0.00 0.00 Middle 50Middle 149.9 14.00 68.00 6.00 2.00 10.00 0.00 0.00 Smallest 50 30.1 17.65 65.20 9.80 0.00 0.00 7.35 0.00 Middle 50Middle 274.3 41.18 38.23 11.76 2.94 1.47 4.41 0.00 Smallest 50 58.0 16.00 65.00 0.00 2.00 2.00 15.00 0.00 Middle 50Middle 722.2 18.00 47.00 4.00 0.00 20.00 11.00 0.00 Smallest 39 33.5 11.76 71.57 4.90 0.00 2.94 8.82 0.00 Middle 40Middle 227.8 14.29 47.14 11.43 0.00 15.71 8.57 2.86 of its own stocks.of its own Miscellaneous: Charities, trusts, voting employees, cooperatives, investors. foreign or minority that do Companies held at the control threshold. at the control held firm Cross-holdings:The firm, by another Y is controlled Y, by contro that is controlled or directly (domestic or foreign), (county, local authority municipality, etc.), corporation: agency. held or government f Widely A non financial firms, and the smallest 50 firms. with less than 120 firms, countries For “Largest 20” the in the largest 20 firms we group group, a divide Widely held financial institution: threshold. at the control held financial held (SIC 6000-6999) that is widely firm Widely State: A financial A nationa corporations according to their size (market capitalization), (market to their size according corporations and, each variable, for the largest 20 firms, presents means for 50 m Controlling shareholders are classified into six types. into are classified shareholders an ind Controlling (including Family: A family Concentration of control and firm size. and firm of control Concentration are used to construct table, this corporations traded to 5,232 publicly Data relating whi shareholder controlling at least 20% of votes are classified as widely held. as widely are classified at least 20% of votes controlling shareholder Germany Largest 20 27,510.2 45.00 15.00 10.00 0.00 12.50 5.00 12.50 France Largest 20 16,467.1 60.00 30.00 0.00 0.00 10.00 0.00 0.00 Finland Largest 20 2,310.2 38.10 33.33 28.57 0.00 0.00 0.00 0.00 Belgium Largest 20 5,386.4 20.00 37.50 0.00 0.00 27.50 15.00 0.00 Table 7.5 Country Category (m US$) held Family State corporation financial Misc. Cross-holding Austria Largest 20 1,593.0 10.00 32.50 42.50 0.00 5.00 10.00 0.00 ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 179 Middle 295.0 29.28 48.82 6.53 2.01 8.65 4.25 0.46 Smallest 50 4.2 42.00 38.67 0.00 0.00 8.00 11.33 0.00 Middle 50Middle 151.7 64.71 19.60 0.00 0.00 13.73 1.96 0.00 Smallest 50 58.0 28.00 55.00 7.33 2.67 6.00 1.00 0.00 Middle 50Middle 352.2 18.00 58.00 10.00 2.00 5.00 7.00 0.00 Smallest 50 44.0 32.00 63.00 0.00 0.00 3.00 2.00 0.00 Middle 50Middle 397.9 49.02 42.16 2.94 0.00 0.98 4.90 0.00 Smallest 50 65.3 36.00 56.00 2.00 0.00 6.00 0.00 0.00 Middle 50Middle 302.4 34.00 46.00 8.00 2.00 10.00 0.00 0.00 Smallest 33 42.8 18.75 60.94 1.56 0.00 6.25 12.50 0.00 Middle 34Middle 170.4 28.57 60.00 2.86 0.00 5.71 2.86 0.00 Smallest 50 62.4 40.00 35.67 7.40 0.00 7.33 4.40 5.20 Middle 50Middle 310.8 38.00 34.83 16.50 0.00 3.83 5.67 1.17 Smallest 50 70.4 14.00 67.00 6.00 2.00 10.00 1.00 0.00 Middle 50Middle 320.4 8.00 63.00 6.00 6.00 14.00 2.00 1.00 Smallest 24 37.0 68.00 24.00 4.00 0.00 4.00 0.00 0.00 Middle 25Middle 198.9 50.00 29.16 0.00 4.17 8.33 8.33 0.00 Smallest 38.9 25.32 59.66 3.72 0.90 5.17 4.80 0.44 All Largest 12,027.2 45.24 27.24 12.20 1.09 8.46 4.42 1.35 UK Largest 20 40,563.4 90.00 0.00 0.00 0.00 10.00 0.00 0.00 Switzerland Largest 20 24,589.9 50.00 35.00 0.00 0.00 15.00 0.00 0.00 Sweden Largest 20 9,575.2 80.00 5.00 5.00 0.00 0.00 10.00 0.00 Spain Largest 20 4,919.2 45.00 20.83 10.00 9.17 15.00 0.00 0.00 Portugal Largest 20 2,338.9 15.00 60.00 17.50 2.50 0.00 5.00 0.00 Norway Largest 20 2,302.7 30.00 45.00 20.00 0.00 0.00 5.00 0.00 Italy Largest 20 10,398.8 35.00 20.00 25.00 0.00 15.00 0.00 5.00 Ireland Largest 20 2,260.2 70.00 20.00 0.00 2.50 0.00 7.50 0.00 180 EQUITY OWNERSHIP STRUCTURE AND CONTROL 20% ample); Own able. 20% Dual class 20% Dual Number of Own affect kely to end is in arrears,end is in (2) on ivid their class rights and winding-up the firm,and winding-up and (3) on are li which resolutions may not exceed 25% of stock capital may not exceed may not exceed 50% of stock capital may not exceed may not exceed 50% of stock capital may not exceed is the percentage of firms with outstanding dual class shares. with outstanding of firms is the percentage voting rights”voting share capital reducing for resolutions provided these are givenprovided “adequate when their d Firms may issue “preference shares”“preference may issue Firms (1) stocks voting include preference is the average minimum percent of the book value of equity required to control 20% of votes (as computed for each firm in our s in each firm (as computed for 20% of votes of equity required to control of the book value percent minimum is the average Con Legal restrictions on issuing dual class shares. Data relating to 5,232 publicly traded corporations are used to construct dual class shares. this on issuing restrictions corporations traded Legal to 5,232 publicly Data relating t Dual class shares (%) UK 1968. shares outlawed since Non voting required for rights voting Minimum 1,953 19.14 23.91 France stocks capped of non voting Proportion capital voting) (and limited Non voting 607 19.93 2.64 Spain stocks capped of non voting Proportion capital voting) (and limited Non voting 632 20.00 0.16 Portugal stocks capped of non-voting Proportion capital voting) (and limited Non voting 87 20.00 0.00 Table 7.6 CountryBelgium Restrictions One-share-one-vote Restriction details firms Con 130 shares (%) 20.00 0.00 Panel A: Country-level results ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 181 U, C, P M, C, P U, M, P U, M, C 6.15 53.92 40.11 10.14 unrestricted (U), requires a minimum r supervisory authorities),r supervisory the as from as well its (P) the issuance of non voting/limited-voting ghts, stocks voting the issue of limited but permit 69 18.91 28.07 99 18.96 23.23 214 15.26 51.17 5,232 18.74 19.91 20% Con Dual class shares (%) U, C, P U, M U, M M, C, P may not exceed 50% of stock capitalmay not exceed 208 18.38 41.35 may not exceed 50% of stock capitalmay not exceed 704 18.83 17.61 exceptions indicates that the ratio is significantly different (at the 5% level) from the sample of firms in countries where regulation is regulation where countries in the sample of firms from (at the 5% level) different is significantly that the ratio indicates Panel B: Results by type of restrictions All Sweden ratio vote Minimum ratio: vote Minimum one-tenth 245 9.83 66.07 Switzerland Unrestricted papers discussed in Section 4. Section in discussed papers Finland ratio vote Minimum ratio: vote Minimum one-tenth 129 15.42 37.60 (i.e., (or thei stock shares). exchanges the national preference from is obtained rights about on voting restrictions Information Italy stocks capped of non-voting Proportion capital voting) (and limited Non voting stocks.The UK is not included in any of these groups since its regulations prohibit the issue of stocks that carry no voting ri the issue of stocks no voting that carry prohibit stocks.The its regulations since any of these groups in UK is not included Germany stocks capped of non-voting Proportion capital voting) (and limited Non voting Ireland Unrestricted voting ratio (M), ratio to the total stock stocksvoting capital), (with reference of non-voting a cap (C) to the proportion imposes or prohib Austria Unrestricted Norway rule One-share-one-vote needed for approval Governmental Restrictions 155Unrestricted stocks of non voting Cap on the proportion Prohibited 19.05U, M, C, P 13.16 19.47 Own 16.86 19.56 Minimum voting ratio required ratio voting Minimum 12.26 182 EQUITY OWNERSHIP STRUCTURE AND CONTROL By contrast, Rydqvist’s survey (1992) give these shares “adequate voting rights” reported that 75% of Swedish firms issue on resolutions on (1) dividends in arrears, non voting stocks. Bergström and Rydqvist (2) reducing share capital and winding-up (1990) reported that stocks issued by the firm, and (3) actions affecting their Swedish firms can be restricted or unre- class rights. The voting rights attached to stricted to foreigners, but unrestricted preference shares are defined by the arti- shares cannot exceed 40% of equity and cles of the firm and vary across firms. We 20% of the votes. Swedish firms seem to combine limited voting (i.e., preference) have taken this as a license to issue shares shares with non-voting shares to calculate with inferior voting rights to foreigners. the ratio of votes to share capital.The con- Shares with superior voting rights are often sequence is that 19.91% of firms issue not traded in Sweden. In Switzerland, firms dual class shares and the Own 20% Con can issue different classes of shares: ratio is 18.74%. Bearer (B-shares), registered (R-shares), and non-voting shares. R- and B-shares have identical cash-flow rights; however, 4.2. Pyramiding, holdings through R-shares have more voting power than multiple control chains, and B-shares since they are issued with a lower cross-holdings face value (Gardiol et al., 1997; Horner, 1988). Italian firms can issue limited vot- Table 7.7 displays the control enhance- ing (i.e., preference) shares and non voting ments employed by firms that have a shares. Only eight firms have limited voting controlling shareholder at the 20% shares outstanding. Both limited voting and threshold. Pyramids are used by 19.13% non voting shares have legally prior claims of such firms in our sample, being most on dividends, and on reimbursement in case prevalent in Norway (33.90%) and least of liquidation, see Zingales (1994). Finnish prevalent in Finland (7.46%). Holdings regulations require a control contestant to through multiple control chains are used treat all classes of shares in a “fair and by 5.52% of controlling shareholders, equitable” manner, so bidders need to offer being most prevalent in Norway all classes of shares the same tender price. (20.34%), Italy (8.78%), and Germany The protection of minority shareholders (7.22%) but insignificant elsewhere. during takeover contests may explain why Cross-holdings are used by 0.73% of the small investors are willing to hold limited controlling shareholders, being most prevalent in Germany (2.69%) and voting and non voting shares. 14 At the opposite extreme, dual class Norway (2.04%) but marginal in other shares are rare in Portugal (0%), Belgium countries, whose regulations typically set a (0%), Spain (0.16%), and France 10% cap on these stakes. (2.64%).13 These countries have either a one-share-one-vote rule or a cap on the 4.3. Other control-enhancing proportion of non voting stocks. In mechanisms Norway, departures from the one-share- one-vote principle require governmental A controlling shareholder is said to be approval. This seems readily given since “alone” if no other owner controls at least 13% of firms have multiple classes of 10% of the voting rights. Table 7.7 shows shares. In the UK, non-voting shares have that this is true of 53.99% of the been outlawed since 1968, but firms can firms that are not widely held. Austria issue preference shares that have a prior exhibits the highest percentage of such claim on dividends and cumulative divi- firms (81.82%) and Norway the lowest dends, but limited voting rights. Preference (38.78%). Bennedsen and Wolfenzon shares may have no voting rights at gen- (2000) and Gomes and Novaes (1999) eral meetings. However,listing rules on the discuss the role of the second largest London Stock Exchange require firms to shareholders. ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 183 Table 7.7 Percentage of firms adopting control-enhancing devices.We only include the 3,300 firms with controlling shareholders at the 20% level. Pyramids reports the percentage of firms whose largest controlling shareholder adopts pyramids as control devices. Holdings through multiple control chains reports the percentage of firms whose largest controlling shareholder adopts at least 5% of holdings through multiple control chains as control devices. Cross-holdings reports the percentage of firms whose largest controlling shareholder adopts cross-holdings as control devices. Controlling owner alone reports the percentage of firms that have single controlling owners. Management reports the percentage of firms whose top managers come from the largest shareholder’s family.The table presents the percentage of firms controlled by different controlling owners at the 20% threshold. Holdings Number through multiple Cross- Controlling Country of firms Pyramids control chains holdings owner alone Management

Austria 88 20.78 6.49 1.14 81.82 80.00 Belgium 104 25.00 2.38 0.00 71.15 80.00 Finland 92 7.46 1.49 0.00 41.30 69.23 France 522 15.67 2.87 0.00 64.75 62.20 Germany 631 22.89 7.22 2.69 59.90 61.46 Ireland 26 9.09 0.00 0.00 42.31 77.78 Italy 181 20.27 8.78 1.13 58.76 70.00 Norway 98 33.90 20.34 2.04 38.78 66.67 Portugal 68 10.91 0.00 0.00 60.29 50.00 Spain 465 16.00 5.43 0.22 44.30 62.50 Sweden 149 15.91 0.00 0.67 48.32 73.47 Switzerland 155 10.91 0.91 0.00 68.39 70.00 UK 721 21.13 4.93 0.00 43.00 75.85 Total 3,300 19.13 5.52 0.73 53.99 68.45

A member of the controlling family is 3.22% of family controlled firms said to be in “management” if he/she is the respectively, in 35.32% and 11.01% of CEO, Honorary Chairman, Chairman, or State-controlled firms, and in 27.96% and Vice-Chairman. We assumed that individu- 16.78% of firms controlled by widely held als are in the same family if they have the financial institutions. The State is most same last names, a convention that under- likely to be the controlling owner alone states family affiliation. Nevertheless, in (58.72%), followed by families (54.74%), more than two-thirds of the family con- then by widely held financial institutions trolled firms, the controlling owner is in (44.74%); however, there are significant management. The proportion is highest variations across countries. (above 70%) in Austria, Belgium, Ireland, Italy, Sweden, Switzerland, and the UK, and lowest in Portugal (50%). 5. DISCREPANCY BETWEEN OWNERSHIP AND CONTROL 4.4. Types of control-enhancing instruments by different types of Pyramids, holdings through multiple con- shareholders trol chains, cross-holdings, and deviations from the one-share-one-vote rule all create Table 7.8 details the means of enhancing discrepancies between ownership and control by families, the State, and widely control rights. Table 7.9 shows that, on held financial institutions. Pyramids and average, the largest ultimate controlling holdings through multiple control chains shareholder owns 34.64% of cash-flow are used to enhance control in 13.81% and rights and 38.48% of voting rights. 184 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 7.8 Means of enhancing control in Europe by types of controlling owners. We only include 3,012 firms with controlling shareholders at the 20% (2,332 firms controlled by families, 218 firms controlled by the State, and 462 firms controlled by widely held financial institutions) level. Own 20% Con reports the average minimum percent of the book value of equity required to control 20% of votes. Dual class shares (%) reports the proportion of firms with dual class shares outstanding. Pyramids reports the percentage of firms whose largest controlling shareholder adopts pyramids as control devices. Holdings through multiple control chains reports the percentage of firms whose largest controlling shareholder adopts at least 5% of holdings through multiple control chains as control devices. Controlling owner alone reports the percentage of firms that have single controlling owners. Panel A presents percentages for family controlled firms, Panel B presents percentages for State-controlled firms, and Panel C presents percentages for firms controlled by widely held financial firms. Holdings Number Own 20% Dual class through multiple Controlling Country of firms Con shares (%) Pyramids control chains owner alone

Panel A: Family-controlled firms Austria 51 18.91 21.57 11.76 1.96 86.27 Belgium 67 20.00 0.00 8.96 4.48 61.19 Finland 65 14.82 43.55 6.15 1.54 36.92 France 395 19.96 1.01 13.16 3.04 63.80 Germany 460 18.75 18.26 14.57 3.04 66.96 Ireland 17 19.53 16.67 5.88 0.00 35.29 Italy 122 18.34 42.62 20.34 5.93 54.55 Norway 60 18.33 21.28 23.33 5.00 38.33 Portugal 54 20.00 0.00 16.67 3.70 61.11 Spain 351 20.00 0.00 11.97 5.41 39.32 Sweden 117 9.34 67.12 18.80 0.85 47.01 Switzerland 106 13.52 68.57 7.55 1.89 65.09 UK 467 19.16 18.84 13.70 2.14 46.47 Total 2,332 18.64 17.61 13.81 3.22 54.74

Panel B: State-controlled firms Austria 16 19.67 12.50 37.50 18.75 68.75 Belgium 3 20.00 0.00 33.33 0.00 100.00 Finland 19 16.81 21.05 5.26 0.00 47.37 France 30 19.93 3.33 36.67 3.33 73.33 Germany 41 19.90 4.88 43.90 4.88 17.07 Ireland 1 20.00 0.00 0.00 0.00 100.00 Italy 22 18.50 40.91 27.27 9.09 77.27 Norway 24 20.00 0.00 70.83 58.33 37.50 Portugal 5 20.00 0.00 0.00 0.00 80.00 Spain 26 20.00 0.00 57.69 3.85 88.46 Sweden 12 10.92 62.50 0.00 0.00 75.00 Switzerland 16 18.09 25.00 6.25 0.00 81.25 UK 3 20.00 0.00 33.33 33.33 0.00 Total 218 18.98 13.17 35.32 11.01 58.72

Panel C: Firms controlled by a widely held financial firm Austria 9 18.29 55.56 22.22 22.22 88.89 Belgium 16 20.00 0.00 0.00 0.00 93.75 Finland 0 n.a. n.a. n.a. n.a. n.a. France 69 19.94 2.90 15.94 1.45 60.87 ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 185 Table 7.8 (Continued) Holdings Number Own 20% Dual class through multiple Controlling Country of firms Con shares (%) Pyramids control chains owner alone Germany 65 18.85 21.54 53.85 40.00 33.85 Ireland 3 20.00 0.00 33.33 33.33 100.00 Italy 27 18.39 33.33 29.63 14.81 40.74 Norway 5 20.00 0.00 40.00 20.00 40.00 Portugal 3 20.00 0.00 0.00 0.00 0.00 Spain 74 20.00 0.00 20.27 9.46 51.35 Sweden 6 13.72 60.00 33.33 33.33 16.67 Switzerland 19 15.51 31.58 15.79 0.00 68.42 UK 166 18.77 31.93 27.71 18.67 35.54 Total 462 18.98 20.67 27.96 16.78 44.74 n.a.: Not available.

These averages are computed over firms largest average number of firms controlled where at least one owner owns at least 5% by a single family (1.46), while Finland of the control rights. The largest ultimate (1.05) and Switzerland (1.10) have the owners average the highest cash-flow rights smallest. in Germany (48.54%), Austria (47.16%) Another perspective on the concentra- and France (46.68%), but the lowest tion of family control is provided by the cash-flow rights in Ireland (18.82%) and percentage of total market capitalization the UK (22.94%), followed by the controlled by the top families in each coun- Scandinavian countries of Norway try. For each family, we sum the market (24.39%), Sweden (25.15%), and Finland capitalization of all firms in which the (32.98%). family is the largest controlling shareholder, The largest ultimate owners average the then divide by the market capitalization of highest voting rights in Germany (54.50%) all firms in our sample from that country. and Austria (53.52%), but the lowest in We then rank families by the share of Ireland (21.55%) and the UK (25.13%), market capitalization that they control and again followed by the Scandinavian coun- calculate the share of market capitalization tries of Sweden (30.96%), Norway controlled by the top 1, 5, 10, and 15 (31.47%), and Finland (37.43%). families. Table 7.9, Panel C shows that the largest The top family controls 17.89% of total ultimate shareholder’s average ratio of market capitalization in Switzerland and cash-flow to voting rights is 0.868, being 10.40% in Italy,15 but only 1.40% in the lowest in Switzerland (0.740), Italy UK, 1.66% in Spain and 3.40% in (0.743), and Norway (0.776), and highest Ireland. The top 15 families control in Spain (0.941), Portugal (0.924), and 36.77% of total market capitalization in France (0.930). Portugal and 36.63% in Belgium, but only To measure the concentration of corpo- 6.55% in the UK, 13.48% in Spain, and rate control in families,Table 7.10 displays 15.38% in Ireland. the average number of firms controlled by each family via control chains that include at least 10% of the control rights at each 6. CONCLUSION link. For example, if a family controls 20% of Firm A, Firm A controls 15% of Firm In this paper, we document the ultimate B, and Firm B controls 9% of Firm C, then ownership of 5,232 listed firms in 13 we include only A and B amongst the firms Western European countries. Widely held controlled by that family. Italy has the and family controlled firms predominate. 186 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 7.9 Cash-flow and control rights. This table includes data relating to 4,806 publicly traded corporations (including both financial institutions and non financial institutions) where the largest controlling owner has at least 5% of voting rights. Cash-flow rights represents the ultimate ownership stake held by the largest controlling shareholder. Control rights the percentage of voting rights controlled by the largest controlling shareholder. Number of Standard Country firms Mean deviation Median 1st quartile 3rd quartile

Panel A: Cash-flow rights Austria 95 47.16 23.52 50.00 25.50 65.00 Belgium 120 35.14 24.96 36.10 14.98 51.81 Finland 119 32.98 23.94 27.60 14.60 49.91 France 604 46.68 26.69 48.98 24.69 66.00 Germany 690 48.54 31.46 48.89 21.05 75.00 Ireland 68 18.82 17.32 14.24 6.76 26.03 Italy 204 38.33 25.13 39.68 16.61 56.83 Norway 149 24.39 21.26 19.42 8.91 36.12 Portugal 86 38.42 20.45 39.31 19.83 52.00 Spain 610 42.72 30.46 32.55 18.50 64.91 Sweden 244 25.15 23.06 17.30 9.45 33.55 Switzerland 189 34.66 24.69 29.00 12.85 51.00 UK 1,628 22.94 17.87 16.21 10.96 29.66 Total 4,806 34.64 26.76 25.90 13.02 51.00

Panel B: Control rights Austria 95 53.52 22.77 54.70 34.00 75.00 Belgium 120 40.09 23.20 39.56 19.49 55.86 Finland 119 37.43 22.44 33.70 20.80 52.36 France 604 48.32 25.55 50.00 28.70 66.00 Germany 690 54.50 28.70 50.76 27.00 76.91 Ireland 68 21.55 16.39 16.64 10.39 26.56 Italy 204 48.26 21.00 50.11 31.39 63.15 Norway 149 31.47 20.18 27.78 15.10 43.59 Portugal 86 41.00 19.18 44.95 22.28 52.30 Spain 610 44.24 29.59 35.73 20.00 65.03 Sweden 244 30.96 22.37 24.90 14.50 40.55 Switzerland 189 46.68 25.97 50.00 22.50 63.00 UK 1,628 25.13 17.87 18.02 13.28 30.19 Total 4,806 38.48 26.10 30.01 15.96 53.98

Panel C: Ratio of cash flow to control rights Austria 95 0.851 0.224 1.000 0.704 1.000 Belgium 120 0.779 0.360 1.000 0.596 1.000 Finland 119 0.842 0.246 1.000 0.800 1.000 France 604 0.930 0.189 1.000 1.000 1.000 Germany 690 0.842 0.267 1.000 0.709 1.000 Ireland 68 0.811 0.321 1.000 0.683 1.000 Italy 204 0.743 0.337 0.971 0.548 1.000 Norway 149 0.776 0.341 1.000 0.532 1.000 Portugal 86 0.924 0.218 1.000 1.000 1.000 Spain 610 0.941 0.178 1.000 1.000 1.000 Sweden 244 0.790 0.339 1.000 0.526 1.000 Switzerland 189 0.740 0.290 0.830 0.468 1.000 UK 1,628 0.888 0.228 1.000 0.907 1.000 Total 4,806 0.868 0.255 1.000 0.852 1.000 ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 187 Table 7.10 How concentrated is control by families/unlisted firms? This table provides summary statistics on the concentration of control by families and unlisted firms. Average number of firms per family refers only to firms in the sample. Percent of total market value of listed corporate assets that families control is the aggregate market value of common equity of firms controlled by the largest, the top 5, the top 10, and the top 15 families divided by the total market value of common equity for all firms included in the sample for a given country.

Average Percent of total market value of listed corporate assets that families control number of firms per Top Top 5 Top 10 Top 15 Country family family families families families Austria 1.16 5.64 15.59 19.47 22.15 Belgium 1.22 5.63 20.38 30.45 36.63 Finland 1.05 3.94 13.98 21.82 25.90 France 1.18 5.94 22.04 29.18 33.80 Germany 1.24 5.43 15.66 21.29 25.01 Ireland 1.16 3.40 11.88 14.46 15.38 Italy 1.46 10.40 16.83 20.18 21.92 Norway 1.29 3.73 16.01 23.13 27.18 Portugal 1.23 6.11 24.59 34.23 36.77 Spain 1.19 1.66 6.97 10.92 13.48 Sweden 1.27 3.37 9.29 13.26 15.66 Switzerland 1.10 17.89 24.35 28.94 30.93 UK 1.17 1.40 4.11 5.85 6.55

Widely held firms are more important University of Singapore, Washington University amongst financial and large firms, while at St Louis, the Capacity Building Seminar in families are more important for non finan- Manila (sponsored by the Asian Development cial and small firms. In some continental Bank), the European Financial Management European countries, the State controls a Association meeting (Athens), the European Finance Association meeting (London), the significant number of larger firms. We also Financial Management Association meeting document the means whereby owners gain (Seattle), and the Scottish Institute for control rights in excess of their cash-flow Research in Investment and Finance rights. The use of multiple class voting (Edinburgh).We also thank Bolsa de Valores de shares contribute only marginally to this, Lisboa, Commerzbank, Helsinki Media Blue except in just a few countries. The use of Book, Hugin, the Union Bank of Switzerland, pyramids and holdings through multiple and the Vienna Stock Exchange for generously control chains is also marginal. providing us with their data sets. Mara Faccio acknowledges research support from Università Cattolica, Milan, and MURST (research grant #MM13572931). Larry Lang acknowledges NOTES research support from a Hong Kong Earmarked Grant. * Corresponding author. 1 Japan stands apart from the rest of † We are grateful to Marco Bigelli, Lorenzo East Asia in this regard: the top Japanese Caprio, Edith Ginglinger, Arun Khanna, John family controls only 0.5% of total market McConnell, Stefano Mengoli, Robert Pye, capitalization. Gordon Roberts, Bill Schwert (the editor), 2 In addition, Greek shares are often held in George Tian, and especially Andrei Shleifer and bearer form, masking the identities of their an anonymous referee for providing helpful owners. comments. We benefited from comments from 3 La Porta et al. (1999) use Hugin for workshop participants at the National Denmark. However,Hugin covers less than 20% 188 EQUITY OWNERSHIP STRUCTURE AND CONTROL of all listed firms and we could not locate 11 Voting caps are used, for example, by information from other official sources. BASF (2.62%), Bayer (5%), Deutsche Bank 4 The Stichting Toezicht Effectenverkeer (5%), Linde (10%), Mannesmann (5%), (STE, the Securities Board of the Netherlands), Phoenix (10%), Schering (3.51%), and responded to our request for data as follows: Volkswagen (20%) in Germany; and Telefonica “The implementation of the 1992 Act has (10%) in Spain. In Italy, the law requires vot- resulted in the list of disclosed notifications ing caps for cooperative banks; they are rather being no more than an historical overview that common in recently privatized firms such as has been overtaken by countless events and no Comit (3%) and Credit (3%).Voting caps also longer provides the desired transparency of the obtain in France, e.g., in Alcatel, Danone, and (Dutch) stock market. Furthermore, the file Pernod Ricard. In Switzerland, 12 out of the corruption will become greater with the passage 50 largest firms have voting caps. Examples of time. Given the above and the fact that from the ten largest corporations include the 1996 Act does not empower the STE to take Novartis (2%), Nestlé (3%), UBS (5%), any measures in this regard which would lead to Swiss Bank Corporation (5%), and Zurich an up-to-date overview, the STE does not issue Insurance (3%). this list to third parties.” See also De Jong et al. 12 As reported in Crespi-Cladera and Garcia- (1998). Cestona (1998), the State holds these golden 5 Data are from 1996 for France, Germany, shares in some recently privatized Spanish Italy, Switzerland, and the UK; from 1997 for firms, such as Repsol, Telefonica, and Endesa. Spain and Portugal; from 1998 for Sweden and The use of golden shares is also popular among Norway; and from 1999 for Austria, Belgium, privatized firms in Italy, as emerged from the Finland, and Ireland. vicissitudes of Telecom Italia and Enel in the 6 Becht and Boehmer (1998) for Germany; late 1990s. Bianchi et al. (1998) for Italy; Bloch and 13 Similar evidence is reported for Spain in Kremp (1998) for France; Crespi-Cladera and Crespi-Cladera and Garcia-Cestona (1998) and Garcia-Cestona (1998) for Spain; De Jong et al. for France in Muus (1998). (1998) for the Netherlands; Renneboog (1998) 14 These figures are, respectively, 2.62% and for Belgium; Goergen and Renneboog (1998) 2.27% if we also include second ultimate for the UK. owners (see Table 7.3). 7 Our sample coverage of countries is: Austria: 15 Brioschi et al. (1989) show that, in the 100.00 (%), Belgium: 89.04, Finland: 87.76, mid-1980s, more than a quarter of total market France: 89.26, Germany: 99.15, Ireland: 82.14, capitalization could be traced to the control of Italy: 100.00, Norway: 72.77, Portugal: three single families. 100.00, Spain: 100.00, Sweden: 94.96, Switzerland: 100.00, and the UK: 95.69. The low figure for Norway is due more to limited data coverage than to our screening. REFERENCES 8 AIDA is a private database provided by Bureau Van Dijk. The AIDA database Barclay, M., Holderness, C., 1989. Private provides accounting information for about benefits from control of public corporations. 130,000 Italian firms and ownership information for 25,314 Italian firms. Firms with Journal of Financial Economics 25, ownership data in the AIDA database are not 371–395. ranked in any order. We traced ownership to Barth, J.R., Caprio Jr., G., Levine, R., 2000. ultimate owners starting from the first firm Banking systems around the globe: do listed in the AIDA database using ownership regulation and ownership affect performance information contained in that database. A firm and stability? Unpublished working paper, was included if we could trace its ultimate own- World Bank, Washington DC. ers. We stopped when our sample reached 15% Becht, M., Boehmer, E., 1998. Ownership and of firms with ownership data. voting power in Germany. Unpublished work- 9 Their results were based on data compiled ing paper, European Corporate Governance by the French central bank (Fiben), that is unavailable to the public. Network. 10 The figures are based on a sample of Bennedsen, M., Wolfenzon, D., 2000. The 12,600 unlisted firms from the Jordan’s balance of power in closely held corporations. database, which is compiled by a private data Journal of Financial Economics 58, vendor, Bureau Van Djik. 113–139. ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 189 Bergström, C., Rydqvist, K., 1990. Ownership of Gardiol, L., Gibson-Asner, R.,Tuchschmid, N.S., equity in dual-class firms. Journal of Banking 1997. Are liquidity and corporate control and Finance 14, 255–269. priced by shareholders? Empirical evidence Berle, A., Means, G., 1932. The Modern from Swiss dual class shares. Journal of Corporation and Private Property. Corporate Finance 3, 299–323. MacMillan, New York. Goergen, M., Renneboog, L., 1998. Strong man- Bianchi, M., Bianco, M., Enriques, L., 1998. agers and passive institutional investors in Pyramidal groups and the separation between the UK. Unpublished working paper, ownership and control in Italy. Unpublished University of Reading. working paper, European Corporate Gomes, A., Novaes, W., 1999. Multiple large Governance Network. shareholders in corporate governance. Bloch, L., Kremp, E., 1998. Ownership and con- Unpublished working paper, University of trol in France. Unpublished working paper, Pennsylvania, Wharton. INSEAD Business School. Holderness, C., Kroszner,R., Sheehan, D., 1999. Bolsa de Valores de Lisboa, 1998. Sociedades Were the good old days that good? Changes Cotadas 1997. Lisbon Stock Exchange, Lisbon. in managerial stock ownership since the great Brioschi, F., Buzzacchi, L., Colombo, M.G., depression. Journal of Finance 54, 435–470. 1989. Risk capital financing and the separa- Horner, M.R., 1988. The value of the corporate tion of ownership and control in business voting right: evidence from Switzerland. groups. Journal of Banking and Finance 13, Journal of Banking and Finance 12, 69–84. 747–772. Il Sole 24 ore, 1997. Il taccuino dell’azionista. Claessens, S., Djankov, S., Lang, L.H.P., 2000. Il Sole 24 Ore Radiocor, Milan. The separation of ownership and control in Jensen, M., Meckling, W., 1976. Theory of the East Asian corporations. Journal of Financial firm: managerial behavior, agency costs, and Economics 58, 81–112. ownership structure. Journal of Financial Claessens, S., Djankov, S., Fan, J., Lang, L.H.P., Economics 11, 5–50. 2002. Disentangling the incentive and Johnson, S., La Porta, R., Lopez-de-Silanes, F., entrenchment effects of large shareholdings. Shleifer, A., 2000. Tunnelling. American Journal of Finance, forthcoming. Economic Review Papers and Proceedings Comision Nacional del Mercado de Valores, 90, 22–27. 1998. Participaciones significativas en La Porta, R., Lopez-de-Silanes, F., Shleifer, A., sociedades cotizadas. (Data file). CNMV, Vishny, R.W., 1997. Legal determinants of Madrid. external finance. Journal of Finance 52, Commerzbank, 1997. Wer gehört zu wem, 19th 1131–1150. Enlarged Edition. Commerzbank AG, La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Frankfurt-am-Main. 1999. Corporate ownership around the world. Commissione Nazionale per le Società e la Journal of Finance 54, 471–518. Borsa, CONSOB, 1997. Bollettino - edizione La Porta, R., Lopez-de-Silanes, F., Shleifer, A., speciale n. 4/97 - Compagine azionaria delle Vishny, R.W., 2002. Investor protection and società quotate in borsa o ammesse alle corporate valuation. Journal of Finance 57, negoziazioni nel mercato ristretto al 1147–1170. 31 dicembre 1996. CONSOB, Milan. Lease, R.C., McConnell, J.J., Mikkelson, W.H., Crespi-Cladera, R., Garcia-Cestona, M.A., 1983.The market value of control in publicly 1998. Ownership and control: a Spanish sur- traded corporations. Journal of Financial vey. Unpublished working paper, European Economics 11, 439–471. Corporate Governance Network. Lins, K., Servaes, H., 1999a. International De Jong, A., Kabir, R., Marra, T., Roell, A., evidence on the value of corporate 1998. Ownership and control in the diversification. Journal of Finance 54, Netherlands. Unpublished working paper, 2215–2239. European Corporate Governance Network. Lins, K., Servaes, H., 1999b. Is corporate Financial Times, 1997. Extel Financial, diversification beneficial in emerging Company Research CD-rom #36/1997. FT, markets? Unpublished working paper,London London. Business School. 190 EQUITY OWNERSHIP STRUCTURE AND CONTROL London Stock Exchange, 1997. The London Renneboog, L., 1998. Shareholding concentration Stock Exchange Yearbook. LSE, London. and pyramidal ownership structures in Megginson, W., 1990. Restricted voting stock, Belgium: stylized facts. Unpublished working acquisition premiums, and the market value paper, European Corporate Governance of corporate control. The Financial Review Network. 25, 175–198. Rydqvist, K., 1992. Dual-class shares: a review. Muus, C.K., 1998. Non-voting shares in Oxford Review of Economic Policy 8, 45–57. France: an empirical analysis of the voting Shleifer, A., Vishny, R., 1997. A survey of cor- premium. Unpublished working paper, porate governance. Journal of Finance 52, Johann Wolfgang Goethe-Universitat, 737–783. Frankfurt-am-Main. The Herald Tribune, 1997. French Firm Muus, C.K., Tyrell, M., 1999. An empirical Handbook 1997. SFB-Paris Bourse. analysis of the voting premium in Germany. Union Bank of Switzerland, 1998. Swiss Stock Unpublished working paper,Johann Wolfgang Guide 96/97. UBS, Zurich. Goethe-Universitat, Frankfurt-am-Main. Wiener Börse, 2000. Yearbook 1999. Vienna Nenova, T., 2000. The value of corporate votes Stock Exchange, Vienna. and control benefits: a cross-country analysis. Zingales, L., 1994.The value of the voting right: Unpublished working paper, Harvard a study of the Milan stock exchange experience. University. Review of Financial Studies 7, 125–148. Chapter 8

Marc Goergen and Luc Renneboog WHY ARE THE LEVELS OF CONTROLS (SO) DIFFERENT IN GERMAN AND UK COMPANIES? EVIDENCE FROM INITIAL PUBLIC OFFERINGS

Source: Journal of Law, Economics and Organization, 19(1) (2003): 141–175.

ABSTRACT We analyze why the control of listed German and U.K. companies is so different. As shareholders in Germany are less protected and control is less expensive, German investors prefer controlling stakes. We also focus on economic factors such as profitability, risk, and growth to predict the probability of occurrence of different states of control six years after the flotation. Large U.K. companies become widely held, whereas in large German firms new shareholders control significantly larger stakes. Wealth constraints become binding for U.K. shareholders, whereas German shareholders can avoid this by using pyramids. We find substantial differences between a takeover by a concentrated shareholder and one by a widely held company. For the United Kingdom, the probability of the former increases when the company is risky, small, and poorly performing. Conversely, the latter is more likely when the target is large, fast growing, and profitable. Poor performance and high risk require control and monitoring by a concentrated shareholder.Conversely, high growth and profitability attract widely held companies. Founders are less inclined to dilute their stake to retain private benefits of control.When German firms are profitable and risky, control is likely to go to a concentrated shareholder, but growth and low profitability increase the probability of a control acquisition by a widely held firm.

1. INTRODUCTION Becht and Mayer,2001). Not only does the concentration of control differ between THERE IS A WELL DOCUMENTED—BUT SO FAR these countries, but so does the nature of largely unexplained—discrepancy in the levels ownership: Germany is characterized by of control concentration between listed intercorporate equity relations and family continental European and Anglo-American control, whereas institutional shareholders firms. For example, about 90% of compa- hold most of the voting rights in the United nies listed on the London Stock Exchange Kingdom. Also, German firms are on do not have a major shareholder owning average more than 50 years old when they 25% or more of the voting rights, whereas are floated, whereas U.K. initial public 85% of the listed German companies have offerings (IPOs) are only 12 years old. It such a shareholder (La Porta et al., 1999; comes as a surprise that such substantial 192 EQUITY OWNERSHIP STRUCTURE AND CONTROL differences are still observable, especially between takeovers by concentrated in the wake of high product-market shareholders and those by widely held competition and economic globalization. companies. For the United Kingdom, the This study contributes to a better under- probability of a transfer of control to a standing of the reasons for the differences concentrated shareholder increases when a in control between Germany and the United company is risky, small, and poorly Kingdom. To this aim, we analyze a unique performing. A U.K. firm is more likely database of recently floated German and to be taken over by a widely held firm if the U.K. firms containing detailed data on the target firm is large, fast growing, and control structure for a period of up to six profitable. So, for the United Kingdom, years after the IPO. poor performance and high risk necessitate We argue that the relative cost of a high level of control and tight monitoring holding large control stakes differs across by a concentrated shareholder. Conversely, countries. In countries with a low protection high growth and profitability attract widely of shareholder rights, shareholders can pre- held companies whose management may vent a violation of their rights by building well be driven by an “empire building” up controlling stakes and by taking advan- acquisition programme. We find that high tage of a higher level of private benefits of growth also leads to more diffuse owner- control (Dyck and Zingales, 2001). A thor- ship in the United Kingdom. However, this ough investigation of minority shareholder is less likely when the founder family is still protection regulation, control disclosure, involved in the company. When German multiple-class shares, fiduciary duties and firms are profitable and risky, control is composition of the board of directors, own- more likely to go to a concentrated share- ership structure, voting rules and practice, holder, but growth and low profitability the arm’s length relation with large share- trigger a control acquisition by a widely holders, and mandatory takeover thresh- held firm. If the founder of a German firm olds reveals that investors in the United is still a shareholder at the IPO and if Kingdom are substantially better protected there are nonvoting shares outstanding, than the ones in Germany and that the control is likely to remain with the initial relative cost of control is lower in Germany. shareholders. This is not surprising as This result is in line with the findings of founding families often extract (nonpecu- La Porta et al. (2000). We pursue our niary) private benefits of control and analysis of the impact of the legal and nonvoting shares enable them to raise regulatory aspects on control by investigating additional equity while maintaining con- the role of inheritance tax and listing trol. Large U.K. companies evolve toward requirements on the stock exchanges. a more widely held equity structure, After examining the legal and regulatory whereas in large German firms, new share- differences which may explain differences holders control significantly larger voting in the level of control concentration, we stakes. The reason is that wealth analyze the corporate characteristics constraints may become binding for U.K. (such as profitability, size, risk, and shareholders, whereas German shareholders growth), that is, the economic determi- can avoid this by using pyramidal owner- nants that may trigger changes in control. ship structures. These ensure control while More specifically, we intend to predict allowing for a dispersion of cash flow whether or not the initial shareholders rights. Age has the opposite impact on retain control or transfer control to a new control concentration of German and U.K. widely held shareholder or to a concen- firms: initial shareholders of older German trated bidder or to a large number of new firms tend to retain smaller voting stakes diffuse shareholders. We also analyze than those of relatively older U.K. firms. whether or not U.K. and German control The article proceeds as follows. In changes are subject to different economic Section 2 we state some conjectures factors. We find substantial differences regarding the impact of the differences WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 193 in inheritance tax, stock exchange We expect that lower levels of control at regulation, and legal rules on German and the float are to be found in the country U.K. control levels. In addition, we analyze where the stock exchange imposes the the potential impact of regulation on the highest minimal free float (conjecture 1). relative costs of holding large voting Hence, as the level of control held by the stakes. In Section 3 we formulate our initial shareholders in U.K. firms is signifi- propositions regarding the relation between cantly lower than that in German companies, specific corporate characteristics and the we expect that the London Stock Exchange dynamics of control concentration in both requires that a higher percentage of the countries. Section 4 describes our data equity is offered to the public. However,this sources and the methodology. In Section 5 is not the case: Goergen (1998) shows that we present some stylized facts on control in the 1980s and 1990s the admission which provide further empirical justification requirements were very similar in terms of for the research objective of this article. minimum size, minimum ownership disper- The empirical results of the Tobit and sion, and trading history. Consequently the multinomial logit models are presented in listing requirements cannot account for Section 6. Section 7 concludes. differences in initial control. This refutes conjecture 1. High inheritance taxes may force owners to float a company to be able to afford the 2. REGULATORY DETERMINANTS taxes (Hay and Morris, 1984). Thus we OF THE LEVEL AND EVOLUTION expect that a lower degree of ownership OF CONTROL retention will occur in the country with the higher tax and a higher average In this section we compare the German corporate age (conjecture 2). The higher and U.K. legal corporate governance initial control concentration in Germany rules, stock exchange regulations, and could be partially explained by higher inheritance taxes and formulate conjec- inheritance taxes in the United Kingdom tures about the impact of these on the and by a higher corporate age of U.K. level of control concentration in both firms. We find precisely the opposite: countries. We analyze whether differences German firms at flotation are on average in regulation and legislation can explain 50 years old, compared to a mere 12 years the stronger concentration of control for U.K. firms (see infra). Furthermore, in German companies compared to U.K. during the 1980s and 1990s the levels of firms. inheritance tax in Germany exceeded those in the United Kingdom (Goergen, 1998). This evidence fails to support 2.1 Stock exchange listing conjecture 2. requirements and inheritance tax

As reported in the introduction and 2.2 Protection of shareholder rights detailed in Section 5, there are striking discrepancies in control concentration When shareholder rights are not suffi- between Germany and the United Kingdom. ciently protected or cannot be easily The differences in control may depend on enforced in court, shareholders may the initial differences at the float and such increase their control to levels that make differences may disappear over time. At the them no longer vulnerable to expropria- IPO, large shareholders of German firms tion or to levels that ensure large private own on average a supermajority (76% of benefits of control. For example, a share- the voting stock) versus 63% in the United holder or coalition of shareholders who Kingdom. This may be caused by owns a (combined) stake of 20% and is differences in listing requirements and afraid of minority shareholder expropria- inheritance taxes. tion by a majority shareholder has an 194 EQUITY OWNERSHIP STRUCTURE AND CONTROL interest to increase the stake to at least 2.2.2 Control-induced tender offers 25%. Such a minority stake enables blocking of changes to the statutes, The City Code on Takeovers and Mergers including changes to voting rights and and the U.K. Company Law protect minority their distribution. We expect a higher and dispersed shareholders by creating control concentration in Germany given obstacles to building controlling stakes. the lower degree of shareholder rights. When a stake of 30% or more has been Consequently, if the initial shareholders acquired in a U.K. firm, a tender offer for decide to sell out, new large shareholders all remaining shares is mandatory. arise and the free float will remain low Therefore some shareholders limit their (conjecture 3). To make this case we control stake to just below the 30% level investigate legal origin, the threshold of (Goergen and Renneboog, 2001). In compulsory tender offers, control disclo- Germany, a similar takeover code was sure, board representation, fiduciary introduced only as recently as 1995 (just duties of directors, complex shareholding after our sample period which ends in structures, proxy voting, the arm’s length 1994), but minority protection is still clause between the major shareholder and weaker. A mandatory tender offer only has the company, and the possibility to limit to take place when a large shareholder voting rights (see Table 8.1). More pre- obtains a 75% control stake (Boehmer, cisely, we show that the value attached to 1999). Furthermore, Wenger, Hecker, and partial control is expected to be higher in Knoesel (1996) show that in 60% of the Germany than in the United Kingdom. offers to minority shareholders, the offer price is below the market value, whereas in the United Kingdom the tender price has to 2.2.1 Legal origin be at least as high as the price at which the bidder acquired the shares over the previ- La Porta et al. (1997, 1998, 2000) ous 12 months. To conclude, shareholders investigate the relation between legal origin in the United Kingdom without the inten- (common law versus civil law), quality of tion to acquire a company may consider a investor protection, and quality of law 29.9% stake as an upper limit. German enforcement, on the one hand, and company shareholders only face a 75% tender characteristics (among them, control threshold, but can make the tender offer concentration), on the other hand, for unattractive by setting the offer price 49 countries. The authors find that the below the market value. average company in civil law countries (such as Germany) has concentrated control. In contrast, the common law system of the United 2.2.3 Board representation and Kingdom corresponds to, on average, widely directors’ fiduciary duties held firms. La Porta et al. (1998) also find a close correspondence between the legal Another incentive for shareholders to build origin and shareholder protection: the com- up large blocks in German firms but not in mon law system provides higher guarantees U.K. ones is created by board representa- for shareholder rights and the enforceability tion and the definition of directors’ fiduciary of such rights is also better. Consequently duties. On the one-tier U.K. board, about the findings of La Porta et al. (1998) pro- 60% of the directors are nonexecutives vide empirical support for conjecture 3, (Franks et al., 2001), most of which are which states that the lower degree of share- independent from management and are not holder protection in Germany is related to a direct representatives of specific (major) higher control concentration and to a lower shareholders. The need for independent occurrence of widely held firms, whereas a directors was emphasized by the Code of high quality of investor protection allows Best Practice of the Cadbury Commission shareholders to hold smaller voting stakes in 1992, which the London Stock in the United Kingdom. Exchange has endorsed for all listed firms WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 195 Table 8.1 Comparison of Regulatory Issues Germany United Kingdom

Ownership disclosure Minimum disclosure level 5% of voting rights 3% (5% prior to 1989) (25% prior to 1995) Minimum threshold of None 15% control to reveal strategic intent (takeover intentions) Dual-class shares Nonvoting shares: No nonvoting shares common Voting caps Common at 5–10%, No but caps were abolished in 1998 with a grandfather clause for existing voting caps in by-laws Control structure Complex, multilayered Simple, one ownership cascades of tier ownership levels Takeover code (since 1995) Minimum level 75% 30% of control for obligatory tender offer on all shares Minimum price of tender Can be lower than Not lower than market value the highest share price during last 12 months Using voting rights Casting votes Presence at AGM Allowed by mail, fax, or required in person Internet. But, in practice, or representative presence (in person or by representative) is required for show of hands. Proxy votes only matter in poll Registration of votes Deposit of shares Shares need to be with notary, registered depository bank, or company itself Degree of minority Weak Strong protection Boards Two-tier board One-tier board with a large- with independent shareholder nonexecutive directors dominated supervisory board Corporate aim of Pursuing stake- Maximizing share- management holder interest holder value (continued) 196 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 8.1 Continued Germany United Kingdom Fiduciary duty of directors Duty of care and Wider and more duty of loyalty developed than (Treuepflicht) in Germany New equity issues Via rights Via rights Listing requirements Similar to the United Similar to Germany Kingdom Inheritance tax Higher than in the Lower than in United Kingdom Germany

since 1993. In Germany, in contrast, the directors of the subsidiary must be Aufsichtsrat (supervisory board of the independent from the parent firm and two-tier board system) is not independent: minority shareholders have the right to be it represents shareholders and employees consulted about, and approve, transactions and is dominated by large shareholders.1 with the parent firm (Franks et al., 2001).3 Baums (2000) compares the fiduciary The effect of these rules is to increase the duties (duty of care and loyalty) in German costs of holding equity stakes.They explain and U.K. corporate law and concludes that why almost all bids are made conditional “the range of fiduciary duties in the on being accepted by 90% or more of the English law system seems wider and more target’s shareholders. The remainder can developed than in its German counterpart” be purchased at the original bid price using (p. 8).The relatively stronger independence a squeeze-out rule under the 1948 of U.K. nonexecutive directors and their Companies Act. German law also has an wider fiduciary duties enable shareholders arm’s length concept (para. 76, Stock to hold smaller stakes, as their rights Corporation Act): the management board are better safeguarded. This supports of a subsidiary is not allowed to follow conjecture 3. instructions or take measures which are not in the interest of the subsidiary and is not allowed to be compensated by the hold- 2.2.4 Relation with major shareholders: ing company. Still, it is doubtful how the Arm’s length? relation between a company and its con- In the United Kingdom, minority protection trolling shareholders can be at arm’s length is based on the “property rule,” which with a shareholder-dominated supervisory prevents any transaction from proceeding board. The efficiency of the arm’s-length without the consent of the minority owners regulation in German firms is eroded by the (Goshen, 1998). In addition, the rules of fact that large shareholders nominate the London Stock Exchange prohibit representatives to the supervisory boards. controlling shareholders owning more than Such actions may ensure that large 50% of shares from having too large an shareholders safeguard their private impact on the firm: “a firm must be capa- benefits of control. This is further support ble at all times of operating and making for conjecture 3. decisions independently of any controlling shareholder and all transactions and 2.2.5 Control through pyramids relationships in the future between the applicant and any controlling shareholder The regulation discussed above encourages must be at arm’s length and on a normal shareholders in Germany to hold a large commercial basis.”2 A majority of the control stake.To compensate for the cost of WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 197 owning large equity stakes in terms of lost issued nonvoting shares converted them liquidity, an intricate web of multilayered into voting shares under the pressure of the tiers of shareholdings is often used by London Stock Exchange and institutional families or companies.The main reason for investors during the early 1990s. Whereas building such pyramids is control leverage.4 German corporate law (Aktiengesetz, para. As companies can issue up to 50% of the 139) prohibits the issue of multiple voting equity as nonvoting shares, a mere 25% of rights, nonvoting shares up to the amount the cash flow rights (50% of the voting of ordinary shares outstanding are explic- rights) at every tier of shareholdings is suf- itly allowed.8 As such, the issue of nonvot- ficient for a shareholder to retain control ing shares increases the relative power of over a target company while the capital the initial shareholders and allows them to investment is minimized. In addition, a retain control with limited investment.9 shareholder (e.g., a family) could also hide The only regulatory element which his identity prior to the introduction of discourages shareholders in Germany from the disclosure regulation.5 While pyramids holding large equity stakes is that voting are not explicitly forbidden in U.K. corpo- limitations can be imposed by the by-laws. rate law, it is surprising that this control Usually the voting power of a shareholder leverage technique is not used at all. The is limited to 5% or 10%.10 Only recently main reason is that ownership disclosure have voting caps been prohibited in regulation in the United Kingdom not only Germany (Act on Control and Transparency applies to individuals or companies but also of Enterprises of 1998 (KonTraG)), but a to individuals and companies with voting grandfather clause applies for existing agreements. Such voting agreements voting caps. Voting caps in listed U.K. consist of obligations or restrictions companies are not allowed. Hence, the between shareholders with respect to the one-share-one-vote principle is better use, retention, or disposal of their stakes. A upheld in the United Kingdom than in coalition of shareholders with a voting Germany. agreement will be considered by the regulatory authorities as a single shareholder. This implies, for instance, that if the com- 2.2.7 Proxy voting and voting practice bined direct and indirect shareholdings of a Another important element of the protec- coalition amount to at least 3%, disclosure 6 tion of shareholder rights is the voting is compulsory. Furthermore, a coalition procedure. In both the United Kingdom and controlling directly and indirectly 30% or Germany, a person or legal entity can more of the equity will be obliged to make 7 represent a shareholder at the AGM. At a tender offer for all shares outstanding. first sight there seems to be less need to This section has shown that the costs of resort to proxy voting in the United holding large share stakes are reduced by Kingdom because exercising one’s voting pyramids that allow the combination of rights seems substantially easier. However, strong control with limited investment. the general voting practice requires that a shareholder of a U.K. firm or his represen- 2.2.6 Nonvoting shares and voting caps tative is present at the general meeting in order to cast his votes. The U.K. voting By issuing nonvoting shares, initial owners practice is captured by this statement from can dilute their cash flow rights in the firm the IPO prospectus of Compel plc: “At a without relinquishing control. Although general meeting every member present in nonvoting shares are in principal admitted person shall, on a show of hands, have one by the London Stock Exchange, issues of vote and every member present in person or nonvoting shares have been actively dis- by proxy shall, on a poll, have one vote for couraged (Brennan and Franks, 1997). every ordinary share of which he is the Goergen and Renneboog (2001) report holder.” On uncontroversial issues, voting that the few listed U.K. companies that had takes place by a show of hands, which 198 EQUITY OWNERSHIP STRUCTURE AND CONTROL requires that shareholders be present and protection of shareholder rights in the all shareholders present in person have a United Kingdom makes holding blocks vote regardless of their number of voting relatively more costly than in Germany. rights (see Stapledon, 1996; Goergen and Third, owning large blocks in Germany is Renneboog, 2001). However, every share- less costly than in the United Kingdom holder (present) can ask for a poll on any because (i) pyramids can be used in item of the agenda. Only then does the Germany, (ii) nonvoting shares can be number of voting rights owned by every issued, and (iii) supervisory board represen- shareholder matter and the proxy votes are tatives can safeguard shareholders’ private counted. In spite of the fact that legal rules benefits of control.The only dissuasive factor seem to facilitate voting in U.K. firms, against holding a high level of control in there is not much of a difference between Germany is the possibility that voting rights Germany and the United Kingdom in terms are restricted in the corporate by-laws.Table of the requirement that shareholders or a 8.1 summarizes the results from this section. representative be present at the AGM.11 Both Baums and Schmitz (2000) and Boehmer (1999) point out that the 3. ECONOMIC DETERMINANTS OF German proxy voting system comprises sig- CONTROL CONCENTRATION AND nificant deficiencies. Banks can receive EVOLUTION proxy votes from the shareholders who have deposited the shares in the bank Whereas the previous section has shown (Depotstimmrecht) such that these control that legal rules and stock exchange regula- rights substantially outweigh the banks’ tions provide incentives for a stronger own control rights. Typically the amount concentration of share ownership in of debt held by banks exceeds the amount of Germany (compared to the United equity held in the firm by a factor greater Kingdom), these rules do not explain how than 10. Consequently banks have little control evolves from the float onwards. incentive to act on behalf of other share- Several years subsequent to the IPO, control holders. Finally, banks virtually always vote concentration can result from a high in favor of management proposals (Baums retention rate by initial shareholders, from a and Fraune, 1995) and seem to have little transfer of a controlling block to a new impact on corporate performance shareholder, or from a full takeover. In the (Chirinko and Elston, 1996). In the United case of a transfer of control, the (potential) Kingdom, the management can solicit agency costs are different when control is proxy votes in support of managerial plans acquired by either a widely held firm, or by (Stapledon, 1996). Thus, although the a closely held firm, a family, or an individual regulation on voting procedures seems to (as the ultimate shareholder). In the former plead more in favor of minority protection case, management has a lot of discretion, in the United Kingdom, the voting practice whereas in the latter,large shareholders may in Germany and the United Kingdom is not influence corporate actions, possibly even at substantially different.This fails to support the expense of minority shareholders. This conjecture 3.12 section develops the relation between To summarize, this section has discussed corporate characteristics and control, and the many incentives for shareholders to formulates the expected differences between hold relatively larger voting stakes in Germany and the United Kingdom. Germany than in the United Kingdom. Specifically, these hypotheses will be tested First, we investigated whether the initial on a sample of German and U.K. firms of (and lasting) differences in equity concen- which the level of and changes in control are tration in German and U.K. IPOs can be collected for the six years after the float. explained by differences in stock exchange The first economic determinant of control listing requirements and by inheritance tax concentration is firm size. Demsetz and law, which neither can. Second, the higher Lehn (1985) argue that wealth limitations WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 199 as well as portfolio diversification needs remain among the major shareholders in restrict the existence of substantial share subsequent years. Morck, Shleifer, and stakes in large firms. Therefore we expect Vishny (1988) argue that the founder of a that the larger the firm, the higher is the firm may provide essential leadership probability that wealth and diversification skills, especially in younger firms, and constraints of the initial shareholders should therefore retain control. Founder become binding. Thus the control retention commitment at the time of the IPO is by the initial shareholders will be lower in included in the model as a proxy for the pri- larger firms and these firms will evolve vate benefits (which may be nonpecuniary) toward widely held control and have fewer that the founder extracts from controlling new large shareholders. Whereas the nega- the firm (Johnson et al., 2000). Given that tive relation between ownership and size large shareholders are more likely to may be equally valid for the United extract private benefits13 in German firms, Kingdom and Germany, the relation we expect higher control concentration in between control and size may differ. In the Germany (Proposition 3). Zingales (1995) United Kingdom, share ownership is and Mello and Parsons (1998) show that equivalent to control as a result of the the optimal path of selling control to max- one-share-one-vote principle. Conversely imize the proceeds consists of two stages. this principle may be violated in German First, the initial shareholders should take companies because nonvoting shares are the firm public and sell some cash flow frequently issued (in 38% of our sample rights to a large number of investors in firms).This enables the initial shareholders order to retain control.14 Second, they to retain control more easily, while the firm should sell control to a new controlling can at the same time attract additional shareholder at a premium. Morck, capital without diluting the initial share- Strangeland, and Yeung (2000) distinguish holders’ control. In addition, pyramids (see between large shareholder stakes resulting above) also limit the direct investment from entrepreneurial investment and those needed to keep controlling stakes. resulting from inherited wealth. The Consequently the negative relation between authors find that heir-controlled Canadian size and control is expected to be weaker in firms have lower levels of financial German than in U.K. firms (Proposition 1). performance, labor-capital ratios, and We also expect that control retention by research and development (R&D). In the initial shareholders decreases with contrast to entrepreneurial control, rising corporate age. Over time, these initial concentrated, inherited ownership impedes shareholders (or their heirs) may face growth and shows signs of entrenchment personal liquidity needs and hence decide and political rent seeking. to liquidate (part of) their stakes. As the Bolton and von Thadden (1998) model average age of German firms at the float is the evolution of control from its initial level substantially higher than the one of U.K. toward its optimal level and find that high- firms, we expect the control retention in risk firms will end up being widely held. U.K. firms to be higher than the one in Their proposition has been supported by German firms (Proposition 2). In the cases several empirical studies: for example, a where control is reduced by the initial negative relationship between control and shareholders of German firms, we expect risk is found by Demsetz and Lehn (1985) control stakes to be transferred and not for the United States and Leech and Leahy dissipated, as blocks are valuable (Shleifer (1991) for the United Kingdom. and Vishny, 1986). This implies that we Alternatively, high risk may trigger a expect to see a higher incidence of new control transfer to a shareholder with a large shareholders in German firms. stronger ability to monitor the firm. If the founder family is still involved in Therefore we propose that the probability the firm (in terms of voting rights) at the that the firm’s original owners will float, the odds are higher that they will reduce their control stake increases with 200 EQUITY OWNERSHIP STRUCTURE AND CONTROL risk and high risk triggers control transfers Profitable firms generating sufficient to large shareholders with strong monitoring cash flow allow the initial shareholders to skills (Proposition 4). As German firms are retain control for a longer period than older at the float (and have more stable poorly performing firms (Dennis and Sarin, and less risky cash flows than their U.K. 1999). Poor performance may reflect not counterparts) and as shareholders can only weak management but also poor mon- control a German firm with a lower itoring, which may activate the market for percentage of voting rights, Proposition 4 corporate control or persuade another is more likely to hold for U.K. firms. shareholder with superior monitoring skills Growth may also force changes in con- to acquire control (Franks et al., 2001). trol: rapidly growing firms may have to Thus we expect poor performance to trig- resort to attracting more external equity ger changes in control (Proposition 7). than firms operating in mature industries. Such changes in control are likely to occur Thus the stronger the growth, the higher is differently in the United Kingdom than in the probability that control by the initial Germany. In the former,the market for cor- shareholders is diluted and that either a porate control consists in full takeovers new shareholder emerges (by a partial or (Franks and Mayer, 1996), whereas in full takeover) or that the firm becomes Germany it operates via partial takeovers widely held (Crespi, 1998) (Proposition 5). or a transfer of a controlling stake to a new We expect this relation to hold more major shareholder (Jenkinson and strongly for U.K. than for German firms Ljungqvist, 2001). because, in the latter, pyramids enable shareholders to attract external capital while limiting control dilution. 4. DATA SOURCES, DESCRIPTION OF In line with DeAngelo and DeAngelo VARIABLES, AND METHODOLOGY (1985) for the United States, voting shares in German IPOs are normally held by the 4.1 Sample description and data initial owners, whereas nonvoting shares sources are issued to the outsiders in the IPO (Goergen, 1999). Immediately after the Even though the German economy is about IPO, the family shareholder owns on 1.8 times larger in terms of gross domestic average 57% of the voting rights, but only product (GDP) than the U.K. economy, its 47% of total equity (voting plus nonvoting capitalization is substantially smaller, as shares). Thus, we expect that issuing non- the number of quoted companies on the voting shares at the IPO leads to higher German exchanges is less than one-third control retention by initial shareholders the number of listed firms on the London and that control retention will be higher in Stock Exchange (which is around 2,000 Germany than in the United Kingdom firms, including 550 financial institutions (Proposition 6). For those cases where and investment funds). Similarly the num- control is transferred, we expect new share- ber of IPOs on the London Stock Exchange holders to be able to acquire control more is much larger.In the United Kingdom, 764 easily in Germany than in the United firms went public during the period Kingdom. For instance, suppose that there 1981–1988, of which 284 were floated on are two firms with a different equity struc- the Official Market and 480 on the ture but which are otherwise identical: one Unlisted Securities Market (USM), that is, firm has only voting shares outstanding, the secondtier market. Over the same while the other has issued 50% voting and period, a total of only 96 German firms 50% nonvoting shares. If the initial share- went to the stock exchange, 51 of which holders decide to sell a controlling stake, were listed on the Official Market and 45 the new controlling shareholder will only chose a listing on the Regulated Market need half of the capital to acquire control (the second-tier market). This study con- in the firm with the nonvoting equity.15 centrates on domestic IPOs listed on the WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 201 official and secondary markets, as data for The U.K. firms are riskier (at the 1% level lower market tiers16 are usually not available. of significance), measured by the standard To ensure comparability across the deviation of their monthly share returns. German and U.K. IPOs, in this study we German firms also have a higher cash flow only retain those IPOs controlled by an both as a proportion of their book value of individual or a group of persons, such as a assets (CF1) and as a proportion of the family or unrelated associates. Thus we do sum of the market value of equity and the not include privatizations or equity carve- book value of debt (CF2). outs. Consequently this study analyzes Information on the identity of initial more than 90% of all German IPOs with shareholders and on pre- and post-IPO available data.17 holdings was obtained from the IPO The distribution of the population of prospectuses. Substantial share stakes 96 German and 764 U.K. IPOs across were traced over the period after the IPO industries reveals significant differences.18 through the company reports as well as the Although the industry with the highest fre- London Stock Exchange Yearbooks for the quency of IPOs is the same for both coun- United Kingdom and the Saling tries (electricals, electronics, and office Aktienführer for Germany. Information on equipment), there are proportionally more both the direct and ultimate voting stakes German IPOs in mature industries (such as was collected for the German firms. To mechanical engineering with 15.5% of the identify the ultimate shareholder and his total number of IPOs and motor compo- level of control, the ownership pyramids nents with 5.2%). Conversely, there is a were reconstructed on a year-by-year basis. higher proportion of U.K. IPOs in cyclical Higher tiers of ownership were traced for service industries with 29% of the sample those large shareholders owning 25% or (service agencies with 9.0%, property with more of the voting rights as long as the 6.0%, leisure with 5.7%, chain stores with control chain in the ownership cascade was 3.6%, and construction with 4.9%). not interrupted. The ultimate level of con- Within each industry, U.K. IPOs are also trol is reached when the ultimate share- usually smaller than German IPOs. holder is either an individual or a family, or Table 8.2 shows that there are also is a widely held company (i.e., a firm that marked differences between the population does not have a shareholder owning at least of German and U.K. IPOs in terms of size, 25% of its voting rights).Whereas ultimate age, industry, and risk. The German IPOs control is only relevant for German compa- are twice as large as the U.K. IPOs, with nies, ownership in the United Kingdom is market capitalizations at the end of the equivalent to control, as multiple or nonvoting first day of listing of £113 million and shares are not used. Also, U.K. ownership £56 million, respectively.19 On average, the structures are simple and pyramids are rare German IPOs were founded 51 years prior (Goergen and Renneboog, 2001). to the float whereas U.K. firms were set up Share prices were collected from the only 14 years before the IPO. The age of Karlsruher Kapitalmarktdatenbank the German and U.K. sample IPOs (KKMDB) and the London Share Price matched by size is similar to the popula- Database (LSPD). IPO characteristics tion: German and U.K. firms are, respec- (age and industry) and the closing market tively, 50 and 12 years old prior to the capitalization for the first day of listing float. The Herfindahl index demonstrates were obtained from the Deutsche Börse AG that voting rights in the German firms at and the London Stock Exchange. the IPO are more concentrated than in their Accounting information was collected from U.K. counterparts (the difference is the IPO prospectuses, company reports, the significant at the 1% level; see Table 8.2, Extel Financial Company Research and panel B). German founder families own Global Vantage CD-ROMs for both coun- shares at the float in 92% of the firms, tries, and from Datastream and the Extel while this is the case in 85% of U.K. firms. Microfiches for the United Kingdom. 202 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 8.2 Summary Statistics of Independent Variables Age Size Herfin- Risk CF1 CF2 Variable (years) (£m) dahl Founder (%) (%) (%)

Panel A: Mean, median, proportion 1, minimum, maximum and sample size Germany Mean 49.4 56.4 0.92 — 9.1 17.7 11.0 Median 48.0 28.4 1.00 — 9.0 17.1 11.3 Proportion 1 — — 43 92.2 — — — Min 0.0 5.3 0.22 — 4.8 5.5 6.2 Max 171.0 296.0 1.00 — 15.9 40.7 22.0 Sample size 55 55 53 51 54 48 48 United Kingdom Mean 11.8 58.4 0.44 — 12.8 15.8 9.2 Median 6.0 28.2 0.39 — 13.1 14.8 9.6 Proportion 1 — — 2.0 85.2 — — — Min 0.0 5.3 0.03 — 4.8 2.0 4.0 Max 84.0 313.6 1.00 — 22.4 64.8 17.3 Sample size 55 55 50 54 54 46 43 Panel B: t-statistics for the difference in sample means 6.826*** 0.147 9.826*** 1.122 5.797***0.954 1.885*

The table is based on a sample of German and U.K. IPOs matched by size (market capitalization).The Z-test in panel B is a two-tailed test for the equality between two proportions following a binomial distribution. Age is the firm’s age since registration. Size is in million pounds sterling and is the market capitalization. Concentrated ownership is measured by a Herfindahl index of all pre-IPO stakes. Founder is a dummy variable set equal to one when the founder or his heirs still own a large equity stake in the firm at the float. Risk is the standard deviation of the monthly share price return over the five years following the IPO. CF1 is the annual cash flow defined as the published profit gross of depreciation, interest, taxes, and changes in provisions divided by the sum of the book values of equity and debt. CF2 is annual cash flow divided by the market value of equity and the book value of debt. ***, **, * represent the level of statistical significance at the 1%, 5%, and 10% level, respectively. Sources: IPO prospectuses, company reports, Datastream, Extel Financial, and London Stock Exchange.

4.2 Methodology and model For the IPO samples, the following description cross-sectional industry-fixed effects are estimated: The direct comparability of control in the two countries was enhanced by matching Percent voting controli the German firms with U.K. IPOs by industry 1Countryi 2Sizei and size.This way we select twin companies 3Founderi 4Riski whose control evolution we analyze while 5Growthi 6Nonvotingi controlling for factors such as risk, 7Profiti 8Agei 9Countryi *Sizei profitability, and growth. The size-matched 10Countryi * Founderi sample consists of 54 German and U.K. 11Countryi * Riski firms, whereas the industry-matched sample 12Countryi * Growthi contains 58 IPOs from each country.20 The 13Countryi * Profiti average difference in size between a German 14Countryi * Agei time dummies IPO and its matched U.K. IPO is 2.7%, with industry dummies i. a median of 0.5% and a standard deviation The percent voting control is the percent- of 4.7%. For seven German firms, a close ages of voting rights held six years after match, defined as a match within 25% the float by (i) the initial shareholders, difference in size, could not be found, neither (ii) the new large shareholders, and (iii) was it possible to find a U.K. industry match small shareholders (the free float), for three German IPOs. The two German respectively. As the dependent variable is samples have 52 firms in common.21 censored (the variable is zero in 33%, WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 203 27.4%, and 22.2% of the cases in the correlated with Age, Profit, Nonvoting, models with initial shareholders, new and Country, and (ii) Age is related to shareholders, and the free float as Country, Growth, and Nonvoting. These dependent variables, respectively), we seven correlation coefficients are statisti- estimate Tobit models. cally significant but are not large in Firm size (Size) is the natural logarithm absolute terms. As these correlations may of the market capitalization at the closing cause multicollinearity problems, we price on the first day of trading, converted subtract from the variables Age and Risk into 1985 pounds sterling. Founder involve- the country averages of these variables. ment is captured by the dummy variable Table 8.A1 in the appendix shows that the Founder, which equals one if the founder or statistical significance of the correlations his heirs hold a control stake in the firm at between the variables Riskminuscountryavg on the IPO and zero otherwise. The level of a the one side, and Age, Country, and firm’s risk (Risk) is measured by the stan- Nonvoting on the other has disappeared, as dard deviation of monthly stock returns well as the one between Ageminuscountryavg on over the five-year period after the flotation. one side and Country, Growth, and The growth rate (Growth) is defined as the Nonvoting on the other side. The high cor- average annual growth rate of total assets relation between Nonvoting and Country over the first five years after the float. cannot be eliminated, as nonvoting shares Nonvoting shares (Nonvoting) is a dummy are only issued by German firms. variable capturing the issue of nonvoting For the sake of robustness, we also use a shares at the time of going public (dummy set of alternative independent variables. equals one). Twenty-three and 25 German Size is now measured by the book value of firms issued nonvoting shares in the size- total assets at the end of the financial year and industry-matched samples, respec- covering the IPO date. Founder stands for tively. The profit rate (Profit) is defined as founder involvement in the management of the annual cash flow divided by the book the firm: the dummy variable equals one if value of total assets with annual cash flow the founder or a member of the founder measured as the profit gross of deprecia- family is an executive director of the firm, tion, interest, taxes, and changes in tax, and zero otherwise. Risk is the standard pension, and special provisions. Age meas- deviation of the cash flows at the IPO and ures the age of the firm in years at the time during the five years subsequent to the IPO. of the IPO. Nonvoting is now a continuous variable As both German and U.K. IPOs are measuring the proportion of nonvoting included in the model, a dummy variable shares in the firm’s total equity immedi- Country is used, which equals one for a ately after the IPO. For Growth, the same German company and zero otherwise. The definition is used as for the first set of inde- coefficient on this dummy registers a pendent variables. The profit rate (Profit) possible difference in the intercept is now the cash flow over the sum of the between German and U.K. IPOs. In order market value of equity and the book value to determine whether there is a country- of debt. specific effect for each of the variables As the dependent variables of the three described above, interactive terms Tobit models are interrelated, we also consisting of the product of each of the estimate multinomial logit models. Here above variables with the Country dummy we distinguish between four states of is included in the model. The interactive control six years before the float: firm i is terms pick up the differential effect (i.e., (i) controlled by the original owners (SC), the differential slope coefficient) for the (ii) widely held (WH), (iii) taken over by a German firms. closely held bidder (TC), or (iv) taken Table 8.A1 in the appendix shows the over by a widely held bidder (TW). Six Pearson correlation coefficients for the years after the float, two-thirds of the independent variables and the p-values for German firms in the size-matched sample the coefficients. We find that (i) Risk is were still controlled by their initial owner. 204 EQUITY OWNERSHIP STRUCTURE AND CONTROL This proportion is three times higher than sample matched by size. Initial control is for the U.K. sample. However, the number statistically different in the German and of (full and partial) takeovers is similar: 19 U.K. firms immediately after the IPO: the and 22 German and U.K. firms, respec- initial shareholders in German firms retain tively. As in most U.K. and German firms, 76.4% versus only 62.8% in U.K. compa- the initial shareholder retains majority nies. Initial shareholder control is diluted control at the float, we use this state as the much more rapidly in the United Kingdom benchmark case in the multinomial logit than in Germany.The original shareholders regressions.22 The specifications include a of British IPOs lose majority control on differential intercept and interactive slope average only two years after the float, coefficients, which pick up possible differ- whereas their German counterparts retain ential effects for the German IPOs. For the majority control up to five years. Although size- and industry-matched samples, the the reduction in control is slower in following cross-sectional multinomial German IPOs, as much as 35% of these model is estimated by voting rights change hands during the six-year period.The differences in initial shareholder U ’z and for ij ij ij concentration in German and U.K. IPOs are j 1,2,...,J: statistically significant at the 1% level for each of the six years following the float. Pr(Y 0) 1 J x At the float of U.K. firms, most of the k i 1 k1e shares are purchased by small shareholders, as the rationing schemes for share distribu- xi e k tion in case of oversubscription often favor Pr(Y j) J kx k small shareholders for control reasons 1 k1e (Brennan and Franks, 1997). Consequently the free float is significantly higher (at the More specifically, Uij is the state of control j in company i: 1% level of significance) in U.K. firms (37.2%) than in German ones (22.2%), State of controlij but remains relatively stable in both 1Countryi 2Sizei countries over the six-year period after the 3Founderi 4Riski IPO. Over this period control is transferred 5Growthi 6Nonvotingi from the initial shareholders to new large 7Profiti 8Agei shareholders who, on average, control 9Countryi * Sizei about 30% of the voting rights in both 10Countryi * Founderi the German and U.K. samples six years 11Countryi * Riski subsequent to the float. 12Countryi * Growthi 13Countryi * Profiti 5.2 Changes of control and control 14Countryi * Agei time dummies states industry dummies . ij Whereas Table 8.3 shows how control evolves over time, Table 8.4 reports the 5. CONTROL EVOLUTION AFTER THE control state six years after the float. Panel FLOAT: STYLIZED FACTS A of Table 8.4 shows that full takeovers are uncommon in Germany, with only one case 5.1 Control evolution by the initial in our sample.23, 24 In contrast, 35% of the and new large shareholders U.K. companies were acquired in a full takeover within six years after the float. In The voting power held by the three categories Germany, control is less frequently transferred of owners (initial, new large, and new small via a full takeover than via the sale of large shareholders) over the six-year period after voting blocks (which in fact constitutes a the IPO is recorded in Table 8.3 for the “partial” takeover). Such partial takeovers WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 205 54 54 52 52 nd U.K. IPOs * ** *** ** 0.1 5.5 54 nificance level,nificance respectively, of a 13.3 26.4 28.8 32.129.2 52 48 ontrol by the new shareholder as of the by the new shareholder ontrol its original shareholders. its original *** *** ** ** ** ** *** 37.2 43.1 39.5 36.0 37.6 36.5 40.8 *** *** *** *** *** *** *** , London Stock Exchange, and Extel. Hoppenstedt , company reports, Saling Initial Free New large Initial Free New large German sample U.K. sample Proportion of Voting Rights Held by Initial Shareholders, by Initial Held Rights by New Large Shareholders,Voting of a Proportion Recent German in and by Small Shareholders indicate that the average stake in U.K. in stake at the 1%, that the average firms indicate German in the stake from is statistically different firms 5%, and 10% sig * , ** , -test on the means. Sources: IPO prospectuses, 2 69.6 25.0 5.4 54 47.3 1345 73.7 64.9 59.4 50.7 24.0 25.3 25.0 2.4 26.3 9.8 15.5 23.1 54 54 54 54 51.4 37.7 33.6 31.4 Immediately 76.4 22.2 1.5 54 62.8 t year of the takeover. If a company is taken private by its original shareholders, by its original 100% by be recorded as a company owned private of the takeover. it will If a company is taken year Table 8.3 IPO (years) (%)6 (%) with the U.K. IPOs are matched by size German The IPOs. and delisted, over If a company is taken it is classed as a case of 100% c *** (%) 45.0 size 24.8 (%) 30.2 (%) 54 (%) 30.0 size Time after shareholders float shareholders Sample shareholders float shareholders Sample 206 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 8.4 State of Control of IPOs Six Years After Flotation

Size sample Industry sample Germany U.K. Germany U.K.

Panel A: Number of firms by state of control Full takeover 1 19 1 18 Partial takeover 18 5 20 4 Widely held (25%) 1 17 1 16 Still controlled by initial 34 13 36 20 shareholders Total 54 54 58 58 Panel B: Average number of years before reaching the new state of control Full takeover 6.0 3.7 6.0 3.5 Partial takeover 4.4 3.0 4.3 3.5 Widely held (25% def.) — 1.8 — 1.7 Panel C: Ultimate shareholder in targets of full and partial takeovers (size-matched sample) Targets full Targets partial All full and partial takeover takeover takeovers Germany U.K. Germany U.K. Germany U.K.

Widely held 1 14 8 — 9 14 Closely held — 5 10 5 10 9

For panels A and B, the disclosed state is the one six years after the IPO or the one the last year of listing if the firm was delisted before the sixth year after the float. A full takeover is a takeover of the entire voting rights of the firm, followed by the delisting of the firm. A partial takeover is a change of the major shareholder, the major shareholder being the largest shareholder holding at least 25% of the voting equity. A firm is widely held if no single shareholder owns more than 25% of the voting rights. A firm is still controlled by an initial shareholder if the initial shareholder is the largest shareholder and holds more than 25% of the votes. In panel C, a firm is classed as widely held if its ultimate largest shareholder is widely held, that is, is not controlled by a person or family holding more than 25% of the votes. A firm is classed as closely held if the ultimate largest shareholder is a person or a family. Sources: McCarthy microfiches, Saling Aktienführer, Financial Times, company reports. are infrequent in the United Kingdom control across the remaining German and because of the legal requirement to make a U.K. firms is still substantially different, as tender offer for the entire equity of the firm 62% of the German firms remain in the hands as soon as an investor acquires 30% or of their initial owners, compared to only 24% more of a firm’s equity.25 Thus, if share- of the U.K. firms in the size-matched sample holders in a U.K. company do not want to (and 35% in the industry-matched sam- end up with 100% control, they will delib- ple). Only one German company becomes erately remain below the 30% threshold widely held, whereas 32% of U.K. firms are (Goergen and Renneboog, 2001). already widely held six years after the IPO. As a matter of fact, partial takeovers in The reason why we follow the control Germany and full takeovers in the United evolution for a six-year period subsequent Kingdom bring about similar results in to the float can be found in panel B of Table terms of control change (Goergen, 1998). 8.4.26 Control changes in U.K. IPOs take Panel A of Table 8.4 shows that, after amal- place three to four years after the float, gamating all control changes (both full and whereas in German IPOs they happen within partial takeovers), the percentage (36% a four-to-six year period. and 38%) is almost equal in German and Even though the percentage of German U.K. IPOs. However, the distribution of and U.K. firms with a control change WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 207 Table 8.5 Voting Rights in Excess of 25%, Six Years After Flotation Germany United Kingdom

A: companies without a large shareholder 1.85% 33.33% B: companies with a large shareholder 98.15% 66.67% 1. Another domestic company 7.41% 29.63% 2. A foreign company 22.22% 5.56% 3. An insurance company 1.85% 0.00% 4. A trust/institutional investor 0.00% 3.70% 5. A family group 66.67% 27.78% 6. A bank 0.00% 0.00% Total 100.00% 100.00%

This table is based on a sample of German and U.K. IPOs matched by market capitalization. The sample is unbalanced, that is, if a firm is delisted prior to its sixth year after going public, the largest shareholder in the last year prior to the delisting is reported.

(full and partial takeovers) may be roughly companies which are fully or partially similar,the question whether the (ultimate) taken over, 14 are taken over by a bidder bidder is widely held or is concentrated with dispersed control. remains important. The potential agency Not only are there differences in the problems in widely held (and hence man- evolution of control between Germany and agement-controlled) firms are different the United Kingdom, but there are also from those in closely held firms (Bratton differences in terms of who owns the votes. and McCahery, 1995; Shleifer and Vishny, Table 8.5 shows the identity of the major 1997). For the former, the target firm will shareholder six years after the float for the be monitored by the bidder’s management, size-matched sample. In German firms, which may have objectives in conflict with families control two-thirds of the voting the maximization of shareholder value. For rights and other companies (mainly foreign the latter, it is the ultimate large share- ones) control about one-third. For U.K. holder who is expected to monitor the tar- firms, other firms (mainly domestic ones) get. Thus the objectives of the bidder may and families control 35% and 28% of the differ depending upon who ultimately voting rights, respectively. controls the bidder. For example, the management of the bidder with dispersed ultimate control may be more interested in 6. PREDICTION OF CONTROL IN acquiring large, profitable, low-risk firms, GERMAN AND U.K. FIRMS whereas the bidder with a strong shareholder may prefer a more-risky, 6.1 Determinants of the evolution of high-growth firm. Empirical evidence control suggests that it is the ultimate shareholder at the top of the ownership pyramid who Table 8.6 investigates the determinants disciplines management rather than the of the evolution of control six years after shareholders at the first ownership tier (see the float.We not only analyze the dilution Renneboog, 2000). Panel C of Table 8.4 of stakes held by the initial shareholders, differentiates between takeovers by an but also the evolution of the free float ultimately dispersed or concentrated and the emergence of new large share- bidder.The bidder in 52% of the German holders. All shareholdings which are partial takeover targets in the size- directly or indirectly controlled by the matched sample is ultimately closely held. same (ultimate) owner are aggregated. The remaining German companies are Large U.K. companies (measured by ultimately widely held. Out of the 24 U.K. market capitalization) tend to evolve 208 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 8.6 Determinants of the Evolution of Control Concentration in Recent German and U.K. IPOs (Tobit Regressions)

Parameter estimates (1) Initial (2) Free (3) New shareholders float shareholder

Constant 0.074 0.058 1.016*** (0.839) (0.780) (0.006) Country 0.230 0.837*** 0.887 (0.636) (0.004) (0.103) Size 0.043 0.162*** 0.158** (0.542) (0.000) (0.024) Size * Country 0.111 0.189*** 0.240** (0.312) (0.004) (0.045) Ageminuscountryavg 0.005 0.003 0.007 (0.175) (0.125) (0.102) * ** Ageminuscountryavg * Country 0.008 0.003 0.009 (0.065) (0.180) (0.045) Founder 0.189 0.263*** 0.279 (0.269) (0.008) (0.115) Founder * Country 0.550* 0.230 0.511* (0.055) (0.180) (0.098) *** * Riskminuscountryavg 1.281 2.322 2.387 (0.378) (0.004) (0.089) *** Riskminuscountryavg * Country 3.334 10.950 6.752 (0.417) (0.000) (0.152) Growth 0.351 0.006 0.204 (0.112) (0.958) (0.315) Growth * Country 0.992** 0.116 1.117** (0.048) (0.696) (0.042) Nonvoting 0.487*** 0.278*** 0.375** (0.000) (0.001) (0.013) Profit rate 0.155 0.203 0.618 (0.820) (0.574) (0.327) Profit rate * Country 0.258 0.426 0.504 (0.795) (0.487) (0.652) Observations 84 84 84

This table shows the determinants of the concentration of voting equity 6 years after flotation for the size-matched sample of German and U.K. IPOs using Tobit regressions.The dependent variables Free float and New shareholder are the shares held by small shareholders and the proportion of shares held by new shareholders (large shareholders who did not hold shares prior to the IPO), respectively. Stakes below 25% are ignored.The dummy variable Country equals one for a German company and zero otherwise. Size is the natural logarithm of the market capitalization at the closing price on the first day of trading and converted into 1985 pounds sterling. Founder equals one if the founder or his heirs held an equity stake in the firm immediately prior to the IPO, and equals zero otherwise. Risk is measured by the standard deviation of monthly stock returns over a five-year period subsequent to the IPO. Growth is the average annual growth rate of total assets over the first five years after the float. Nonvoting is a dummy variable which is set to one if nonvoting shares were issued at the time of going public, and zero otherwise. Profit is the annual cash flow standardized by the book value of total assets. Age is the number of years since the registration of the firm. From the variables Age and Risk, the average by country was subtracted in order to avoid multicollinearity.The p-values are shown in parentheses. ***, **, * represent statistical significance at the 1%, 5%, and 10% level, respectively. Industry and time dummies are included. WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 209 toward a widely held equity structure as is higher than in firms with below average wealth constraints put a limit on holding risk. Moreover, the free float in risky U.K. large blocks of equity (column 2). However, firms is lower.High-risk German companies these wealth constraints do not seem to be also require a strong new shareholder with binding for German firms, as the market potentially strong monitoring skills capitalization does not influence control (column 3). Still, column 2 also shows that concentration (the parameter estimates of a high degree of risk also leads to a higher 0.162 and 0.189 largely cancel out). free float in German firms. Thus control Larger U.K. firms also have fewer new concentration is related more strongly to large shareholders, but this is not the case risk in the United Kingdom than in for larger German firms in which new Germany, and this partially supports shareholders control significantly larger Proposition 4. equity stakes. Thus these findings support Although we do not find any impact of Proposition 1, stating that there is a growth on the control of U.K. firms, strong stronger negative relation between control growth in German firms leads to a reduc- and corporate size for U.K. firms. tion in the initial shareholders’ control and To control for the fact that the average control is transferred to new large share- German firm is substantially older,we have holders.The findings support Proposition 5 subtracted the country average from Age. for the German firms, but not for the U.K. Table 8.6 shows that the initial sharehold- ones. None of the U.K. firms had issued ers of German firms with an above-average nonvoting shares versus 38% of the age tend to retain smaller voting stakes German firms. (Nonvoting shares enable than those of relatively older U.K. firms. the initial shareholders to raise additional This supports Proposition 2. The control equity capital while maintaining control.) relinquished by the initial shareholders in Table 8.6 confirms that the voting rights of the relatively older German firms is not dis- the initial shareholders do not tend to be sipated but is taken over by new large dissipated after the IPO when the company shareholders (column 3). has nonvoting outstanding shares (column 1). Table 8.6 also investigates the influence Thus both a transfer of control to a new of the founder family on control six years large shareholder and a dissipation of con- after the float. The table reports different trol are less likely (columns 2 and 3). This results for U.K. and German IPOs. If the supports Proposition 6. Proposition 7 founder of a German firm is still a share- states that when sufficient cash flows are holder at the IPO, control is likely to remain generated, there is a higher probability that tight. Given that the initial shareholders the initial shareholder retains control. retain large stakes, it is less likely for large Unlike Heiss and Köke (2001), we do not new shareholders to acquire controlling find any such evidence in Table 8.6 and stakes in German firms (column 3). In con- reject Proposition 7.27 trast, the presence of the founder family To conclude: in fast-growing, older among the initial shareholders in the German firms, the initial shareholders sub- United Kingdom has little impact on the stantially reduce their controlling share degree of control exerted by the initial and stakes over the six years after the float, new shareholders, although the likelihood provided they are not part of the founder that the firm evolves toward widely held family. In contrast, the initial shareholders ownership decreases. These findings sup- of German firms tend to retain control in port Proposition 3. firms with nonvoting shares. New share- Demsetz and Lehn (1985) found a nega- holders acquire control in larger, older, and tive relation between risk and ownership strongly growing German firms in which concentration for U.S. firms. Our findings the founder family is no longer involved and support this picture for U.K. firms: in com- which have no nonvoting equity. A higher panies with risk above the average 12.8%, free float tends to occur in highrisk German the control held by new large shareholders firms. In U.K. firms, we find a strong 210 EQUITY OWNERSHIP STRUCTURE AND CONTROL relation between control and size: the to reject Proposition 2. If the founder or larger the firm, the lower is the control by his heirs are involved in the firm at the the new shareholders and hence the larger IPO, the likelihood that they retain control is the free float.When the founder does not over the six-year period subsequent to the sell out at the float, the probability that IPO is large and hence the likelihood that control is dissipated is low. High risk seems the firm evolves toward diffuse control is to necessitate high control by the new large significantly reduced. This supports shareholders.28 Proposition 3. The fact that founder involvement at the flotation does not 6.2 Multinomial logit prediction of influence the probability of being taken the state of control over provides some support for the predictions of the models by Zingales (1995) and Whereas in the previous section we investi- Mello and Parsons (1998) and is also gated the determinants of the control levels consistent with the empirical findings for the of initial and new shareholders, we now United States (Chung and Pruitt, 1996). estimate the likelihood that six years after In companies with high risk, we expect the float the following four different states the initial shareholders to reduce control in of control arise: initial shareholder control favor of new large, stable shareholders with (SC), diffuse control (WH), takeover by good monitoring abilities.Table 8.7 reveals a closely held bidder (TC), or takeover by a that high risk significantly increases the widely held bidder (TW). This analysis is probability of full and partial takeovers performed for two reasons. First, it sheds relative to the possibility that the firm some light on how control concentration remains controlled by the initial sharehold- comes about (via different types of ers (panel A) and relative to the state of takeover, or via control retention). Second, diffuse ownership (panel B). Panel C also the Tobit models estimated in the previous reveals an interesting result: high risk section (Table 8.6) assume that the increases the likelihood of a takeover by a dependent variables (control held by initial concentrated bidder rather than by a widely or new large shareholders or diffuse con- held firm. So it seems that tight control trol) are independent, which is not the with increased monitoring is the best solu- case. Therefore we estimate multinomial tion to high risk. This contradicts logits predicting the control state.29 Proposition 4 and the predictions of Bolton Table 8.7 confirms that there is a higher and von Thadden’s (1998) model.While we probability for large firms to end up with hypothesized that Proposition 4 would hold diffuse control (panel B). This supports more strongly for the U.K. firms, we find no Proposition 1. In addition, large firms are differences in the risk-control relation more likely to be taken over by a widely between the United Kingdom and Germany. held firm than by a concentrated bidder. High growth could lead to control dilu- This is expected for two reasons. First, a tion resulting from the limited availability concentrated bidder (a family or individ- of internally generated funds and the ual) is more likely to be wealth constrained implied need to attract external capital to than a widely held firm or may be less able finance investment opportunities.30 Table 8.7 to raise external funds to finance the confirms Proposition 5: both diffuse control acquisition. Second, the reason for control and a takeover by a widely held a takeover by a widely held firm may be company are the more likely “equilibrium” agency related: the management of a states six years after the float for growth widely held firm may be tempted to “build companies. This proposition is expected to an empire” when it is insufficiently moni- hold more strongly for the U.K. firms than tored (Berkovitch and Narayanan, 1993; for the German ones. Panel C of Table 8.7 Goergen and Renneboog, 2003). shows that this is the case: high growth Age does not seem to matter for either triggers more partial takeovers, but not country’s control levels such that we have necessarily by widely held bidders.The reason WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 211 is that pyramids enable shareholders to of acquisitions by closely held bidders than attract external capital while limiting by widely held bidders. control dilution. Finally, it should be noted that the For German firms, the consistently nega- dummy variable Country is not statistically tive sign of the dummy variable Nonvoting significant. This strongly suggests that the indicates that firms with nonvoting shares country interaction terms are capturing all are less likely to undergo a change in the economic determinants of the control control. Still, in contrast to the results in states. Table 8.6, the parameter coefficients are not statistically significant, such that we fail to support Proposition 6. 7. CONCLUSION Table 8.7 supports the proposition by Burkart, Gromb, and Pannunzi (1997) that This article analyzes why the levels of con- different states of the world require spe- trol are so different in Germany and the cific control structures (Proposition 7). United Kingdom. A first reason for share- Some states of the world (such as the ones holders to hold larger voting stakes in with a low profitability) require a large sta- German firms is found in the differences in ble shareholder,while others cope well with the regulatory and legal environment. dispersed control, allowing for sufficient A detailed analysis of the regulation of managerial discretion. We find that a low the German and U.K. stock exchanges, of profit rate increases the probability of a the rules on minority shareholder protection, U.K. firm being acquired by a closely held of informational transparency, and of the bidder.Thus it seems that incumbent share- takeover codes shows that there is lower holders sell out to a concentrated bidder, if shareholder protection in Germany. The the former are not able to provide sufficient voting practice at annual meetings, the monitoring to improve the firm’s perform- composition of the board of directors, and ance. Indeed, poor performance not only their fiduciary duties further reinforce this results from poor managerial performance, relative weakness of shareholder rights in but also from a breakdown of corporate Germany. As a consequence, control is governance. These findings support more valuable to shareholders of German Proposition 7 for the United Kingdom. firms either to avoid expropriation of their Conversely the higher the profitability, the investments or to take advantage of the higher the odds that the firm is taken over higher levels of private benefits. by a widely held bidder. The fact that Furthermore, holding large control stakes widely held companies prefer profitable is less expensive in Germany relative to the takeover targets can be explained in an United Kingdom because ownership agency-cost setting: managers of widely pyramids, the possibility of issuing nonvoting held firms may find it easier to boost finan- stock, and the possibility of nominating cial performance by taking over profitable one’s representatives to the board of firms rather than by improving the financial directors ensure that control can be main- results of their current businesses.31 tained with relatively low levels of cash Alternatively, widely held bidders may not flow rights. be able to provide the level of monitoring, Although the legal environment predicts which is required to turn around a badly stronger levels of control in Germany, it performing target firm. For Germany we does not explain how the difference in con- find a much weaker relation between trol concentration comes about. Both U.K. control states and profitability. First, as in and German firms are floated on the stock the United Kingdom, poor performance exchange with high levels of initial control, entails a higher likelihood that control is and this raises the question as to what trig- transferred than diluted. Second, unlike our gers subsequent changes in control.To answer findings for the United Kingdom, strong this question we investigate the economic performance leads to a higher occurrence factors that determine control retention 212 EQUITY OWNERSHIP STRUCTURE AND CONTROL ) TC /P * * TW 6.678 0.028 5.498 1.791 73.717 42.283 ) Log (P WH /P TW 0.020 0.036 0.185 2.067 6.102 34.430 53.431 29.851 8.254 ) Log (P WH /P * * TC 2.253 0.056 27.974 38.187 ) Log (P SC /P TW 1.068 48.385 0.016 31.793 60.541 20.285 27.047 ) Log (P SC * /P TC 1.753 0.314 0.052 25.323 31.804 26.097 37.311 31.730 35.509 ) Log (P sc /P ** WH 7.170 41.215 0.499 4.983 0.017 0.070 0.043 0.057 0.030 0.004 1.883 0.014 2.231 0.440 2.356 0.565 (0.417) (1.000) (1.000) (1.000) (1.000) (1.000) (0.879) (0.638) (0.577) (0.616) (0.464) (0.746) (1.000) (1.000) (1.000) (1.000) (1.000) (1.000) (1.000) (0.833) (0.185) (1.000) (1.000) (0.314) (1.000) (0.539) (0.196) (1.000) (1.000) (0.813) (1.000) (1.000) (1.000) (1.000) (1.000) (1.000) (0.036) (1.000) (1.000) (1.000) (1.000) (1.000) (0.328) (0.131) (0.607)(0.658) (0.062) (0.059) (0.758) (0.954) (0.100) (0.051) (0.744) (0.100) (1.000) (0.221) (0.672) (1.000) (1.000) (0.373) 34.405 20.627 30.522 54.378 64.399 10.103 11.056 13.176 Log (P Panel A Panel B Panel C Country Country * * Country Prediction of State of Control for German and U.K. German for of State Control Prediction Logit) IPOs (Multinomial * Country * minuscountryavg minuscountryavg minuscountryavg minuscountryavg Constant 2.126 Country Size Age Founder Table 8.7 Risk Founder Size Risk Age WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 213 Size ** * ** ** 59.030 15.996 is the number of is the number Age equals 1 if the founder or equals 1 if the founder -values are in parentheses. are in -values p is a dummy variable capturing the capturing is a dummy variable Founder 0.648 1.518 6.051 34.570 37.674 23.823 an 25%]), by a over is taken TW (the firm ith four possible states of control: possible ith four SC (control Nonvoting ds sterling. ** equals one for a German company and zero otherwise. company and zero a German equals one for 4.112 2.640 6.753 2.166 30.184 35.207 Country is measured by the standard deviation of monthly stock returns over a over stock returns of monthly is measured by the standard deviation Risk ** * * 1.546 7.080 9.945 25.544 16.530 is the annual cash flow standardized by the book value of total assets. by the book value standardized is the annual cash flow Profit , multicollinearity.The by country to avoid was subtracted the average * ** Risk 0.919 7.671 3.064 and 21.144 Age * 5.031 1.215 9.040 0.222 8.917 0.110 33.487 (0.083)(1.000) (0.764)(0.226) (0.169)(1.000) (0.013) (0.096) (0.294) (0.041) (0.190) (0.053) (1.000) (0.080) (0.215) (0.031) (1.000) (1.000) (0.039) (1.000) (0.059) (1.000) (0.010) (0.006) (1.000) (0.230) (0.250) (1.000) (1.000) (0.589) is the average annual growth rate of total assets over the first five years after the float. after years five the first of total assets over rate annual growth is the average (51,0.000) (51,0.000) (51,0.000) Growth -value) 111 111 111 p Country * Country * represent statistical significance at the 1%,represent statistical significance 5%, and 10% level, respectively. the model. in are included Industry dummies and time * , ** , years since the registration of the firm. the registration since years the variables From issue of nonvoting shares at the time of going public (dummy equals 1). public of going shares at the time issue of nonvoting five-year period after the IPO. after period five-year his heirs held an equity stake in the firm immediately prior to the IPO, prior immediately otherwise. the firm and equals zero in an equity stake heirs held his is the natural logarithm of the market capitalization at the closing price on the first day of trading, on the first price at the closing capitalization 1985 poun into of the market converted logarithm is the natural widely held company), held widely and shareholder).The by a concentrated dummy variable over is taken TC (the firm Growth Observations w model and U.K. logit sample of German a size-matched a multinomial for using of control firms the probability shows table This 84*** 84 84 remains in the hands of the initial [pre-IPO] shareholders), WH (the firm is widely held [no shareholder owns a stake of more th a stake owns [pre-IPO] shareholders), [no shareholder held the hands of initial in remains is widely WH (the firm Profit rate % of correct predictions% of correct 75.00 58.33 67.86 Growth Nonvoting Profit rate (d.f., Chi-squared 214 EQUITY OWNERSHIP STRUCTURE AND CONTROL are the equals one minuscountryavg is measured by Age Country minuscountryavg Risk Age , and minuscountryavg is the annual cash flow standardized is the annual cash flow 0.080 1 Growth minuscountryavg Profit float.The dummy variable Growth minuscountryavg 0.0510.134 1 minuscountryavg Risk 0.107 1 0.067 is the average annual growth rate of total assets over the first five years five the first of total assets over rate annual growth is the average Growth 0.380 0.024 0.093 0.038 0.833 0.009 1 0.043 0.079 0.281 0.118 0.005 0.989 0.081 0.124 0.250 0.056 0.180 1 0.015 0.035 0.000 0.098 1 is the natural logarithm of the market capitalization at the closing price on the first day of trading and converted into 1985 into and converted day of trading on the first price at the closing capitalization of the market logarithm is the natural 0.016 1 Size is the number of years at the IPO since the creation of the company. the creation at the IPO since of years is the number 0.099 0.177 0.070 1 0.061 0.191 0.069 0.099 Age is a dummy variable capturing the issue of nonvoting shares at the time of going public (dummy equals 1). public of going shares at the time the issue of nonvoting capturing is a dummy variable equals one if the founder or his heirs held an equity stake in the firm immediately prior to the IPO and equals zero otherwise. to the IPO and equals zero prior immediately the firm in an equity stake heirs held or his equals one if the founder 0.533 0.117 0.154 0.121 0.081 1 0.541 0.152 0.094 0.009 0.0860.000 0.0900.000 0.875 0.094 0.062 0.026 1 0.191 0.342 0.003 0.176 , is subtracted.The parentheses. the country of which is in average statistical and Age variables significance (0.461) (0.189) (0.183) (0.000) (0.337) (0.403) (0.000) (0.136) (0.500) (0.278) (0.035) (0.480) (0.000) (0.217) (0.121) (0.257) (0.311) (0.000) (0.172) (0.123) (0.006) (0.478) (0.080) (0.286) (0.053) (0.244) (0.000) (0.135) (0.296) (0.133) (0.139) (0.291) (0.367) (0.277)(0.000) (0.008) (0.218) (0.154) (0.129) (0.194) (0.000) (0.080) (0.015) (0.273) (0.483) (0.500) (0.168) (0.267) (0.114) (0.221) (0.170) (0.358) (0.000) (0.083) (0.220) Country Size Founder Risk Growth Nonvoting Profit Age Risk Founder Nonvoting Pearson Correlation Coefficients of Independent Coefficients Correlation Pearson Variables Risk, Growth minuscountryavg minuscountryavg minuscountryavg after the float. after by the book value of total assets. by the book value the standard deviation of monthly stock returns over a five-year period subsequent to the IPO. period a five-year over stock returns of monthly the standard deviation pounds sterling. Founder Risk Growth Nonvoting for a German company and zero otherwise. company and zero a German for Growth Age APPENDIX Table 8.A1 CountrySize 1 the after six years of ownership of the concentration of the determinants coefficients correlation the Pearson shows table This Profit respective Age Risk WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 215 by large initial shareholders, dissipation of of growth on the control concentration of control among many small shareholders, U.K. firms, strong growth in German firms and control transfers whereby we make a leads to the initial shareholders transfer- distinction between widely held and ring control to new large shareholders.The concentrated bidders. The article uses a impact of risk on control is stronger in the unique database of IPOs with data over the United Kingdom than in Germany. In U.K. period 1981–1994. Industry and size companies with above average risk, the effects are controlled for by creating size- degree of control held by new large share- and industry-matched samples of “twin” holders is higher and the free float is lower. German and U.K. firms whose evolution of The multinomial logit models, which pre- control was followed over a six-year period. dict the occurrence of different states of We find that not only do the initial share- control (initial shareholders retain control, holders in the average German company control is diluted, control is transferred to own much larger stakes than their U.K. a concentrated shareholder or to a widely counterparts, they also lose majority held firm), show that specific corporate control only six years after the public characteristics lead to different “equilibrium” offering. In contrast, initial owners in U.K. control states six years after the float. companies lose majority control two years We find substantial differences between after going public. takeovers by a concentrated shareholder We have found strong evidence that cor- and takeovers by widely held companies. porate characteristics lead to differences in For the United Kingdom, the probability of control evolution across companies but a transfer of control to a concentrated also between the United Kingdom and shareholder increases when a company is Germany. The Tobit models that estimate risky, small, and poorly performing. A U.K. the percentage of control held by the initial firm is more likely to be taken over by a and new large shareholders and the size of widely held firm if it is large, fast-growing, the free float six years subsequent to the and profitable. So, for the United Kingdom, float show that size is an important deter- poor performance and high risk necessitate minant of control concentration in the a high level of control and tight monitoring United Kingdom but not in Germany. Large by a concentrated shareholder.High growth U.K. companies evolve toward a more and profitability attract widely held compa- widely held equity structure, whereas in nies whose management may follow an large German firms, new shareholders hold “empire building” acquisition program. We significantly larger voting stakes.The reason found that high growth also leads to more is that wealth constraints become binding diffuse control, which in turn is less likely for U.K. shareholders, whereas German when the founder family is still involved in ones can avoid this effect by leveraging the company. Founding families may be less control via pyramids. Age has an inverse inclined to dilute their stake in order to impact on the control of German and U.K. retain private benefits of control. When firms: initial shareholders of older German German firms are profitable and risky, confirms tend to retain smaller voting stakes trol is more likely to be acquired by a con- than those of relatively older U.K. firms. centrated shareholder, but growth and low If the founder of a German firm is still a profitability increase the likelihood of being shareholder at the IPO and if there are acquired by a widely held firm. nonvoting shares outstanding, control is likely to remain tight in the hands of the initial shareholders. This is not surprising, NOTES as founding families often extract (nonpecuniary) private benefits of control and * We are deeply indebted to Alan Schwartz, the nonvoting shares enable them to raise editor,as well as to two anonymous referees.Their additional equity capital while maintaining comments and suggestions have substantially control.Whereas we do not find any impact improved this article. We are also grateful to 216 EQUITY OWNERSHIP STRUCTURE AND CONTROL Rafel Crespi, Julian Franks, Carles Gispert, Uli is likely to reduce uncertainty with respect to Hege, Alan Hughes, Jens Köke, Colin Mayer,Joe expropriation and increase the value of affected McCahery, Josep Tribo, Eddy Wymeersch, and firms.” the participants at the 2001 meeting of the 6. Section 204 of the Companies Act 1985. European Finance Association in Barcelona, the 7. See the City Code on “concert parties.” It participants at seminars at Oxford University, should be noted that this regulation does not UMIST, Tilburg University, CUNEF (Madrid), apply for “ad hoc” coalitions which are tempo- and the participants at the conference on rary and formed with one particular aim, for “Convergence and Diversity in Corporate example, the removal of poorly performing man- Governance Regimes and Capital Markets” in agement (Stapledon, 1996). Eindhoven for stimulating comments. 8. Nonvoting shares have dormant voting rights 1. Full-parity determination for the sharehold- that are triggered by two consecutive omitted ers and employees only exists in the steel and dividend payments (Aktiengesetz, par. 140). coal sector. In small companies with more than 9. It should be noted that in neither country 500 but less than 2,000 employees, one-third of can the control stakes be diluted by seasoned the supervisory board consists of labor repre- equity offerings. In the United Kingdom, sea- sentatives. In larger firms with more than 2,000 soned new equity must be in the form of rights employees, a system of quasi-parity codetermi- issues (Section 89(1) of the Companies Act nation exists as employee representatives make 1985). These rights may only be waived if a up half of the supervisory board, but the chair- supermajority (of at least 75%) votes to do so. man who is a shareholder representative has a Even where shareholders vote to drop their pre- casting vote in case of stalemate. emption rights, the discount of any new issue 2. See sections 3.12 and 3.13 of Chapter 11 must not exceed 10% of the market price at the of the Listing Rules. time of the issue’s announcement (para. 4.26, 3. See sections 11.4 and 11.5 of the Listing Stock Exchange Rules, 1999). These rules Rules. ensure that the stakes of the initial shareholders 4. Correia da Silva, Goergen, and Renneboog cannot be diluted by the ones of new sharehold- (2003) argue that the taxation of dividends in ers. They are reinforced by guidelines, issued by Germany does not provide an incentive to con- the National Association of Pension Funds and struct shareholding pyramids. the Association of British Insurers, limiting 5. In the United Kingdom, the Company Act companies to raise 5% of their share capital requires the identity of shareholders purchasing each year by any method apart from rights share blocks in excess of 3% (5% prior to issues, and 7.5% in any rolling three-year period. 1989) to be disclosed to the target company. German corporate law is similar: it requires Germany was the last European country to seasoned equity issues expanding the equity by introduce such a disclosure regulation (for 10% or more to take place via rights issues. stakes of 5% or more): the European 10. Voting caps in the by-laws do not apply to Transparency Directive (88/627/EEC) was only the proxies which banks have collected (Baums implemented in Germany in 1995 (the and Schmitz, 2000). Wertpapierhandelsgesetz). Prior to 1995, there 11. The fact that bearer shares are common was only compulsory disclosure of control at the in German firms, whereas nominal shares are level of 25%. In the United Kingdom, share- used in the United Kingdom, does not lead to holders owning 15% or more of the equity of a differences in voting procedure. In both coun- U.K. firm must make public their intentions with tries, shares need to be registered if one wants regard to launching a takeover.There is no such to exert voting rights (Baums, 1997). requirement in Germany. Becht and Boehmer 12. Given the competition in a globalized (2001) argue that the efficiency of the 1995 economy, it is surprising that corporate gover- disclosure regulation is very low, as ultimate nance regulation has not converged more. ownership cannot easily be inferred from Bebchuk and Roe (1999) claim that the rigidity published filings, there is no disclosure on who to changes in control concentration hinges to a exercises proxy votes, and accumulated votes held large extent on the structures with which the by business groups cannot easily be determined. economy started (structure-driven path depend- They conclude:“the low transparency of control ence).The efficient choice of a corporate control is likely to increase the cost of capital to structure is influenced by sunk costs. affected German corporations relative to their Furthermore, there is an endowment effect, as international competitors listed in markets that there are advantages to using the dominant are more transparent. Full disclosure of control form in the economy, which is the one which WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 217 most players are familiar with. Internal rent of the 1980s and computer and software seeking by parties who participate in corporate manufacturers were first assigned to group 19, control may also explain why such parties would then to group 69, and only later to group 35. attempt to impede change toward a more 19. Market capitalization is adjusted for U.K. efficient control structure. For example, the inflation by the annual GDP deflator (base year management of widely held companies may 1985) of the International Monetary Fund prefer to retain a diffuse control structure, as (IMF). Several German firms in our sample this enables them to maintain their private ben- have dual-class shares of which one class is not efits at the expense of shareholders. Likewise, listed. The market capitalization for these firms Bebchuk (1998) argues that concentrated was computed by multiplying the total number ownership—and hence uncontested corporate of shares by the market price of the listed class control—prevails in continental Europe because of shares. the lax corporate-governance regulation allows 20. It may be argued that a different type of large shareholders to reap substantial private matching based on firm age should also have benefits of control. Hence, as long as continental been performed. However, a reasonable match European regulation does not change, control (plus or minus two years of difference) could concentration will resist change (Bratton and only be found for about 19 German firms. McCahery, 1999). However, we control directly for age in the 13. See the discussion on minority protection regressions. Similarly we tried to match firms in Section 2. Evidence of the extraction of pri- simultaneously by size and industry. However, vate benefits of control by large shareholders in again for more than three-quarters of the firms, Germany is given by Franks and Mayer (2001), a satisfactory match within the 25% range and by Dyck and Zingales (2001) in an interna- could not be found. tional comparison. 21. The German size-matched sample includes 14. Brennan and Franks (1997) show that two companies which are not included in the control retention is an important reason for industry-matched sample, because there were no underpricing: share-rationing schemes enable U.K. IPOs in these specific industries during the the original shareholders to disperse shares to period of study. The German industry-matched atomistic subscribers. sample comprises six firms not included in the 15. This argument assumes that the voting size-matched sample. share prices in both companies are equal. 22. As log PWH /PSC log PSC/PWH, we 16. During the period of the study, lower-tier report results for (i) log(PWH/PSC), markets were the Unregulated Unofficial Market log(PTC/PSC), log(PTW/PSC), (ii) log(PTC/PWH), and the OTC for Germany and the Third Market log(PTW/PWH), and (iii) log(PTW/PTC). and OTC for the United Kingdom. 23. Franks and Mayer (1998) report that 17. Control and ownership concentration of hostile takeovers have been a rare phenomenon 61 of these 80 German firms could be traced in Germany. Since World War II and prior to reliably over time. For most of the other IPOs, 2000, there have only been three attempts, two the identity of the shareholders was available, of which failed. Since then there has been a but not the exact size of their holdings. fourth successful hostile takeover (the 18. The industry classification is based on the Mannesmann takeover by Vodaphone in 2000). two-digit U.K. SE groups, the industrial classifi- 24. Whereas panel A considers only direct cation used by the London Stock Exchange to holdings, panel C takes both direct and indirect compile its quarterly lists of new issues. The holdings into account. groups are covered by the F.T. Actuaries 25. The Takeover Panel may grant an excep- Investment Index Classification with the amal- tion to this rule, which makes a tender offer gamation of certain related groups. For each mandatory to a party which reaches the 30% German firm the industry description at the time threshold of equity (or voting rights), in the case of the IPO in the Satling Aktienführer was where a shareholder takes a large stake in a recorded. Subsequently German firms were financially distressed company. reclassified into two-digit U.K. SE groups. We 26. Expanding the analyzed time period of six have merged specific groups which only had a years subsequent to the IPO to eight years small number of IPOs: for example, groups 27 changes neither the data presented in Table 8.3 (Misc. Mechanical Engineering) and 28 nor the results of this article. (Machine and Other Tools) were merged. Groups 27. As a robustness check, similar Tobit 19 (Electricals), 35 (Electronics), and 69 regressions but with alternative independent (Office Equipment) were also merged since variables (for the definition, see Section 4.2) groups 35 and 69 did not exist at the beginning were estimated. The country-dummy variables 218 EQUITY OWNERSHIP STRUCTURE AND CONTROL confirm that the share stakes of initial discussed above are robust to alternative shareholders are higher in Germany, whereas variable specifications. the percentage controlled by new large share- 28. All the results in this section—apart from holders is higher in the United Kingdom (as a the conclusions concerning corporate growth— result of the more frequent occurrence of full are also valid for the industry-matched sample. takeovers). Firm size (book value of total These results are not reported in the article, but assets) is positively related to the control held are available upon request. The only difference by small shareholders and by new large share- in results is that the Growth coefficient is not holders, which confirms the findings of this significantly different from zero at the 10% section. We find that, if the founder is on the level for the industry sample. This is probably management board (Vorstand) of a German due to the fact that firms matched by industry firm, control by the initial shareholders is less have more comparable growth rates than those likely to be reduced over the six years after the matched by size. IPO. These initial shareholders retain a higher 29. It should be noted that the state of control percentage of the voting rights than those in is based on ultimate control. The predictive firms without founder involvement. As power of these models with control based on expected, a higher retention rate goes hand in direct shareholding was lower than the one hand with a reduced presence of new share- based on ultimate control. Statistical signifi- holders. Growth, in contrast, is strongly related cance for the direct ownership models was to control concentration. High growth in total always worse for the case of the size-matched assets leads to a reduction in the stakes of the sample, and worse or similar for the industry- initial shareholders in the United Kingdom and matched sample. even more so in Germany. Furthermore, a 30. All results in this section—apart from higher proportion of nonvoting shares in the conclusions concerning corporate growth— Germany allows the initial shareholders to are also valid for the industry-matched sample. retain their control. In contrast to the findings These results are not reported in the article, presented in Table 8.6, we find different results but are available upon request.The only differ- for our risk and profitability measures. We do ence in results is that the Growth coefficient is not find a relation between the risk measured as not significantly different from zero at the the standard deviation of cash flows over the 10% level for the industry sample. This is five-year period after the IPO and control. This probably due to the fact that firms matched by may be due to the fact that our original risk industry have more comparable growth measure (based on the volatility of stock rates than those matched by size (see also returns) is the better one. Whereas the prof- Goergen, 1999). itability measure (cash flow/book value of total 31. This finding is confirmed by a study on assets) was not related to ownership concentra- hostile takeovers in the United Kingdom by tion in Table 8.6, the alternative profit rate Franks and Mayer (1996). (cash flow over market value of equity plus book value of debt) shows a statistically significant relation with control concentration.These REFERENCES results support the proposition by Burkart, Gromb, and Panunzi (1997) that different states of the world require different control Baums, T. 1997. “Shareholder Representation structures. In some states of the world (such as and Proxy Voting in the European Union: the one with low profitability), a major, stable A Comparative Study,” working paper, shareholder is needed, while in other cases a University of Osnabrück. dispersed share structure provides the neces- ——. 2000.“Corporate Governance in Germany— sary managerial discretion. High profitability in System and Current Developments,” working the United Kingdom ensures that the initial paper,University of Osnabrück. shareholders retain control and consequently ——, and C. Fraune. 1995. “Institutionelle that the voting blocks held by new shareholders Anleger und Publikumsgesellschaft: Eine in the more profitable firms are smaller. For Germany, in contrast, we find the opposite Empirische Untersuchung,” 40 Die result: initial shareholders seem to consider Aktiengesellschaft 97–112. high profitability as an opportunity to transfer ——, and R. Schmitz. 2000. “Shareholder control stakes to new large shareholders. We Voting in Germany,” working paper 76, conclude that the findings of Table 8.6 University of Osnabrück. WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 219 Bebchuk, L. 1998. “A Theory of the Choice Compensation: A Unifying Framework,” 20 Between Concentrated and Dispersed Journal of Banking and Finance 1135–59. Ownership of Corporate Shares and Votes,” Correia da Silva, L., M. Goergen, and working paper, Harvard Law School. L. Renneboog. 2002. Corporate Governance ——, and M. Roe. 1999. “A Theory of Path and Dividend Policy. Oxford: Oxford Dependence in Corporate Governance and University Press. Ownership,” 52 Stanford Law Review Crespi, R. 1998. “Determinantes de la 127–70. Estructura de Propiedad: Una Aproximacion Becht, M., and E. Boehmer. 2001. “Ownership al Caso Espanol con Datos de Panel,” 206 and Control in Germany,” in F. Barca and Moneda y Credito 115–43. M. Becht, eds., The Control of Corporate DeAngelo, H., and L. DeAngelo. 1985. Europe. Oxford: Oxford University Press; “Managerial Ownership of Voting Rights—A 128–53. Study of Public Corporations with Dual Becht, M., and C. Mayer. 2001. “Introduction,” Classes of Common Stock,” 14 Journal of in F.Barca and M. Becht, eds., The Control of Financial Economics 33–69. Corporate Europe. Oxford: Oxford University Demsetz, H., and K. Lehn. 1985.“The Structure Press; 1–45. of Corporate Ownership: Causes and Berkovitch, E., and M. P. Narayanan. 1993. Consequences,” 93 Journal of Political “Motives for Takeovers: An Empirical Economy 1155–77. Investigation,” 28 Journal of Financial and Dennis, D., and A. Sarin. 1999. “Ownership and Quantitative Analysis 347–62. Board Structures in Publicly Traded Boehmer, E. 1999. “Corporate Governance in Corporations,” 52 Journal of Financial Germany: Institutional Background and Economics 176–223. Empirical Results,” working paper,Humboldt Dyck, A., and L. Zingales. 2001. “Private University, Berlin. Benefits of Control: An International Bolton, P., and E.-L. von Thadden. 1998. Comparison,” working paper, University of “Liquidity and Control: A Dynamic Theory of Chicago. Corporate Ownership Structure,” 154 Franks, J., and C. Mayer. 1996. “Hostile Journal of Institutional and Theoretical Takeovers in the UK and the Correction of Economics 177–211. Managerial Failure,” 40 Journal of Financial Bratton, W. W., and J. A. McCahery. 1995. Economics 163–81. “Regulatory Competition, Regulatory ——. 1998. “Bank Control, Takeovers and Capture, and Corporate Self-Regulation,” 73 Corporate Governance in Germany,” 22 North Carolina Law Review 1861–948. Journal of Banking and Finance 1385–403. ——. 1999. “Comparative Corporate ——. 2001. “Ownership and Control of German Governance and the Value of the Firm: The Corporations,” 14 Review of Financial Case Against Global Cross-Reference,” 38 Studies 943–77. Columbia Journal of Transnational Law ——, and L. Renneboog. 2001.“Who Disciplines 213–237. Management of Poorly Performing Brennan, M., and J. Franks. 1997. Companies?,” 10 Journal of Financial “Underpricing, Ownership and Control in Intermediation 209–48. Initial Public Offerings of Equity Securities in Goergen, M. 1998. Corporate Governance and the UK,” 45 Journal of Financial Economics Financial Performance: A Study of German 391–413. and UK Initial Public Offerings. Cheltenham: Burkart, M., D. Gromb, and F. Panunzi. 1997. Edward Elgar. “Large Shareholders, Monitoring and the ——. 1999. “Insider Retention and Long-Run Value of the Firm,” 112 Quarterly Journal of Performance in German and UK IPOs,” Economics 693–728. working paper, UMIST. Chirinko, R., and J. Elston. 1996. “Finance, ——, and L. Renneboog. 2001. “Strong Control and Profitability: The Role of West Managers and Passive Institutional Investors German Banks,” working paper,WZB Berlin. in the UK,” in F. Barca and M. Becht, eds., Chung, K., and S. Pruitt. 1996. “Executive The Control of Corporate Europe. Oxford: Ownership, Corporate Value and Executive Oxford University Press; 258–84. 220 EQUITY OWNERSHIP STRUCTURE AND CONTROL Goergen, M. and L. Renneboog. 2003. the Performance of Large British “Shareholder Wealth Effects of Large Companies,” 101 Economic Journal European Takeover Bids,” in C. Cooper and A. 1418–37. Gregory, eds., Advances in Mergers and Mello, A., and J. Parsons. 1998.“Going Public Acquisitions, vol. 2. Amsterdam: JAI Press. and the Ownership Structure of the Firm,” Goshen, Z. 1998. “Liability Rule or Property 49 Journal of Financial Economics Rule: Voting in Conflict of Interests in 79–109. Corporate Law and Minority’s Protection,” Morck, R. K., A. Shleifer, and R. Vishny. mimeo. Jerusalem: Hebrew University of 1988. “Management Ownership and Jerusalem. Market Valuation. An Empirical Analysis,” Hay, D., and D. Morris. 1984. Unquoted 20 Journal of Financial Economics Companies: Their Contribution to the United 293–315. Kingdom Economy. London: Macmillan. Morck R. K., D. A. Strangeland, and B. Yeung. Heiss, F., and J. Koke. 2001. “Dynamics of 2000. “Inherited Wealth, Corporate Control Ownership and Firm Survival: Evidence and Economic Growth: The Canadian from Corporate Germany,” Discussion Paper, Disease,” in R. Morck, ed., Concentrated 01-63, Center for European Economic Corporate Ownership. Chicago: University of Research (ZEW). Chicago Press. Jenkinson, T., and A. Ljungqvist. 2001. “The Renneboog, L. 2000. “Ownership, Managerial Role of Hostile Stakes in German Corporate Control and the Disciplining of Poorly Governance,” 7 Journal of Corporate Finance Performing Companies Listed on the Brussels 397–446. Stock Exchange,” 24 Journal of Banking and Johnson, S., R. La Porta, F. Lopez-de-Silanes, Finance 1959–95. and A. Shleifer. 2000. “Tunnelling,” working Shleifer, A., and R. W. Vishny. 1986. “Large paper 7523, NBER. Shareholders and Corporate Control,” 95 La Porta, R., F. Lopez-de-Silanes., A. Shleifer, Journal of Political Economy 461–88. and R. Vishny. 1997. “Legal Determinants of ——. 1997. “A Survey of Corporate External Finance,” 52 Journal of Finance Governance,” 52 Journal of Finance 737–83. 1131–50. Stapledon, G. 1996. Institutional Shareholders ——. 1998. “Law and Finance,” 106 Journal of and Corporate Governance. Oxford: Political Economy 1113–55. Clarendon Press. ——. 1999. “Ownership Around the World,” 54 Wenger, E., R. Hecker, and J. Knoesel. 1996. Journal of Finance 471–517. “Abfindungsregeln and Minderheitenschutz ——. 2000. “Investor Protection and Corporate bei Börsennotierten Kapitalgesellschaften,” Governance,” 58 Journal of Financial working paper, University of Würzburg. Economics 3–27. Zingales, L. 1995. “Insider Ownership and the Leech, D., and J. Leahy. 1991. “Ownership Decision to Go Public,” 62 Review of Structure, Control Type Classifications and Economic Studies 425–48. Chapter 9

Mara Faccio1 and M. Ameziane Lasfer* DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES IN WHICH THEY HOLD LARGE STAKES?

Source: Journal of Corporate Finance, 6(1) (2000): 71–110.

ABSTRACT In this paper we analyze the monitoring role of occupational pension funds in the UK.We argue that because of their objectives, structure and overall share holding, occupational pension funds are likely to have more incentives to monitor companies in which they hold large stakes than other financial institutions. By comparing companies in which these funds hold large stakes with a control group of companies listed on the London Stock Exchange, we show that occupational pension funds hold large stakes over a long-time period mainly in small companies. However, the value added by these funds is negligible and their holdings do not lead companies to comply with the Code of Best Practice or outperform their industry counterparts. Overall, our results suggest that occupational pension funds are not effective monitors.

1. INTRODUCTION Agrawal and Knoeber (1996) show that the relationship between large institutional IT IS NOW WIDELY RECOGNIZED THAT COMPANIES shareholding or blockholding and corpo- have to set a number of mechanisms to rate performance as measured by Tobins Q control the agency problems, which arise is weak.2 In contrast, other studies show whenever managers have incentives to pur- that large shareholders play a significant sue their own interests at the expense of role in top management turnover (e.g., those of shareholders. In an extensive sur- Franks and Mayer, 1994; Kaplan and vey of corporate governance, Shleifer and Minton, 1994; Kang and Shivdasani, Vishny (1997) show that legal protection 1995), in take-overs (Shleifer and Vishny, alone is not sufficient to ensure investor 1986; Agrawal and Mandelker, 1990; protection and that other mechanisms, such Sudarsanam, 1996), in the certification of as ownership concentration, could be the initial public offerings (Lin, 1996) and solution to these, so called, agency that block purchases by large shareholders problems. However, the empirical evidence are typically followed by an increase in provided to date on the role and effective- value, in top management turnover, in ness of such alternative mechanisms is financial performance and in asset sales mixed. For example, Demsetz and Lehn (e.g., Mikkelson and Ruback, 1985; Shome (1985) find no cross-sectional relationship and Singh, 1995; Bethel et al., 1998). between the concentration of shareholdings Other studies that specifically analyze share- and the accounting rates of return. Similarly, holder activism also yield mixed results.3 222 EQUITY OWNERSHIP STRUCTURE AND CONTROL For example, while Strickland et al. that could be used to finance growth (1996) show that monitoring by share- opportunities (e.g., Hutton, 1995).4 In holders enhances firm value, Karpoff et al. contrast, other studies show that pension (1996) do not find evidence that share- funds do not get involved in corporate holder proposals increase firm value or monitoring because they find it easier and influence firm policies. cheaper to sell their holdings, they do not The purpose of this paper is to extend want to sit on the board for fear of getting previous research that documents the price sensitive information or because of impact of large holdings on corporate the agency problems within the funds performance by analyzing the monitoring themselves.5 At the same time, policy role of occupational pension funds in the makers in the UK tend to rely on these UK. We identify separately these funds institutions to promote corporate governance from other financial intermediaries (e.g., Cadbury, 1992; Greenbury, 1995).6 because of the large dimension of their Second, previous studies were mainly overall stakes in the UK market, the undertaken under the US framework. particular structure of their portfolios and Franks and Mayer (1997) show that, their investment objectives. As defined by despite the fact that the US and the UK Brickley et al. (1988; 1994), occupational governance systems are both market- pension funds are typical pressure-resistant based, the two countries differ in two major institutions as opposed to other institu- respects: the US has more quoted compa- tions, such as banks, investment and unit nies than the UK and, while the largest trusts and insurance companies, which are category of shareholders in the UK is pressure-sensitive. Unlike previous studies pension funds, most of the equity in the US (e.g., Strickland et al., 1996; Bethel et al., is held by individuals (if each different type 1998) we do not analyze block purchases of institutional shareholder is treated as a because of event date uncertainty and we different category). In addition, unlike US do not concentrate on formal targeting by pension funds where investments and pension funds because we could not find activism programs are developed and any event where occupational pension implemented by fund staff then overseen funds in the UK target companies, i.e., and approved by trustees (Del Guercio and make particular proposals at the Hawkins, 1999), the UK pension funds, company’s annual meetings or negotiate despite the size of their holdings, are not privately some corporate governance issues known for their monitoring and hardly vote (as in Carleton et al., 1998). Instead, we at the annual general meetings (NAPF, compare firms in which occupational 1996b; Financial Times, 1999). Thus, the pension funds hold large stakes against a testing of the empirical hypotheses in a control group of companies listed on the pension fund dominated market such as the London Stock Exchange and test the UK where corporate governance issues are hypotheses that monitoring increases with debated and companies suffer from the ownership concentration and, as a result, same free cash flow problems as their US these funds reduce agency conflicts and counterparts, will strengthen the evidence lead companies to better performance. provided to-date.7 Third, since our analysis Further research in this area is is centered on pension funds that are tax warranted for a number of reasons. First, exempt and hold large stakes, the evidence the issue of involvement of pension funds in we provide should be of relevance to tax and the running of companies is controversial. market regulators and to policy makers The popular belief is that pension funds are involved in the growth of the UK economy. short-termists and impose their views We construct from the financial on companies in which they invest. In statements of all UK quoted non-financial particular, given their tax-exempt status, companies a test sample of companies in UK pension funds are criticized for which pension funds hold more than 3% of making companies pay high cash dividends the issued share capital and a control sample DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 223 by matching our test firms by industry and pension funds. A large proportion of these size.8 We find that pension funds hold large holdings is concentrated in the portfolios of stakes in 289 out of 1640 (18%) compa- large funds. In this section we review the nies. These holdings are mainly in small literature on the role of large shareholders, companies and they did not change signifi- discuss the literature on pension funds cantly over the 1992–1996 sample period. activism, and analyze the structure of the We show that our test companies are not UK occupational pension funds, the impor- more likely to restrain management com- tance of equities in their portfolios, the pensation and/or adopt the Code of Best relatively high concentration of their indus- Practice, i.e., split the roles of chairman try and the management approaches and and CEO, have more non-executive direc- objectives. tors and/or narrow the size of their board than our control firms. These results are not consistent with the recommendations 2.1. Review of the literature of Cadbury (1992) and Greenbury (1995). Corporate governance deals with how In addition, we find that our test companies companies are managed in the long-term are not more profitable and do not pay interest of their shareholders.The literature higher dividends than the control firms has identified two main corporate governance despite the tax credit pension funds could systems that predominate in the developed claim during the sample period. We report economies: the market-based systems of weak and even negative relationship the UK and the US characterized by liquid between occupational pension funds block- markets and unconcentrated company own- holdings and firm value, and, over a longer ership, and the relationship-based systems time period, our test firms do not overper- of Japan and Germany where ownership is form their peers. Our overall results are concentrated and markets are relatively consistent with previous US evidence (e.g., illiquid.9 The issue, although not trivial, has Romano, 1994 and Wahal, 1996) and cast been considered in the literature only doubt on the effectiveness of UK occupa- recently when agency theorists argue that tional pension funds’ monitoring role. At public corporations suffer from excessive the same time, our results provide support costs as managers pursue their own to the proposition of Coffee (1991) and interests rather than the interests of share- imply that pension funds do not follow an holders (e.g., Jensen, 1986). As a result, ‘exit’ policy which is increasingly more there is a need for setting up mechanisms expensive because they must accept to make managers maximize shareholder substantial discounts in order to liquidate wealth. These mechanisms include share- their holdings. Thus, once “locked in”, holding of managers, intermediaries and occupational pension funds avoid costly large blockholders (McConnell and monitoring and refrain from selling their Servaes, 1990; Morck et al., 1988), out- large stake for fear of losing from large side directors (Cotter et al., 1997), debt discounts and/or conveying information to policy (Lasfer, 1995; McConnell and the market. Servaes, 1995; Lang et al., 1996), the The rest of the paper is structured as market for corporate control and incentive follows. Section 2 presents the theoretical contracts (Hart and Holmstrom, 1987; background. Section 3 describes the data. Hart, 1995), large intermediaries Section 4 presents the results. Conclusions (Diamond, 1984; Admati et al., 1994), and are in Section 5. long-term relationships (Ayres and Cramton, 1993). In theory, Diamond (1984) suggests that 2. THEORETICAL FRAMEWORK a large intermediary can represent a better solution to agency conflicts because of In recent years, more than a third of all economies of scale and diversification. listed equities in the UK are held by Admati et al. (1994) argue that when 224 EQUITY OWNERSHIP STRUCTURE AND CONTROL monitoring is costly, the intermediary will of pension funds. They find that, unlike monitor only if this will result in a modifi- proposals sponsored by externally-managed cation in firm’s payoff structure and lead to funds, those made by internally managed net gains. When the intermediary does not funds are not associated with general hold all the firm’s equity and the transac- increases in governance-related events at tion costs are not excessive, the level of target firms. Other studies that looked at commitment will be sub-optimal even when the characteristics of companies in which optimal risk-sharing is attained. Maug institutions hold large stakes find that the (1998) extends this analysis and argues relationship between such holdings and that liquid markets reduce large shareholders firm value or accounting rates of return is incentive to monitor because they can sell weak (e.g., Demsetz and Lehn, 1985; their holdings easily, but such markets Agrawal and Knoeber, 1996). make corporate governance more effective In sum, the primarily US-based studies as it is cheaper and easier to acquire and provide mixed results on the monitoring hold large stakes. Kahn and Winton (1998) role of pension funds. The testing of these distinguish between liquidity, speculation hypotheses under a different institutional and intervention and argue that interven- framework is, thus, warranted. tion is a function of the size of the institution’s stake, firm specific factors and institution’s trading profit. Shleifer 2.2. Institutional settings and Vishny (1997) and Agrawal and In this section we describe the pension funds Knoeber (1996) suggest that large industry and the corporate governance investors, because of the relevance of the system in the UK and set up the hypotheses. resources invested, have all the interest and the power to monitor and promote better governance of companies. 2.2.1. The UK pension fund system Previous empirical studies show that institutions behave differently from individ- The UK pension fund system includes, in uals in sponsoring initiatives (Jarrell and addition to the public pension scheme, Poulsen, 1987; Karpoff et al., 1996) and occupational pension schemes, which are that the institutional behavior is not homo- organized and sponsored by employers, and geneous as it depends on the sensitivity to individual pension schemes offered by managerial pressure (Brickley et al., 1988; financial institutions.The occupational pen- Gordon and Pound, 1993). However, they sion schemes are usually defined-benefits disagree on the effectiveness of share- (DB) where the amount of benefits relates holder activism.10 In particular, Wahal to the final salary of the member while (1996) and Karpoff et al. (1996) find little individual pension schemes are defined- evidence that operating performance of contribution (DC) where pension benefits companies that are the target of pension depend on the contributions paid during the funds proposals improves.These results are working life of the members and the consistent, among other things, with the returns realized on the investment (Blake, arguments of Murphy and Van Nuys 1995).This difference has significant impli- (1994) and Romano (1994) that pension cations on these two schemes’ investment funds are not effective monitors because of policies. With defined contribution plans, the agency problems within the funds them- individuals bear the investment risk and selves. In contrast, Nesbitt (1994) and require a more cautious investment strat- Smith (1996) find that companies targeted egy. As a result, the proportion of shares by large pension funds, such as CalPERS, relative to total assets of occupational pen- increase significantly their performance. sion schemes ranges between 70% and More recently, Del Guercio and Hawkins 80% while defined contribution pension (1999) show that the monitoring effective- plans usually hold no more than 25–30% ness depends on the investment strategies of shares (NAPF, 1996b). In 1997 the DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 225 overall value of individual pension schemes institutional activism are weaker in the UK assets was £190 billion (ABI, 1998) because active shareholders that hold a while occupational pension funds assets “block of shares” are not subject to any (including insured schemes) reached £635 filing requirements, such as the 13D Form billion.11 with the SEC in the US, and they cannot be The proportion of UK pension funds sued for breaching any duty of disclosure of assets invested in equities is the highest their plans or proposals.This rule applies in amongst OECD countries (Davis, 1995). the US to shareholders who act together For example, in 1993, 78% of assets were (e.g., in the case of co-ordinated activism), invested in equities, 12% in fixed income and to investors who, individually or jointly securities and the remaining in cash and hold 5% or more of a firm’s equity (Black, property (Business Monitor, 1997). This 1998). Moreover, unlike in the US, UK trend is likely to reflect the overall invest- pension funds have no legal duty to vote. ments of all pension funds as Blake (1995) The pension fund industry is highly shows that asset structure is not signifi- concentrated. For example, in 1994, the cantly associated with the size of pension largest five in-house managed occupational funds assets. The preference of equities pension funds managed assets worth £65.8 over fixed income securities can be related billion, 14.8% of all occupational pension to the tax-exempt status of pension funds funds assets (NAPF, 1996a) and British who, like charities, are not subject to Telecommunications, accounted for £17.2 capital gains tax and claim back the tax billion. The largest 68 schemes, whose credit, referred to as the advanced corpora- assets value exceeds £1 billion in 1995, tion tax, when they receive dividends. accounted for 57.3% of all occupational Lasfer (1996) shows that this tax credit pension funds assets (Pension Funds and discriminates in favor of dividends relative their Advisers, 1996).The industry of fund to capital gains. managers is also highly concentrated. At In contrast, US pension funds invest a the end of 1996, the top 20 segregated lower proportion of their assets in equities. fund managers managed assets worth For example, in 1990, out of the total £285.7 billion on behalf of occupational assets of $2,491 billion, 38.6% are pension funds and, as a whole, managed invested in equities (Charkham, 1995). assets of some £1029.2 billion (Financial Davis (1995) shows that the proportion of Times, 1997a). assets invested in equities has not changed Over the last three decades, the over the 1970–1990 period. However,these aggregate share ownership in the UK has investments are not evenly distributed but changed substantially. While in 1963 concentrated mainly in large firms individuals were the main shareholders (Charkham, 1995; Stapledon, 1996; with 58.7% of all UK listed equities and Brancato, 1997). pension funds held only 7%, by 1993 In managing these assets, the UK pension funds stakes increased to 34.7% pension funds are subject to trust law and (London Stock Exchange, 1995). In implicitly follow the prudent-man concept. contrast, in the US, individuals held 50% in There is no explicit prudent-man rule and 1990 followed by pension funds with the pension trust law is very flexible. 20.1%, increasing to 25.4% in 1995 However, the duty of prudence to trustees (Prowse, 1994; Brancato, 1997). can be interpreted as requiring the pension We use our sample firms (detailed funds money to be invested for the sole below) to analyze further the ownership benefit of the beneficiaries. The recent leg- structure of UK companies. The results, islation for pension fund management reported in Figure 9.1, show that managers (e.g., the 1995 Pension Fund Act) is simi- and block shareholders (including occupa- lar to the American Employee Retirement tional pension funds) own 53% of equity, Scheme Act with a basic view towards implying a rather concentrated ownership prudence. The legal barriers against structure. However, within these blocks, no 226 EQUITY OWNERSHIP STRUCTURE AND CONTROL

Dispersed Managers + shareholders blockholders 46.98% 53.20%

Managers Blockholders 26.91% 73.09%

Non-financial blockholders Financial blockholders 37.91% 62.09%

Internally managed occupational pendsion funds 14.46% Externally managed occupational pendsion funds 9.34% Managed funds (for clients other than OPFs) 28.02%

All other institutions 48.18%

Figure 9.1 Structure of block share ownership in our sample firms. Dispersed shareholders are defined as those owners who individually own less than 3% of ordinary shares, excluding managers. Blockholders are owners who individually own at least 3% of ordinary shares (excluding managers). Non-institutional blockholdings are block shareholdings attributable to individuals and families, non-financial companies, public bodies, foreign investors and nominees. Institutional blockholdings refer to block shareholdings held by financial institutions. Internally managed occupational pension funds are those pension funds whose assets are, wholly or in part, managed directly by the fund’s trustees. Externally managed pension funds holdings refer to block shareholdings attributable to the largest 20 segregated pension fund managers, times the incidence of pension fund assets to the total asset managed by the segregated fund. Managed funds (for clients other than OPFs) refer to the residual proportion of block shareholdings managed by the top 20 segregated pension fund managers, once accounted for the assets managed on behalf of occupational pension funds (previous variable). Finally, all other institutional holdings are computed as residual item, and refer to block shareholdings held by merchant banks, unit and investment trusts, insurance companies, venture capital firms, and all other financial institutions. All percentages reported are computed assuming the relative upper line to be equal to 100%. single class of shareholders clearly dominates funds account for 14% of all financial the others. Financial blockholders account blockholdings as they hold an average of for 62% of block shareholdings by holding 3% of equity. Externally managed occupa- 24% of our firms’ equity, while the remain- tional pension funds account for 9% by ing 38% are attributable to individuals and holding 2% of equity. The assets managed families, non-financial companies, public by external pension fund managers on authorities, foreign investors and nominees. behalf of clients other than pension funds Internally managed occupational pension represent 28% of these stakes, i.e., 7% of DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 227 equity. Finally, all other financial institu- nominating committee which is responsible tions (including merchant banks, insurance for making recommendations for board companies, unit and investment trusts) membership, they should be the sole or wholly account for 48% of institutional majority members of the remuneration holdings as they hold about 12% of equity. committee which makes recommendations to the board on the pay of executive direc- 2.2.2. The UK corporate governance tors, and of the audit committee whose system function is to advise on the appointment of auditors, to insure the integrity of the com- The importance of corporate governance in pany’s financial statements and to discuss the UK has been emphasized by recent with the auditors any problems arising concerns about the way in which remuner- during the course of the audit.13 ation packages for senior executives have The report has also highlighted the been determined, the spectacular collapse responsibilities of institutional investors of a number of large companies and by the such as pension funds and suggests that such fraudulent use of the pension fund of institutions should be encouraged to make Mirror Group Newspapers to finance an greater use of their voting rights and to seek illegal scheme for supporting the share contacts with companies at a senior executive price of Maxwell Communications level. In particular, they should monitor Corporation. These cases have highlighted boards where there is a concentration of instances where directors do not act in the power in the hands of the chief executive, best interest of shareholders. seek to promote the influence of non-executive In the train of these and other scandals, directors and they are expected to bring the Committee on the Financial Aspects of about changes in underperforming compa- Corporate Governance (referred to as the nies rather than dispose of their shares. Cadbury Committee after its chairman) Although the report recognizes that closer was set up to look at the changes needed in relations with managers can result in these corporate governance in the UK and pub- investors gaining price sensitive information lished a report in December 1992 which makes them insiders, it does not go so (Cadbury, 1992). The report is similar to far as to recommend the formation of share- the Statement on Corporate Governance holders’ committees and the participation of released in September 1997 in the US by shareholders in the appointment of directors the Business Roundtable, an association of and auditors. CEOs of large companies. At the heart of This emphasis on institutions is driven by the report is the Code of Best Practice the size of their holdings but also by the which details the role and composition of fact that these institutions, such as pension the board of directors, the appointment funds, do not target companies and they of non-executive directors, the disclosure of rarely caste their vote at the annual general the remuneration of executive directors and meetings, making them the object of public the renewal of their contracts, and the way criticism (e.g., Mallin, 1997). This passive companies should report and audit their stance has not changed even after the accounts.The main recommendation is that advent of the Cadbury Report. More the offices of the chairman and the chief recently, commenting on the first major cor- executive officer should be separated to porate governance action by pension funds, prevent excessive concentration of power in namely the ousting of the head of Mirror boardrooms and that companies should Group, a media company which has under- appoint independent non-executive direc- performed the market by 34% over the last tors with high caliber so that their views 5 years, the Financial Times (1999) wrote: will carry weight in board discussions.12 The code defines the various roles non- Unlike in the US, investor mutinies in executive directors should play. For example, Britain are still a relatively rare event. they are to be in a majority on the Complacency, reflected in the dismal 228 EQUITY OWNERSHIP STRUCTURE AND CONTROL attendance at annual general meetings, Blake, 1995; NAPF, 1996a). In addition, is the norm. This state of affairs has its Stapledon (1996) reports that many large critics.The Treasury, for example, has let companies managed internally their pension it be known that it regards funds schemes in the early 1990s. For example, in managers as a breed of idle “fat cats” 1993, 14% of occupational pension who are partly to blame for the nations schemes were managed wholly internally economic ills. but these account for some 38% of the total assets invested.14 Thus, following the The main reservation of the code centers arguments of Admati et al. (1994) and on the issue of compliance and enforcement. Diamond (1984), we would expect inter- The corporate governance system in the UK nally managed occupational pension funds has traditionally stressed the importance of to be more active in corporate monitoring internal controls and the importance of because of their size and the magnitude of financial reporting and accountability as their holdings.Their size and expertise min- opposed to a large amount of external imize the monitoring costs. They are likely legislation. In this spirit, the Code of Best to understand when activism is necessary Practice is voluntary and lacks effective and they are large enough to make moni- sanctions. Nevertheless, as a continuing toring effective. Their large holdings are obligation of listing, the London Stock expected to alleviate the free-rider problem Exchange requires all companies regis- that makes atomistic shareholders’ action tered in the UK to state, after June 1993, non-rational and inefficient. whether they are complying with the code Second, internally managed occupa- and to give reasons for any areas of non- tional pension funds are expected to compliance. monitor companies in which they hold large stakes because they control directly or 2.2.3. Pension fund management style indirectly the investment and the voting and corporate monitoring decisions. Their objective is likely to maximize the value of funds in order to minimize Del Guercio and Hawkins (1999) show that the company’s contributions and, possibly, the level of monitoring by pension funds use any pension fund surplus to inflate depends significantly on the way their company’s profits (Short and Keasey, assets are managed. In the UK pension 1997). In contrast, funds that delegate funds assets can be managed in three their investment functions to external man- different ways: self-managed, externally agers effectively disconnect their activism managed and insured. Within self-managed efforts from their investment actions (Del schemes, the trustees of the scheme define Guercio and Hawkins, 1999). As a result, asset allocation, portfolio selection policies given that they will not be able to trade and directly invest pension fund assets. profitably on any private information that Within externally managed schemes, the results from their activism, they are not investment power is wholly or in part likely to monitor or to publicize their delegated to one or few external managers. activism efforts. The level of monitoring In the case of insured schemes, the funds role of externally managed pension funds are invested in insurance policies or man- will depend on the content of the contract aged through fund contracts taken out with with the trustees and the level of competi- an insurance company. tion among fund managers. Given that we We expect internally managed occupa- do not have data on these contracts and on tional pension funds to have a stronger the investments made on behalf of pension incentive to monitor than externally funds, we cannot expect all externally managed pension funds for a number of managed funds to monitor. reasons. First, internally managed pension Internally managed occupational funds are larger than externally-managed pension funds may not monitor individual and insured schemes (e.g., Minns, 1980; companies if they find it easier to sell, if DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 229 they do not want to gain access to price invested in these companies.We expect large sensitive information or if they themselves pension funds and those with significant are subject to agency problems. In addition, commitments to have a higher incentive to given that these funds are defined benefits monitor companies in which they hold large schemes, they are likely to be indexed and stakes. In the long run, companies in which passive to minimize their management risk, occupational pension fund carry on holding transaction costs and to fit the needs of large stakes are expected to outperform long-term pension investors (Tomlinson, their industry counterparts. 1998). Del Guercio and Hawkins (1999) argue that such passive management style will lead pension funds to monitor by 3. SAMPLE CONSTRUCTION AND promoting spill-over effects that boost the DEFINITION OF PROXY VARIABLES performance of the stock market overall rather than specific stocks. In our analysis We search all the 1640 non-financial we focus on internally-managed pension companies quoted in the London Stock funds to test whether size and passive Exchange in 1995–1996 for those where management styles define the level of mon- occupational pension funds hold large itoring activity but we control also for the stakes. We exclude financial companies monitoring roles of external pension funds because of the specificity of some of their and other blockholders. ratios such as leverage, which cannot be The monitoring activity can take the related to the level of their risk and/or form of selecting the board structure resolution of their agency conflicts.To avoid and/or improving performance. The moni- survivorship bias, our sample includes all toring of the board relates to the size of the companies for which the relevant data is board and its composition so that it available even if they are currently becomes more accountable to the share- extinct.17 holders (Cadbury, 1992).15 Jensen (1993) We started by gathering information on argues that as the size of the board shareholding by category from the annual increases, its ability to control manage- reports but we find that only a handful of ment decreases and the communication companies disclose this information in their and co-ordination problems increase. Analysis of Ordinary Shareholdings section Consistent with this proposition, Yermack of their accounts. This data would have (1996) and Eisenberg et al. (1998) find a been ideal but it would have given us only negative correlation between board size the aggregate holding of pension funds and firm performance.Thus, if occupational without distinguishing between funds that pension funds are effective monitors, they are internally managed from those that are are expected to mitigate this agency con- externally managed. Instead, we rely on any flict by restricting the size of the board. At disclosed holding above 3% threshold in the same time, they will lead companies to the accounts and reported in Extel adopt the Code of Best Practice defined by Financial18 and define these as occupational Cadbury (1992), i.e., to split the roles of pension funds holdings or blockholders. chairman and chief executive officer, to All the holdings of directors are disclosed have a high proportion of non-executive even if they are zero (Company Act 1985). directors and to restrain executive pay.16 Although this cut off point of 3% is The monitoring of firm performance constraining, it is the only data available. implies that companies in which occupa- We, nevertheless, posit that the holding of tional pension funds hold large stakes have 3% or above is significant to warrant a higher value than widely held companies monitoring and to allow us to test directly and/or companies held by other blockholders. the Diamond (1984) and Admati et al. This level of monitoring is a function of the (1994) propositions. We searched for the size of the pension funds and the relative keywords pension, pension funds, retire- importance of the amount of assets ment, superannuation and superannuation 230 EQUITY OWNERSHIP STRUCTURE AND CONTROL schemes and for known pension funds, such As in Agrawal and Knoeber (1996), we as Hermes, to define the holdings of compute Tobin’s Q as the ratio of market occupational pension funds. We identified value of equity plus book value of debt over 289 individual companies (18% of our total assets to measure firm value. The total sample) with at least one occupational results are simulated using other measures pension fund holding above 3%, and 356 of firm value such as market-to-book, M/B, large stakes held by 99 individual occupa- and market-to-sales, M/S, ratios (Lins and tional pension funds. We split the other Servaes, 1999). These variables are, how- major disclosed holdings (other than man- ever, ambiguous measures of value-added agerial holdings) into externally managed by pension funds investments, since they pension funds, institutional and non- can also capture the value of future invest- institutional blockholding depending on the ment opportunities. As in Yermack (1996), identity of the shareholders. We compare we control for growth opportunities by the performance of our test firms against using P/E ratio. In addition, we use various all other remaining companies and against accounting rates of returns and a one year a control group of companies with similar share price return, Ri, t 12 to t’ to measure size and industry characteristics. We performance and market value of equity, collect all the relevant accounting and ME, or total assets,TA, to control for size. financial data from Extel Financial and We follow previous literature (e.g., Lang from each company’s accounts. et al., 1996) and use both the market value Table 9.1 reports the proxy variables of leverage, Mlev, defined as long-term debt used to test our hypotheses. We use five over market value of equity plus long-term definitions of pension fund holdings: (i) the debt, and the book value of leverage, Blev, first largest stake held by pension funds, defined as long-term debt over shareholders LPF; (ii) the sum of all pension fund funds plus long-term debt, to test for the holdings, TPF, to analyze pension funds’ monitoring role of debtholders. We use individual and collective monitoring roles; Split (a dummy variable equal to 1 if the (iii) the pension fund incidence, IPF, the role of chairman and CEO are differenti- ratio of pension fund investment in our test ated), Nechair (a dummy equal to 1 if the sample over their total assets to assess the position of chairman is covered by a non- magnitude of such investment in their executive director), number of directors, portfolio;19 (iv) the number of occupational #DIR, the proportion of non-executive pension funds in our test firms, NPF, and directors in the board, %NED, and the rel- (v) the logged value of the largest pension ative remuneration of directors, Dir fund’s asset, LNPFA. Directors’ ownership, rem./Sal, to assess the pension fund role in Dir, is used to control for the managerial monitoring board composition and com- entrenchment hypothesis.20 Any other pensation policy. Finally, payout ratio is large stake, Block, is used to control for the used to test the hypothesis that pension monitoring role of blockholders. This funds prefer to invest in companies that variable is split into institutional block pay high dividends. shareholdings, IBlock, which includes all financial institutions’ stakes (excluding internally- and externally-managed occupa- 4. EMPIRICAL RESULTS tional pension funds’ stakes), externally- managed pension funds, EPFM, i.e., the 4.1. Characteristics of occupation stakes held by the largest 20 external pension funds holdings pension fund managers (either on behalf of occupational pension funds and other Table 9.2, Panel A, reports the distribution clients), and NIBlock, the holdings of of companies with pension funds holdings. Out individuals and families, non-financial of 289 test companies, 175 (61%) reported companies, public bodies, foreign investors an overall holding of between 3% and 6%, and nominees.21 and 70 (24%) had between 6 and 10%. DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 231 sign ends ivid Firm size Firm Proxying for (Hypotheses) E Firm size Firm ends performance Firm ivid ) are for the Logit regressions the Logit ) are for sign E shareholders’ funds of debtholders role Monitoring market value of equity value market of debtholders role Monitoring plus plus book value of debt over sales of debt over book value value Firm book value of debt over total assets of debt over book value value Firm plus plus ends over earningsends over of d Monitoring 1 if the role of Chairman is split from that of CEO from is split of Chairman 1 if the role board composition Monitoring 1 if the role of Chairman is attributed to a non-executive director to a non-executive is attributed of Chairman 1 if the role board composition Monitoring other than internally and externally managed pension funds managed pension and externally than internally other companies Year-end share price over earnings per share per earnings over share price Year-end opportunities Growth Market value of equity value Market Market value of equity over shareholders’ of equity over value Market funds value Firm Market value of equity value Market 1-year share price return, share price 1-year stock adjusted issues and d for Description of proxy variables. of proxy Description ( Expected signs t 12 to i, t R ROS sales and tax over interest before Profit performance Firm ROE shareholders’ over Earnings funds performance Firm ROA total assets and tax over interest before Profit performance Firm P/E M/S M/B Q NIBlock shareholders by non-financial equity owned of outstanding Proportion Blockholders’ to monitor incentive IBlock institutions, by financial equity owned of outstanding Proportion financial by other to monitor Incentive %NED directors of non-executive Proportion board composition Monitoring EPFM fund managers by the top 20 external equity owned of outstanding Proportion managers’ External to monitor incentive #DIR and non-executives) of directors (both executives Number board size Monitoring Split Dummy Block large shareholders by other equity owned of outstanding Proportion Blockholders’ to monitor incentive Nechair Dummy Dir by directors equity owned of outstanding Proportion Management entrenchment Dir rem./Sal Directors’ monetary pay (excl. sales over options) directors’ Monitoring pay NPFLNPFA fund with the largest stake of the pension our test Log of the asset value companies in large stakes funds holding pension of occupational Number to monitoring fund commitment Pension to monitoring fund commitment Pension Blev debt long-term debt over Long-term IPF funds’ pension Occupational to their assets relative test companies in holdings to monitoring fund commitment Pension Mlev debt long-term debt over Long-term META of equity at balance sheet date value Market assets Total TPF funds pension by all occupational equity owned of outstanding proportion The to monitoring fund commitment Pension Table 9.1 VariablesLPF Definition funds pension by largest occupational equity owned of outstanding Proportion to monitoring fund commitment Pension Payout Ordinary divid 232 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 9.2 Distribution of pension funds holdings in test companies The sample includes 289 test companies where occupational pension funds hold 3% or more of ordinary shares and 356 individual pension fund block shareholdings. LPF is the proportion of outstanding equity owned by largest occupational pension funds; TPF is the proportion of outstanding equity owned by all identified pension funds; IPF is the ratio of occupational pension funds holdings over their total assets; NPF is the number of occupational pension funds; EPFM is the proportion of shares owned by the largest 20 external pension fund managers, either on behalf of occupational pension funds or other clients; NC is number of companies; and NStakes is for number of stakes; NPension funds is the total number of pension funds.

(Panel A) Distribution of companies by total pension funds holdings (NC 5 289) % holdings (TPF) 3–6% 6–10% 10–20% 20%

Number of companies 175 70 36 8 % of total 60.6 24.2 12.5 2.8

(Panel B) Distribution of companies by number of pension funds’ stakes (NC 289;NStakes 356) Number of pension funds’ stakes (NPF) 1 2 3 4 5

Number of companies 238 40 8 1 2 % of total 82.3 13.8 2.8 0.3 1.0

(Panel C) Distribution of pension funds by number of stakes held (NStakes 356; NPension funds 99) Number of stakes held 1 2–4 5–9 10–19 20–39 40

Number of pension funds 68 18 4 4 3 2 % of total 68.7 18.2 4.0 4.0 3.0 2.0 (Panel D) Descriptive statistics of pension funds and block holdings % Mean Median Minimum Maximum

LPF 6.01 4.70 3.00 56.90 TPF 6.96 5.06 3.00 56.90 NPF 1.23 1.00 1.00 5.00 IPF 0.15 0.03 0.01 14.75 EPFM 8.99 4.79 0.00 61.88

Thus, more than 84% of our test companies interest; 40 companies (14%) have two have pension funds holdings between 3 and relevant pension funds, 8 (3%) have three 10%. In 36 companies (12%) pension pension funds, 1 company had four pension funds hold between 10% and 20% and in funds and 2 reported five relevant pension 8 companies (3%) the holdings exceed funds holdings, a total of 356 stakes. 20%. Although the pension fund deeds Table 9.2, Panel C, reports the distribu- often set up upper limits of the proportion tion of the 99 pension funds that hold these of shares that a fund is allowed to hold in 356 stakes. The vast majority of these an individual company, the law in force funds hold a small number of stakes: 68 does not establish any limitations. The only pension funds hold just one large stake, and limit is that a pension fund cannot invest 18 pension funds hold less than five large more than 10% of its assets value in the stakes.These 86 pension funds hold a total sponsoring company’s shares. of 115 large stakes. Although not reported Table 9.2, Panel B, shows that 238 com- in the table, we find that the magnitude of panies (82%) have only one pension fund the holdings is positively correlated with DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 233 the size of individual occupational pension total, rather than the individual, incidence funds. The nine occupational pension funds is likely to be high. In particular,the largest that hold at least 10 large stakes, hold a fund, British Telecommunications/Post total of 211 stakes (59% of our sample of Office pension fund, holds 53 large stakes, stakes) and, with the exception of Mars implying an average incidence of 7.95% Pension, the market value of their managed (53 0.15%). Thus, while, the aggregate assets exceeds £4 billion. British blocks may represent, individually, a small Telecommunications-Post Office Pensions, proportion of pension fund assets, for each the largest occupational pension fund in pension fund the total stake may be sub- the UK, holds 53 stakes. stantial, suggesting that their monitoring Table 9.2, Panel D, reports the descrip- role should be beneficial. tive statistics of occupational pension Overall, the results reported in Table 9.2 funds holdings and their incidence. On are striking and suggest that, despite their average, occupational pension funds hold aggregate holdings, individually, internally- individually 6% and collectively 6.96% of managed occupational pension funds hold our test firms equities. Although these large stakes in a small number of compa- holdings range between 3 and 57%, the nies, most of these stakes are between 3 vast majority of our companies reported and 10%, the vast majority of our test holdings of between 3 and 6% (Panel A). firms report only one occupational pension These holdings are relatively large comparing fund holding and these holdings represent a to those used in previous studies. For relatively small proportion of pension example, the average stake of institutional funds’ assets. investors in target firms is 0.3% in Wahal Table 9.2, Panel D, shows that external (1996) and between 0.4% and 2.3% in pension fund managers, EPFM, as a whole, Del Guercio and Hawkins (1999). hold, on average, 8.99% (median 4.79%) To assess the impact of these stakes on of our sample firms’ equity, but these hold- each individual pension fund assets, we ings are not statistically different from compute Incidence on Pension Funds those of internally managed occupational (IPF), the ratio of these investments over pension funds (t-statistic of the difference total assets of each pension fund. As stated in means 1.07). However, Financial above, the data on the portfolio of the Times (1997a) reports that only about assets managed is not available for all pen- 25% of these funds’ assets are managed on sion funds. The denominator of this ratio behalf of occupational pension funds. This which includes other investments in fixed implies that the large stakes held by these income securities, property, overseas equity external managers on behalf of occupa- and cash, is likely to be larger than pension tional pension funds represent on average fund equity investment.Thus, this ratio will roughly 2.25% (8.99% * 0.25) of firms’ understate the actual incidence level.22 equity, assuming that these funds invest in Keeping this drawback in mind, Table 9.2, the same way pension funds assets and Panel D shows that, on average, the invest- other assets they manage. This suggests ment of pension funds in our test compa- that the impact of these investors is likely nies represents 0.15% of their total assets, to be marginal. Unfortunately, we do not ranging between 0.01 to 14.75%. These have data on the way these stakes are allo- proportions are larger than those reported cated among the different clients, but we by Del Guercio and Hawkins (1999) who have included them in our analysis to show that pension funds holdings represent assess the extent to which externally man- between 0.17 and 0.34% of their invested aged pension funds monitor companies. portfolio. However, given that, in our sam- In Table 9.3, Panel A and B, we report ple, 211 blocks out of 356 are held by only changes in the holdings of occupational nine funds, the combination of the pension pension funds in our test firms over the fund incidence statistics with this high con- 1992–1996 period and the t-statistics of centration of holdings implies that the differences in means and medians. Data on 234 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 9.3 Annual changes in occupational pension fund holdings and list of occupational pension funds investing in their own companies

(Panel A) Distribution of occupational pension funds annual holdings Mean Median Minimum Maximum

1992 6.21 5.00 3.00 41.6 1995 5.93 4.70 3.00 41.6 1996 6.09 4.60 3.00 56.9

(Panel B) t-statistics of annual differences in means and medians t-statistics of Mann– differences in means Whitney (p-value in parentheses) p-value 1992 vs. 1995 0.80 (0.43) 0.193 1992 vs. 1996 0.32 (0.75) 0.149 1995 vs. 1996 0.44 (0.66) 0.844

(Panel C) The holdings of Pension fund in their parent company Proportion of Company equity held (%)

Ensor Holdings 4.70 Eve group 4.48 Fuller, Smith and Turner 3.30 Garton Engineering 3.09 Gibbs and Dandy 10.86 Pension Trustees LPA Industries 4.82 Lucas Industries 6.11 MS International 10.90 Oliver Group 4.26 Rexmore 3.60 Walker,Thomas 5.11

pension fund stock ownership in 1992 is confirm that pension fund holding did not taken from the Hambros Company Guide change over the 1992–1996 period. The and from the London Stock Exchange results are consistent with WM (1996) Yearbook,23 since Extel Financial provides findings and suggest that pension funds are data only for the latest year. The table long-term investors and that our analysis is shows that 241 companies (83% of our not sample dependent. test companies) reported large pension We check whether pension funds invest in funds holdings in 1992 and 1996. On their own companies where they may have average, occupational pension funds held different objectives and monitoring roles individually 6.21% of shares in 1992, than when they invest in other firms. 5.93% in 1995 and 6.09% in 1996. The Table 9.3, Panel C, lists the 11 companies difference in means and in median across (3.8% of our test firms) with such stakes. these years is not statistically significant. The number of these companies and the We repeat the exercise using total pension magnitude of their holdings are not substantial fund holdings and number of pension funds and the inclusion of these companies in our and the results (not reported in Table 9.3) test sample did not alter our results. DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 235 4.2. Financial characteristics of our directors remuneration than other companies. test companies Interestingly, the difference in payout ratios between our test and control firms is not Table 9.4 reports the descriptive statistics statistically different. This comparison is, of financial attributes of the test and however, likely to be driven by size differ- control firms. We compute the t-statistics ences across the two samples.To overcome to test for differences in means and the this bias we construct a control sample, by Mann–Whitney p-value to test for differ- matching every test company with a similar 24 ences in medians. In Panel A, we test for company from the same industry and size, a number of attribute differences between using year-end market value of equity. our test and all the remaining non-financial Panel B, shows that compared to the companies listed on the London Stock industry and size-adjusted control group, Exchange. Companies in which pension the block and managerial holdings in our funds hold large stakes are, on average, test firms are not statistically different. very small. Their average (median) market Blockholders hold an average of 34% in value is £97 million (£28 million) com- our test firms and 36.5% in our control pared to £534 million (£43 million) for the sample. Insiders hold an average of 14.5% remaining sample.The test statistics of the in our test firms and 14% in our control differences in means and in medians are sample.The test statistics for differences in both significant at the 1% level, suggesting means and medians are not statistically that, on average, our test firms are signifi- significant at any confidence level. The cantly smaller than all the remaining firms results imply that our analysis is not in the UK market. The results based on affected by block or managerial ownership. total assets as an alternative measure of However, compared to the results reported size confirm that our test companies are in Table 9.2, insiders and blockholders hold substantially smaller than the control significantly larger stakes in our test firms firms. However, our test companies are not than occupational pension funds, suggest- all small. Out of the 289 test companies, ing that these funds do not invest a large 188 (65%) are part of the FTA All Share proportion of their assets in our test Index, the 800 most traded companies in companies. the London Stock Exchange, with 8 com- Table 9.4, Panel B, shows that the value panies (1.4% of our test sample) that are of our test firms is significantly lower than in the UK top 100 companies (FTSE 100 that of our control group. The average Index). The remaining 101 companies are Tobin’s Q of our test firms is 1.16 while from the mid-capitalization and the small that of the control group is 1.62.The same companies indices. The findings that pen- conclusion is reached when we use market- sion funds did not change their holdings to-book and market-to-sales as alternative over the 1992–1996 period reported above measures of firm value. At the same time, are not likely to be the result of illiquidity our test companies have lower leverage of our test companies. Bethel et al. (1998) than our control firms. However,in terms of report also a large number of small com- accounting and market rates of return and panies in their sample of firms in which P/E ratio, our test companies are not large investor acquire large stakes. different from the control firms. However, given that pension funds hold, at In terms of internal governance struc- the aggregate, more than a third of the UK ture, our test and control companies have equity market and that our test companies exactly the same number of directors of are relatively small, we can tentatively con- about 6, ranging between 2 and 15, and the clude that the vast majority of companies same proportion of non-executive directors in which pension funds invest (but without of about 40%. The proportion of our holding significant stakes) are large. control companies that split the roles of Panel A, Table 9.4, shows that our test chairman and CEO of 88% is higher than companies have lower leverage but higher that of our test firms (84%). Finally,Table 9.4 236 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 9.4 Descriptive statistics on means of selected data on the test and control firms This table provides the descriptive statistics of the proxy variables. The test sample includes all companies that reported pension fund holding above 3% in 1995–1996. In Panel A, the control sample includes all other quoted companies in the London Stock Exchange. In Panel B, the control sample includes industry and size matched firms. Block is the proportion of outstanding equity owned by blockholders other than directors and occupational pension funds; Dir is the proportion of outstanding equity capital owned by directors; ME is market value of equity at balance sheet date; TA is total assets; Q is the ratio of market value of equity plus book value of debt over total assets; M/S is the market value of equity plus book value of debt over sales; M/B is the market value of equity over book value of equity; Ri, t12 to t is a 1-year stock return; P/E is the price-earnings ratio at the balance sheet date; Blev is the ratio of long-term debt over long-term debt plus shareholders funds; Mlev is the ratio of long-term debt over long-term debt plus market value of equity; Dir rem./Sal is the ratio of directors’ remuneration over sales; Split is a dummy variable equals to 1 if the roles of chairman and CEO are split; #DIR. is the number of directors in the board; %NED is the proportion of non-executive directors; Payout is the ratio of dividends over earnings.

Test sample Control sample t-statistics Mann– of difference Whitney Variables Mean Median Mean Median in means p-value

Panel A: Size and other variables relative to all companies in the UK (NTe s t 586; NControl 2702) ME (£ million) 96.7*** 28.1*** 534.10*** 42.87*** 3.38 0.000 TA (£ million) 125*** 33.4*** 571.61*** 49.54*** 2.41 0.000 Q 1.16 0.87 1.48 0.87 1.32 0.520 ROA (%) 6.13 8.9 6.73 8.93 1.15 0.304 ROE (%) 1.10 11.30 24.97 11.72 0.75 0.960 Blev (%) 14.58*** 9.90*** 19.10*** 14.60*** 3.88 0.001 Dir rem./Sal (%) 2.40 1.00*** 2.56 0.79*** 0.25 0.000 Payout (%) 31.78 33.0 36.03 33.55 1.41 0.120 Panel B: Size and other variables relative to industry and size matched control firms (N 586) Block 34.00 32.48 36.54 34.82 1.29 0.135 Dir 14.53 6.02 14.01 7.38 0.32 0.875 ME (£ million) 96.7 28.1 121.1 31.1 1.26 0.875 TA (£ million) 125 33.4 122 32.0 0.12 0.519 Q 1.16*** 0.87*** 1.61*** 0.94*** 2.21 0.005 M/S 1.33* 0.60* 2.65* 0.65* 1.68 0.051 M/B 2.78* 1.68*** 3.44* 1.87*** 1.68 0.004 ROA (%) 6.13 8.90 4.10 8.70 1.03 0.550 ROE (%) 1.10 11.30 11.95 12.00 1.16 0.113 ROS (%) 5.49 6.80 0.00 6.50 1.11 0.758 Ri, t12 to t % 23.57 15.40 20.04 5.75 0.74 0.133 P/E 9.68 8.25 9.33 8.36 0.48 0.101 Mlev (%) 11.34*** 5.80** 17.90*** 12.75** 3.21 0.012 Blev (%) 14.58*** 9.90 21.01*** 13.30 3.08 0.458 Dir rem./Sal (%) 2.4 1.00 2.30 0.90 0.10 0.469 Split 0.84 1.00 0.88 1.00 1.61 0.121 #DIR 5.92 6.00 5.80 6.00 0.85 0.382 % NED 0.40 0.40 0.38 0.40 1.31 0.334 Payout (%) 31.78 33.00 33.10 30.70 0.40 0.399

Notes:*** Significant at 0.01 levels.** Significant at 0.05 levels.* Significant at 0.10 levels. DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 237 shows that, our test firms do not exhibit Column (6), Table 9.5, controls for lower directors’ remuneration than the differences in other monitoring mecha- control firms.These results are striking and nisms to account for the fact that these suggest that pension funds large holdings various mechanisms may be substitutes. do not increase the likelihood of compliance The results show that the main difference with the Cadbury (1992) and Greenbury between our test and control firms is the (1995) recommendations. measure of firm value, Q, suggesting that Table 9.4, Panel B, shows that our test our test firms have lower value than the firms pay an average of 32% of their control firms.25 The coefficient of block- earnings as dividends. This is not statisti- holding is also negative and significant at cally different from the average payout the 10% confidence level, suggesting that, ratio of 33% of our control firms. The after controlling for all other differences results are not consistent with the short- between our test and control firms, the test termism arguments and indicate that companies have lower blockholding than occupational pension funds do not our control firms. No single pension fund- necessarily demand high payouts from monitoring variable is significant at any companies in which they hold large stakes. confidence level. These results cast doubt Following Jensen (1986) arguments, the on the monitoring role of pension funds in results also imply that pension funds are the UK.26 not monitoring these companies as they do In Table 9.6 we analyze the causality in not make them disgorge the free cash the relationship between board structure flow. The results are, nevertheless, and occupational pension funds holdings. consistent with previous studies who do We run a set of regressions where the not find dividend tax clientele in the UK various dependent variables which proxy (e.g., Lasfer, 1996). for board structure are measured in the 1996–1997 financial year and the explana- 4.3. Do pension funds affect firm tory variables such as pension funds holdings value and board structure? are in 1995–1996 financial year. We hypothesize that pension funds require Table 9.5 reports the results of the Logit companies in which they hold large stakes regressions where the various agency in 1995–1996 to adopt the Cadbury’s variables are considered simultaneously. (1992) board structure in 1996–1997, The dependent variable is equal to 1 if i.e., to split the roles of chairman and chief company i is in the test sample and 0 if it executive officer,to appoint a non-executive is part of the control sample. The table director as a chairman and to have a large shows that, with the exception of the vari- proportion of non-executive directors on ables that proxy for firm value that are the board. In addition, we follow significantly lower for our test firms, there McConnell and Servaes (1995) and focus is no statistical difference between our test separately on low and high growth firms and control firms. As shown in Table 9.4, using E/P ratio as a proxy for growth companies in which occupational pension opportunities. Firms with E/P ratio above funds hold large stakes are not more (equal or below) the median are classified profitable and do not pay higher dividends as low (high) growth. We expect a positive than the control firms. In addition, our test relationship between pension funds hold- firms are not more likely to split the roles ings and the adoption of the Cadbury of chairman and chief executive officer, (1992) recommendations on the composi- have less directors or more non-executive tion of the board of low growth companies, directors than our control group.These two which are more likely to suffer from the last issues were the main focus of the free cash flow problem. recommendations of Cadbury (1992), The first three columns of Table 9.6 which relied on pension funds for their report the results of the Logit regression implementation. where the dependent variable is equal to 1 238 EQUITY OWNERSHIP STRUCTURE AND CONTROL 2.20) 1.85) ted log 0.63) 0.09) 0.95) ( 0.00) 0.19) (6.93) ( * *** *** 0.002 0.00 ( 0.019 ( 0.008 (0.70) 0.02 ( 0.005 ( 0.015 0.025 ( 0.005 (0.01) 1.42 ry and size matched ry and size ctors and occupational he number of directors on he number 3.48) 1.72) 0.16) ( ( (1.84) 0.79 * * *** *** 0.001 (0.10) 0.041 ( 0.108 1.65 0.006 (0.03) 16.53 ends over earnings;ends over of Dir is the proportion ivid 2.51) is a 1-year share price return; share price is a 1-year Dir rem./Sal is the ratio t (32.16) 0.47 ( *** *** * 12 to t , i 0.013 (0.75) 0.05 (0.64) 0.01 2.81 R is the ratio of market value of equity plus book value of debt over total of debt over of equity plus book value value of market is the ratio Q 2.14) (31.91) 0.52 ( *** * *** 0.01 0.31) (6.69) 0.52 *** 0.017 ( 1.55) 0.27) (6.64) 0.61 *** is market value of equity plus book value of debt over sales; of debt over of equity plus book value value is market 0.002 ( 0.003 (0.28) 0.005 (0.53) 0.014 ( 1.05 0.66 2.33 , the chi-squared, the restricted and the unrestric by comparing are zero regression is used to test that all slopes of the Logit (1) (2) (3) (4) (5) (6) 2 M/S -values are in parentheses. are in -values t Logit regressions of the probability that pension fund holdings exceeds 3% of shares 95–96 exceeds fund holdings that pension of the probability regressions Logit t 12 to Significant at the 0.01 level. Significant Significant at the 0.05 level. Significant Significant at the 0.10 level. Significant i, t 2 pension funds; pension Block outstanding equity held by directors; equity held outstanding than dire other by blockholders equity capital held of outstanding Block is the proportion Dir #DIR 0.011 (1.18) 0.016 (1.58) %NED 0.04 (0.37) 0.042 (0.36) R Payout 0.006 (1.10) 0.006 (1.10) ROS the board; the board; directors in of non-executive is the proportion %NED of d is the ratio Payout ROA 0.04 (0.92) assets; funds; debt plus shareholders long-term debt over of long-term Blev is the ratio M/S Dirrem./Sal 0.123 (0.56) Q Blev The dependent variable is equal to 1 for companies that reported pension fund holding above 3% in 1995–1996 and to 0 for indust 1995–1996 and to 0 for 3% in above fund holding that reported pension companies is equal to 1 for dependent variable The Split Constant 0.71 control firms. control of directors remuneration over sales; over and CEO are split; of chairman equals to 1 if the roles of directors remuneration is a dummy variable Split #DIR is t Table 9.5 likelihoods. *** ** * DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 239 if the roles of CEO and Chairman are split externally-managed pension fund monitoring and zero otherwise. The results show that of the board structure. At the same time, the coefficients of the pension fund vari- board structure is unrelated to market ables are positive but not significant except performance, suggesting that the recent for pension fund size (LNPFA). When we trend towards the adoption of the split our sample into high and low growth Cadbury’s prescriptions was not related to companies none of the pension funds vari- the presence of agency conflicts, but rather ables is significant. We tested for possible dictated by some “need of visibility” by multicollinearity problem by running the companies. Franks et al. (1998) also regressions with each single variable. We report a similar relationship. We do, how- find (but do not report) that the holdings of ever, report some evidence consistent with directors, Dir, and the coefficient of nonmonitoring role of other than pension funds institutional block shareholders, NIBlock, institutional shareholders and of debtholders, are negative and significant. None of the at least with regards to the appointment of occupational pension funds variables is sig- non-executive directors within low growth nificant. We find that the coefficient of the firms. There is also some evidence that external pension fund managers variable, managerial ownership is used to entrench EPFM, is positive and significant for the the position of incumbent managers as the whole sample, though it is not when the coefficient of directors’ ownership is, in sub-samples of high and low growth firms most cases, negative and significant. are separated. Finally, our results are consistent with the Columns 4 to 6 provide the results of the hypothesis that institutional blockholders Logit regressions where the dependent vari- (other than pension funds) lead high able is equal to 1 if the company has growth firms to appoint a non-executive appointed a non-executive director as chairman. We simulated our results using chairman and zero otherwise. Here again growth in profits, market returns and Q as none of the coefficients of the pension fund proxies for growth opportunities and/or variables are significant. However,when we presence of agency conflicts. The results run the regressions with a single independent are qualitatively similar to those reported variable, we find that the coefficient of in Table 9.6. pension funds incidence, IPF,is positive and significant (1.36 with t 1.71) for the 4.4. Pension funds investments and whole sample. For the low-Q companies, we firm value find that the coefficient of total pension funds,TPF,is positive and significant (0.06 In this section we focus only on our test and t 1.93).The coefficients of directors companies and test for the relationship ownership, Dir, and non-institutional between firm value and ownership structure. blockholders, NIBlock, are negative and Table 9.7 reports the Pearson correlation significant while those of institutional matrix.The correlation between the various blockholders and external fund managers measures of occupational pension fund are positive and significant. holdings and firm value is, in most cases, The last three columns present the weak and negative. An exception is results of the OLS regressions where the represented by the correlation between the dependent variable is the proportion of market-to-sales ratio and the pension fund non-executive directors in the board. The incidence variable. Similarly the correlation results show that none of the pension fund between firm value, blockholding and direc- variables is significant. When the regres- tors holdings is not significant. However, sions are run separately, we find that only ownership variables are negatively corre- size, Ln(TA), and leverage, Blev, are posi- lated with size as measured by total assets, tive and significant. suggesting that occupational pension The overall results do not provide funds, directors, institutional blockholders strong support for the occupational and (other than pension fund managers) and 240 EQUITY OWNERSHIP STRUCTURE AND CONTROL onal measures 2 0.46) 0.03) 0.00003 0.0004 0.002 1.25) R rtunities are rtunities he dependent variable. d (i.e., 1995–1996).The s holdings over their total over s holdings rs; EPFM is the proportion ns the Pseudo- 0.09) ( 0.0001 equity held by non-instituti equity held the proportion of equity overallthe proportion held by 0.24) ( 0.0002 0.0002 0.001 ** 0.18) (0.73) (1.00) ( 1.28) ( 0.019 0.003 0.001 0.001 0.03) (0.02) (0.60) (0.50) (0.13) 1.44) ( 1.42) ( 0.165 0.121 0.163 0.213 0.048 0.020 0.015 0.067 0.025 ** ** 0.024 1.46) ( 2.17) ( 0.017 1.11) (0.02) ( 6.447 0.091 0.000 * ratio.The first six columns are Logit regressions where Split or Nechair are equal to 1 if the company has split the roles or Nechair are equal to 1 if the company has split Split where regressions ratio.The six columns are Logit first 0.010 0.028 0.039 0.019 0.002 1.79) (0.03) ( 1.11) ( 0.88) (0.19) ( 0.19)0.068 (0.77) (2.34) (0.74) (2.06) (0.21) (0.74) ( E/P is a 1-year stock return. is a 1-year 0.004 0.96) ( 0.68) ( 0.44) ( 2.974 0.611 0.011 (1.13) ( i,t – 12 to t rminants of board structure for low and high growth firms growth and high low of board structure for rminants All Low growth High growth All Low growth High growth All Low growth High growth ( ( ( Split Nechair %NED R Dete of its chairman and chief executive officer or has a non-executive director as a chairman, director or has a non-executive otherwise. officer and zero executive and chief of its chairman In these regressio IBlock 0.030 measured using the median measured using institutional block shareholders other than (both internally and externally managed) pension funds; managed) pension and externally of NIBlock is the proportion than (both internally other shareholders block institutional Dir assets; fund; of the pension is the log of asset value LNPFA by the directo equity owned of outstanding Dir is the proportion clients); funds or other on behalf of pension (either fund managers pension IBlock represents by external of equity held TPF is the proportion of outstanding equity owned by all identified pension funds; pension by all identified equity owned of outstanding fund TPF is the proportion pension of occupational IPF is the ratio the goodness of fit. In the last three columns we run OLS regression with the proportion of non-executive on the board %NED as t of fit.the goodness on the board %NED of non-executive with the proportion In the last OLS regression three columns we run In each regression the dependent variable is computed for the period 1996–1997, lagge the period is computed are 1-year for the dependent variable variables all independent In each regression while all 250 companies, for separately are run regressions All, (119), companies low-growth (131). companies growth and high oppo Growth Constant NIBlock Table 9.6 blockholders; DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 241 0.35) 0.001 0.04) ( 0.0001 0.0010.10) 0.10 (1.55) 0.01) (0.27) 0.080.02 0.17 0.01 0.01) 0.005 0.0010.36) 0.003 (1.9) * ** 0.23) ( 0.0002 0.64) ( 0.01 ( 0.05) ( 0.001 0.004 0.02 0.94) (0.76) (0.83) (0.18) 0.01) (2.19) (1.40) (0.76) 0.17) ( 0.34 0.01 0.02 0.005 0.05 0.14 0.06) ( 0.68) ( 0.02 0.25) (0.28) ( 0.004 0.006 0.02 0.82) ( 0.12) ( 0.01) (0.50) ( 0.21 0.01 0.13 0.06 0.26 0.63 1.54 1.56 0.11 1.18 0.10 ** ** 0.47 0.09 0.07 1.56) (0.24) (0.18) ( 0.03 0.837.29% 15.92% 0.11 4.87% 15.27% 11.98% 19.82% 0.95% 0.00% 3.17% (0.43) (1.10) (0.05) ( (1.26) (0.46) (0.45) (0.94) (1.13) (0.77) (0.04) ( (0.08) (1.23) (0.07) ( (2.15) (0.12) (0.94) ( (2.17) (0.32) (0.63) (1.61) (0.10) (0.78) (1.70) (0.99) (1.61) (0.38) (0.86) (0.51) (1.38) (0.71) (1.03) (0.79) ( ( 2 t R 2 12 to Significant at the 0.01 level. Significant Significant at the 0.05 level. Significant i, t Significant at the 0.10 level. Significant or R Blev 6.25 EPFM 0.016 0.259 0.002 IPF 1.48 1.46 0.75 0.74 1.88 1.05 0.0005 LN(TA) Pseudo- LNPFA 0.53 TPF 0.02 0.10 0.03*** 0.05** * 0.03 0.05 0.002 R 242 EQUITY OWNERSHIP STRUCTURE AND CONTROL 0.05 are measured 0.04 plit; is the %NED M/S ables are measured in ables and lock is the proportion of lock is the proportion epresents the proportion Q occupational pension funds pension occupational erm debt plus shareholders erm ned by the directors; EPFM 0.06 0.03 0.06 0.01 0.20 0.13 0.13 0.17 0.05 0.46 0.25 0.12 0.13 0.03 0.38 0.21 0.10 0.17 0.27 IBlock NIBlock ln(TA) Blev No.Dir Split %NED 2 0.17 0.13 0.15 0.10 0.06 0.13 0.13 0.07 0.12 0.20 0.25 0.06 0.23 0.02 0.03 0.26 0.21 0.13 0.21 0.32 0.02 0.25 0.03 0.34 0.27 0.29 0.95 0.04 0.25 0.05 0.21 is the price–earnings ratio at the balance sheet date. ratio is the price–earnings P/E 0.11 0.07 0.10 0.08 0.15 0.07 0.02 0.11 0.30 0.01 0.22 0.36 0.04 0.05 0.08 0.02 0.07 0.04 0.09 0.28 0.01 0.01 0.00 0.04 0.02 0.04 0.04 0.01 0.07 0.04 0.02 0.05 0.04 0.13 0.18 0.29 0.05 0.01 0.02 0.04 0.16 0.04 0.03 0.09 0.02 0.16 0.05 0.04 0.10 0.08 0.04 0.11 0.69 0.07 0.02 0.11 0.06 0.07 0.07 0.02 0.10 0.12 0.04 0.10 0.15 0.18 0.59 0.11 0.37 0.26 Pearson correlation coefficients between the variables used between the variables coefficients correlation Pearson Q M/S TPF IPF NPF LNPFA EPFM Dir Dir 2 %NED 0.06 0.04 Split No.Dir 0.09 0.07 0.01 0.01 0.05 0.18 0.07 Blev funds; of directors on the board; #DIR is the number and CEO are s of chairman equals to 1 if the roles is a dummy variable Split Dir In(TA) equity held by non-institutional blockholders; by non-institutional equity held is the log of total assets; In(TA) long-t debt over of long-term Blev is the ratio Dir 0.11 0.00 EPFM NIBlock of equity overall held by institutional block shareholders other than (both internally and externally managed) pension funds; managed) pension of and externally than (both internally NIB other shareholders block by institutional held equity overall LNPFA 0.04 is the proportion of equity held by external pension fund managers (either on behalf of pension funds or other clients); funds or other on behalf of pension (either fund managers pension by external IBlock of equity held is the proportion r NPF IBlock IPF 0.02 0.31 holdings over their total assets; over funds; pension holdings of occupational NPF is the number equity ow of outstanding Dir is the proportion M/S TPF 1995–1996 financial year.TPF is the proportion of outstanding equity owned by all identified pension funds; pension by all identified equity owned of outstanding is the proportion of IPF is the ratio year.TPF 1995–1996 financial The sample includes all companies that reported occupational pension fund holding above 3%. above fund holding pension that reported occupational all companies sample includes The dependent variables The Table 9.7 the board; directors in of non-executive proportion P/E in funds. year. pension by largest occupational equity owned of outstanding LPF is the proportion 1996–1997 financial All the vari DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 243 non-institutional blockholders hold large funds holdings (3). Although the coefficient stakes mainly in small firms. However, the of pension fund incidence, IPF,and the size fund manager variable is positively related of pension fund, LNPFA, are positively to size. Consistent with previous evidence related to firm value, they are not (e.g., Lasfer, 1995), leverage is positively significant. The coefficient of external related to firm size but negatively pension fund managers’ stakes is negatively, correlated with firm value. The number of though not significantly, related to firm value. directors is positively correlated with In contrast to previous studies (e.g., occupational pension fund, externally- Yermack, 1996, and Eisenberg et al., managed pension funds, non-institutional 1998) we report a positive relationship blockholding, leverage and size, while it is between firm value and the number of negatively correlated with institutional directors. The difference in the results blockholders and directors’ ownership. could be due to the fact that our companies The split dummy variable is positively are relatively middle-sized compared to the related to leverage, block ownership, size sample of small companies of Eisenberg and proportion of non-executive directors et al. (1998) and that of large companies in the board, but negatively related to of Yermack (1996). Finally, the coefficients directors’ holdings. This suggests that the of the holdings of directors variable and its larger the company and the higher squared value are not significant suggesting the debt–equity ratio, blockholding and the that there is no linear or non-linear proportion of non-executive directors in the relationship between value and managerial board, the higher its propensity to split ownership. These results are not consistent the roles of chairman and CEO. However, with the findings of McConnell and Servaes none of the occupational pension fund (1990; 1995). measure is statistically related to the split We simulate these results using market dummy, implying that pension funds, value of the firm over sales as a proxy for individually or collectively, do not push firm value (Lins and Servaes, 1999). The companies to split the roles of chairman results, reported in (4) and (5), show a and chief executive officer.Finally, directors’ positive relationship between value and holdings are negatively related to the pension fund incidence, but a negative proportion of non-executive directors in the relationship with the number of pension board, suggesting that such holdings funds.The coefficient of externally-managed exacerbate the potential agency conflicts pension funds is negative but not significant between the board and the management. while that of institutional block ownership In Table 9.8, columns (1) to (3), we is positive and significant.We also simulate report the results of the regressions of firm our results by using sales as a measure of value, as measured by Tobin’s Q in size and capital expenditure over total 1996/97, against various measures of assets as a proxy for growth opportunities. occupational pension funds holdings in the The results, not reported for space reasons, 1995/96 financial year.These results show are qualitatively similar to these reported that, with the exception of total pension in Table 9.8. Overall, our results suggest funds stakes, TPF, none of the various that pension funds do not add value to measures of pension funds holdings explain companies in which they hold large stakes. firm value. The total pension funds stakes variable, TPF, is actually, negative and 4.5. Effects of pension fund holdings significant, suggesting that pension funds on firms’ long-term performance collectively destroy value. Even after controlling for other monitoring mechanisms We analyze the long-term performance of documented in the previous literature (e.g., our test firms by comparing the changes in Agrawal and Knoeber, 1996; Yermack, accounting and stock price performance 1996), such as size and P/E, firm value is over the sample periods 1994–1995 and still negatively related to total pension 1996–1997. As in Karpoff et al. (1996) 244 EQUITY OWNERSHIP STRUCTURE AND CONTROL 3.11) 2.01) 1.02) th inter- (5.41) (4.08) irectors; (2.00) (2.80) ( (1.71) ( *** ** * *** *** ** *** *** 0.28 0.57 0.512 ( 9.35 is the price–earnings sured in 1995–1996. sured in E ; EPFM is the propor- / the number of d the number P sing the market value of value the market sing s holdings over their total over s holdings , value of market the ratio Q ors; n of outstanding equity owned n of outstanding 2.92) 0.18) 0.13) 1.09) 0.79) (3.98) 0.031 1.41) (4.50) 0.372 ( (1.66) 0.013 * *** *** *** *** 0.001 ( 0.001 ( 0.56 0.29 ( 0.000 ( 4.31 0.619 ( M/S 2.57) 0.001 (0.06) 0.33) 0.013 0.53) 0.36) (5.62) 0.032 1.83) 0.03 (0.41) ( 1.47) (1.85) 0.056 (1.32) 0.066 ( (1.77) 0.40 (0.80) ** * *** * * *** 0.22 ( 0.002 ( 0.003 ( 0.002 ( 0.09 (0.56) 0.049 0.000 (0.03) 4.18 0.12 3.07) 0.37) 0.51) 0.36) ( (5.65) 0.035 1.79) 1.53) (1.86) 0.061 Significant at 1% level. Significant ( (1.78) 0.69 *** * *** * * *** *** 0.11 0.002 ( 0.035 0.23 ( 0.003 ( 0.002 ( 0.042 4.84 2.60) -values are in parentheses. are in -values Significant at 5% level. Significant t 0.46) 0.083 (1.34) 0.086 (1.38) 0.36 ( ** 0.64) 0.62 (0.39) 0.57 (0.36) 0.07 (0.04) 0.97 *** ** is its squared value; IBlock represents the proportion of equity overall held by institutional block shareholders other than (bo other shareholders is its squared value; block by institutional held of equity overall IBlock represents the proportion 0.033 0.97 ( 2.99 2 Q (1) (2) (3) (4) (5) Relationship between firm value and pension fund holdings and pension value between firm Relationship Significant at 10% level. Significant % 2.21% 18.08% 17.43% 18.09% 20.63% 2 * R 2 E / by the directors; Dir DirDir 0.004 (1.28) 0.004 (0.42) 0.010 (0.80) ln(TA) tion of equity held by external pension fund managers (either on behalf of pension funds or other clients); funds or other on behalf of pension (either fund managers pension Dir is the proportio by external of equity held tion No.Dir 0.062 Split is a dummy variable equals to 1 if the roles of chairman and CEO are split; of chairman direct equals to 1 if the roles is a dummy variable of non-executive is the proportion Split %NED NPF P Split assets; fund’s asset; of the largest pension value is the log of market LNPFA funds pension of occupational NPF is the number IBlock nally and externally managed) pension funds; managed) pension and externally blockholders; by non-institutional nally of equity held NIBlock is the proportion #DIR is TPF %NED 0.67 EPFM LNPFA 0.110 (1.62) 0.092 (1.46)Adj. 0.095 (1.50) 0.007 (0.09) NIBlock IPF 0.034 ( Blev 0.002 (0.01) 0.020 (0.04) equity to sales as dependent variable.The dependent variables are measured in 1996–1997 while the independent variables are mea variables the independent 1996–1997 while are measured in equity to sales as dependent variable.The dependent variables funds; pension by all identified equity owned of outstanding fund TPF is the proportion pension of occupational IPF is the ratio of equity plus book value of debt over total assets, of debt over of equity plus book value as the dependent variable. u In column (4) and (5) the results are simulated Constant The sample includes all companies that reported occupational pension fund holding above 3%. above fund holding pension that reported occupational all companies sample includes The In columns (1) to (3) we use Table 9.8 at the balance sheet date. ratio F Notes: DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 245 and Del Guercio and Hawkins (1999), we company selected by identifying all firms investigate whether companies in which with a market value of equity between pension funds hold large stakes rebound 70% and 130% of that of the test more quickly from poor performance or company. Then, from this set of firms, we maintain their good performance over a select the company with the market-to- longer time period. We use return on book ratio closest to that of the test firm. assets (ROA) as a proxy for accounting We use the return of this company,rc*,n, as performance. We follow Wahal (1996) and a benchmark. compute the 2-year industry-adjusted This measure of performance accounts Cumulative Abnormal Return On Assets for the fact that occupational pension (CAROA) as funds are long-term investors, as reported in Table 9.3, and does not suffer from T cumulating biases observed in arithmetic CAROAi,t,T (1 ROAi,n) nt mean returns (Conrad and Kaul, 1993; T Wahal, 1996; Barber and Lyon, 1997). We (1 ROAs,n) nt eliminate survivorship bias by including dead companies in our sample. Out of our where ROAi,t is company i return on assets 289 test companies, 34 firms (11.76%) in year t and computed as are excluded because they went public after 1995 and we could not compute the mar- Profit Before Interest and Tax ket performance for the first period. ROA . i,t Book Value of Total Assests The results reported in Table 9.9 show that the industry-adjusted return on assets ROAs,t is the industry s median return on (CAROA) has not changed significantly assets in year t. over the two sub-periods. Companies in We simulate our results by computing in which occupational pension funds hold the same way Cumulative Abnormal large stakes underperform the industry Return On Equity (CAROE) and average (though not significantly) by Cumulative Abnormal Return On Sales 3.38% in 1994–1995 and by 1.50% in (CAROS). All measures of performance are 1996–1997. As in Wahal (1996), we split adjusted by subtracting the industry the sample into overperformers and median. underperformers. Companies with CAROA The abnormal performance is also evalu- below (above) the median are classified as ated using stock market returns.We compute underperformers (overperformers).27 The the Share Price Return (SPR) from a abnormal performance of the overperforming buy-and-hold strategy over the sub-periods companies decreased from 11.62% to 1994–1995 and 1996–1997 as follows: 6.44% and that of the underperformers has increased from 18.5% to 9.51%. T The results are consistent with Wahal SPR i,t,T {1 r i,n} nt (1996) and indicate that both the T under-and overperformers experience a {1 rc*,n} nt convergence to industry means. The results are not qualitatively affected by the use of where other measures of performance such as industry-adjusted return on equity (Panel B) Pi,n or industry-adjusted return on sales ri,n 1. Pi,n1 (Panel C). Panel D and E report the changes in firm We adopt the Barber and Lyon (1997) value as measured by (the industry- methodology, and control for size (market adjusted) Tobin’s Q and changes in abnormal capitalization) and market-to-book. We returns. For the full sample, the industry- match each test company with a control adjusted Q has decreased from 0.163 to 246 EQUITY OWNERSHIP STRUCTURE AND CONTROL Table 9.9 Pension fund holdings and long term accounting and stock price performance The sample includes all companies that displayed (at least) one relevant pension fund holding. Industry-adjusted return on assets (CAROA) is the return on assets of each company in the sample less the median return on asset of the industry. Industry-adjusted return on equity and return on sales and Q are computed in a similar way.The control sample includes firms taken from the same SEC industry whose 1994–1995 performance is the closest to our test firm. Panel E reports the results based on the size and market-to-book adjusted share price returns using the Barber and Lyons (1997) methodology. The sample is dividend into a sample of underperformers and overperformers based on the median performance of the firm in 1994–1995. For example, if the industry-adjusted return on assets for a firm is lower (higher) than the median, the firm is classified as an underperformer (overperformer).

t-statistic Mann– 1994–1995 1996–1997 difference Whitney N Mean Median Mean Median in means p-value

(Panel A) Industry-adjusted return on assets CAROA% Full sample 255 3.38 0.584 1.50 0.31 1.24 0.344 Overperformers 128 11.62*** 6.48*** 6.44*** 3.56*** 3.56 0.000 Underperformers 127 18.50*** 10.30*** 9.51*** 4.76*** 3.59 0.000 (Panel B) Industry-adjusted return on equity CAROE% Full sample 255 4.96 1.412 3.89 0.53 0.36 0.906 Overperformers 128 20.79*** 10.31*** 5.72*** 2.13*** 4.48 0.000 Underperformers 127 30.90*** 16.50*** 13.56*** 6.91*** 3.94 0.000 (Panel C) Industry-adjusted return on sales CAROS% Full sample 255 0.08 0.67 0.33 0.00 0.24 0.811 Overperformers 128 14.99*** 7.13*** 7.78*** 4.92*** 3.32 0.000 Underperformers 127 15.27*** 7.57*** 7.19*** 4.75*** 3.35 0.000 (Panel D) Industry-adjusted firm value Q Full sample 255 0.163*** 0.082*** 0.072*** 0.095*** 2.72 0.009 Overperformers 128 0.674*** 0.346*** 0.480*** 0.15*** 3.22 0.000 Underperformers 127 0.353 0.322 0.339 0.306 0.58 0.237 (Panel E) Size and market-to-book adjusted share price return SPR% Full sample 255 0.49 3.85 3.20 8.76 0.42 0.916 Overperformers 128 68.22*** 43.64*** 6.83*** 13.05*** 6.62 0.000 Underperformers 127 67.76*** 57.19*** 0.45*** 2.09*** 7.01 0.000

* Significant at 10%, 5% and 1% respectively. ** Significant at 10%, 5% and 1% respectively. *** Significant at 10%, 5% and 1% respectively.

0.072.28 The decrease in both the average to 0.48. The decrease in the mean median and median firm value is statistically are statistically significant. In contrast, the significant. The results suggest that, over average and the median Tobin’s Q of the time, the value of companies in which occu- underperforming companies did not change pational pension funds hold large stakes significantly. These companies have under- decreases. As in the above panels, we split performed in 1994–1995 period and car- the sample into overperformers and ried on underperforming in the 1996–1997 underperformers. Overperforming (under- period. These results are striking as they performing) companies are companies with imply that companies in which occupa- Tobin’s Q higher (lower) than the median. tional pension funds hold large stakes The overperforming companies have done decrease in value through time, and those worst over the two sample periods. Their that are already underperforming do not average value has decreased from 0.67 improve. In sum, it appears that pension DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 247 funds are passive investors: they do not companies in which they hold large stakes. make companies in which they hold large Our results cast doubt on the monitoring stakes improve their performance and they role of pension funds which are considered do not sell their holdings in companies that theoretically, on the one hand, to be the are underperforming. main promoters of corporate governance in Panel E reports the results based on the the UK, and, on the other hand, to be short- size and market-to-book adjusted share termist and dictate their rules to companies. price returns using the Barber and Lyon At the same time, we show that, despite (1997) methodology. For the full sample, the relatively poor performance of the companies have decreased their average companies in which they invest, occupa- performance from 0.49% to 3.2% and tional pension funds do not opt for the ‘exit’ the median performance from 3.85% to strategy.We show that there is no apparent 8.76%. However, this decrease is not specific relationship between the funds and statistically significant. The stock price the companies in which they invest and performance of the sub-sample of the over- these companies are not illiquid to make it performing firms declined significantly impossible to exit. One possibility is that from 68.22% to 6.83% (t-statistic of pension funds choose to invest in low Q the difference in means 6.62), while firms to benefit from return reversals on underperforming companies increased their these securities. However, the lack of data performance from 67.76% to 0.45% on the investment styles of these funds does (t-statistic of the difference in not allow us to explore this issue further.At means 7.01). As in Wahal (1996), our this stage, our results suggest that, once results are consistent with mean reversion ‘locked in’ pension funds find it difficult and hypothesis and cannot be attributed to costly to monitor or to sell their holdings for pension funds investments.29 fear of selling at a discount or to convey information to the market. Our results may not come as a surprise. 5. DISCUSSION AND CONCLUSIONS There is a large debate in the UK about the lack of monitoring by pension funds and it In this paper we analyze the performance is only recently that the National of companies in which occupational Association of Pension Funds considered pension funds hold large stakes and test seriously this issue (NAPF, 1996b). In the hypotheses that, because of their addition, occupational pension funds may size, structure and objectives, these funds not have a material effect on the perform- should be effective monitors of UK ance of companies in which they hold large companies. stakes because these blocks tend to repre- We compare the financial performance sent relatively small fractions of the total of companies in which occupation pension values of the funds’ assets. The mean funds hold more than 3% of the issued (median) block represents only 0.15% share capital against all other non-financial (0.03%) of the typical funds assets, and and industry and size-adjusted control with such a small fraction, it is not clear groups. We show that our test firms are that the gains to the funds of expending small and have low value.These companies effort monitoring the sample firms would are also not likely to be more efficient justify the costs. Thus, our results are and/or to pay higher dividends than the consistent with Admati et al. (1994) control group. However, our results show proposition that, given that monitoring is that these holdings constitute a small pro- costly, these funds will not monitor as this portion of the occupational pension funds is not likely to result in a modification in assets, suggesting that most of their funds the firms payoff structure and will not lead are invested in large companies where they to net gains. They may also refrain from do not hold large stakes. We report that intervening publicly for fear of drawing to pension funds do not add value to the public attention the difficulties the company 248 EQUITY OWNERSHIP STRUCTURE AND CONTROL is facing and/or trading on insider information. 3 See Black (1998) and Karpoff (1998) for a However, our results could also imply that survey of the shareholder activism literature. pension funds are passive investors, investing 4 Hutton (1995) argues that “pension most of their funds in the index and these funds . . . have become classic absentee land- investments we analyzed in this paper are lords, exerting power without responsibility and making exacting demands upon companies with- peripheral. out recognizing their reciprocal obligations as While our results are not consistent with owners” (p. 304). the monitoring role of occupational pension 5 For example, Drucker (1976) stipulates that funds, our study does not fill all the gaps in “pension funds are not ‘owners’, they are the literature. Our sample includes only investors. They do not want control . . . If they companies in which occupational pension do not like a company or its management, their funds hold more than 3% of the issued duty is to sell the stock” (p. 82). More recently, share capital. A more desegregated and Porter (1997) argues that institutional comprehensive data is not available. investors, despite their substantial aggregate Similarly, because of data unavailability, we holdings, do not sit on corporate boards and have virtually no real influence on managements have not addressed the question of whether behavior because they invest nearly all their occupational pension funds sponsor initia- assets in index funds rather than directly in tives, whether they meet with companies companies. Short and Keasey (1997) suggest and whether there is a co-operation that once pension funds are locked in, it is costly between various funds. The extent to which to get involved in monitoring and they cannot these additional factors will strengthen or exit in case they are considered to trade on alter our analysis is a matter of further insider information. Murphy and Van Nuys investigation. (1994) argue that pension funds are run by individuals who do not have the proper incentives to maximize fund value. 6 Cadbury (1992) notes that “Because of ACKNOWLEDGEMENTS their collective stake, we look to the institutions in particular, with the backing on Institutional We are grateful to Yakov Amihud, Shareholders’ Committee, to use their influence Francesco Cesarini, John McConnell, as owners to ensure that the companies in which Annette Poulsen (the editor), Jonathan they have invested comply with the Code” (para. Sokobin, Geof Stapledon, Sudi 6.16).The National Associate of Pension Funds also endorses such recommendations (NAPF, Sudarsanam, an anonymous referee, and 1996b). seminar participants at Brunel University 7 A number of studies document the free cash and the 1998 European Financial flow problem in the UK. For example, Franks Management Association meeting in and Mayer (1994) find that UK companies pay Lisbon for valuable insights and helpful high dividends relative to German companies; comments. Faccio acknowledges a research Lasfer (1995) shows that debt mitigates the grant from City University Business free cash flow problem; Lasfer (1997) provides School. All errors are our responsibility. evidence that firms with free cash flow problems pay scrip, rather than cash, dividends. The reports of Cadbury (1992) and Greenbury (1995) are a manifestation of the previous NOTES debate on the various wider corporate governance issues detailed in Charkham (1995), * Corresponding author. Tel.: 44-0-171- Stapledon (1996) and Keasey et al. (1997). 477-8634; fax: 44-0-171-477-8648; e-mail: 8 Companies Act 1995, Sections 198 and 199 [email protected] requires UK companies to disclose in their 1 Tel.: 39-02-7234-2436; e-mail: m.faccio@ accounts the name of any investor who holds iol.it. 3% or more of the issued share capital. 2 Other studies suggest that the relationship 9 See Chew (1997) for a collection of papers between ownership, such as the fraction of dealing with these two corporate governance shares held by insiders and performance is not systems. linear but roof-shaped (e.g., Morek et al., 1988; 10 See Black (1998) and Karpoff (1998) Stulz, 1988; McConnell and Servaes, 1990). for a survey. DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 249 11 Similarly, in the US, the value of DC plans 19 The ideal would be to exclude from the in 1993 was $1068 billion compared to $1248 denominator of this ration other assets such as billion for DB plans. However, DC schemes are property, cash and fixed income securities, but growing at much faster rates of 19% per year, the desegregated data on equity investment is compared to 14% for DB plans (Jepson, 1998). not available. 12 The report states that no “one individual 20 This variable includes directors holdings has unfetted powers of decision.Where the chair- but excludes those of officers. UK quoted com- man is also chief executive, it is essential that panies are required to disclose in their financial there should be a strong and independent element statements the proportion of shares held on the board, with a recognised senior member”. directly and indirectly by executive and non- 13 The Code of Best Practice No 4.3 executive directors (Companies Act 1985). recommended an audit committee of at least However, no similar disclosure applies to offi- three non-executive directors with written terms cers. This legal disclosure requirement means of reference and No 4.30 recommends the that we had to define managerial ownership as institution of a nomination committee as an ownership by members of the board of direc- internal committee within the board. This tors. Although this definition is consistent with committee should be composed of majority of that of Morck et al. (1988), it differs from that non-executive directors and chaired by the of McConnell and Servaes (1990) and Denis chairman of the board. and Sarin (1999) as we do not include shares 14 The 99 in-house managed pension funds owned by corporate officers not members of included in our study manage assets for some the board. £150 billion, which correspond to 26.8% of all 21 Denis and Kruse (1999) and Denis and occupational pension funds’ assets. Sarin (1999) distinguish between unaffiliated 15 The report summarizes the functions of the and affiliated block shareholdings. Unaffiliated board as follows: “The responsibilities of the blockholders would be expected to perform an board include setting up the companys strategic important monitoring role and make firms com- aims, providing the leadership to put them into ply with the Code of Best Practice, while affili- effect, supervising the management of the ated blockholders are more likely to be sensitive business and reporting to shareholders on their to managerial pressure or pursue goals other stewardship”. As to the financial aspects of than share price maximization (Brickley et al., corporate governance, the report mentions: 1988, 1994). In this perspective, our internally “The way in which boards set financial policy managed occupational pension funds are clearly and oversee its implementation, including the “unaffiliated shareholders”. Since our data use of financial controls, and the process does not generally allow us to check for the whereby they report on the activities and intensity of the relationships between firms and progress of the company to the shareholders”. shareholders, we consider together affiliated 16 See John and Senbet (1998) for an exten- and unaffiliated shareholders in the category of sive survey of the monitoring role of corporate blockholders. However, we distinguish between board of directors and Stapledon (1996, pp. institutional and non-institutional blockholders, 138–153), for the monitoring role of non- because these two categories of investors were executive directors. shown to behave differently, i.e., when sponsor- 17 The choice of 1995–1996 sample period is ing initiatives (Jarrell and Poulsen, 1987; driven by data availability.The data on sharehold- Karpoff et al., 1996). ing is inserted manually because Extel Financial 22 One possibility of overcoming this problem provides only the latest data on shareholding in of lack of data of pension fund assets invested in text format and this data is not available in UK equities would be to assume that all pension machine-readable form. Other similar studies use funds are homogeneous in their investment also short time period (see Karpoff, 1998, for a styles, i.e., the aggregate distribution of pension review).We report below that the vast majority of fund assets would apply to each fund. They companies (83% of our test firms) had large would therefore hold an average of 53% of pension funds holdings in both 1992 and 1996 their assets in UK equities and the remaining periods and the magnitude of their holdings has 47% in fixed income securities, overseas equi- not changed significantly.Thus, our results are not ties, property and cash, as reported in NAPF likely to be sample-period dependent. (1996a). However, we consider that this 18 Extel Financial is a financial database assumption is not likely to hold as the average microsystem. The database provides accounting holding is not constant through time. as well as financial and reference data for all 23 Hambros Company Guide is a quarterly UK companies and many international companies. publication, which gives a summary financial 250 EQUITY OWNERSHIP STRUCTURE AND CONTROL data and shareholdings above 3% of UK Journal of Financial and Quantitative companies.The Stock Exchange Yearbook is an Analysis 31, 377–397. annual publication of the London Stock Agrawal, A., Mandelker, G.N., 1990. Large Exchange. It provides a summary data of the shareholders and the monitoring of activities, shareholdings and performance of UK managers: The case of antitakeover charter listed companies. amendments. Journal of Financial and 24 We exclude 58 companies with negative book value of equity due to goodwill write-offs Quantitative Analysis 25, 143–162. when we use the market-to-book ratio, book Ayres, I., Cramton, P., 1993. An agency per- leverage and return on equity. The inclusion of spective on relational investing. Working these companies did not, however, alter our Paper, Stanford University, Stanford, CA. reported results. Barber, B.M., Lyon, J.D., 1997. Detecting long 25 The negative relationship between pension run abnormal stock returns: The empirical fund investments and Q has been widely power and specification of test-statistics. documented in the investment literature. For Journal of Financial Economics 43, 341–372. example, Lakonishok et al. (1994) show that Bethel, J.E., Liebeskind, J.P., Opler, T., 1998. pension funds invest in glamour stocks (low Block share repurchases and corporate per- book-to-market firms), which generally underperform value stocks (high book-to-market formance. Journal of Finance 53, 605–635. firms) because the previous success of the Black, B.S., 1998. Shareholder activism and glamour stocks helps institutions justify their corporate governance in the United States. portfolio selection to their investors. In: Newman, P. (Ed.), The Palgrave 26 The issue of director’s pay system has just Dictionary of Economics and the Law. been considered by the National Association of Blake, D., 1995. Pension Schemes and Pension Pension Funds through their call to vote on Funds in the United Kingdom. Clarendon boardroom pay (Financial Times, 1997b). Press, Oxford. 27 We have also split the sample into over- Brancato, C.K., 1997. Institutional Investors and underperformers on the basis of positive and Corporate Governance: Best Practice for and negative CAROA or other variables used below but the results did not change. Increasing Corporate Value. McGraw Hill, 28 For both 1994–1995 and 1996–1997, the Chicago. Q ratio is computed as arithmetic average. Brickley, J.A., Lease, R.C., Smith, C.W. Jr., 29 We obtained similar results either by using 1988. Ownership structure and voting on the industry median return as benchmark and, antitakeover amendments. Journal of by matching our test firms by industry and prior Financial Economics 20, 267–291. performance. Also, for all accounting measures Brickley, J.A., Lease, R.C., Smith, C.W. Jr., of performance, as well as firm value, the results 1994. Corporate voting: Evidence from were qualitatively similar when we compared charter amendment proposals. Journal of the industry adjusted performance of 1990–92 Corporate Finance 1, 5–30. to that of 1993–95. Similar results are obtained when we concentrate on survived companies, Business Monitor, 1997. Central Statistical and exclude IPO firms. Office, London. Cadbury, A., 1992. Report of the Committee on the Financial Aspects of Corporate Governance. Gee, London. REFERENCES Carleton, W.T., Nelson, J.M., Weisbach, M.S., 1998. The influence of institutions on corpo- ABI, 1998. Research update: How much in rate governance through private negotiations: funded pensions in 1996? Insurance Trends Evidence from TIAA-CREF. The Journal of 16, 27–29. Finance 53, 1335–1362. Admati, A.R., Pfleiderer, P., Zechner, J., 1994. Charkham, J., 1995. Keeping Good Company: A Large shareholder activism, risk sharing, and Study of Corporate Governance in Five financial market equilibrium. Journal of Countries. Oxford University Press, Oxford. Political Economy 102, 1097–1130. Chew, D.H. (Ed.), 1997. Studies in International Agrawal, A., Knoeber, C.R., 1996. Firm per- Corporate Finance and Governance Systems: formance and mechanisms to control agency A Comparison of the US, Japan and Europe. problems between managers and shareholders. Oxford University Press, New York. DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 251 Coffee, J., 1991. Liquidity versus control: The France. Journal of Applied Corporate Finance institutional investor as corporate monitor. 9, 30–45. Columbia Law Review 91, 1277–1368. Franks, J.R., Mayer, C., Renneboog, L., 1998. Conrad, J., Kaul, G., 1993. Long-term market Who disciplines bad managers? Working overreaction or biases in computed returns? Paper, London Business School, London. Journal of Finance 48, 39–64. Gordon, L.A., Pound, J., 1993. Information, Cotter,J.F.,Shivdasani, A., Zenner,M., 1997. Do ownership structure, and shareholder voting: independent directors enhance target share- Evidence from shareholder-sponsored corpo- holder wealth during tender offers? Journal rate governance proposals. Journal of of Financial Economics 43, 195–218. Finance 48, 697–718. Davis, E.P., 1995. Pension Funds. Retirement- Greenbury, R., 1995. Directors Remuneration: Income, Security and Capital Markets. An Report of a Study Group Chaired by Sir International Perspective. Clarendon Press, Richard Greenbury. Gee, London. Oxford. Hart, O., Holmstrom, B., 1987. The theory of Del Guercio, D., Hawkins, J., 1999. The motiva- contracts. In: Bewley, T. (Ed.), Advances in tion and impact of pension fund activism. Economic Theory. Cambridge University Journal of Financial Economics 52, Press, Cambridge, pp. 71–155. 293–340. Hart, O., 1995. Firms, Contracts, and Financial Demsetz, H., Lehn, K., 1985. The structure of Structure. Oxford University Press, Oxford. corporate ownership: causes and conse- Hutton, W., 1995.The State We Are In. Vintage quences. Journal of Political Economy 93, Press, London. 1155–1177. Jarrell, G.A., Poulsen, A.B., 1987. Shark repel- Denis, D.J., Kruse,T.A., 1999. Managerial disci- lents and stock prices: The effects of anti- pline and corporate restructuring following takeover amendments since 1980. Journal of performance declines. Working Paper, Purdue Financial Economics 19, 127–168. University, West Lafayette, IN. Jensen, M.C., 1986. Agency costs of free cash Denis, D.J., Sarin, A., 1999. Ownership and flow, corporate finance, and takeovers. board structures in publicly traded corpora- American Economic Review 76, 323–329. tions. Journal of Financial Economics 52, Jensen, M.C., 1993. The modern industrial 187–223. revolution, exit, and the failure of internal Diamond, D.W., 1984. Financial intermediation control systems. Journal of Finance 48, and delegated monitoring. Review of 831–880. Economic Studies 51, 393–414. Jepson, J., 1998. US Defined Benefits and Drucker, P.F., 1976. The Unseen Revolution: Defined Contribution: Comparison and How Pension Fund Socialism Came to Implications. Paper presented at the 1998 America. Heinemann, London. NAPF Conference, Eastbourne, UK. Eisenberg, T., Sundgren, S., Wells, M.T., 1998. John, K., Senbet, L.W., 1998. Corporate gover- Large board size and decreasing firm value in nance and board effectiveness. Journal of small firms. Journal of Financial Economics Banking and Finance 22, 371–403. 48, 35–54. Kahn, C., Winton, A., 1998. Ownership struc- Financial Times, 1997. FT Survey on Pension ture, speculation and shareholder interven- Fund Investment. May 9, 1997. tion. Journal of Finance 53, 99–129. Financial Times, 1997. Pension Funds to Call Kang, J.K., Shivdasani, A., 1995. Firm perform- for Votes on Boardroom Pay: NAPF Will ance, corporate governance, and top executive Oppose Hampel Committees Findings. turnover in Japan. Journal of Financial September 29, 1997, p. 20. Economics 38, 29–58. Financial Times, 1999. Fighting Fit. January Kaplan, S.N., Minton, B.A., 1994. 29, 1999, p. 17. Appointments of outsiders to Japanese Franks, J.R., Mayer, C., 1994. The ownership Boards: Determinants and implications for and control of German corporations. Working managers. Journal of Financial Economics Paper, London Business School, London. 36, 225–258. Franks, J.R., Mayer,C., 1997. Corporate owner- Karpoff, J.M., 1998. Does Shareholder ship and control in the UK, Germany, and Activism Work? A Survey of Empirical 252 EQUITY OWNERSHIP STRUCTURE AND CONTROL Findings. Working Paper, University of Minns, R., 1980. Pension Funds and British Washington, Seattle, WA. Capitalism: The Ownership and Control of Karpoff, J.M., Malatesta, P.H., Walkling, R.A., Shareholders. Heinemann, London. 1996. Corporate governance and shareholder Morck, R., Shleifer, A., Vishny, R.W., 1988. initiatives: empirical evidence. Journal of Management ownership and market valua- Financial Economics 42, 365–395. tion: An empirical analysis. Journal of Keasey, K.S., Thompson, S., Wright, M., 1997. Financial Economics 20, 293–316. Corporate Governance: Economic, Murphy, K., Van Nuys, K., 1994. State pension Management, and Financial Issues. Oxford fund shareholder activism. Working Paper, University Press, Oxford. Harvard Business School, Boston, MA. Lakonishok, J., Shleifer, A., Vishny, R.W., 1994. NAPF, 1996. Year Book. National Association Contrarian investment, extrapolation, and of Pension Funds, London. risk. Journal of Finance 49, 1541–1578. NAPF, 1996. Good Corporate Governance. Lang, L.H.P., Ofek, E., Stulz, R.M., 1996. National Association of Pension Funds, Leverage, investment, and firm growth. London. Journal of Financial Economics 40, 3–29. Nesbitt, S.L., 1994. Long-term rewards from Lasfer, M.A., 1995. Agency costs, taxes and shareholder activism: a study of the debt: The UK evidence. European Financial ‘CalPERS effect’. Journal of Applied Management 1, 265–285. Corporate Finance 6, 75–80. Lasfer,M.A., 1996.Taxes and dividends:The UK Pension Funds and their Advisers, 1996. evidence. Journal of Banking and Finance 20, London. 455–472. Porter, M.E., 1997. Capital choices: Changing Lasfer, M.A., 1997. On the motivation for pay- the way American invests in industry. In: ing script dividends. Financial Management Chew, D.H. (Ed.), Studies in International 26, 62–80. Corporate Finance and Governance Lin, T.H., 1996. The certification of large block Systems: A Comparison of the US, Japan shareholders in initial public offerings: The and Europe. Oxford University Press, New case of venture capitalists. Quarterly Journal York, pp. 5–17. of Business and Economics 35, 55–65. Prowse, S., 1994. Corporate governance in and Lins, K., Servaes, H., 1999. International evi- international perspective: A survey of corpo- dence on the value of corporate diversifica- rate control mechanisms amongst large firms tion. Journal of Finance 54, 2215–2239. in the United States, the United Kingdom, London Stock Exchange, 1995. Survey on Japan and Germany. BIS Economic Papers, Share Ownership, Stock Exchange, London. No. 41, Bank of International Settlements, Mallin, C.A., 1997. Investors voting rights. In: Basel. Keasey, K., Wright, M. (Eds.), Corporate Romano, R., 1994. Public pension fund activism Governance: Responsibilities, Risks and in corporate governance reconsidered. Remuneration. Wiley: West Sussex. Columbia Law Review 93, 795–853. Maug, E., 1998. Large shareholders as moni- Shleifer, A., Vishny, R.W., 1986. Large share- tors: is there a trade-off between liquidity and holders and corporate control. Journal of control? Journal of Finance 53, 65–98. Political Economy 94, 461–488. McConnell, J.J., Servaes, H., 1990. Additional Shleifer,A.,Vishny, R.W., 1997. A survey of cor- evidence on equity ownership and corporate porate governance. Journal of Finance 52, value. Journal of Financial Economics 27, 737–783. 595–612. Shome, D.K., Singh, S., 1995. Firm value and McConnell, J.J., Servaes, H., 1995. Equity own- external blockholdings. Financial ership and the two faces of debt. Journal of Management 24, 3–14. Financial Economics 39, 131–157. Short, H., Keasey, K., 1997. Institutional share- Mikkelson, W.H., Ruback, R.S., 1985. An holders and corporate governance in the empirical analysis of the interfirm equity United Kingdom. In: Keasey, K.S.,Thompson, investment process. Journal of Financial S., Wright, M. (Eds.), Corporate Governance: Economics 14, 523–553. Economic, Management, and Financial DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 253 Issues. Oxford University Press, Oxford, Sudarsanam, S., 1996. Large shareholders, 18–53. takeovers and target valuation. Journal of Smith, M.P., 1996. Shareholder activism by Business Finance and Accounting 23, institutional investors: evidence from 295–314. CalPERS. Journal of Finance 51, 227–252. Tomlinson, L., 1998. Should the Future Be All Stapledon, G.P., 1996. Institutional Passive? Paper presented at the 1998 NAPF Shareholders and Corporate Governance. Conference, Eastbourne, UK. Clarendon Press, Oxford. Wahal, S., 1996. Pension fund activism and Strickland, D., Wiles, K.W., Zenner,M., 1996. A firm performance. Journal of Financial and requiem for the USA: Is small shareholder Quantitative Analysis 31, 1–23. monitoring effective? Journal of Financial WM, 1996. UK Pension Fund Industry Results. Economics 40, 319–338. WM, Edinburgh (various issues 1994–1996). Stulz, R.M., 1988. Managerial control of voting Yermack, D., 1996. Higher market valuation of rights: Financing policies and the market for companies with a small board of directors. corporate control. Journal of Financial Journal of Financial Economics 40, Economics 20, 25–54. 185–211.

Part 3 Corporate governance, underperformance and management turnover

INTRODUCTION

ART 3 COMPRISES PAPERS THAT DEAL WITH THE effectiveness of corporate Pgovernance mechanisms in disciplining top managers. A measure of such discipline is the removal or turnover of such managers in response to poor corporate performance. Where such managers are entrenched owing to their ownership of the company’s voting shares, connections to controlling shareholders or both, this form of disciplining is unlikely and the corporate governance mechanism is rendered ineffective. The controlling large block shareholder may in theory have an incentive to improve corporate performance, if necessary by disciplining top managers of underperformers but this concern may be overridden by their enjoyment of private benefits of control. Minority shareholders, however, do not enjoy such countervailing private benefits. Large shareholding thus may align the interests of managers and large shareholders but create another agency problem (i.e. between large, controlling shareholders and minority shareholders). The phenomenon whereby controlling shareholders manage to receive private benefits at the expense of minority shareholders is called ‘tunnelling’. In the first of the papers in this part,Volpin (Ch.10) uses the interesting Italian setting to test the disciplinary effect of ownership structure on management turnover. Italy features low on the ranking of shareholder rights as noted by La Porta et al. in their paper (see Part 1, Ch.3). Corporate ownership structure in Italy is characterised by large controlling, often family, shareholders, voting syndicates binding significant shareholders to vote together, thereby achieving joint control, and pyramidal ownership in which a controlling block of voting rights allow control of a hierarchy of group companies with less than majority ownership. Italy also has a substantial incidence of voting versus non-voting shares separating voting rights from mere cash-flow rights to dividends. Volpin explores the impact of controlling shareholders, voting syndicates and pyramidal controls on the sensitivity of management turnover to underperformance with a large sample of firms listed in Milan and over a long sample period. His results suggest that there is poor governance, as measured by a low sensitivity of turnover to performance and a low Q ratio, when (i) the controlling shareholders are also top executives, (ii) the control is fully in the hands of one shareholder and is not shared by a set of core shareholders, and (iii) the controlling shareholders own less than 50 percent of the firm’s cash-flow rights. He also reports further interesting insights. Large cash-flow rights are important even for controlling shareholders and help to align controlling and minority shareholder interests. Furthermore, where control is contestable, as in a syndicate of significant shareholders, management turnover becomes sensitive to underperformance. Normally corporate governance is made up of more than a single mechanism with extant mechanisms working in tandem (complementary mode) or as substitutes (substitutory mode). 256 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER In the latter case, some of the mechanisms may be redundant although, ex ante, determining which ones are redundant may be difficult. In the second paper in this part, Luc Renneboog (Ch.11) examines how corporate control is exerted in companies listed on the Brussels Stock Exchange. There are several alternative corporate governance mechanisms, which may play a role in disciplining poorly performing management: block-holders (holding companies, industrial companies, families and institutions), the market for partial control, creditors, and board composition. Even if there is redundancy of substitute forms of discipline, some mechanisms may dominate. It is found that top managerial turnover is strongly related to poor performance measured by stock returns, accounting earnings in relation to industry peers and dividend cuts and omissions. Tobit models reveal that there is little relation between ownership and managerial replacement, although industrial companies resort to disciplinary actions when performance is poor. When industrial companies increase their share stake or acquire a new stake in a poorly performing company, there is evidence of an increase in executive board turnover, which suggests a partial market for control.There is little relation between changes in ownership concentration held by institutions and holding companies, and disciplining. Still, increased disciplining also follows high leverage and decreasing solvency and liquidity variables, as are a high proportion of non-executive directors and the separation of the functions of CEO and chairman. The third paper by Julian Franks, Colin Mayer and Luc Renneboog (Ch.12) follows a similar line of empirical enquiry into the relative effectiveness of different disciplinary devices but this time in the UK context and the focus is on financially distressed firms. Economic theory points to five parties disciplining management of poorly performing firms: holders of large share blocks, acquirers of new blocks, bidders in takeovers, non-executive directors, and investors during periods of financial distress. This paper reports a comparative evaluation of the role of these different parties in disciplining management.The authors find that, in the United Kingdom, most parties, including holders of substantial share blocks, exert little disciplining and that some, for example, inside holders of share blocks and boards dominated by non-executive directors, actually impede it. Bidders replace a high proportion of management of companies acquired in takeovers but do not target poorly performing management. In contrast, during periods of financial constraints prompting distressed rights issues and capital restructuring, investors focus control on poorly performing companies. These results stand in contrast to the United States, where there is little evidence of a role for new equity issues but non-executive directors and acquirers of share blocks perform a disciplinary function.The different governance outcomes are attributed to differences in minority investor protection in two countries with supposedly similar common law systems.This paper makes the interesting link between the relative effectiveness of different corporate control mechanisms and the larger legal environment in which these mechanisms operate. In 1992, the Cadbury Committee issued the Code of Best Practice (the Code). It recommends that boards of UK corporations include at least three outside (i.e. independent) directors and that the positions of chairman and CEO be held by different individuals (the non-duality principle). The underlying presumption is that these recommendations would lead to improved board oversight. Jay Dahya, John McConnel and Nick Travlos (Ch.13) empirically investigate the impact of the Code by comparing the relationship between CEO turnover and corporate performance before and after the Code. CEO turnover increased following issuance of the Code; the negative relationship between CEO turnover and performance became stronger following the Code’s issuance; and the increase in sensitivity of turnover to performance was concentrated among firms that adopted the Code. This Code is now part of the listing requirements on the London Stock Exchange on the ‘comply or explain’ (non-compliance) basis. Dahya et al.’s study is an interesting example of empirical methodology for evaluating externally imposed corporate governance regimes in their impact on corporate behaviour and performance. Chapter 10

Paolo F.Volpin GOVERNANCE WITH POOR INVESTOR PROTECTION: EVIDENCE FROM TOP EXECUTIVE TURNOVER IN ITALY*

Source: Journal of Financial Economics, 64(1) (2002): 61–90.

ABSTRACT This paper studies the determinants of executive turnover and firm valuation as a function of ownership and control structure in Italy, a country that features low legal protection for investors, firms with controlling shareholders, and pyramidal groups. The results suggest that there is poor governance, as measured by a low sensitivity of turnover to performance and a low Q ratio, when (i) the controlling shareholders are also top executives, (ii) the control is fully in the hands of one shareholder and is not shared by a set of core shareholders, and (iii) the controlling shareholders own less than 50% of the firm’s cash-flow rights.

1. INTRODUCTION high ownership concentration to help solve the managerial agency problem because THE “LAW AND FINANCE” APPROACH, controlling shareholders have the incen- RECENTLY advocated by La Porta et al. tives and the power to discipline manage- (1998, 2000), emphasizes the important ment. On the other hand, they create the role of laws and institutions protecting conditions for a new agency problem investors for the development of a country. because the interests of controlling and These authors argue that a firm’s ability to minority shareholders are not perfectly raise external capital and grow is limited by aligned. For instance, the controlling share- the extent to which control can be effec- holders can expropriate minority ones via tively separated from ownership without targeted issues and repurchases of securi- increasing the risk that investors are expro- ties, transfers of assets, entrenchment, and priated by management. Better legal pro- exploitation of a business relationship with tection for investors reduces the risk of affiliated companies through transfer pricing. expropriation, allows more separation Johnson et al. (2000) call this form of between ownership and control, and agency problem “tunneling”. increases growth. From a theoretical point of view, As shown in La Porta et al. (1999a), in Bebchuk (1999), Wolfenzon (1998), and several countries “plagued” by low investor Bebchuk et al. (1998) argue that the protection some separation of ownership balance between the two forces (namely, from control is obtained via pyramidal the reduction in managerial discretion groups and nonvoting shares. On the one due to the presence of a controlling hand, these institutions preserve sufficiently shareholder and the potential conflict of 258 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER interest between controlling and minority between voting and cash flow rights (for an shareholders) is likely resolved in favor of example of pyramid, see Figure 1). the second one, leading to a magnification The first finding in the analysis is that rather than a reduction of the agency problem. controlling shareholders are entrenched. This paper provides direct evidence on the Indeed, the probability of turnover and its potential costs of pyramidal groups and sensitivity to performance are significantly nonvoting shares by analyzing executive lower for top executives who belong to the turnover and market valuation in Italian family of the controlling shareholder than listed companies. for other executives. Second, the larger the Italy represents an ideal setting to fraction of cash-flow rights owned by the address these issues because it features controlling shareholder, the more sensitive weak legal protection of creditors and turnover is to performance. This result shareholders, inefficient law enforcement, suggests that incentives matter and that high ownership concentration, and an abun- governance improves when the controlling dance of pyramidal groups and nonvoting shareholder internalizes the consequences shares. There is suggestive evidence that of his actions. The third finding is that the size of the private benefits of control is turnover is more sensitive to performance particularly large in Italy. Zingales (1994) when a voting syndicate controls the firm. finds an average voting premium of 82% in A voting syndicate is a coalition of relevant Italian companies with dual-class shares, shareholders who sign a binding agreement while the average voting premium in the to vote together for a few years. About 15% US is 10%, in the UK is 13%, in Canada of the companies in the sample have a voting 23%, and in Switzerland is 27%. syndicate. These coalitions help the largest The existing literature indicates two shareholder to control a company when his strategies to assess the effectiveness of a stake would not be large enough to do so by corporate governance system.The first one, himself. This result suggests that turnover following Kaplan (1994a) and Coffee becomes more sensitive to performance (1999), is to test whether executive when control is, to some extent, contestable, turnover increases as a firm’s performance as in the case of a voting syndicate. declines. The second one, derived from These findings are confirmed by the Morck et al. (1988) and McConnell analysis of the firm’s Q ratio: Q is signifi- and Servaes (1990), is to analyze the cantly smaller in firms where the control- firm’s valuation in relation with similar ling shareholders are among the top companies. executives, is significantly larger when con- Accordingly, this paper will first study trol is partially contestable as in the case the determinants of executive turnover in of a voting syndicate controlling the firm, Italian publicly traded companies, by and increases with the fraction of cash-flow focusing on how the ownership and control rights owned by the controlling shareholder. structure of a firm affects the sensitivity of Within pyramidal groups, I find a signif- the firm’s executive turnover to perform- icant lower Q ratio (between 13% and ance. Then, it will evaluate the effect of 27%) in firms at the bottom of a pyramid. these same factors on the firm’s Q ratio. This result is consistent with the argument Both analyses are based on a large data that pyramids increase agency problems by set, which covers all traded companies in creating a wedge between voting and cash- Italy (banks and insurance companies flow rights. A possible explanation is that excluded) during the period 1986–1997 good managers are promoted to a higher and contains information on ownership, layer of the pyramidal group against the board, and capital structures. For all com- interests of investors in the firms at the panies in the sample, I was able to trace bottom of the pyramid. Indeed, the rela- back the control chain, identify the ulti- tionship between turnover and performance mate owner, and determine his ownership is weaker in pyramidal groups although the stake in the company, distinguishing difference is not statistically significant. GOVERNANCE WITH POOR INVESTOR PROTECTION 259 Regarding the role of large minority Hence, the basic empirical hypothesis to shareholders, they do not seem to play an be tested in this paper is the following. important monitoring role within the firm. Basic Hypothesis. Top executive turnover is Specifically, the results in this paper negatively related to performance. suggest that minority shareholders have a If the basic hypothesis is verified on the governance role only if their votes are whole sample, the second step is to study necessary to the controlling shareholder whether there are significant differences to control the firm, as is the case in voting across firms. An important factor that may syndicates. Except for their role in a voting affect turnover is the ownership structure syndicate, large minority shareholders do of a firm. not improve the firm governance since Firms on the Italian stock market typi- they do not increase the sensitivity of cally have a clearly identifiable controlling turnover to performance nor the firm’s shareholder who controls at least 20% of valuation. the voting rights. Exceptions are banks and Finally, turnover is much lower in the insurance companies. However, these are company at the top of a pyramid (6%) excluded from the sample because their than in its subsidiaries (16%). This result accounting measures of performance are not may be explained by the fact that the con- directly comparable with the other firms. trolling shareholders of the group sit as Hence, Italian traded companies can be executives of their holding companies and classified into four large categories according they are entrenched in control. They do so to their ultimate owner: the state, a foreign because the benefits of control are larger in company, a set of banks, and one or more the holding company, as suggested by the Italian individuals (I define this last finding that the voting premium in the holding category as family-controlled firms). One companies is significantly higher than in may expect to find differences across these the subsidiaries. groups in the sensitivity of turnover to per- The structure of the paper is as follows. formance. For example, in state-controlled Section 2 formulates the hypothesis to test companies management turnover can be by overviewing the literature and describing more affected by political than economic the Italian corporate governance system. factors. Therefore, I first check whether Section 3 describes the data set. Section 4 there are differences across these contains and discusses the results on the categories. However, in order to use a determinants of top executive turnover and homogeneous and large set of observations, firm valuation. Section 5 extends the analysis I focus the remainder of the analysis on the to pyramidal groups and evaluates the role set of firms that are controlled by Italian of the market for corporate control. The individuals. conclusion is in Section 6.

2. EXISTING LITERATURE AND 2.1. Ownership structure and TESTABLE HYPOTHESES executive turnover Within family-controlled firms, there may Studying top executive turnover and the be significant differences across firms sensitivity of turnover to performance is depending on the relationship between the one way to assess the quality of the corpo- management and the controlling share- rate governance standards within a firm or holder of the firm. Denis and Denis (1994) within a country. The reason is, as argued find that in the US, majority-owned firms by Coffee (1999), that successful gover- experience significantly lower turnover for nance systems penalize managers of firms given performance than widely held ones. with poor stock performance and with low Also, they find that in majority-owned cash flows. This statement is supported by firms the controlling shareholder typically large international evidence.1 sits as top executive of the firm. 260 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Consistent with the result above, Denis legally binding only in the former case. It is et al. (1997) show that the probability of important to notice that the degree of top executive turnover (and the sensitivity entrenchment of the controlling share- of turnover to performance) is negatively holder could be much lower if he or she correlated with the ownership stake held needs a voting syndicate to control the by officers. Most Italian traded companies firm.With a voting syndicate, the controlling are majority owned. Moreover, the size of shareholder does not have a lock on control the private benefits of control is extremely and control is partially contestable. large in Italy, as shown by Zingales An example is given by the turnover in (1994). It is possible that the controlling Olivetti in 1996 (Il Sole 24 Ore, September shareholder is entrenched as a top execu- 4, 1996). Carlo De Benedetti, the long-time tive against the interest of the other Chairman and President of the Board, was shareholders, in order to preserve his ability the relative majority shareholder in Olivetti to extract those benefits. Hence, the with 15% of the voting rights.Thanks to a first main hypothesis to test is the voting syndicate he controlled another entrenchment hypothesis. 25% of the votes. In January 1996, after several years of very poor performance at Entrenchment Hypothesis. Top executive Olivetti, the voting syndicate broke down, turnover is lower and less sensitive to per- and in September of the same year De formance if the controlling shareholder is Benedetti was forced to step down from all an executive. executive roles in the company. This case In Italy, the separation between owner- suggests that executive turnover is more ship and control is enhanced by the wide- sensitive to performance if the controlling spread use of traded pyramids and shareholder does not have absolute control nonvoting shares. The sensitivity of over the company, that is, in the instances turnover to performance can be proportion- where there exists a voting syndicate. ate to the fraction of cash-flow rights However, it is also conceivable that voting owned by the controlling shareholder. One syndicates sustain collusive agreements immediate rationale for this hypothesis is among large families aiming at preserving that monitoring the management may the stability of control. In this second case, come at a cost. Hence, the higher the voting syndicates do not necessarily fraction of cash-flow rights owned by increase turnover or the sensitivity of the controlling shareholder, the larger the turnover to performance. controlling shareholder’s incentive to Pagano and Roell (1998) argue from a monitor the management. theoretical viewpoint that large minority shareholders play a role in monitoring the Incentive Hypothesis. Top executive controlling shareholder.For the US, Denis et turnover is more sensitive to performance if al. (1997) show that the probability of top the controlling shareholder owns a larger executive turnover is positively correlated to fraction of cash-flow rights. the presence of an outside blockholder. In about 15% of Italian traded companies, a coalition of important shareholders Outside Monitoring Hypothesis. Top execu- helps the controlling party control the com- tive turnover is higher and more sensitive to pany. These shareholders are kept together performance in companies with a large by explicit agreements to vote together, minority shareholder and/or a voting which are called voting syndicates (“sindacati syndicate. di voto”). These agreements are publicly An alternative way to test the quality of announced on national newspapers, last for the corporate governance within a firm is to a fixed number of years (usually three) and look at the valuation of the firm. For the can be renewed. A voting syndicate can US, Morck et al. (1988) and McConnell decide on its actions either unanimously or and Servaes (1990) find a nonlinear rela- by majority rule. According to law experts tionship between Q ratio and managerial (see Galgano, 1997), these agreements are ownership. A similar approach is employed GOVERNANCE WITH POOR INVESTOR PROTECTION 261 also in Yermack (1996) and in La Porta Since Italian firms typically have a et al. (1999b). controlling shareholder, a sale of the If the absence of sensitivity of turnover controlling stake is a simple proxy for a to performance is an indicator of bad change in the firm’s ownership structure governance, this should be reflected in the and should be associated with an increase firm’s valuation. Hence, I will test the in top executive turnover.However,since the following hypotheses on the Q ratio. sale can only happen with the consent of the controlling shareholder, control is not Entrenchment Hypothesis. Q is lower if the contestable, and takeovers in Italy can have executives are controlling shareholders. a more limited disciplinary role than in the Incentive Hypothesis. Q is higher if the con- US or UK. trolling shareholder owns a larger fraction Gilson (1989), for the US, and Franks of the cash-flow rights. et al. (2001), for the UK, find that turnover is higher in firms at the onset of a financial Outside Monitoring Hypothesis. Q is higher crisis, when the firm’s creditors increase in firms where there are outside blockhold- their pressure on the management and seize ers and/or a voting syndicate. control. Transfer of the control to creditors should then be associated with an increase in 2.2. Pyramidal groups top executive turnover in the Italian sample. The impact of transfers of control as a The above hypotheses will be tested in determinant of executive turnover will be Section 4 of this paper and will help char- evaluated in Section 5.2. acterize the relationship between executive turnover and ownership structure in Italian firms. However, as mentioned before, many 3. DESCRIPTION OF THE DATA SET firms in Italy (more than half of the firms traded on the Milan stock exchange) are The sample is collected from several issues organized in pyramidal groups. Pyramids of of “Il Taccuino dell’Azionista”, an annual traded firms magnify the separation publication edited by Il Sole 24 Ore. This between ownership and control because source provides basic balance sheet data, they allow the controlling shareholder of information about the ownership structure, the holding company at the top of the and the names of the individuals sitting on pyramid to control the companies in the the Board of Directors (“Consiglio di pyramid by owning a small fraction of their Amministrazione”) of all companies traded capital. By the same token, they magnify on the Italian stock market. Data cover a the potential conflict of interest between period of 12 years, from 1986 to 1997. controlling and minority shareholders. The From the set of all companies traded on the impact of pyramidal groups on executive Milan stock exchange over that period, turnover and firm valuation is studied in I exclude banks, insurance companies, and Section 5.1. foreign companies because of different accounting rules. Foreign companies are 2.3. Transfer of corporate control firms incorporated abroad, while foreign- controlled companies are firms incorpo- Finally, the paper will try to evaluate the rated in Italy even though owned by relative role of external and internal gover- foreigners. The latter ones are included in nance forces. Martin and McConnell the sample. I exclude all companies that (1991), for the US, and Franks and Mayer were traded on the Milan stock exchange (1996), for the UK, find that turnover for less than three years because I need at increases following takeovers. Barclay and least three years of data to compute the Holderness (1991), for the US, and Franks measures of performance and turnover. et al. (2001), for the UK, find a similar Hence, the sample used in the regres- increase in turnover following block trades. sions covers 205 firms and contains a total 262 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Table 10.1 Descriptive statistics: sample of all firms traded on the Milan Stock Exchange The table reports the number of observations, the mean, median, standard deviation, minimum, and maximum for some of the variables used in the analysis. Top executive turnover is a dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1. The sample includes all companies traded on the Milan Stock Exchange, excluding banks, insurance, and pure financial companies, in the years 1987 through 1996.The number of observations is 1,611.

Variable Mean Median Std. Dev. Min. Max.

Ownership structure Fraction of cash-flow rights owned by 38.0 40.5 25.0 0.3 99.4 controlling shareholder (%) Fraction of voting rights controlled by 56.4 53.8 14.7 20.1 100 controlling shareholder (%) Board composition Number of directors 10.4 10 3.3 3 25 Number of top executives 3.34 3 0.68 1 6 Turnover data Fraction of top executives replaced in 16.9 0 25.3 0 100 a year (%) Top executive turnover 0.14 0 0.35 0 1

of 1,611 observations. Table 10.1 reports ultimate owner is the state or a government summary statistics on ownership structure, agency (216 observations). In 55 observations board composition, and executive turnover the controlling party is a group of banks, for this sample. All variables used in the since the company defaulted and the banks analysis are defined in Appendix A. took control. The rest of the observations Italian law limits the extent of cross- (1,234) have one or more private Italian holdings to 2% of voting rights among citizens or an Italian family as the ultimate traded companies. Moreover, controlled owner (I will call this last set family- firms cannot exercise the voting rights controlled firms). eventually owned in their parent company. In Table 10.1, two variables describe the Hence, it is simple to identify the control firm ownership structure. The fraction of chain once ownership data are available for cash-flow rights owned by the largest all companies. An example of a common shareholder is defined as the product of the control structure is represented in Figure fraction of voting rights along the controlling 10.1. The figure shows the structure of the path.This number is corrected for nonvoting Pesenti group at the end of 1995. The shares by assuming that the ultimate owner Pesenti family is defined as the controlling owns none of them. For example in Figure. shareholder because an individual, Rosalia 10.1, the fraction of cash-flow rights owned Radici Pesenti, owns the controlling stake by the Pesenti family in Italmobiliare is in the holding company (44.87%). 29.4%, while the fraction of voting rights The observations have been classified is 44.8% because about a third of into four categories according to the Italmobiliare’ s equity is made up of nonvot- information available on the controlling ing shares. The fraction of cash-flow rights shareholder. A firm is classified as foreign- owned by the Pesenti family in Italcementi controlled in a given year if the ultimate is 9.5%, which is the product of 29.4% (the owner is a foreign company (106 observa- fraction of cash-flow rights directly owned tions); and is defined as state-controlled if the by the family in the holding company GOVERNANCE WITH POOR INVESTOR PROTECTION 263

Pesenti Family

29.41% cash-flow 44.87% voting rights (CF) rights (V)

Italmobiliare Giampiero Pesenti (P and CD)

32.41% CF 54.26% V 62.25% CF and V (9.53% CF) (44.87% V) (18.31% CF and 44.87% V)

Italcementi Franco Tosi Giovanni Giavazzi (P), Giampiero Pesenti (CD), and Jacques Conseil (P and CD), Giampiero Gianfranco Barabani (VP) Pesenti (VP), and Giorgio Falck (VP)

73.72% CF and V 77.57% CF and V (7.03% CF and 44.87% V) (7.39% CF and 44.87% V)

Cementerie Siciliane Cementerie di Sardegna Diego Scotti (P) and Ettore Rossi (VP) Diego Scotti (P) and Ettore Rossi (VP)

Figure 10.1 Structure of the Pesenti’s group as of 12/31/1995.The figure shows the ownership and control structure of the traded companies controlled by the Pesenti family. Each box represents a traded company. Inside each box are the name of the company and the names of the top executives with their role on the board of directors: P President, CD CEO (“Consigliere Delegato”), VP Vice-President. The arrow indicates the direction of control. The numbers above each box represent the percentages of cash- flow (CF) and voting (V) rights directly owned by the controlling party (individual or company). In parenthesis are the fractions of cash-flow and voting rights directly or indirectly owned by the ultimate owner (the Pesenti family).These are computed according to the definitions reported in Appendix A.

Italmobiliare) and 32.4% (the fraction of exercise its voting rights on Italmobiliare. cash-flow rights owned by Italmobiliare in This represents the burden of having an Italcementi). The fraction of voting rights extra layer of control. Even if Italmobiliare controlled by the controlling shareholder is owns 54.26% of the voting rights in computed as the minimum share of voting Italcementi, the Pesenti family owns only rights controlled by the controlling share- 44.87% of the voting rights in holder along the control path. Italmobiliare. Hence, the effective voting The rationale for this definition is power of the Pesenti family in Italcementi explained as follows with the help of Figure is equal to 44.87%. The logic of this 10.1. Let us consider the second layer example simply generalizes to more complex firm Italcementi and try to evaluate the control structures. In the case of a syndi- voting rights of the Pesenti family in this cate among shareholders, the controlling company. The Pesenti family controls party is assumed to control all the voting Italcementi through the holding company rights that belong to the syndicate. Italmobiliare. In order to exercise the As measure of executive turnover, I use voting rights that Italmobiliare owns in the fraction of top executives replaced Italcementi, the Pesenti family needs to within a year.This is preferable to the CEO 264 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER turnover because in Italian firms all top 99.4%. Disciplinary turnover (measured as executives (Presidente, Vice-Presidente, top executive turnover) happens in 14% of and Amministratore delegato) have similar the observations.The median number of top executive power and there is no clear rank- executives in the sample is three, while the ing in authority among them (especially median size of the board is ten directors. between Presidente and Amministratore delegato). Also, some firms do not have a CEO; others have many CEOs. The data 4. EMPIRICAL ANALYSIS have been cleaned of the cases of retirement identifiable through LEXIS-NEXIS In this section, I will test each of the (42 observations are relabeled as non- hypotheses discussed in Section 2, starting turnover in this way). However,Italian com- with the relationship between turnover and panies do not provide information about performance in the whole sample. the age and tenure of their executives in the annual report. Since I was able to collect 4.1. Turnover and performance data on age and tenure for only a subset of the executives, I exclude these two vari- In Table 10.2, the observations are classified ables from the regressions.2 In order to according to the ultimate owner in family-, reduce the biases due to potential mislabel- state-, foreign-, and bank-controlled firms. ing of individual cases of turnover, in the The table shows that turnover is negatively regressions I use as dependent variable a related to performance in the sample as a dummy variable (Top executive turnover) whole. More precisely, the observations are that takes value one in year t when at least sorted into quintiles according to the firm’s half of the top executives are replaced past performance. I then compare turnover between t and t 1. between the worst quintile and the best Two alternative measures of perform- one. In Panel A, performance is proxied by ance are used in the analysis: the change in the change in the ratio of EBIT over assets, the ratio of earnings before interest and while in Panel B is measured by the stock taxes (EBIT) and total assets between return. t 1 and t and the annual stock return In the whole sample, the probability of between t 1 and t. The former is the top executive turnover is significantly lower measure of performance used in Denis and if performance is good than if is bad. In Kruse (2000). The latter one, the stock Panel A, the average turnover increases return, is the standard measure of perform- from 14.5% in best performing firms to ance used in studies of executive turnover 22.5% in worst performing ones. Similarly, since the contributions by Warner et al. in Panel B, turnover increases from 14.5% (1988) and Weisbach (1988). In unre- in best performing firms to 19.3% in worst ported regressions I performed the performing ones. In Panel A, the average analysis with the addition of lagged turnover decreases monotonically as measures of performance without finding performance improves across quintiles with any difference. the significant jump being the one As illustrated in Table 10.1, all compa- between the first and second quintiles. In nies in the sample have a controlling share- Panel B, the average turnover is not monot- holder,that is a shareholder who controls a onic and the third and fourth quintiles are fraction of voting rights larger than 20%. not significantly different from the first The fraction of voting rights held by the one, as far as turnover is concerned. controlling shareholder is on average 56%, When examining turnover across types of while it ranges between 20.1% and 100%. ultimate owner, one finds a negative and There is a large variability in the fraction of significant relationship between turnover and cash-flow rights owned by the controlling performance according to either perform- shareholder.The average is 40%, while the ance measure only for family-controlled minimum is 0.3% and the maximum firms. It is interesting to notice that in GOVERNANCE WITH POOR INVESTOR PROTECTION 265 Table 10.2 Turnover and performance: family-, state-, foreign-, and bank-controlled firms The table reports the average fraction of top executives replaced by quintiles of performance. The observations are sorted in five classes according to their past performance (1 low, 2–5 high). The average turnover is compared between the firms with low performance and those with high performance using a two-tailed t-test.The observations are also classified according to the type of ultimate owner (a family, the state, a foreign company, a bank).

Average fraction of top 1 Low 5 High Test: executives replaced performance 2 3 4 performance (1) (5)

Panel A: Observations are sorted according to change in EBIT/total assets All firms 22.5 16.6 15.4 15.3 14.5 *** [323] [322] [322] [322] [322] Family-controlled firms 21.6 13.5 13.5 13.9 12.6 *** [259] [250] [222] [243] [260] State-controlled firms 34.6 34.1 19.8 17.2 22.4 0 [26] [46] [74] [44] [26] Foreign-controlled firms 20.0 14.9 20.6 22.3 19.9 0 [30] [14] [17] [22] [23] Bank-controlled firms 21.9 18.1 16.7 23.1 28.2 0 [8] [12] [9] [13] [13] Panel B: Observations are sorted according to stock return All firms 19.3 15.3 18.0 17.6 14.4 ** [320] [320] [319] [320] [319] Family-controlled firms 19.1 13.7 15.2 15.4 12.0 *** [238] [245] [246] [254] [239] State-controlled firms 22.9 21.4 31.5 22.4 22.0 0 [35] [37] [54] [42] [47] Foreign-controlled firms 19.6 16.3 13.9 31.9 17.6 0 [28] [21] [12] [18] [25] Bank-controlled firms 14.0 23.0 17.9 33.3 37.5 ** [19] [17] [7] [6] [8]

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. Zero denotes no significant difference. In the first panel performance is measured by the change in EBIT/total assets between t 1 and t; in the second panel performance is measured by the excess stock return between t 1 and t. bank-controlled firms turnover is the types of ultimate ownership. In significantly higher after a good perform- family-controlled firms, the relationship ance than otherwise (when stock return is between turnover and performance is the measure of performance). This result strongly significant and negative. For the can be explained by noticing that bank- remaining firms, the results in Table 10.2 controlled firms are firms that recently are confirmed, although in a weaker sense. defaulted. For troubled companies, the While the relationship between turnover replacement of the management could be a and performance is weak in all but family- positive signal that liquidation can be controlled firms, the regression fails to find avoided and the firm can emerge from significant differences across ultimate owners. reorganization (see, e.g., Gilson et al., 1990). The only exception is the case of bank- Table 10.3 presents the results of two controlled firms for which the sensitivity of regressions that characterize the relation- turnover to performance is significantly ship between turnover and performance different from family-controlled firms when after controlling for year and industry performance is proxied by the stock return. dummies and firm size. Interactive dum- In Table 10.3, the coefficients of the pro- mies allow for different coefficients across bit model are transformed to simplify their 266 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Table 10.3 Turnover and performance: regression analysis of the whole sample Probit regressions.The dependent variable (Top executive turnover) is a dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1. Size is the logarithm of total assets (in millions of Liras). Performance is the change in the ratio of EBIT and total assets between year t 1 and year t in regression (1) and the stock return between t 1 and t in regression (2). Robust standard errors (in parentheses) control for correlation and clustering at firm level. Year and industry dummies are included but their coefficients are not reported.

Performance change Performance stock in EBIT/TA return (1) (2)

Performance 0.646*** 0.063** (0.188) (0.026) Performance * State ownership dummy 0.158 0.008 (0.467) (0.070) Performance * Foreign ownership dummy 0.447 0.055 (0.422) (0.058) Performance * Bank ownership dummy 0.834* 0.394*** (0.481) (0.153) State ownership dummy 0.098*** 0.103*** (0.034) (0.034) Foreign ownership dummy 0.038 0.034 (0.040) (0.040) Bank ownership dummy 0.010 0.051 (0.038) (0.054) Size 0.002 0.004 (0.008) (0.006) Pseudo R2 0.050 0.044 No. of observations 1,611 1,598

Performance 5 change Performance 5 stock in EBIT/TA return (1) (2) (3) (4)

Performance 0.581*** 0.283 0.107*** 0.097*** (0.211) (0.201) (0.032) (0.032) Performance * Owner-manager dummy 0.387 0.279 0.011 0.007 (0.395) (0.413) (0.058) (0.058) Performance * Owner with high incentives dummy 0.505* 0.702** 0.059 0.051 (0.280) (0.295) (0.044) (0.045) Performance * Large minority shareholders dummy 0.136 0.003 (0.361) (0.055) Performance * Voting syndicate dummy 0.741** 0.062 (0.357) (0.060) Owner-manager dummy 0.065*** 0.065*** 0.069*** 0.069*** (0.018) (0.019) (0.018) (0.018) Owner with high incentives dummy 0.002 0.000 0.013 0.012 (0.019) (0.021) (0.021) (0.021) Large minority shareholders dummy 0.018 0.019 (0.020) (0.019) Voting syndicate dummy 0.028 0.012 (0.024) (0.028) Size 0.008 0.007 0.010 0.009 (0.007) (0.007) (0.007) (0.007) Pseudo R2 0.071 0.074 0.056 0.056 No. of observations 1,234 1,234 1,222 1,222

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively.The reported coefficients are transformed to represent the change in probability for an infinitesimal change in each independent variable evaluated at the mean values. consistent with the definition of turnover, The second dummy variable (owner with the owner-manager dummy variable is set high incentives) identifies the companies in equal to one when at least half of the top which the controlling shareholder owns a executives belong to the family of the con- large fraction of the cash-flow rights. More trolling shareholder.This happens in about specifically, I define as such all companies 30% of the observations. The purpose of in which the controlling shareholder owns this first dummy variable is to address the more than 50% of the cash-flow rights, a Entrenchment Hypothesis described in requirement that roughly selects 30% of Section 2. the observations. The purpose of this 268 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER second dummy variable is to address the The results in Table 10.4 are as follows. Incentive Hypothesis. Regressions (1) and (2) show that turnover According to the Outside Monitoring does not always decrease with performance Hypothesis discussed in Section 2, when the latter is proxied by the change in turnover should be more sensitive to earnings. Specifically, in regression (1), a performance when the company has large 10% decrease in earnings over assets minority shareholders.The intuition is that increases the probability of executive large minority shareholders can play an turnover by 5.8%. However, in regression active role in corporate governance. This (2), the increase in the probability of exec- hypothesis is tested in Table 10.4 by using utive turnover for a similar change is only two alternative proxies for minority share- 2.8% and is not statistically significant. holders.The first is an indicator of whether When the controlling shareholder sits as the second largest shareholder owns a top executive of the firm, the relationship fraction larger than 5% of the voting between turnover and performance rights (large minority shareholder). This becomes even weaker, as shown by the pos- happens in about 40% of the observations. itive (although not significant) coefficient The second one identifies cases where the on the first interactive term. The level of firm is controlled by a voting syndicate, as turnover is also significantly lower when a happens in 15% of the observations member of the family of the controlling (voting syndicate). shareholder sits among the top executives. The methodology I use to address the The probability of executive turnover hypotheses above is to run a probit regression decreases with 6.5% when the controlling with the dummies just described entering shareholder is a top executive.These results both additively and in interaction with the are weakly in favor of the Entrenchment measure of performance. Eq. (1) describes Hypothesis. the probit model estimated in Table 10.4: The first two regressions also show that the sensitivity of turnover to performance Pr(Turoverit 1) Φ(t 0 Performanceit increases significantly when the controlling 4 shareholder owns a large stake in the firm. (j Performanceit If the largest shareholder owns more than j1 50% of the cash-flow rights, a 10% j j)Dit Sizeit decrease in earnings over assets increases Φ Industury dummyit turnover by 10%. This result is in favor of it), (1) the Incentive Hypothesis. Regarding the Outside Monitoring where Φ(.) is the cumulative gaussian Hypothesis, regression (1) suggests that distribution, t is the year dummy, and the presence of a large minority share- firm’s size and an industry dummy (firms holder does not increase the sensitivity of are classified in nine industries according turnover to performance. However, regres- to the official classification on the Milan sion (2) shows that a voting syndicate does stock exchange) are introduced as control indeed increase the sensitivity of turnover variables.3 to performance. After a 10% decrease in In the regression, I cannot control for earnings, executive turnover is 7% more firm effects, as with traditional linear models, likely if the firm is controlled via a voting because a fixed-effect estimator is not syndicate than otherwise. The intuitive available in a probit model. Since the obser- explanation for these results is that minority vations are likely not to be independent, the shareholders have the power to play a standard errors are corrected for correlation governance role (and they do so) only when and clustering at the firm level. In an control is not locked in the hands of the unreported regression, I estimated a linear controlling shareholder.That is, only if the OLS model with fixed effects at firm level, controlling shareholder needs a voting finding similar results to the ones reported. syndicate to control the firm. GOVERNANCE WITH POOR INVESTOR PROTECTION 269 Regressions (3) and (4), in which (1988) and McConnell and Servaes performance is proxied by the stock return, (1990). This analysis is performed in do not produce any significant coefficient Table 10.5. The results of four fixed-effect on the interactive terms. The signs on the regressions are represented in the table: coefficients are consistent with the Outside regressions (1) and (2) include firm Monitoring Hypotheses, but against effects; regressions (3) and (4) include the Entrenchment and the Incentive only industry effects but report standard Hypothesis.The results in these two regres- errors corrected for clustering at the firm sions can be explained by the fact that the level. The rationale to present the second stock return is not likely an ideal measure pair of regression together with the first of performance in the sample of Italian one is that controlling for firm effects, as in firms. The stock return is a noisy measure (1) or (2), may reduce the significance of of performance for many companies in the the coefficients because it eliminates most sample because many stocks suffer a lack of the variability in the firm ownership of liquidity and infrequent trades. structure, which is relatively constant across years. As done in Table 10.4, in 4.3. Valuation regressions (1) and (3) the power of minority shareholders is measured by the dummy An alternative procedure to assess the variable large minority shareholders; in efficiency of a governance regime is to eval- regressions (2) and (4) minority shareholders’ uate the impact of ownership structure on power is proxied by the dummy variable firm valuation, as done by Morck et al. voting syndicate.

Table 10.5 Analysis of the firm’s Q ratio The dependent variable is the firm’s Q ratio. Size is the logarithm of total assets (in millions of Liras). Owner-manager is a dummy variable that identifies the cases when at least half of the top executives belong to the family of the controlling shareholder. Owner with high incentives is a dummy variable that identifies the cases when the controlling shareholder owns more than 50% of the cash-flow rights. Voting syndicate is a dummy variable that takes value 1 when the firm is controlled by a voting syndicate. Large minority shareholders is a dummy variable that takes value 1 when the second largest shareholder owns a fraction larger than 5% of the firm’s voting rights. Robust standard errors (in parentheses) control for correlation and heteroskedasticity.Year dummies are included but the coefficients are not reported.

(1) (2) (3) (4)

Owner-manager dummy 0.067* 0.116*** 0.092*** 0.083*** (0.036) (0.044) (0.027) (0.028) Large minority shareholders dummy 0.072 0.000 (0.048) (0.027) Voting syndicate dummy 0.400*** 0.141* (0.117) (0.083) Owner with high incentives dummy 0.026 0.080* 0.015 0.006 (0.052) (0.048) (0.031) (0.032) Size 0.059** 0.068*** 0.092*** 0.092*** (0.025) (0.024) (0.027) (0.027) Fixed effects Industry Industry Firm Firm Adjusted R2 0.199 0.273 0.654 0.656 No. of observations 1,234 1,234 1,234 1,234

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. 270 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER The results on Q match one-to-one those that turnover is significantly higher for obtained in the analysis of turnover: firms other managers (18.4%) than for owner- where turnover is unaffected by perform- managers (6.5%). Similarly, the sensitivity ance are discounted by the market. Indeed, of turnover to past performance is signifi- Table 10.5 shows that Q is significantly cantly higher for other managers than for smaller (between 8% and 11% discount) owner-managers. In Panel B, the turnover for firms in which relatives of the control- of family and non-family executives is com- ling shareholder are among the top pared by running a regression on both sets executives. This is consistent with the of data with a dummy variable (entering results in Table 10.4 and with the both additively and in interaction with per- Entrenchment Hypothesis. Table 10.5 also formance) identifying whether the data on shows that Q is significantly larger when turnover refers to family or other execu- control is partially contestable as in tives.The findings are that both the level of the case in which a voting syndicate con- turnover and its sensitivity to performance trols the firm (between 14% and 40% is significantly lower for family managers. premium).This result is consistent with the In regression (1), a 10% decrease in earn- findings in Table 10.4 and with the Outside ings over assets implies a 6% increase in Monitoring Hypothesis. Finally, the table the probability of turnover for a non-family shows that Q increases only weakly with executive and only a 2% increase for a the fraction of cash-flow rights owned family executive. In regression (2), a real- by the controlling shareholder: if the ization of the stock return equal to 10% controlling shareholder owns more than causes the probability of turnover to 50% of the cash-flow rights, the firms increase by 0.6% for non-family managers trades at a premium estimated between and only by 0.3% for family ones. These 0% and 8%.This final result supports the results suggest that, when a company is in Incentive Hypothesis. Consistent with the trouble, non-family managers are likely to finding in Table 10.4, the dummy for large be replaced, but family managers are likely minority shareholder does not affect the Q to stay, that is, family managers are ratio, as shown by regressions (1) and (3) entrenched. of Table 10.5. 4.5. International comparison 4.4. More on the Entrenchment Hypothesis This section compares the findings on Italy with evidence available on other countries. An alternative test of the Entrenchment Few studies have produced results that are Hypothesis is to measure directly the directly comparable across countries probability of turnover and its sensitivity to because of differences in the econometric performance (for executives who belong to procedures and in the performance and the family of the controlling shareholder turnover measures used. A partially com- and those who do not), and compare them. parable set of coefficients can be obtained This is done in Table 10.6. from Kaplan (1994a, b) for Germany, The executives in the sample are classi- Japan, and the US. In Table 10.7, I repro- fied into two groups: family-executives duce the results in these two papers and (that is, firm’s executives who belong to the add comparable results obtained from the family of the controlling shareholder of the sample of (family-owned) Italian firms. firm) and other executives (that is, execu- The coefficients on Germany are tives who do not belong to the family of the extracted from Kaplan (1994b),Table 10.2. controlling shareholder). For each group, The measure of turnover is the fraction of year, and company, I compute the fraction members of the management board of executives replaced in one year. The replaced in one year. Similarly, turnover in average turnover is reported in the third the Italian sample is measured by the frac- column of Table 10.6, Panel A, which states tion of top executive replaced in one year. GOVERNANCE WITH POOR INVESTOR PROTECTION 271 Table 10.6 Family executives versus other executives Panel A reports the average turnover and the sensitivity of turnover to performance for owner-managers (executives who belong to the family of the controlling shareholder) and other managers. Standard deviations are in parentheses. Performance is measured by the change in EBIT/total assets and by the stock return. Panel B presents the result of an OLS regression with the fraction of executives replaced in a year as dependent variable. Size is the logarithm of total assets (in millions of Liras). Family-executive is a dummy variable that identifies the executives belonging to the family of the controlling shareholder. Robust standard errors (in parentheses) control for correlation and clustering at firm level. Year and industry dummies are included but the coefficients are not reported.

Panel A: Descriptive statistics Number of Average turnover Correlation between turnover and observations (%) performance Change in EBIT stock return

Family-executives 805 6.5 20.061* 20.015 (23.4) Other executives 1,184 18.4 20.119*** 20.053* (29.6)

Panel B: Regression analysis Performance Performance change in EBIT/TA stock return (1) (2)

Performance 0.572*** 0.060*** (0.174) (0.021) Performance * Family executive dummy 0.316* 0.033* (0.171) (0.020) Family executive dummy 0.112*** 0.111*** (0.011) (0.012) Size 0.007 0.009 (0.006) (0.006) Adjusted R2 0.060 0.051 No. of observations 1,989 1,977

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively.

The results on Japan and US are from Italy as much as in the other countries.This Kaplan (1994a),Table 10.2, where turnover finding holds across all performance is measured over a two-year period and measures. Also, the magnitude of the coef- refers to representative directors in Japan ficients is not significantly different across and executive directors in the US. Due to the countries. However, in the Italian sample difference in the time horizon of the two performance explains a much smaller frac- turnover measures, the results are only par- tion of the variability in turnover than in tially comparable. Indeed, both turnover and Germany, Japan and US. The R2 ranges performance are computed over a two-year between 2% and 4% in Italy, between period for Japan and the US, and over a one- 12% and 15% in Japan and the US, and year period for Germany and Italy. between 5% and 10% in Germany. This The results in Table 10.7 show that suggests that, with respect of the other turnover is correlated with performance in countries, executive turnover in Italy is 272 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Table 10.7 International comparison OLS regressions.The dependent variable is the fraction of top executives replaced in a year for Italy and the fraction of members of the management board replaced for Germany. For Japan the dependent variable is the fraction of representative directors replaced in a two-year period and for the US the fraction of executive directors replaced over the same horizon. Stock returns, sales growth, earnings growth, and negative net income dummies are used as performance measures. A separate regression is run for each performance measure where the latter enters contemporaneously and once lagged. For Italy and Germany, one lag is one year; for Japan and the US, one lag is two years. In Panel D for Japan and US, the regression does not contain the lagged negative income dummy. Robust standard errors are in parentheses. Year dummies are included but the coefficients are not reported.

Italy Germany Japan US

A. Stock return Contemporaneous 0.034 0.080* 0.056 0.072** (0.026) (0.046) (0.038) (0.029) Once lagged 0.043** 0.103** 0.100*** 0.081*** (0.022) (0.046) (0.037) (0.030) R2 0.020 0.102 0.139 0.148 B. Sales growth Contemporaneous 0.063*** 0.066 0.177** 0.171*** (0.015) (0.083) (0.076) (0.083) Once lagged 0.064** 0.057 0.064 0.031 (0.022) (0.086) (0.076) (0.050) R2 0.040 0.054 0.133 0.150 C. Change in earnings/assets Contemporaneous 0.160 0.494 0.121 0.208 (0.197) (0.376) (0.460) (0.147) Once lagged 0.636*** 0.205 0.707* 0.219 (0.173) (0.485) (0.447) (0.152) R2 0.030 0.058 0.127 0.130 D. Negative net income Contemporaneous 0.035* 0.095*** 0.148*** 0.074*** (0.020) (0.034) (0.035) (0.028) Once lagged 0.066*** 0.042 (0.022) (0.028) R2 0.026 0.059 0.155 0.137 Mean dependent variable 0.152 0.099 0.285 0.234

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. The results on Germany are from Kaplan (1994b,Table 2, p. 150); the results for Japan and the US are from Kaplan (1994a,Table 2, pp. 526–527). explained by other factors than performance, two extensions. First, more than half of such as ownership and control structure, the companies traded on the Milan stock personal relationships, family ties, and exchange belong to pyramidal groups. It is nonobservable performance measures. interesting to see whether the pyramidal structure affects the dynamics of the executive turnover in specific ways not 5. EXTENSIONS captured by the analysis in Section 4. The second extension explores whether the So far,the paper has evaluated the impact sensitivity of turnover to performance is of the firm’s ownership structure on top due to internal or external governance executive turnover. This section considers forces. GOVERNANCE WITH POOR INVESTOR PROTECTION 273 5.1. Turnover in pyramidal groups pyramidal group (that is, firms that are controlled directly or indirectly by a This section evaluates the impact on company of type 3). The table presents the turnover and valuation of the firm organi- mean of a selection of variables conditional zational structure distinguishing between on the type of ownership structure. stand-alone firms and companies that About 55% of the observations in the belong to pyramidal groups. sample come from pyramidal groups. The Table 10.8 provides detailed information remaining firms in the sample belong to on the organizational structure of family- horizontal groups (42 observations only) or controlled firms. The observations are are stand-alone firms (all the rest). The divided into four categories: (1) first impact of the pyramidal structure is to stand-alone firms and firms that belong to create a wedge between the fraction of horizontal groups (that is, firms that are cash-flow and voting rights owned by the not controlled by any other traded controlling shareholder. As shown in Table company and do not own a controlling 10.8, the separation of voting and cash- stake in any other traded company); (2) flow rights (as measured by the ratio of firms that are at the top of a pyramidal voting to cash-flow rights) increases from group (that is, firms that are not controlled 1.4 at the top of the pyramid to seven at by any other traded company and do con- the bottom of it. The difference between trol at least one other traded company); voting and cash-flow rights in cate- (3) firms that are at level 2 of a pyramidal gories (1) and (2) is due to nonvoting group (that is, firms that are directly con- shares (azioni di risparmio) and voting trolled by a company of type 2); and (4) syndicates. firms that are at level 3 or higher of a

Table 10.8 Separation of ownership from control in family-controlled firms The table reports conditional means.The observations are classified into four categories according to the structure of control.The first category comprises stand-alone firms and firms that belong to horizontal groups, the other three categories are for firms that belong to vertical (or pyramidal) groups. Companies belonging to pyramidal groups are divided in three groups depending on their position in the pyramid. Level 1 are the holding companies, Level 2 are the companies controlled by Level 1 firms, and Level 3 and higher are all the others. Top executive turnover is a dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1. Modified top executive turnover is a dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1 and leave all traded companies of the group to which the firm belongs.

Horizontal Pyramidal Pyramidal Pyramidal groups and stand group: group: group: alone Level 1 Level 2 Level 3

Number of observations 541 173 243 277 Fraction of cash-flow rights owned 49.4 42.2 22.5 6.0 by controlling shareholder (%) Fraction of voting rights controlled 58.2 59.8 52.5 42.6 by controlling shareholder (%) Percentage of relatives of the 38.2 50.1 26.6 7.4 controlling shareholder among the top executives (%) Top executive turnover (%) 12.0 5.8 14.0 15.5 Modified top executive turnover (%) 7.9 4.0 9.5 10.1 Q ratio 1.13 1.01 1.09 1.00 Voting premium (%) 77.2 70.2 59.5 32.5 [Number of observations] [184] [118] [108] [127] 274 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Regarding the top executives, turnover is controlling shareholders entrench significantly lower at the top of pyramidal themselves as executives of the holding groups than in all other firms. At the top of companies is to enjoy higher private a pyramid, the probability that at least half benefits of control. of the executives are replaced in one year Each of the hypotheses suggested by the is less than 6%.This suggests the possibility discussion of Table 10.8 is addressed more of entrenchment. Moreover, the table carefully in Tables 10.9 and 10.10. Table shows that members of the family of the 10.9 tests whether turnover follows controlling shareholder are more likely to different dynamics in firms that belong to sit as top executives in firms at the top of pyramidal groups than in stand-alone a pyramid. Turnover at the bottom of a ones. Table 10.10 does the same for the pyramid is significantly higher (about Q ratio. 16%) than at the top. One possible The regressions reported in Table 10.9 explanation is that the management of show that the sensitivity of turnover to subsidiaries is replaced after good per- performance is not significantly different formance and promoted to higher layers of in pyramidal groups and in stand-alone the pyramidal group. In order to address firms. This result holds independently of this concern, I relabeled as nonturnover all the measure of performance (change in cases in which an executive leaves a com- earnings or stock return) and turnover pany to move to another traded company (top executive turnover or modified top within the same group. I will refer to this executive turnover) used. When looking at variable as modified top executive the sensitivity of turnover to performance turnover. Table 10.8 suggests that within pyramids (by testing the hypothesis turnover according to this measure is sig- that the sum of the coefficients on nificantly lower.Hence, I find some support the regressors Performance and of the hypothesis that the higher turnover Performance * Pyramidal level is equal at the bottom of the pyramid is due to the to zero), turnover is unaffected by per- dynamics in the internal managerial mar- formance at the top of a pyramid and at ket in that well-performing managers are the bottom of it, when the accounting promoted to higher level of the pyramid measure of performance (change in while poor-performing ones are replaced. EBIT) is used. In the internal managerial market, the The weak results found in Table 10.9 firms at the bottom of the pyramid work as suggest that the firm’s organizational a screening device for managers.They help structure does not affect the sensitivity of the controlling shareholder select the best turnover to performance directly. As shown managers and discard the bad ones. An in Section 4, the latter is affected directly important caveat of this result is that I by the firm ownership and control structure cannot control for cases in which the man- as measured by the fraction of cash-flow ager moves to a privately held company rights owned by the controlling shareholder, that belongs to a group, because I do not by the presence of a voting syndicate, and have data on nonlisted firms. by whether the controlling shareholder is Table 10.8 also reports the average Q also a top executive in the firm. In an unre- ratio for each type of ownership struc- ported regression, I added these three indi- ture. Q is significantly lower at the bottom cators to the regressions presented in Table and at the top of pyramidal groups than in 10.9. The result is that the coefficients on stand-alone firms. Interestingly, the Q the variables describing the organizational ratio does not decrease monotonically as structure (namely, the pyramidal level) are one proceeds down the pyramid. On the not significant, while those capturing other hand, the voting premium is higher directly the firm’s ownership and control at the top than at lower layers of the structure are. pyramid (as also shown by Nicodano, Table 10.10 shows that there is a 1998). This suggests that the reason why significant discount (between 13% and 27%) GOVERNANCE WITH POOR INVESTOR PROTECTION 275 Table 10.9 Executive turnover and pyramidal groups Probit regressions.The dependent variable is Top executive turnover in regressions (1) and (2) and modified top executive turnover in regressions (3) and (4). Size is the logarithm of total assets (in millions of Liras). Performance is the change in the ratio of EBIT and total assets between year t 1 and year t in regressions (1) and (3) and the stock return between t 1 and t in regressions (2) and (4). Pyramidal level 1 is a dummy variable that identifies the firms that belong to a pyramidal group and are not controlled by any other traded company. Pyramidal level 2 is a dummy variable that identifies the firms that are directly controlled by a pyramidal level 1 company. Pyramidal level 3 is a dummy variable that identifies the firms that are indirectly controlled by a pyramidal level 1 company. Robust standard errors (in parentheses) control for correlation and clustering at firm level. Year and industry dummies are included but the coefficients are not reported.

Dependent variable: top Dependent variable: modified top executive turnover executive turnover Performance Performance Performance Performance change in EBIT/TA stock return change in EBIT/TA stock return (1) (2) (3) (4)

Performance 0.759*** 0.070** 0.491*** 0.030 (0.201) (0.032) (0.144) (0.020) Performance * Pyramidal level 1 dummy 0.602 0.047 0.479 0.066 (0.550) (0.063) (0.393) (0.044) Performance * Pyramidal level 2 dummy 0.056 0.023 0.263 0.026 (0.395) (0.055) (0.355) (0.037) Performance * Pyramidal level 3 dummy 0.441 0.025 0.221 0.056 (0.537) (0.053) (0.429) (0.039) Pyramidal level 1 dummy 0.055* 0.061** 0.025 0.032 (0.025) (0.025) (0.022) (0.022) Pyramidal level 2 dummy 0.034 0.028 0.027 0.031 (0.028) (0.026) (0.021) (0.021) Pyramidal level 3 dummy 0.050* 0.040 0.040* 0.031 (0.030) (0.030) (0.026) (0.025) Size 0.009 0.013* 0.009 0.013** (0.006) (0.007) (0.006) (0.006) Pseudo R2 0.065 0.049 0.063 0.044 No. of observations 1,234 1,222 1,234 1,222

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. The reported coefficients are transformed to represent the change in probability for an infinitesimal change in each independent variable evaluated at the mean values. in firms at the bottom of a pyramid (those the analysis of the determinants of the in the category “Pyramidal level 3 ”). executive turnover, the source of the This is consistent with the recent theoreti- agency costs created by the separation of cal literature on pyramidal groups (see, ownership and control is to be found in e.g.,Wolfenzon, 1998; Bebchuk et al., 1998). other areas (like, for example, the firm’s Since this difference does not show up in investment policy). 276 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 5.2. The market for corporate turnover to performance increases with a control change in control. For example, in regression (1) the interaction term of performance This section evaluates the importance of and change of control is negative and the external forces in the Italian corporate significant. A 10% decrease in earnings governance system.To do so, I run a simple over assets implies a 13% increase in the regression of turnover on past performance probability of executive turnover if there is where the changes of the controlling a change of control. By contrast, the shareholder are identified by a dummy variable increase in the probability of executive that is introduced both additively and in turnover equals only 4% if there is no interaction with performance. If the market for change of control. In regression (2), where corporate control plays a role in the corporate performance is measured by the stock governance regime, we expect the interactive return, the coefficient on the interaction variable to be negative and significant, as term is non significantly different from zero. the sensitivity of turnover to performance In a regression that is not reported, I find should increase when there is change of control. a higher probability of executive turnover The results are presented in Table 10.11. and sensitivity of turnover to performance As in previous tables, I use the change in in forced changes of control (i.e., those earnings over assets and the stock return as changes of control due to the firm’s default) alternative measures of performance. In than in voluntary ones (i.e., those due to the both regressions, I find that changes of sale of the controlling stake). However, the control increase the level of turnover by difference between forced and voluntary more than 50%. However, the increase in changes of control is not significantly executive turnover due to changes of different from zero. This latter finding can control is largely independent of past be due to the relatively smaller number of performance. Indeed, the sensitivity of cases of forced changes of control. Since

Table 10.10 Internal and external governance forces Probit regressions.The dependent variable (Top executive turnover) is a dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1. Size is the logarithm of total assets (in millions of Liras). Performance is the change in the ratio of EBIT and total assets in regression (1) and the stock return in regression (2). Change of control is a dummy variable that takes value 1 in year t when the firm changes controlling shareholder between t and t 1. Robust standard errors (in parentheses) control for correlation and clustering at firm level. Year and industry dummies are included but the coefficients are not reported.

Performance change in EBIT/TA Performance stock return (1) (2)

Performance 0.426*** 0.080*** (0.152) (0.024) Performance * Change of control dummy 0.882** 0.104 (0.446) (0.076) Change of control dummy 0.547*** 0.592*** (0.064) (0.060) Size 0.004 0.006 (0.006) (0.006) Pseudo R2 0.170 0.161 No. of observations 1,234 1,222

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. The reported coefficients are transformed to represent the change in probability for an infinitesimal change in each independent variable evaluated at the mean values. GOVERNANCE WITH POOR INVESTOR PROTECTION 277 Table 10.11 Q ratio and pyramidal groups OLS regression with fixed-effects. The dependent variable is the firm’s Q. Size is the logarithm of total assets (in millions of Liras). Pyramidal level 1 is a dummy variable that identifies the firms that belong to a pyramidal group and are not controlled by any other traded company. Pyramidal level 2 is a dummy variable that identifies the firms that are directly controlled by a pyramidal level 1 company. Pyramidal level 3 is a dummy variable that identifies the firms that are indirectly controlled by a pyramidal level 1 company. Robust standard errors (in parentheses) control for correlation and heteroskedasticity.Year dummies are included but the coefficients are not reported.

(1) (2) Pyramidal level 1 dummy 0.086 0.013 (0.065) (0.075) Pyramidal level 2 dummy 0.030 0.010 (0.097) (0.114) Pyramidal level 3 dummy 0.268* 0.129* (0.156) (0.070) Size 0.085*** 0.058*** (0.028) (0.027) Fixed effects Firm Industry Adjusted R2 0.656 0.201 No. of observations 1,234 1,234

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. out of the 65 cases of changes of control, shareholder. This result suggests that 16 qualify as forced and 49 as voluntary. governance improves when the controlling Overall, the results in this section suggest shareholder’s objectives are more aligned that the market for corporate control does with those of minority share-holders. Third, not play an important role in the corporate turnover is more sensitive to performance governance regime operating in Italy. The when control is, to some extent, con- main reason is probably that control is not testable, as in the case of a voting syndicate. really contestable, since all firms in the Fourth, aside from the role of a voting sample have a controlling shareholder. syndicate, large minority shareholders do not seem to improve governance. These findings are confirmed by an analysis 6. CONCLUSION of the firms’ Q ratio. I find that Q increases with the fraction of cash-flow rights owned by the This paper provides direct evidence on the controlling shareholder, is significantly smaller performance of the corporate governance for firms in which controlling shareholders are regime operating in countries characterized among the top executives, is significantly larger by low legal protection for investors, firms when a voting syndicate controls the firm, and with large controlling shareholders, and is unaffected by the presence of large minority some separation of ownership and control shareholders. The combined results on execu- created via pyramidal groups and nonvoting tive turnover and Q suggest a one-to-one shares. Studying Italy, a country that shares correspondence between bad governance (that all these features, I find that the probability is, low sensitivity of turnover to performance) of turnover and its sensitivity to performance and low valuation (that is, low Q). are significantly lower for top executives who The paper also provides evidence on belong to the family of the controlling share- pyramidal groups. I find that executive holder than for other executives.This result is turnover is not significantly affected by the evidence that controlling shareholders are firm’s organizational structure. Indeed, the entrenched. Second, turnover is more sensitive sensitivity of turnover to performance is not to performance the larger the fraction of significantly different in stand-alone firms cash-flow rights owned by the controlling and in firms that belong to pyramidal groups. 278 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Within pyramidal groups, turnover is very suggest that the market for corporate control low in the holding company, while it is still plays a very limited corporate governance relatively high in the subsidiaries. This result role. According to Bebchuk (1999), large suggests that pyramidal groups help the con- shareholders stay firmly in control wherever trolling shareholder select the best managers control is extremely valuable. Ultimately, as whereby well-performing managers are pro- argued by La Porta et al. (1999a), control is moted to higher level of the pyramid while valuable because low legal protection for poor-performing ones are replaced. Finally, investors enables the controlling shareholder there is a significant discount in firms at the to enjoy large private benefits. This suggests bottom of a pyramid since the Q ratio is that the solution lies at a political level in new between 13% and 27% lower in those firms legislation to improve the quality and the if compared to similar standalone companies. enforcement of investor protection. The This is consistent with the recent theoretical so-called Draghi Reform introduced in Italy literature on pyramidal groups, suggesting in 1998 effectively increased the protection that the separation of ownership and control of minority shareholders and may be one step created by these organizational structures is in this direction. likely to generate large agency problems The diffusion among firms of codes of because of the conflict of interests between best practice is another way to improve controlling and minority shareholders. corporate governance. Their effectiveness What can be done to improve corporate has been testified by a recent paper on the governance in a country like Italy? The UK. Employing a methodology similar finding that voting syndicates increase the to the one in this paper,Dahya et al. (2000) sensitivity of turnover to performance and show that the Cadbury report in the United increase the firm’s Q suggests that the Kingdom improved corporate governance creation of a more competitive and active by increasing executive turnover and its market for corporate control may be the sensitivity to performance, especially in the answer. However, an effective market for firms that adopted the code. A similar corporate control requires control to be con- phenomenon could take place in Italy and testable, otherwise control can be transferred other countries, given the increasing atten- only with the consent of the controlling share- tion to corporate governance throughout holder or if the company defaults. Consistent the world and the growing pressure from with this view, the results in this paper the international capital markets.

APPENDIX A

Definition of the variables used in the analysis.

Size Logarithm of total assets expressed in billions Liras. Sizeit 0.5TAit 0.5TAit1 Change in EBIT Change in earnings before interest and taxes normalized by size: Change in EBITit (EBITit EBITit1)/Sizeit Stock return Stock return between year-end t-1 and t Stock returnit (Pit Pit1)/(0.5Pit 0.5Pit1) Dividend yieldit.To reduce the impact of outliers, I have set the excess return equal to 1 when smaller than 1 and equal to 2 when larger than 2, for a total of 23 changes Cash-flow rights of the Fraction of the firm’s equity (voting and nonvoting controlling shareholder shares) owned by the ultimate owner of the firm. If a firm A is controlled indirectly via another traded firm B, the fraction of cash-flow rights of A owned by the controlling shareholder is equal to the product of the cash-flow rights owned by the controlling shareholder in B times the GOVERNANCE WITH POOR INVESTOR PROTECTION 279 fraction of cash-flow rights owned by firm B in firm A. This algorithm can be generalized to more layers of controls and more complex control structures Voting rights of the Fraction of the shares with voting rights of a company controlling shareholder controlled by its ultimate owner. If a firm A is controlled indirectly via another traded firm B, the fraction of voting rights of A in the hands of the controlling shareholder is equal to the minimum between the voting rights owned by the controlling shareholder in B and the voting rights owned by firm B in firm A. This algorithm can be generalized to more layers of controls and to more complex control structures Ultimate owner indicator Indicator that classifies the firm in four groups: family- controlled firms, state-controlled firms, foreign- controlled firms, and bank-controlled firms Fraction of top executives Fraction of top executives of the firm at year t who are replaced not executives of the firm any more at year t 1. All discovered cases of retirement have been excluded Top executive turnover Dummy variable that takes value 1 in year t if the fraction of top executives replaced is at least one half Modified executive turnover Dummy variable that takes value 1 in year t if at least half of the top executives are replaced between t and t 1 and at t 1 do not hold any executive position in any traded companies of the group to which the firm belongs Family-executives Firm’s top executives who belong to the same family of the controlling shareholder of the firm Owner-manager Dummy variable that takes value 1 if the fraction of relatives of the controlling shareholder among the top executives is at least 0.5 Owner with high incentives Dummy variable that takes value 1 if the controlling shareholder owns a fraction of cash-flow rights in the firm larger than 50% Large minority shareholder Dummy variable that takes value 1 if there is a minority shareholder with at least 5% of the voting shares Voting syndicate Dummy variable that takes value 1 if a coalition of relevant shareholders is held together in a voting syndicate Pyramidal level 1 Dummy variable that takes value 1 if the firm belongs to a pyramidal group and is not controlled by any other traded company Pyramidal level 2 Dummy variable that takes value 1 if the firm is directly controlled by a “Pyramidal level 1” company Pyramidal level 3 Dummy variable that takes value 1 if the firm is directly or indirectly controlled by a “Pyramidal level 2” company Q ratio The ratio of market value of the firm ( market value of equity book value of debt) over book value of total assets Change of control Dummy variable that takes value 1 if the firm’s controlling shareholder changes between t and t 1 Voting premium It is defined only for firms with both voting and nonvoting shares: the percentage difference in the price of a voting share with respect to a nonvoting one

APPENDIX B

The matrix of correlations is given in Table 10.12. 280 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Table 10.12 Matrix of correlations

Top Size Change Stock Owner- High- Voting executive in EBIT/ return manager incentive syndicate turnover TA dummy dummy dummy

Size 0.043 Change in 0.135*** 0.002 EBIT/TA Stock return 0.049* 0.102*** 0.163*** Owner-manager 0.104*** 0.041 0.019 0.015 dummy High-incentive 0.006 0.202*** 0.001 0.068** 0.145*** dummy Voting syndicate 0.021 0.073** 0.018 0.004 0.116*** 0.219*** dummy Large minority 0.030 0.118*** 0.012 0.030 0.043 0.089*** 0.306*** dummy

*, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. The number of observations is 1,611.

NOTES infinitesimal change of each independent variable evaluated at the mean values of the data. * I thank an anonymous referee, Julian Franks, Rafael La Porta, Marco Pagano, Henri Servaes, REFERENCES Andrei Shleifer, and participants at seminars at Harvard University, London Business School, and London School of Economics for helpful Barclay, M., Holderness, C., 1991. Negotiated comments. I also thank Richard Frost and block trades and corporate control. Journal Samanta Padalino for editing suggestions, and of Finance 46, 861–878. Marcello Bianchi for providing some of the data. Bebchuk, L., 1999. A rent-protection theory of I acknowledge financial support from a National corporate ownership and control. NBER Science Foundation and a Luciano Jona graduate working paper, No. 7203. fellowships, and the JP Morgan Chase research Bebchuk, L., Kraakman, R., Triantis, G., 1998. fellowship at London Business School. Stock pyramids, cross-ownership, and dual 1 The first studies on US data are Coughlan class: the creation and agency costs of and Schmidt (1985), and Warner et al. (1988); on Japan, Kaplan and Minton (1994) and Kang separation between control and cash flows and Shivdasani (1995); on Germany, Kaplan rights. NBER working paper, No. 6951. (1994b), and Franks and Mayer (2001); on the Coffee, J., 1999. The future as history: the UK, Franks and Mayer (1996), and Franks prospects for global convergence in corpo- et al. (2001); on Belgium, Renneboog (2000); rate governance and its implications. on eight developing countries, Gibson (1999). Northwestern University Law Review 93, 2 When the econometric analysis is performed 641–708. on the subset of observations with complete data Coughlan, A., Schmidt, R., 1985. Executive on age and tenure, representing about a third of compensation, managerial turnover, and firm the total sample, I find no change in the main performance: an empirical investigation. results and I find that the coefficients on age and tenure are not significantly different from zero. Journal of Accounting and Economics 7, 3 As already mentioned, the coefficients 43–66. reported in the tables are not the , , , and Dahya, J., McConnell, J., Travlos, N., 2000. The ø in Eq. (1) because they have been transformed Cadbury committee, corporate perform- to represent the change in probability for an ance and top management turnover. GOVERNANCE WITH POOR INVESTOR PROTECTION 281 Unpublished working paper, Cardiff Kaplan, S., Minton, B., 1994. Appointments of University, Cardiff, UK. outsiders to Japanese boards: determinants Denis, D., Denis, D., 1994. Majority owner- and implications for managers. Journal of manager and organizational efficiency. Financial Economics 36, 225–258. Journal of Corporate Finance 1, 91–118. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Denis, D., Kruse,T., 2000. Managerial discipline Vishny, R., 1998. Law and finance. Journal of and corporate restructuring following Political Economy 101, 678–709. performance declines. Journal of Financial La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Economics 55, 391–424. 1999a. Corporate ownership around the Denis, D., Denis, D., Sarin, A., 1997. Ownership world. Journal of Finance 54, 471–517. structure and top executive turnover. Journal La Porta, R., Lopez-de-Silanes, F., Shleifer,A., of Financial Economics 45, 193–221. Vishny, R., 1999b. Investor protection Franks, J., Mayer,C., 1996. Hostile takeovers in and corporate valuation. Unpublished the UK and the correction of managerial working paper, Harvard University, failure. Journal of Financial Economics 40, Cambridge, MA. 163–181. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Franks, J., Mayer, C., 2001. Ownership and Vishny, R., 2000. Investor protection and control of German corporations. Review of corporate governance. Journal of Financial Financial Studies 14, 943–977. Economics 58, 3–27. Franks, J., Mayer,C., Renneboog, L., 2001.Who Martin, J., McConnell, J., 1991. Corporate disciplines management in poorly performing performance, corporate takeovers and companies? Journal of Financial management turnover.Journal of Finance 46, Intermediation 10, 209–248. 671–687. Galgano, F., 1997. Diritto Commerciale. McConnell, J., Servaes, H., 1990. Additional Zanichelli, Bologna, Italy. evidence on equity ownership and corporate Gibson, M., 1999. Is corporate governance value. Journal of Financial Economics 26, effective in emerging markets? Unpublished 595–612. working paper, Federal Reserve Board, Morck, R., Shleifer, A., Vishny, R., 1988. Washington, DC. Management ownership and market valua- Gilson, S., 1989. Management turnover and tion. Journal of Financial Economics 20, financial distress. Journal of Financial 293–315. Economics 25, 241–262. Nicodano, G., 1998. Corporate groups, dual- Gilson, S., Kose, J., Lang, L., 1990. Troubled class shares and the value of voting rights. debt restructurings: an empirical study of Journal of Banking and Finance 22, private reorganizations of firms in default. 1117–1137. Journal of Financial Economics 27, Pagano, M., Roell, A., 1998.The choice of stock 315–353. ownership structure: agency costs, monitor- Johnson, S., La Porta, R., Lopez-de-Silanes, F., ing, and the decision to go public. Quarterly Shleifer, A., 2000. Tunneling. American Journal of Economics 113, 187–275. Economic Review Papers and Proceedings Renneboog, L., 2000. Ownership, managerial 90, 22–27. control and the governance of companies Kang, J., Shivdasani, A., 1995. Firm perform- listed on the Brussels stock exchange. ance, corporate governance, and top executive Journal of Banking and Finance 24, turnover in Japan. Journal of Financial 1959–1995. Economics 38, 29–58. Warner, J., Watts, R., Wruck, K., 1988. Stock Kaplan, S., 1994a. Top executive rewards and prices, event prediction, and event studies: an firm performance: a comparison of Japan and examination of top management restructur- the United States. Journal of Political ing. Journal of Financial Economics 20, Economy 102, 510–546. 461–492. Kaplan, S., 1994b. Top executives, turnover, and Weisbach, M., 1988. Outside directors and firm performance in Germany. Journal of Law, CEO turnover. Journal of Financial Economics and Organization 10, 142–159. Economics 20, 431–460. 282 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Wolfenzon, D., 1998. A theory of pyramidal Journal of Financial Economics 40, ownership. Unpublished working paper, 185–211. Harvard University, Cambridge, MA. Zingales, L., 1994.The value of the voting right: Yermack, D., 1996. Higher market valuation of a study of the Milan stock exchange experi- companies with a small board of directors. ence. Review of Financial Studies 7, 125–148. Erratum Paolo F.Volpin* Erratum to ‘Governance with poor investor protection: evidence from top executive turnover in Italy’ [Journal of Financial Economics 64 (2002) 61–90†]

Due to a mistake an error was made The corrected Table 10.5 is on the in Table 10.5 on p. 269. Two minus- following page. signs were accidentally put in the The Publisher apologizes for the row of “Voting syndicate dummy”. inconvenience caused.

* Tel.: 44-20-72625050; fax:44-20-77243317. E-mail address: [email protected] (P.F.Volpin). †PII of the original article S0304-405X(02)00071-5 Table 10.5 Analysis of the firm’s Q ratio

The dependent variable is the firm’s Q ratio. Size is the logarithm of total assists (in millions of Liras). Owner-manager is a dummy variable that identifies the cases when at least half of the top executives belong to the family of the controlling shareholder. Owner with high incentives is a dummy variable that identifies the cases when the controlling shareholder owns more than 50% of the cash-flow rights. Voting syndicate is a dummy variable that takes value 1 when the firm is controlled by a voting syndicate. Large minority shareholders is a dummy variable that takes value 1 when the second largest shareholder owns a fraction larger than 5% of the firm’s voting rights. Robust standard errors (in parentheses) control for correlation and heteroskedasticity. Year dummies are included but the coefficients are not reported.

(1) (2) (3) (4)

Owner-manager dummy 0.067* 0.116*** 0.092*** 0.083*** (0.036) (0.044) (0.027) (0.028) Large minority shareholders dummy 0.072 0.000 (0.048) (0.027) Voting syndicate dummy 0.400*** 0.141* (0.177) (0.083) Owner with high incentives dummy 0.026 0.080* 0.015 0.006 (0.052) (0.048) (0.031) (0.032) Size 0.059** 0.068*** 0.092*** 0.092*** (0.025) (0.024) (0.027) (0.027) Fixed effects Industry Industry Firm Firm Adjuster R2 0.199 0.273 0.654 0.656 No. of observations 1,234 1,234 1,234 1,234

*, **, and ***, denote significance at the 10%, 5%, and 1% level, respectively. Chapter 11

Luc Renneboog OWNERSHIP,MANAGERIAL CONTROL AND THE GOVERNANCE OF COMPANIES LISTED ON THE BRUSSELS STOCK EXCHANGE

Source: Journal of Banking & Finance, 24(12) (2000): 1959–1995.

ABSTRACT This paper examines how corporate control is exerted in companies listed on the Brussels Stock Exchange. There are several alternative corporate governance mechanisms which may play a role in disciplining poorly performing management: blockholders (holding companies, industrial companies, families and institutions), the market for partial control, debt policy, and board composition. Even if there is redundancy of substitute forms of discipline, some mechanisms may dominate. We find that top managerial turnover is strongly related to poor performance measured by stock returns, accounting earnings in relation to industry peers and dividend cuts and omissions.Tobit models reveal that there is little relation between ownership and managerial replacement, although industrial companies resort to disciplinary actions when performance is poor.When industrial companies increase their share stake or acquire a new stake in a poorly performing company, there is evidence of an increase in executive board turnover, which suggests a partial market for control. There is little relation between changes in ownership concentration held by institutions and holding companies, and disciplining. Still, high leverage and decreasing solvency and liquidity variables are also followed by increased disciplining, as are a high proportion of non-executive directors and the separation of the functions of CEO and chairman.

1. INTRODUCTION corporate governance – the Dutch Peeters report (1997), the French Viénot report WHEREAS IN ANGLO-AMERICAN COUNTRIES, (1995) and UK Cadbury report (1992) – managerial performance is maintained by the Belgian policy recommendations of the complementary intervention of both 1998 by the Stock Exchange Commission, internal and external control mechanisms the Association of Employers (VBO) and (see Shleifer and Vishny, 1997, for an the Commission for Banking and Finance overview), the disciplinary function of the focus on the effectiveness of internal (hostile) take-over market in Belgium, and corporate control mechanisms.1 most other Continental European countries, This paper investigates whether or not is limited. Recent Belgian legislative poor corporate performance triggers board changes with regard to ownership disclosure restructuring and whether disciplinary laws and anti-take-over procedures have actions are initiated by internal gover- further reduced the likelihood of take-overs nance. This paper also examines whether as a corporate control mechanism. the accumulation of shares into large Consequently, as in recent codes of good blocks of shares mitigates the problems of 286 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER free riding in corporate control, permitting incumbent management in the UK even in control to be exerted more effectively. The the wake of poor performance. relation between the nature of ownership Consistent with Shleifer and Vishny and incidence of disciplinary turnover when (1986) and Grossman and Hart (1980), corporate performance is poor is also studied. we find that higher board turnover is posi- Besides ownership concentration, capital tively correlated with strong concentration structure choice may be an instrumental in ownership which limits free riding on monitoring variable as it can be a bonding control. Still, this relation is limited to device triggering corporate control actions. industrial and commercial companies and Such creditor monitoring is expected to be family shareholders. Considering that the intensified in case of low interest coverage ownership structure is typically complex and low liquidity. with stakes held through multiple tiers of We also analyse whether a market for ownership, we find that the decision to sub- share stakes arises. In Continental Europe, stitute top management of poorly perform- such a market might play a role equivalent ing companies is taken by ultimate to the role of external markets in the UK shareholders (industrial companies and and the US. If a company underperforms, families) who control either directly or able monitors can increase their voting indirectly, via affiliated companies, a large rights to reach a control level allowing percentage of the voting rights. However, them to nominate a new management neither large institutional investors nor team. holding companies seem to be involved in We find that poor company performance active corporate monitoring, which further precedes increased board restructuring questions the role and need for ownership (turnover of executives, of the management cascades involving holding companies. committee and of CEO and executive chair- Although, an active market in share man). This is consistent with findings stakes exists, it is only weakly related to reported by, among others, Denis and Denis performance. Specific shareholder classes (1995) and Warner et al. (1988) for the (industrial and commercial companies) US, by Franks and Mayer (1998) and with superior monitoring abilities or with Kaplan (1994) for Germany and by Franks private benefits of control, increase their et al. (1998) for the UK. voting stake to better position themselves to The composition of the board also has an replace management. Such a market for important impact on the internal corporate blocks of control also exists in the UK and control system. A high fraction of non- in Germany, as detailed in Franks et al. executives on the board and the separation (1998) and Franks and Mayer (1998). of the functions of CEO and (non-executive) Shareholders who increase their holdings do chairman increases the turnover of so with a clear intention to assume an active executive directors of underperforming monitoring role since management turnover companies. Weisbach (1988) also reports significantly increases in subsequent periods. that outside directors of US firms play a We also find that high leverage and low larger role in monitoring management than interest coverage are related to increased inside directors. Franks and Mayer (1998) board restructuring which suggests that show that, in German companies with con- creditors intervene as the risk of financial centrated ownership, supervisory board distress increases. However, because this representation goes hand in hand with own- interpretation is not corroborated in inter- ership or large shareholdings. For Japan, views with monitors; liquidity and solvency- Kaplan and Minton (1994) show that related indicators may act as monitoring board appointments of directors represent- triggers for directors or shareholders. ing banks and corporations are followed by Finally, management replacement is increases in top management turnover. In followed by modest improvements in contrast, Franks et al. (1998) report that growth of dividends per share over a period non-executive directors seem to support of two years after turnover. However, board OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 287 turnover is followed by decreases in earnings. of the CEO and of the management The earnings decline may result from new committee when performance is poor. management’s decision to expense large Separating the functions of CEO and chair- costs while earnings reductions can still be man facilitates disciplining of underper- attributed to predecessors, thus lowering forming management, and such dual the benchmark and allowing for substantial control should lead to higher turnover. improvements in subsequent years (Murphy Hypothesis 3. (a) When performance is and Zimmerman, 1993). poor, the presence of large shareholdings The remainder of this paper is organised is followed by higher board turnover. as follows. Section 2 explains the hypotheses. (b) However, disciplining of underperform- Section 3 presents the data and methodo- ing management is accomplished by logy. Section 4 provides stylised facts about those large shareholders with superior the ownership structure in Belgian listed monitoring abilities. Conflicts of interest companies and Section 5 discusses the main dissuade institutions to monitor whereas results of the governance models. Finally, holding companies, industrial companies, Section 6 summarises the findings. and families and individuals discipline management. 2. RELATIONSHIP BETWEEN Hypothesis 4. Managerial disciplining DISCIPLINING AND ALTERNATIVE decisions are taken by the decision maker at the top of an investor group GOVERNANCE MECHANISMS pyramid, called ‘ultimate or reference’ shareholder. Few of the tasks which good corporate governance consists of, like strategy devel- Hypothesis 5. In companies without opment or control, are visible to non-insiders sufficiently large shareholders or with to the corporation. Minutes of board or shareholders who take a passive stance committee meetings or the outcome of concerning monitoring, poor performance shareholder-management meetings are not gives rise to changes in the ownership disclosed. Hence, one of the few occasions structure. Hence, increases in share- to study corporate control actions (or the holdings are associated with higher mana- lack of them) is poor corporate perform- gerial turnover in the same year or the ance or a financial crisis.The paper studies year following the monitors’ disciplinary several substitute forms of discipline and, actions. where there is redundancy, whether some Hypothesis 6. Management of poorly per- 2 This forms dominate others consistently. forming companies with high leverage and section provides an overview of the poor liquidity and solvency face increased hypotheses after which each of these are monitoring. further expanded. Hypothesis 7. Management and board Disciplining of top manage- Hypothesis 1. restructuring, triggered by poor performance, ment is triggered by poor company results in improvements of company performance: directors, CEOs, top man- performance, but performance improve- agers and executive chairmen are replaced ments are not expected in the year of following poor share price performance management substitution but are expected and/or low accounting earnings and in later years. dividend cuts and omissions. Hypothesis 2. The greater the proportion 2.1. Corporate performance and of non-executive directors, the lower potential disciplinary corporate governance board domination by management and the actions higher the monitoring ability of the non-executive directors. This is reflected in To the extent that share price and increased turnover of executive directors, accounting returns are influenced by the 288 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER quality of managerial inputs and actions, incentives to develop reputations as corporate performance provides useful decision control experts whose human information on managerial performance capital depends on performance (Fama (Joskow and Rose, 1994). However, both and Jensen, 1983). Consequently, directors market prices and accounting data present themselves face an external labour measurement problems of managerial market which provides some form of quality. On one hand, the relation between disciplining for passive leadership, as (executive) board restructuring and share reported for the US by Kaplan and price performance may be weaker because Reishus (1990) and Gilson (1990). share prices already incorporate market Separating the role of CEO and of non- expectations regarding managerial replace- executive chairman is also supposed to ment. On the other hand, accounting data strengthen the board’s monitoring ability can (temporarily) be manipulated by the since a non-executive chairman could choice of accounting policies (see e.g. ensure more independence from manage- Moses, 1987;Teoh et al., 1998).Therefore, ment.3 Consequently, we expect both a the impact of both share price returns, and high proportion of non-executive directors levels of and changes in operating and net and the separation of the functions of CEO accounting earnings, on turnover are and chairman to be positively correlated included in testing Hypothesis 1. Besides with turnover (Hypothesis 2). share price and earnings performance, we also examine dividend changes. Such changes may be an important critical 2.3. Ownership concentration, the performance measure as management is costs of free riding on control and generally reluctant to reduce dividends superior monitoring abilities unless a reduction is unavoidable Monitoring management may be prohibi- (Michaely et al., 1995). Consequently, tively expensive for small shareholders as a dividend cuts or omissions are associated monitor pays all the costs related to with unusually poor stock price and earn- his control efforts but only benefits in ings performance (Healey and Palepu, proportion to his shareholding (Grossman 1988) and are expected to be negatively and Hart, 1980, 1988; Demsetz, 1983). In related to turnover. contrast, the costs of shirking are shared by all the shareholders. Therefore, 2.2. The impact of board composition monitoring will only be cost effective if a and structure on the board’s ability to single party becomes large enough to monitor performance internalise the costs of corporate control (Hypothesis 3a). A balanced board including both The incentives to monitor and correct executives and non-executives reduces managerial failure depend not only on the the potential conflicts of interest among concentration of ownership, but also on its decision makers and residual risk nature (category of shareholder). Specific bearers. It also reduces the transaction classes of owners may value control or agency costs associated with the differently as the source of the control separation of ownership and control premium is the additional compensation (Williamson, 1983). There are several and perquisites the controlling security reasons why non-executives are (ex ante) holders can accord themselves (Jensen and expected to exert a control task. Non- Meckling, 1976). Barclay and Holderness executives are legally bound to monitor (1989) argue:“In absence of private gains, due to their fiduciary duty. Moreover, in blocks of shares ought to be sold at a an equity market with strong ownership discount due to the greater risk exposure concentration, many non-executives are and due to the monitoring costs. However, appointed by and represent large share- blocks are usually sold at a premium which holders. Thus, non-executives have suggests the presence of private gains”. OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 289 That different classes of owner have (Hypothesis 4). Hence, the relation between different abilities to extract control rents is turnover and direct ownership (voting rights) empirically supported for the US by by category of owner is expected to be less Demsetz and Lehn (1985), Barclay and statistically significant than the one between Holderness (1991) and Holderness and turnover and ownership concentration Sheehan (1988). Holding companies are whereby the direct equity stakes (voting prevalent in Belgium and their private rights) are reclassified based on the share- benefits and reasons for control accumulation holder category of the ultimate owner. are manifold: capturing tax reductions by facilitating intercompany transfers, reducing transaction costs by offering economies of 2.5. The disciplining role of the scale or by supplying internal sources of market for share stakes funds (Banerjee et al., 1997). Likewise, Burkart et al. (1997) argue that the degree corporate shareholders may hold substantial of voting right concentration acts as a share stakes in a target that may be a commitment device to delegate a certain supplier or customer, in order to influence degree of authority from shareholders to and/or capitalise on the target’s strategic management. They show that the use of decisions. In contrast, there is little or no equity implements state-contingent control: systematic evidence of monitoring actions in states of the world with decreasing by institutions (investment funds, banks, corporate profitability, close monitoring insurance companies ...).In Belgium, resulting from strong ownership concentra- many institutions are affiliated with tion is desirable. In other states of the financial institutions and are legally obliged world, it may not be optimal to have close to avoid conflicts of interest (Renneboog, monitoring as this may reduce managerial 1997). No such impediments hinder moni- discretion and hence management’s effort toring by holding companies, industrial and (also in Bolton et al., 1998). Hence, when commercial companies, individual investors performance is poor, a partial corporate or families. We therefore expect a positive control market may arise, consisting of relation between turnover and ownership large (controlling) blocks. Furthermore, concentration held by holding companies, poor performance may reflect not simply industrial and commercial firms, individuals poor management but also ineffective and families and no relation between monitoring and control. If this is the case, turnover and institutional shareholder share poor performance may lead low quality concentration (Hypothesis 3b). monitors to sell their stakes and new (controlling) shareholders could improve 2.4. Ultimate ownership and dilution future corporate performance by substituting of control incumbent management (Hypothesis 5). Shleifer and Vishny (1986) show that once Ownership structures are frequently a block of shares is assembled, the position complex and pyramidal, and are con- is unlikely to be dissipated. It is in the large structed for reasons of control leverage shareholder’s interest to wait until (Wymeersch, 1994). Therefore, decisions someone who values control expresses about disciplining management may not be interest in this block because if the block is taken by direct investors but rather by the broken up and sold on the open market, ultimate shareholders4 who control these part of the firm’s value arising from the direct shareholders directly or through possibility of value-increasing monitoring multiple tiers of ownership. Monitoring is is lost. not performed by intermediate holding companies which are investment vehicles of 2.6. Leverage as a bonding device controlling industrial companies or individuals and families, but by these industrial Creditor intervention may be expected companies and families themselves when the probability of defaulting on debt 290 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER covenants increases or when the company 3. DATA AND METHODOLOGY needs to be refinanced. The choice of gearing can be considered as a bonding 3.1. Data sources mechanism for management (e.g. in Aghion and Bolton, 1992; Berkovitch et al., 1997) 3.1.1. Sample description such that high turnover is positively related to high gearing (Hypothesis 6). Dennis and The sample consists of all Belgian companies Dennis (1993) infer creditor monitoring listed on the Brussels Stock Exchange from the fact that high leverage combined during the period 1989–1994. In 1989 and with managerial ownership improves 1994, respectively, 186 and 165 companies shareholder returns. were listed.6 Bankrupt companies and IPOs over the period 1989–94 were included until the year of bankruptcy and from the 2.7. Post-disciplining corporate year of floatation.7 About 40% of the performance Belgian listed companies are holding com- For internal and external control mecha- panies with multi-industry investments, 13 nisms to be effective, the replacement of percent are in the financial sector (banking, underperforming top management should insurance and real estate) and 47% are be followed by performance improvements industrial or commercial companies. (Dennis and Dennis, 1995) (Hypothesis 7). However, it is unclear which performance 3.1.2. Ownership data variables are expected to improve. As anticipations about future performance of Data on the ownership structure over the a new management team will be reflected period 1989–1994 were collected from the in share price returns at the latest at Documentation and Statistics Department the announcement of the replacement, of the Brussels Stock Exchange. Ownership abnormal returns over periods subsequent data are only available since 1989, follow- to the announcement effect are not ing the introduction of the Ownership expected to be significantly positive. Disclosure Legislation (of 2 March 1989). Furthermore, Murphy and Zimmerman To capture a company’s ownership position (1993) conclude that ‘earnings manage- at the end of its fiscal year and the yearly ment’ 5 is more likely to occur if the out- changes in shareholdings, about 5000 going CEO is terminated following poor hardcopy Notifications of Ownership performance since it is more credible for Change from 1989 till 1994 were consulted. the new CEO to blame the previous CEO With this information about major direct for past mistakes. Moreover,by constantly shareholdings and about indirect control overstating losses attributable to prede- which is complemented with details from cessors, management improves account- annual reports, the multi-layered (pyramidal) ing expectations about the future and ownership structures were reconstructed lowers the benchmark against which its for each company over the period own accounting performance will be 1989–1994. As different classes of share- measured (Elliott and Shaw, 1988). holders may have different information, Hence, performance improvements are not monitoring competencies and incentives, all expected in the year of management shareholders with stakes of 5 percent or substitution but potentially only in later more are categorised into 8 classes: (i) time periods. A competing hypothesis holding companies, (ii) banks, (iii) investment states that if performance leading to companies (pension funds, investment management replacement is poor, the funds), (iv) insurance companies, (v) industrial success of managerial disciplining may not and commercial companies, (vi) families just be inferred from performance and individual investors, (vii) federal or improvements but rather from the avoidance regional authorities, (viii) realty investment of bankruptcy. companies. The yearbooks of Trends OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 291 20,000, which comprise industry sector 3.2. Methodology classification and financial data for most listed and non-listed Belgian companies, were A panel of data is formed for the six year used to classify all Belgian investors into period 1989–94 with each firm-year repre- ownership categories. Foreign investors were senting a separate observation.The relation classified with information from Kompass. between board restructuring, performance, ownership, leverage, board structure is examined in the following model: 3.1.3. Share price and accounting data 3 Monthly (from 1980) and weekly (from RESTRUCi,t i,t i,kPERFi,tk 1986) share price returns, corrected for k1 stock splits and dividend pay-outs, and a Performance (lagged) value-weighted index of all companies 8 listed on the Brussels Stock Exchange were i,l CONCi,l,t1 l1 provided by the Generale Bank. Accounting 8 data (total assets, equity, operating income, i,l CONCi,l,t1PERFi,l,t1 earnings after tax, dividends per share, l1 debt–equity structure) were collected from Ownership concentration and interaction annual reports and from the database of 8 Central Depository of Balance Sheets at i,l INCCONCi,l,t1 the National Bank of Belgium. l1 8 i,l INCCONCi,l,t1PERFi,l,t1 3.1.4. Data on the board of directors and l1 the management committee Market in share stakes and interaction 2 The database of the National Bank of im,l DEBTi,m,t1 Belgium also contains data on the board of m1 directors.Turnover data were compiled and 2 i,l DEBTi,m,t1PERFi,m,t1 reasons for directors to leave the company m1 were collected from the notes in the annual Debt policy and interaction reports. Natural turnover due to retirement, death or illness is usually reported 2 im,l BOARDi,n,t and is used to correct the turnover data. n1 Other reasons for turnover are rarely men- 2 tioned in either the annual reports or the i,l BOARDi,n,t PERFi,n,t1 n1 financial press. When no grounds or non- informative reasons8 were given for Board composition and interaction turnover,forced turnover due to disciplining 15 actions or due to company policy disputes log(SIZEi,t) i,p industry p1 was assumed. Data on size and turnover of 5 the management committee were gathered i,q year ␧i,t from the annual reports. When the annual q1 report did not explicitly mention the exis- Size, industry and time dummies tence of a management committee, the yearbooks Memento der Effecten and the i company, t year, l classes of Jaarboek der Bestuurders (Yearbook of owner, m number of debt policy vari- Directors) were consulted to determine ables, n number of board composition whether or not directors had executive variables. functions. If the annual reports or other RESTRUC Board restructuring, meas- public sources did not reveal the data ured by (1) executive board turnover, needed, companies were contacted by fax (2) CEO or executive chairman turnover, and phone to supplement lacking data. (3) management committee turnover. 292 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER PERF performance variable measured Logit models are used if the dependent by lagged (1) market adjusted returns, variable is a dummy (in the case of CEO (2) changes in earnings after tax, (3) earn- turnover). For executive director and ings losses, (4) ROE, (5) ROE – industry management committee turnover, GLS median ROE (with earnings after tax), (6) models and OLS models with a logarithmic ROA, (7) ROA – industry median ROA transformation of the dependent variable (with earnings from operations before are used and the estimation is conducted interest and taxes), (8) changes in with heteroscedasticity consistent covariance dividends, (9) changes in ROE, (10) changes matrix estimator (White, 1980). Tobit in ROE – industry median of ROE changes, models are also used to address that fact (11) changes in cash flow on equity, that the dependent variable (executive and (12) changes in cash flow on equity – industry committee turnover) is censored. Industry median of changes, (13) changes in cash and time effects are accounted for by flow margin, (14) changes in cash flow including industry and time dummies, margin equity – industry median of changes. respectively. Corporate board size and firm CONC ownership concentration (%) size are included as control variables.9 The by class of owner: (i) holding companies, relations are also tested including corporate (ii) banks, (iii) investment companies dummies and taking innovations to remove (pension funds, investment funds), firm-specific effects. In order to address the (iv) insurance companies, (v) industrial and endogeneity problems lagged data for commercial companies, (vi) families and ownership, performance and debt policy individual investors, (vii) federal or were utilised in the models. Over- or under- regional authorities, (viii) realty investment performance in relation to industry peers was companies. Both the percentages of owner- measured by correcting performance ship by category of owner and the percentage variables for the median industry performance. held by the largest shareholder are In Section 5, Tobit models are shown, but included (in separate regressions). Both tables with other estimation methods are direct shareholdings by category of owner available and the robustness of the results are included as are the direct shareholdings across estimation techniques is discussed. reclassified into the categories of owner based on the category of the ultimate (reference) shareholder (in separate 4. OWNERSHIP STRUCTURE AND regressions). Herfindahl indices of the CONTROL OF BELGIAN LISTED largest 3 shareholders by category of owner COMPANIES: STYLISED FACTS are also used as concentration measures. INCCONC purchases of share stakes 4.1. Ownership concentration (in %) by category of owner. Both direct shareholdings and reclassified ones based on In a nutshell, the characteristics of Belgian ultimate shareholder are included, see CONC. corporate ownership can be summarised as DEBT debt policy and debt structure follows: (i) few–only 165–Belgian compa- variables: debt/equity ratio, current ratio, nies are listed, (ii) there is a high degree of quick ratio, interest coverage (EBIT/interest ownership concentration, (iii) holding com- expenses). In each model, gearing was only panies and families, and to a lesser extent included along with one of the other industrial companies, are the main investor variables in order to avoid multicollinearity. categories, (iv) control is levered by pyram- BOARD board composition (% of idal and complex ownership structures and non-executive directors), separation of the (v) there is a market for share stakes. functions of CEO and chairman (1 no Properties (i) to (iv) imply that Belgium separation), board size, tenure of CEO. can be portrayed as a Continental European blockholder system rather than a SIZE logarithm of total assets or of market based system (Bratton and total employees. McCahery, 1999). However, typical for OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 293 Belgium is the importance of holding stakes are level 1-shareholdings). companies which are often part of pyramidal Ownership cascades are usually used to ownership chains and are used to lever dilute the one-share-one-vote rule: a chain control (Renneboog, 1997; Daems, 1998). with intermediate holdings of e.g. 50% The sum of the share stakes held by large allows de facto majority control with lim- shareholders (owning at least 5% of ited cash flow rights. As a proxy for control outstanding shares) amounts to, on average, leverage via ownership cascades, the ratio more than 65%. The largest direct share- of the direct largest shareholding and its holder controls 43% in the average listed levered shareholding (the multiplication of company. The three most important direct the shareholdings on consecutive ownership investor classes are holding companies, tiers) is used. For instance, company A, industrial and commercial companies, and whose shares are widely held, owns 40% families and individual investors. They own, of company B which, in turn, owns 40% of respectively, 33%, 15% and 4% of the company C. In this example, the ultimate voting rights. However, taking into account shareholder level is 2, the direct largest ownership cascades to reclassify the direct shareholding (of B in C) is 40%, the ulti- share stakes according to the shareholder mate shareholding amounts to 16% (40% category of the ultimate owner10 reveals that 40%), and the leverage factor (largest holding companies control directly and indi- direct shareholding/levered share stake) is rectly an average of 26.7% of direct voting 2.5 (40/16). For our sample companies, rights in listed Belgian companies whereas the average largest direct share stake the category of industrial and commercial amounts to about 55%, whereas the companies controls an average stake of levered shareholding is 39%. The smaller 11%. Individual and family investors do not the shareholdings with which control is generally hold shares directly in Belgian maintained through intermediate levels and companies, but use intermediate companies11 the larger the number of intermediate as investment vehicles with which they ownership tiers, the higher the control control an average shareholding of 16%. leverage factor or the more considerable Table 11.1 illustrates the high level of the violation of the one-share-one-vote ownership concentration and gives the rule. Table 11.2 discloses that since 1989 percentage of Belgian listed companies the control leverage factor decreased from with voting rights concentration of at least 3.6 to 2.7. Since the average ultimate a blocking minority (25%), an absolute ownership level and the ultimate levered majority and a supermajority (75% and shareholding do not change significantly more). Panel A reveals that a voting rights over this time, the decline of the control majority exists in more than half (56%) of leverage factor indicates that control on the listed companies. In 18% of the intermediate levels has become more Belgian companies, a supermajority gives concentrated. absolute control to one shareholder(group) since blocking minorities cannot be formed. Shareholdings of 25% or more are present 4.3. The market for corporate in 85% of all companies.The concentrated control ownership pattern is similar in the subsam- Although a market for corporate control ples of listed holdings companies, financial (commonly defined as a (hostile) take over and institutional companies, and industrial market) is usually associated with the US and commercial corporations. and the UK,Table 11.3 shows that a partial control market or a market in substantial 4.2. Ownership cascades and the share blocks exists in Belgium. In more than 22% of the listed companies, violation of one share-one vote rule substantial changes (of more than 5%) in Table 11.2 shows that the ultimate owner- ownership concentration take place and in ship tier averages 2.2 (where direct share 7.6% of firms blocking minorities are sold. 294 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER % of companies with a stake % of companies ries. MIN control (minimal 50% of the voting rights) 50% of the voting (minimal control areholders holding stakes of 25% or more. stakes holding areholders mate investor: direct shareholdings belonging to the mate investor: belonging shareholdings direct % of companies with a stake of 75% or larger. with a stake % of companies Direct shareholdings. the direct stands for a 20) N ( 73) N ( 157) 64) N ( N ( % of companies with a stake of 50% or larger, with a stake % of companies SUP Blocking minority, shareholdings Blocking and supermajority majority All investors Holding co’s Families Indus. co’s Belgian investors Foreign investors : Notifications. and Ownership based on BDPart Own calculations Dir. and 93 55 16 34 19 3 29 18 4 24 11 7 61 37 8 32 18 8 indirect Panel D: Industrial and commercial companies Direct 86 47 15 48 25 4 0 0 0 28 15 8 66 37 7 20 10 8 DirectDir. and 80 75 55 50 15 10 40 35 15 15 0 0 5 0 5 0 0 0 5 5 5 5 5 5 48 62 33 40 10 10 32 13 22 10 5 0 Dir. and 83indirect 59Panel C: Financial sector (banking, insurance, real estate) 20 50 36 13 22 13 2 9 6 3 45 30 11 38 30 13 Direct 79 39 14 50 23 8 5 2 2 17 9 2 59 31 11 20 8 3 Dir. and 85indirect 56Panel B: Holding companies 18 41 26 6 23 14 3 15 8 5 51 33 9 34 23 10 Source A shareholder. as one single are considered subsidiaries owned of fully chain Direct 82 45 14 48 23 5 2 1 1 21 12 5 63 36 9 19 9 5 Panel A: All sample companies of 25% or larger, MAJ Dir.and indirect refers to the fact that the direct shareholdings are classified according to the shareholder class of the ulti to the shareholder according are classified Dir.and to the fact refers shareholdings that the direct indirect same ultimate; summed. subsequently were group investor based on (i) a majority is control and indirect) (direct control Ultimate or (ii) shareholdings; pyramid of the ownership sh tier the absence of other in of at least tier 25% on every ownership on every Table 11.1 1994 MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP MIN MAJindirect SUP MINa catego shareholder by the main held shareholdings with a minority, or supermajority of the sample companies majority Percentage MAJ SUP OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 295 Table 11.2 Largest direct and ultimate (direct and indirect) levered shareholdings, and the control leverage factora

1989 1990 1991 1992 1993 1994

Sample size 160 156 156 156 156 158 Ultimate ownership level 2.2 2.2 2.1 2.1 2.0 2.0 (1.364) (1.290) (1.188) (1.159) (1.098) (1.020) Direct largest shareholding 55.1 56.4 57.2 57.8 56.3 55.6 (19.737) (19.509) (19.923) (20.632) (20.341) (19.987) Levered shareholding 38.0 38.5 40.3 41.7 42.0 39.4 (22.524) (22.906) (23.988) (24.600) (23.657) (21.454) Control leverage factor 3.6 3.6 3.0 2.9 2.8 2.7 (direct/levered shareholding) (8.391) (8.650) (6.756) (6.710) (6.432) (6.356) a This table presents the ultimate ownership level, defined as the highest level of ownership in an uninterrupted control chain (direct shareholdings are level 1). Ultimate control is control based on (i) a majority control (minimal 50% of the voting rights) on every ownership tier of the ownership pyramid or (ii) shareholdings of at least 25% on every tier in the absence of other shareholders holding stakes of 25% or more. A chain of fully owned subsidiaries are considered as one single shareholder.The direct largest shareholding is the average direct largest share stake of at least 25%. The levered shareholding is calculated by multiplying the share stakes of subsequent ownership tiers.The control leverage factor is the ratio of the direct shareholding divided by the ultimate levered shareholding. For instance, company A, whose shares are widely held, owns 40% of company B which, in turn, owns 40% of company C.The ultimate shareholder level is 2, the direct largest shareholding (of B in C) is 40%, the ultimate shareholding is 16% (40% 40%), and the leverage factor is 2.5 (40/16).There was no direct shareholding of at least 25% in 17 sample companies, which were not included in this table. Standard deviation in parentheses. Source: Own calculations based on data from the BDPart database and the Notifications of Ownership.

Twenty-eight majority stakes changed hands.12 5. RESULTS These findings suggest that this market for share stakes is not insignificant. Table 11.3 Belgian companies have a one-tier board also discloses that the holding companies are system with average board size amounting the main sellers and purchasers of share to 10 directors for the period 1989–1994 stakes. Institutional investors, mainly banks and with a median of 9. Yearly, between and insurance companies, acquire 49 share- 9% and 12% of the directors leave the holdings of more than 5% and sell 43 stakes board. Annual turnover among executive of similar size. Families and individuals sell directors in this period is high: between 17 stakes of blocking minority size and 27% and 41%, whereas only about 7% of more, while 10 such stakes are purchased. the non-executive directors is replaced. Most of the exchanges of the largest blocks The yearly replacement of the CEO (called of shares are negotiated deals and take place ‘delegated’ or managing director) amounts ex exchange.13 to 18%. A third measure of top management restructuring consists of replace- 4.4. Capital structure ment in the management committee. Although such a management committee Belgian listed companies are relying to a is no legal requirement, 65% of the large extent on short term debt: long term companies mention in their annual reports debt on equity amounts to 28% whereas such committees, which count on average short term debt (including trade credit) on 3.6 members (median of 4). The executive equity is 53%. Holding companies carry directors are always members of this more long term debt (39% on equity) than committee and have an average of 2.4 industrial and commercial firms (with only members (median of 2). Annual turnover 12%). Average current ratios are 4.1 for of the management committee totals industrial companies and 5.4 for holding 17%. Although managerial turnover is companies. corrected for natural turnover related to 296 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER

Table 11.3 The market in share stakes over the period 1989–1994a

Number of increases and decreases stakes 1989–1994 [1–5%] [5–10%] [10–25%] [25–50%] [50–100%] Total

Panel A: Purchases for all sample companies Purchases: all 113 103 66 40 21 343 shareholders Purchases: holding 50 51 26 22 4 153 companies Purchases: institutional 39 25 13 5 6 88 investors Purchases: industr. 10 14 13 7 7 51 and commerc. co’s Purchases: families 14 13 14 6 4 51 and individuals Panel B: Sales for all sample companies Sales: all shareholders 119 78 81 45 33 356 Sales: holding 40 47 46 17 20 170 companies Sales: institutional 49 13 15 12 3 92 investors Sales: industr. and 5 4 7 3 6 25 commerc. Co’s Sales: families and 25 14 13 13 4 69 individuals a This table gives the size distribution of purchases and sales of large shareholdings by category of owner over the period 1989–1994. All changes are given excluding changes in government stakes and real estate as these categories are minor.Purchases and sales are calculated by comparing the share stakes of a shareholder category of a fiscal year to the shareholdings of previous year. Institutional investors consists of banks, investment and pension funds and insurance companies.Total number of firm-years over the period is 1024. Source: Own calculations based on BDPart and Ownership Notifications. retirement age, death or illness of directors, from disciplinary actions.We also investigate the turnover data may still contain some when such corporate governance actions non-conflictual turnover since corpora- are undertaken and whether disciplining tions do not generally release information takes place at an early stage, i.e. rapidly regarding management replacement or do after earnings, cash flows or share price so in euphemistic terms. declines or,rather late when the company is no longer able to generate profits or has to cut dividends? Including lagged 5.1. Board restructuring in industrial performance up to three years prior to and commercial companies turnover allows us to investigate the reaction time of board restructuring.14 Warner 5.1.1. Executive board turnover et al. (1988) and Coughlan and Schmidt (1985) report that US boards react 5.1.1.1. Corporate performance and quickly to poor performance in their decision to replace management because share disciplining of management. performance lagged up to two calendar A first question is whether or not turnover, years helps predict current-calendar-year corrected for natural turnover, is related to management changes. Share price poor corporate performance and results performance may underestimate the true OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 297 relation between performance and executive models yields weaker correlations with turnover given that share prices reflect board restructuring. current profitability as well as expected Levels of performance as well as changes future opportunities including the potential in performance, corrected by industry performance improvements under new medians, are analysed as it may well be management (Weisbach, 1988). As that it is not just low earnings which trig- accounting earnings depend on discre- ger managerial disciplining but peer group tionary managerial accounting choices, we (industry) underperformance. Morck et al. use a combination of accounting, dividend, (1989) find that when a firm significantly cash flow measures and market adjusted underperforms its industry, the probability share returns as performance benchmarks of complete turnover of the top manage- in the Tobit models of Tables 11.4 and ment team rises. Table 11.4 shows that 11.5. Operating earnings before interest both industry adjusted levels and changes and taxes (standardised by total assets) in ROE and in cash flow are negatively are used as they are not sensitive to correlated to management changes prior to financing policy, tax regime, windfall profits turnover, but more so for changes than or extra-ordinary losses.The use of operat- for levels. ing income rather than net earnings after All in all, the evidence of Table 11.4 tax reduces the impact of the described fails to reject Hypothesis 1: it shows that ‘earnings management’ (Dennis and the poorer the performance, the higher is Dennis, 1994). ROE is taken after interest, the turnover of the executive board. These extraordinary results and taxes. The indus- results are consistent through different try medians are substracted from both the estimation techniques (Tobit and OLS levels of and the changes in ROE and cash with and without fixed effects). flow on equity.15 Companies only resort to substituting Table 11.4 (lines 2–4) shows that, for executive directors when accounting listed industrial and commercial compa- returns are very weak: when the company nies, there is a negative significant relation was not able to generate profits or was between executive director replacement forced to cut dividends in prior periods. and market adjusted performance in the Furthermore, disciplinary actions are under- three years prior to management substitu- taken when the company under-performs tion. Earnings losses over the fiscal year its industry peers and when market prior to turnover are followed by increased adjusted returns are negative in the period levels of executive board turnover. Warner prior to board restructuring. et al. (1988), amongst others, confirm for the US that unless performance is 5.1.1.2. Ownership concentration. extremely good or bad, their management turnover models have little predictive As a single shareholder(group) controls a value. Another critical performance bench- voting rights majority in more than half of mark, substantial cuts in dividends (of at Belgian listed companies, and as a blocking least 25%) or omissions, also precede minority exists in 85% of firms, the control board restructuring. Given that deviations percentage of the largest block is included from expectations about dividend policy as an explanatory variable.16 The free usually contain signalling information, riding control-hypothesis predicts that management is generally reluctant to large share blocks facilitate disciplining of reduce dividends unless such a reduction is management. However, Table 11.4 (lines unavoidable. Hence, dividend cuts are asso- 5–8) shows that that the presence of large ciated with unusually poor stock-price and share blocks held by holding companies and earnings performance (Healey and Palepu, institutions (banks, investment funds or 1988; Ofer and Siegel, 1987; Marsh and insurance companies) is not related to Merton, 1987). Including changes in board restructuring. In contrast, manage- earnings or dividends into the monitoring ment replacement is influenced by large 298 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 0.00 0.10 0.10 0.00 0.03 0.10 0.00 in *** * * *** ** * *** in CF/Eq– 4.40507 0.08407 0.00043 0.43 0.00018 0.35 0.01 0.04 0.02046 ** ** a 0.00034 0.15 0.00035 2.35075 0.01383 0.61 0.00143 0.85 0.00344 0.00047 0.32 0.00007 0.85 0.00037 0.11 0.05295 0.65 0.03464 0.59 0.02 0.09 0.09 0.000.02 0.00987 0.00087 0.90 0.01169 0.06 0.01 0.00863 0.17 0.01869 1: ** * * *** ** * ** in Cash flow on eq. – 0.05700 0.03032 0.73 2.24604 in ROE– 0.00 0.080.01 0.00136 0.70 0.07 0.01087 0.01491 0.08 0.00037 0.21 *** * ** * * 1 and performance at t 0.09706 0.00042 5.26084 0.00470 0.23 0.02440 0.77 0.00097 0.16 0.00058 0.50 0.00011 0.64 0.00015 0.68 0.00019 0.66 0.03 0.05 0.02 0.01164 0.10 0.00061 0.360.03 0.00100 ): t 1: ** ** ** * ** 1, t ( yes) median (%) industry median industry median (%) industry median 0.00224 0.76 0.00971 0.18 0.02284 0.000070.00066 0.44 2.59715 0.29769 1 0.04 0.09 0.000.00 0.00457 0.002460.09 0.51 0.72 0.02706 0.01330 0.00006 0.17 0.070.00 0.99 0.00180 0.01456 0.00000 0.00590 0.75 0.98 0.00123 0.85 0.00668 0.11 yes) ( ** * *** *** * *** 1 0.01271 0.01753 0.00259 0.63 1.27277 0.31887 0.000020.00007 0.32 0.31 rminants of executive board restructuring in listed industrial and commercial companies and commercial industrial listed in board restructuring of executive rminants 0.09 0.00 0.08 0.07 0.060.08 0.00347 0.01192 0.26 0.01100 0.05 0.03 * *** * * * * ** ** 0.56100 0.78401 0.35742 1.61149 0.23 0.01714 0.82 0.02262 0.06148 0.00696 0.81 0.00016 1.59057 Market adj.return (%) Operating earnings Dividend cuts ROE-indus. losses ( Par.Estim P(Chi) Par.Estim P(Chi)Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Tobit model of the dete model Tobit = 3 1 2 t t t 4 Perf. 5 Hold. co’s 0.00346 0.60 6 Institutions 0.00549 0.44 2 Perf. 3 Perf. 7 Indus. co’s8 Fam/Ind. 0.02212 0.02297 9 Hold. co’s 0.05232 0.11 0.00001 0.77 0.00016 0.37 1 Intercept 13 Hold. co’s 0.01521 0.28 Table 11.4 Performance Share stake held by the largest shareholder category of owner at t Increases in ownership concentration by category of owner within Interaction between share stake held by the largest shareholder category of owner at t 12 Fam/Ind. 0.11505 14 Institutions 10 Institutions 11 Indus. co’s OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 299 0.00 0.08 0.00 0.00 0.00 0.06 0.00 0.00 0.08 *** * *** *** *** * *** *** * Interest coverage, Interest 0.00001 0.39 0.00144 0.00098 0.00066 0.04889 0.00063 0.23 0.00367 0.92 0.14957 0.20 0.03277 26.924 0.01 0.01007 0.07 0.00059 0.09 0.100.01 0.003360.06 0.02212 0.33 0.05 0.00 4.46604 0.00 ** * * * ** * ** *** *** Debt/Equity, Intcov 0.00019 0.00098 0.00093 0.00735 0.11967 0.05741 0.26 0.00640 26.924 0.04 0.01573 0.03 0.01 0.02 0.00 4.17495 0.01 1: ** ** ** ** *** ** 0.00181 0.10365 0.00088 0.41 0.00067 0.38 0.00023 0.80 0.00851 0.88 0.00032 0.30 0.00152 0.64 0.00148 0.00011 0.54 0.00100 0.02100 0.19 0.01779 0.19 24.859 Families and Individuals, D/E Families 0.02 0.01267 0.03 0.000.00 0.00533 0.240.10 0.08 0.01118 0.000040.00 0.22 3.85480 0.00002 0.23 and performance at t ) ** ** *** *** * * *** t 1, t ( 0.00123 0.22 0.00074 0.04384 0.32 0.00086 0.00075 0.35 0.00059 27.947 0.00 0.02014 0.00 0.03 0.02231 0.00 5.20913 *** *** ** *** Indus. and comm. co’s, Fam/Ind. 0.00056 0.15 0.00414 0.81 0.02346 0.38 0.00236 0.89 0.04264 0.28 0.00260 0.82 0.00071 0.00016 0.60 0.05734 0.35 0.03886 0.35 0.00054 0.55 0.00401 0.58 0.02899 0.19 Total number of directors. number Total are of CEO and chairman if the functions equal to 1 is included A dummy variable 28.567 1 1 0.02 0.04032 0.01 0.080.00 0.044010.01 0.00681 0.43 0.07 0.000020.00 0.91 3.44586 0.00 ** ** * *** ** * *** *** 0.00007 0.16 0.03342 0.31 0.00648 0.560.00024 0.00019 0.01877 0.300.00034 0.009680.00059 0.77 0.15899 0.00002 0.98 20.874 Holding co’s,Holding Indus. co’s 0.060.04 0.02296 0.02 0.01 0.040.01 0.01401 0.00 0.000.10 0.000000.00 0.14492 0.32 3.80638 0.34 0.00000 0.08951 0.620.01 0.64 0.00133 0.00007 0.00 0.28 0.18795 0.31 0.14951 0.38 0.04767 0.80 * ** ** ** ** ** *** *** * *** ** *** 0.00056 0.15238 0.13707 0.09332 0.21246 0.30067 0.14 0.00118 0.42 16.773 15.90586 0.23 Percentage non-executive directors, non-executive Percentage Num. dir. 1 1 0.00354 1 1 0.00890 t t Performance, Hold. co’s CEO 0.29686 CEO t t ϶ ϶ combined by one person. combined response Log likelihood % non-ex. 17 Hold. co’s 0.07118 0.11 27 Ch 15 Indus. co’s 0.05496 Weibull aPerf. 29 Num. dir. 0.00453 0.87 0.00057 0.67 0.00650 0.55 0.00046 0.61 25 Ch 16 Fam/Ind. 0.04521 18 Institutions 19 Indus. co’s 20 Fam/Ind. 22 Intcov. Interaction between debt variables and performance at t 23 D/E 24 Intcov. Interaction between board variables and performance at t 28 % nonex. 4.66866 of tot. assets) or neg.Zero 195 192 197 197 197 197 197 21 D/E Interaction between increases in ownership concentration by category of owner within 26 % nonex. 5.31115 30 Size (Log 30 Size 300 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 0.06 0.06 in * * 0.04158 0.03967 0.014750.00557 0.12 0.02355 0.59 0.52 0.000070.00053 0.94 0.54 0.00132 0.72 1.52486 0.40 1.85787 0.31 in CF/Eq– 0.02 0.00 0.06 0.01 0.01 0.014170.00 0.47 0.00 0.00 0.080.04 6.95358 0.41 * ** * ** ** *** *** ** * ** 0.02383 0.39 0.20236 0.22205 0.10645 0.32 0.01220 0.20 0.01645 0.27 4.48663 1.51379 0.31 1.77376 0.02671 0.01 0.00 in Cash flow on eq. – a ** *** 1: 0.30803 0.00591 0.74 0.010480.03376 0.14 0.17 0.03187 0.00506 0.67 0.03388 0.00035 0.50 0.00329 7.97964 0.14 0.42521 0.83 in ROE– 0.10 0.00 0.10 0.010.00 0.01492 0.56 0.000.05 0.000590.03 0.380.01 0.00761 0.00325 0.00071 0.33 0.00890 * *** * ** *** *** * ** ** and performance at t 1 indus. 0.02129 0.14491 1.36375 0.54 0.06358 0.00570 1.53244 0.29 1.01566 0.16 0.00229 0.48 0.02882 0.03182 0.00 0.01 0.00 0.00 0.00395 0.05 0.00464 0.00 0.01735 0.34 ): 1: t *** ** *** *** ** *** 1, yes) median (%) industry median industry median (%) industry median t ( 1 0.00906 0.06775 0.00048 0.45 0.00287 0.34073 0.97 0.02773 0.50 0.000280.00024 0.80 0.00545 0.82 0.74106 0.01479 1.68032 0.59 22.30041 0.35 0.00094 0.09 yes) ( * 1 0.01075 0.24 0.01338 0.44 3.68603 0.12 5.21520 0.29 0.30066 rminants of executive board restructuring in listed holding companies holding listed in board restructuring of executive rminants 0.09 0.01 0.01 0.12457 0.24 0.100.07 0.00160 0.00001 0.24 0.71 0.07 * ** ** * * * 2.37284 0.02472 0.75 0.03263 1.08528 1.30870 0.38 11.49972 0.16 24.25648 0.29 Market adj.return (%) Operating earn. Dividend cuts ROE losses ( Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Tobit model of the dete model Tobit = 2 3 1 t t t 1 Intercept 5 Hold. co’s 0.00067 0.93 3 Perf. 6 Institutions7 Indus. co’s 0.009738 Fam/Ind. 0.41 0.02852 0.00109 0.01048 0.89 0.68 0.01489 0.00248 0.16 0.82 4 Perf. 9 Hold. co’s 0.03150 0.20 0.00007 0.38 2 Perf. 13 Hold. co’s 0.00455 0.87 Share stake held by the largest shareholder category of owner at t Table 11.5 Performance 10 Institutions11 Indus. co’s 0.01317 0.19 0.07345 Increases in ownership concentration by category of owner within 0.00017 0.41 12 Fam/Ind. 14 Institutions Interaction between share stake held by the largest shareholder category of owner at t 15 Fam/Ind. 1.34751 OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 301 0.04 0.00 0.05 ** *** ** Interest coverage, Interest 0.00307 0.25 0.17474 0.44 0.00075 0.61100 0.32 0.12670 0.33 0.10907 19.897 0.06 0.01 0.00022 0.94 0.01 0.000.06 0.000120.00 0.39 5.72372 0.00 * ** ** *** * *** *** Debt/Equity, Intcov 0.02876 0.26 0.03356 0.37 0.00046 0.93 0.00461 0.38 0.00096 0.45780 0.26303 0.14793 0.38 12.893 0.09 0.010.03 0.02260 0.005190.00 0.12 0.03 0.00232 0.04 0.00 6.28175 0.07 1: * ** ** *** ** ** *** * 0.00058 0.93319 0.03945 0.11 0.25987 0.00231 0.00419 0.16702 0.12 0.37400 0.00361 0.61 0.00235 0.32 0.00079 0.91 20.635 Families and Individuals, D/E Families 0.04 0.04 0.00 6.79918 and performance at t ** ** *** ) t 1, t ( 0.00214 0.00395 0.08235 0.75 0.07418 0.11 0.67105 0.36 10.436 0.00 0.01 0.00096 0.760.00 0.00772 0.000010.00 0.97 0.00 4.44313 0.00 0.00 0.00408 0.97 *** ** *** *** *** *** *** Indus. and comm. co’s, Fam/Ind. 0.27650 0.00092 0.79457 0.01853 0.19 0.03948 0.74 0.34014 0.11 0.02154 0.00011 0.56 0.00019 0.51 0.00018 Total number of directors. number are Total of CEO and chairman if the functions equal to I is included A dummy variable 1 1 0.02 0.00170 0.040.04 0.01531 0.001130.05 0.21 0.00 0.00 4.05689 0.02 ** ** ** ** ** ** ** 0.00008 0.06893 0.30 0.05833 0.56 0.01051 0.07609 0.63 0.88579 0.00239 0.00131 0.46 0.00001 0.11 0.00073 0.59 0.00079 0.71 0.00110 0.47 22.042 15.165 Holding co’s,Holding Indus. co’s 0.01 0.00042 0.08 0.00974 0.0 6.03862 0.04 0.05 ** * *** ** * 7.71821 0.07816 0.73 0.00211 0.59 0.20213 12.297 Percentage non-executive directors, non-executive Percentage Num. dir. 1 0.00014 0.84 1 1 0.02280 0.11 1 0.00981 Performance, Hold. co’s CEO CEO 0.65725 0.40 ϶ ϶ 16 Hold. co’s 0.16037 response Log likelihood 20 D/E t Interaction between increases in ownership concentration by category of owner within Weibull aPerf. % non-ex. combined by one person. combined 25 % nonex. 17 Fam/Ind. 16.25936 0.16 Interaction between debt variables and performance at t 18 D/E t 19 Intcov. t 22 Ch 23 % nonex. 5.97179 tot. assets) or neg.Zero 150 148 143 160 160 160 160 21 Intcov. t Interaction between board variables and performance at t 24 Ch 26 Num. dir. 0.00042 0.77 27 Size (Log of 27 Size 302 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER industrial investor shareholdings (in 6 out exception: when industrial companies and of 7 models) and by blocks held by families families obtain substantial share stakes, (5 models). Piecewise regressions – with changes in management are implemented. dummies indicating whether or not the Such board restructuring takes place (lines largest owner holds a blocking minority, 19– 20) when prior performance was poor majority or supermajority (as in Hermalin (negative market adjusted returns, negative and Weisbach, 1991) – reveal that minor- changes or levels of performance), which ity stakes held by industrial companies are suggests a partial corporate control sufficiently large to exert control and to market (Hypothesis 5). It can be observed restructure the board.17 that disciplining underperforming manage- Table 11.4 (lines 9–12) also investigates ment happens in the year of turnover or in whether the ownership structure plays a the subsequent fiscal year.20 performance-induced disciplining role. Concerning the role of ownership None of the categories of large blockholders concentration and the partial market seem to be involved in disciplinary actions for control, it is important to realise that against management when performance is above results were obtained after classifying poor. The lack of institutional investor all blocks of voting rights into ownership involvement is in line with Hypothesis 3 categories based on the identity of the ulti- which states that they abstain from moni- mate or reference owners of each of these toring to avoid conflicts of interest. In blocks. No significant results were attained contrast, the fact that the large holding in a first set of regressions where all own- companies do not seem to monitor is ership variables (levels and increases) were surprising as these often cite superior included by category of shareholder owner corporate governance as one of the core on the direct ownership level. For example, contributions of their stable ownership if a holding company holds 10% of the stakes as ‘reference shareholders’.18 The voting rights in a listed company, this 10% lack of significance of the interaction stake is classified as a stake owned by a terms between large industrial and family holding company.The fact that the interme- owners, and performance, raises doubt diate holding company may be directly or about the fact whether board restructuring indirectly controlled by e.g. a family is is initiated by families or industrial compa- ignored in this first set of regressions. nies as a result of poor performance.19 All However, when we reclassify all direct in all, there is little evidence about the shareholdings (voting rights) in ownership corporate control role of existing large categories based on the identity of the true shareholders. (i.e. ultimate) owner – in the example above the 10% stake is a family controlled stake – we find the conclusions discussed 5.1.1.3. The market in share stakes above: significant results for the presence When performance is poor, shareholders of industrial co’s and families versus without a distinct interest in monitoring insignificant ones for institutions and holding sell stakes, while those with strong moni- companies. This suggests support for toring abilities increase their stakes in Hypothesis 4 and implies that the ultimate order to reinforce their position as (major) or ‘reference’ shareholder, who controls the shareholder. If this were true, we would voting rights of a listed target company by expect positive signs for the increases in ways of a cascade of intermediate holdings, 21 shareholdings (Hypothesis 3). In spite of exerts corporate control. the fact that institutions and holding companies actively trade in share stakes over 5.1.1.4. Gearing as a bonding mechanism 1989–1994 (Table 11.3), ownership increases by these categories are not corre- High leverage encourages management to lated with changes in board structure (lines generate sufficient funds to service the 13–16 of Table 11.4). However,there is one debt commitments. Consequently, a high OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 303 debt–equity ratio is expected to reduce directors on board is a proxy for the management’s discretion and summon control power of a share block and that more intensive creditor monitoring, as is these large shareholder representatives are suggested in Table 11.4 (line 21) where performing their roles as monitors and are executive director replacement is positively executing their disciplining part well. As correlated with a high gearing (6 out only a few companies could (or were willing) of 7 regressions). Executive monitoring to disclose the representative function of increases especially when corporate its board members, this hypothesis could performance is negative (negative market not be tested further. Board size differs adjusted returns, earnings losses, level and substantially across firms; some have small changes in ROE and cash flow adjusted for boards with 6 directors, while others count industry medians).22 Low interest coverage 15 or more. In spite of the fact that large is an important indicator of financial boards may reduce efficiency, board size distress; when the interest cover decreases does not seem to influence managerial below 2, a company typically loses investment disciplining (line 29). grade. Table 11.4 also shows that executive For the US, Weisbach (1988) finds that board restructuring also coincides with low CEO turnover is more sensitive to perform- interest coverage (line 22), but that the ance in firms whose boards are dominated interaction of poor share price and by outsiders. Outsiders are carefully accounting performance is not correlated defined as directors who work neither for to executive board turnover (line 24). This the corporation nor have extensive dealings implies that interest coverage may be con- with that company. In a study on the sidered as another monitoring performance performance effects of the composition of benchmark and important trigger for the board of directors in the US, Baysinger monitoring actions.23 The strong correla- and Butler (1985) conclude that those tion between gearing and interest cover,and firms with stronger independent boards performance (Hypothesis 6) suggests ended up with superior performance enhanced creditor monitoring when per- records, in the form of superior relative formance is poor. Interviews with executive financial performance (an industry and non- executive directors revealed that corrected return on equity). It should be the monitoring role by creditors is considered emphasised that research about the impact limited (unless there is a danger of bond of US board composition on CEO turnover covenant violation).24 is not directly comparable with research on Belgian boards.The emphasis in the US has 5.1.1.5. Board composition and been put on the independence of separation of control ‘outside’ directors, whereas some nonexecutives in Belgium are large shareholder Table 11.4 supports the hypothesis that the representatives and ‘independent or expert’ board structure is instrumental for the non-executive directors’ appointment to the monitoring efficiency of the internal gover- board might be subject to large share- nance mechanism (Hypothesis 2).The more holder approval. independent the non- executive board from Next to the strengthening of the inde- management, proxied by the proportion of pendence of non-executive board members, non-executive directors, the easier it is to another recommendation in the recent replace management when managerial Guidelines for Good Corporate Governance performance is inadequate (lines 26 and (of the Cardon Commission, Stock 28).The fact that the role of large ownership Exchange Commission, Commission for stakes was not supported by our model Banking and Finance) is the separation of (apart from for industrial companies) may be the functions of managing director and of explained to some extent by the importance chairman of the board, but this necessity is of the proportion of non-executives. It may not upheld by the findings of this model well be that the number of non-executive (lines 25 and 27 of Table 11.4). 304 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Although larger companies may have a companies and which internal and external bigger internal managerial labour market control mechanisms trigger such gover- and have better access to the external nance actions. Support for Hypothesis 1 is managerial labour market, corporate size presented in Table 11.5: past poor share is negatively related to executive board price and accounting performance in replacement (line 30).25 Are these holding companies lead to executive board disciplinary actions only directed at top restructuring, although this relation is managers who are on the board or do these weaker than for listed industrial companies. actions extend to other managers, c.q. the Over the period 1988–1994, dividend cuts, members of the direction committee? earnings losses, low industry corrected The results with management committee ROE and cash flows precede executive turnover (executive directors and other top director replacement (lines 2–4). There is managers) as dependent variable gives evidence of leverage related monitoring weaker results than with executive (lines 18–21; Hypothesis 6) as high gearing turnover. This implies that performance ratios result in increased managerial related turnover is targeting the very top removal when performance is poor. In management, namely predominantly those addition, board composition does also seem three managers appointed to the board. to influence corporate control as high executive board turnover is positively related to a high percentage of non-executive 5.1.2. Disciplinary actions against CEO26 directors on board (Hypothesis 2). Although some holding companies are CEO replacement in industrial and commercial widely held, most of the others are con- companies (corrected for natural turnover) trolled by families or other holding compa- is correlated with the presence of large nies. Contrary to Hypothesis 3, there is no shareholdings held by families and individu- consistent relation between the presence of als, institutional investors and industrial large share blocks and executive board companies. The fact that the interaction restructuring of listed holding companies. terms of ownership and performance are The evidence in Table 11.5 even hints significant (10% level) suggests that these (in 3 out of 7 models) that less monitoring replacements are the result of corporate is expected in the presence of a controlling governance actions (Hypotheses 3 and 4). holding company (negative correlation in Furthermore, CEO substitution also follows line 5).28 Furthermore, the negative increases in share stakes held by holding parameter coefficient in line 13 reveals companies and industrial firms, but due to that when there is poor performance and lack of significance in interaction terms it is increases in share stakes controlled by questionable whether this stake accumula- holding companies, there is a reduced man- tion is induced by poor performance agerial disciplining.This raises the question (Hypothesis 5). Internal governance struc- why holding companies with controlling ture like separation of the functions of CEO stakes or holding companies acquiring and chairman, and the presence of a high large stakes do not monitor. Like in France proportion of non-executives facilitates disci- (Banerjee et al., 1997, hence BLV),holding plining of the CEO when the companies companies in Belgium may enhance wealth 27 Finally, results are bad (Hypothesis 2). creation by offering a internal capital there is evidence that low interest coverage market to the companies they control, by leads to higher CEO turnover (Hypothesis 6). offering the possibility to invest indirectly in non-listed companies to individual investors, 5.2. Listed Belgian holding by smoothing tax liabilities of the group or companies by lowering bankruptcy costs by facilitating workouts. However, both the BLV study Table 11.5 investigates whether executive (1997) for France and Daems (1998) for director replacement takes place in holding Belgium fail to find wealth creation by OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 305 holding companies. The share price of expansion strategy and the call to form one Belgian holding companies is even estimated large Belgian commercial bank (merger to be 35–39% lower than the market value with the Generale Bank), GBL has not of all their equity participations.29 favored such strategies.33 All in all, the The potential corporate governance system with strong ownership concentration benefits by holding companies consist of comprises important drawbacks (see also the supply of strategic advice and a reduc- Daems, 1998; De Wulf et al., 1998). tion in agency costs due to economies of scale in monitoring. Still, BLV state that 5.3. Post-disciplining performance “holding companies can create agency problems of their own and, with well- The effectiveness of the corporate control diversified portfolios, it is unclear why they mechanism could be assessed by analysing would be willing or be able to engage in performance following the installation of costly monitoring activities for each of the new management (Hypothesis 7). companies they control.The quality of their Improved performance after executive monitoring activities has never been board restructuring would confirm that the ascertained, not has it ever been compared ousted directors and management had to that provided by other large sharehold- underperformed and that the monitors ers, the external market for corporate were able to attract a management better control, . . . ”. The findings in this paper suited to reorganise the company.The post- corroborate this statement as controlling disciplining earnings evolution is analysed holding companies or holding companies by industry:34 for the companies in which which acquire large share stakes in listed more than 25% of the executive board or Belgian industrial companies (Table 11.4)30 the CEO resigned, the returns on equity two and in listed Belgian holding companies years prior and two years subsequent (Table 11.5)31 do not seem to be involved to executive board restructuring are with performance correcting governance compared. Prior and subsequent ROE are actions. Furthermore, listed Belgian hold- calculated as deviations from the ROE of ing companies may suffer from the very companies which did not experience agency problems they pretend to solve for executive board restructuring of 25% or the companies they control because the CEO replacement. Even after the executive controlling shareholders in holding compa- board restructuring, the level of ROE in nies do not seem to discipline holding most industries subsequent to the board companies’ underperformance. restructuring is still lagging the ROE of This conclusion about holding companies companies without board restructuring. can be extended to the quoted financial There is little evidence of ROE improve- institutions – in this paper defined as a ments in time windows subsequent to board rather heterogeneous sample of banks, restructuring with exception of the industry insurance companies and real estate firms.32 of consumer goods/health/pharma. This Most of these institutions are part of a finding may not come as a surprise as it is holding group of which the monitoring a well documented fact that in US and UK abilities have been questioned. Bank companies, a decrease in earnings often Brussels Lambert (BBL) is often mentioned follows the departure of the CEO because as an example of the break down of new CEOs often write off as many expenses ‘reference’ or ultimate shareholder as possible during their first year (Murphy monitoring. Over whole time period of the and Zimmerman, 1993). Moreover, the paper (1989–94), BBL is controlled by the success of corporate governance actions, Group Brussels Lambert who holds a direct may not be visible in subsequent performance stake of 13% (the largest stake) and controls because, in specific cases, earnings stabili- indirectly via the Royal Belge insurance sation at a low level or avoidance of further group another 9.3%. In spite of the need financial distress or even bankruptcy may for the development of an international be considerable achievements. 306 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Dividend policy (changes in dividends per levered by pyramidal and complex ownership share) are also compared for periods two structures and there is an important years prior and two years subsequent to market for share stakes. Little relation was board restructuring. Similarly, changes in uncovered between ownership structures dividends are defined as the difference in and the disciplining of top management in changes of high turnover companies from listed industrial and commercial compa- the benchmark, namely dividend changes of nies. However, the presence of large indus- companies with low board restructuring trial shareholders (and to a lesser extent of (less than 25% of executive board turnover family shareholdings) is related to high and no CEO turnover). For the sectors of executive board turnover when perform- electrical equipment and electronics, and ance is poor, whereas no evidence was services, there is a significantly higher found for a monitoring role by institutions growth in dividends per share after board or holding companies. restructuring than prior. For holding A fraction of the market for share stakes companies, and the sector chemicals and may be considered as a corporate control materials, this effect is only marginally market because industrial and commercial significant.This result may indicate that, as companies increase their shareholdings (or changes in dividends tend to have a perma- purchase large share blocks) in listed nent character, increases in dividends poorly performing industrial companies in reflect some more confidence in future which there is, subsequently, increased con- profitability after managerial disciplining. flictual management replacement. In the light of disciplining actions undertaken by industrial companies (who are often in the 6. CONCLUSION same industry as the target company), the lack of active corporate control in the wake In this paper the importance of several of poor performance by holding companies internal and external mechanisms of is striking. High ownership concentration corporate governance was analysed in held by a holding company group can lead terms of disciplining management of poorly to strategic deadlocks for quoted compa- performing companies. There may be nies, as illustrated by the BBL case with substitute forms of discipline, and even if the Group Brussels Lambert as reference there is redundancy, one form may domi- shareholder. Daems (1998, p. 65) illus- nate the other consistently. Disciplinary trates this deadlock with another example: actions against management are taken ‘A reference shareholder [belonging to a when market adjusted share returns are holding group] will tend to concentrate on negative and when the company generates the interests of the group as a whole. He operating earnings’ losses or resorts to sub- might be tempted to divide its markets over stantial cuts in dividends in the years prior its subsidiaries such that they do not com- to the restructuring. There is also evidence pete too intensively with each other. Hence, that companies with levels of and changes [the French holding group] Suez, could in ROE and cash flows below those of have an interest in dividing the interna- industry peers are subjected to increased tional [utility] markets over its sub- monitoring. These performance–turnover sidiaries, [the French] Lyonnaise des Eaux relations are much stronger for listed and [the Belgian] Tractebel. This limitation industrial and commercial companies than of strategic freedom of the subsidiaries is for listed holding companies and financial not in the interest of minority shareholders institutions. and investors who are participating in the Belgian equity markets are characterised group via the stock exchange.’ by few listings, a high degree of ownership Although high leverage also seems concentration held by holding companies instrumental to replace poorly performing and families, and to a lesser extent indus- management, there is no direct evidence trial companies. Furthermore, control is that it is bondholders or banks who OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 307 force underperforming management out. Theo Vermaelen, Michel Habib, Ian Cooper, Interviews with non-executive directors Lorenzo Caprio, Marc Goergen, Rafel disclosed that high leverage or low interest Crespi, Rez Kabir, Marco Becht, Piet coverage stimulate actions by the (non- Moerland and Joe McCahery. Helpful executive) board (and as such the share- advice was given by two anonymous holders) rather than creditor intervention. referees, by the participants to the finance We find that the role of the non-executive seminars at the universities of Tilburg, directors is important in the disciplining Antwerp (UFSIA), Barcelona (Autonoma), process: a high proportion of non-executive Venice, Oxford and by discussants at directors leads to increased executive several conferences (EFA, EEA, Financial board turnover.Given that companies usually Markets Symposium of the CEPR in do not reveal the representation of the share- Gerzensee). I am grateful to Mr. Maertens holders of the board of the directors, the of the Department of Information and percentage of non-executive directors was used Statistics of the Brussels Stock Exchange as a proxy for independence of the board for allowing access to some of the data from management, but the board structure used in this study. Financial support was may very well be influenced by ownership provided by the European Commission and concentration. Furthermore, a higher proba- the Netherlands Organisation for Scientific bility of CEO replacement was found when Research (NWO). the tasks of CEO (‘delegated’ director) and (non-executive) chairman are separated. Corporate governance relations in holding NOTES companies and especially financial institutions are much weaker than in listed * Present address: Department of Finance industrial companies. Board composition and CentER, Tilburg University, Warandelaan and leverage have a substantial impact on 2, 5000 Tilburg, Netherlands. Tel.: 31-13- board restructuring, but neither ownership 466-8210; fax: 31-13-466-2875. E-mail concentration nor a partial market for address: [email protected] (L. Renneboog). share stakes leads to increased disciplining 1 The recent changes in legislation on disclosure in holding companies. Belgian listed of voting rights now allow detailed corporate governance studies in Europe. Description of holding companies may to a large extent ownership and voting rights in Europe can be suffer from a lack of corporate control and found in Barca and Becht (2000, forthcoming. seem to have discharged themselves from Who Controls Corporate Europe?, Oxford efficient monitoring of the companies they University Press). The countries covered are control, as seems also be the case for Austria (Gugler, Kalss, Stomper and Zechner), French holding companies (Banerjee et al., Belgium (Becht, Chapelle and Renneboog), 1997). In spite of the presence of a large France (Bloch and Kremp), Germany (Becht shareholder in a vast majority of Belgian and Bohmer), Italy (Bianchi, Bianco and listed companies and of the breakdown – Enriques), Netherlands (De Jong, Kabir, Mara both in terms of performance and and Roëll), Spain (Crespi and Garcia-Cestona), Sweden (Agnblad, Berglof, Hogfeldt and governance – of the ownership cascade Svancar), UK (Goergen and Renneboog, system involving holding companies, it is 2000a,b), US (Becht). problematic that the recent codes for good 2 Still, a priori, it is not certain whether one corporate governance do not encompass specific corporate governance mechanism is any recommendation with regard to large positively related to performance as, even if one shareholder monitoring.35 mechanism may be used more frequently, the existence of other corporate governance devices and their interdependence may result in comparable equilibrium performance (Agrawal and ACKNOWLEDGEMENTS Knoeber, 1996). 3 Such recommendations have been formu- I am indebted to Julian Franks for stimulat- lated in the US Bacon report (1993), the UK ing discussions as well to Colin Mayer, Cadbury Committee report (1992), the French 308 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Viénot report (1995), the Dutch Peeters the kind: “pursuing other interests”, “spending Commission report (1997), the Belgian corpo- more time with the family” or “retirements” at rate governance guidelines by the Stock an age of 62 or below. Exchange Commission, the Association of 9 Including board size controls for the fact Employers and the Commission for Banking and that different governance mechanisms may Finance (all in 1998). prevail in large versus small companies. Large 4 An investor is considered to be the ‘ultimate companies may have a larger internal managerial or reference shareholder’ in an ownership–control labour market and have better access to an chain if control is maintained through multiple external managerial labour market. tiers of ownership. Interlocking ownership via a 10 We define a control relation between an holding company or through a more elaborate ultimate shareholder and a target company if stock pyramid enables a given investor to own (i) there is a series of uninterrupted majority different quantities of voting and cash flow shareholdings on every ownership tier through- rights. For instance, 50.1% of ownership (and out the pyramid or (ii) if there is a large share- voting rights) held by the ultimate shareholder holding of at least 25% on every ownership in an intermediary holding company which, in level in the absence of other shareholders with turn, owns 50.1% of an operating subsidiary stakes of blocking minority size or larger. could guarantee majority control on the 11 Often, Luxembourgian intermediate invest- subsidiary’s board with only a 25.1% interest in ment companies are used. its common stock cash flow. 12 These changes exclude shareholding 5 Following management changes, asset restructuring within investor groups, as these write-offs (Strong and Meyer,1987), changes to changes do not have any impact on control. income reducing accounting methods (Moore, 13 We find a negative correlation (significant 1973) or income reducing accounting accruals at the 1% level) between past corporate (Pourciau, 1993) frequently occur. performance and increases in ownership; the 6 The sample size was reduced by 9 companies lower the performance, the larger the increases in 1989 and by 10 in 1994 as these listed firms, in ownership. Note that all increases, regardless all in coal mining and steel production, were of their size, are taken into consideration involved in a long liquidation process but were because some shareholders only need a small still listed. increase in the percentage of their voting rights 7 The results do not change when we exclude to reach a blocking minority or a majority. from the sample recent IPOs or companies that 14 If the fiscal year end is e.g. March 1994, went bankrupt. Sector codes, dates of introduc- the data of this fiscal year are included in the tion and of delisting are provided by the regressions as 1993 as most of the fiscal year is Documentation and Statistics Department of in 1993. If the fiscal year end is 30 June 1994 the Brussels Stock Exchange. Companies dis- or later in 1994, the data of the year are appearing as a separate entity following included in the regressions as 1994. The yearly absorption by another company as a result of a market adjusted returns are calculated such merger are included until the year prior to that they coincide with the fiscal years of the the merger. corporations. Only lagged performance variables 8 Warner et al. (1988) and Weisbach (1988) are included because a performance variable of also mention that reasons for turnover are often the year coinciding with the year of turnover lacking. Weisbach also only excludes retire- may be a (partial) lead variable especially if the ments if they are age related (63 years or older) turnover takes place early in the fiscal year. which eliminates most of the non-linearity in the 15 Apart from the performance measures turnover–age relationship: “ . . . companies do given in Tables 11.4 and 11.5, models with lev- not announce the true reason behind their els and changes of return (after interest, taxes CEOs’ resignations. Therefore, I ignore the and extraordinary) on assets (both with and stated reasons for resignation in constructing without industry median correction) and cash my sample. I do, however, eliminate the resigna- flow margin were estimated.The results of these tions for which I am able to corroborate the models are in line with the ones discussed. cause independently. Changes in CEOs caused by 16 Including the total share concentration by death and preceding a takeover are excluded class of owner or Herfindahl indices, yields – because these ‘resignations’ are totally verifi- expectedly – similar results. Including squared able.” (p. 438). This bias is also mentioned by, ownership does not yield robust results across among others, Dennis and Dennis (1995) and models. Hermalin and Weisbach (1991). Non-informative 17 Piecewise regressions are not shown, but reasons found for leaving the company are of tables are available. OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 309 18 In the years following the take over battle along with gearing and interest coverage in between the French Suez group and the Italian order to avoid multicollinearity. group of de Benedetti in 1989, the Generale 24 Several directors (among others, M. Davignon Maatschappÿ van België or the Société (Generale Maatschappij van België/Société Générale de Belgique, was restructured using a Générale de Belgique), M. Bodson (Tractebel), focus strategy on 8 industrial and financial M. Samyn (NPM/SNP) ) were interviewed and sectors. The Group Brussels Lambert, another asked to comment on the results of this study. large holding company, has often been criticized They confirmed the limited role of creditor for failing to establish a strategic plan for the monitoring unless a large refinancing takes companies it controlled and is often given as an place and emphasized the role of the large example of a stalemate situation brought about shareholders (for which we only found limited by the reference shareholder model. The fact support) and of the board of directors (for that some of these large holding companies, which we found strong support). which control several listed (and many unlisted) 25 Using the logarithm of the total number of companies, may fail in their monitoring role has employees as size variable, yields less significance an important impact on our conclusions regarding (only 3 regressions out of 7 in Table 11.4). the governance ability of holding companies. 26 Tables with logit models on CEO turnover For a discussion, see Daems (1998) and Dewulf are available upon request. et al. (1998). 27 CEOs with long tenure are less easily 19 The findings described are robust across removed when performance is poor as they may estimation methods. OLS with fixed effects yield have a good track record. However, this conclu- somewhat stronger significance for the presence sion is only based on one fourth of the sample of large shareholdings held by industrial and companies as these data were only disclosed for commercial companies and by families: in three a limited number of companies. regressions, industrial and performance effects 28 The probability that the CEO of listed are significant. holding companies is replaced increases with 20 Unlike the models with executive and CEO the degree of family control. This is especially turnover, regressions with management the case when the market adjusted return is committee turnover only yielded weakly signifi- low and when the company faces losses and cant results. This implies that only those top has to reduce dividends. In contrast to the managers who hold a board seat are held capital structure, board composition is again a responsible and subject to disciplinary corpo- determining factor. Both a large percentage of rate control. non-executive directors and the separation of 21 Although control is exerted by the ultimate the functions of CEO and non-executive shareholder,there is evidence that when control- chairman increases the probability of CEO ling stakes are held through multiple tiers of substitution. ownership and when intermediate shareholdings 29 De Standaard of 7, 8 August 1999. deviate from full ownership, ultimate investor 30 See lines 5, 9, 13, and 17. control is diluted. For instance, sequences of 31 See lines 5, 9, 13, and 16. majority control in the form of e.g. stakes of 32 Note that the reason why the proposed 50.1% throughout the pyramid might not guar- governance model does not fit the financial antee the same degree of control a first tier companies is not due to lack of turnover. The majority holding would give, unless there is dependent variable executive turnover is strong board representation on each ownership between 22% and 37% depending on the year, tier. Consequently, the larger the number of which is similar to turnover in industrial and ownership tiers and the larger the deviation commercial companies. from full ownership, the weaker the relation 33 A similar example of a reference share- between turnover and ownership concentration. holder hampering a fair M&A competition Including levered shareholdings (see Section 4) between Fortis and ABN Amro for control of by category of owner gives similar results but the Generale Bank in 1997 has been detailed reduces significance. extensively in the financial press. 22 Including changes in gearing did not give 34 Due to sample size limitations, several any significant results. Changes in capital struc- sectors are added (based on NACE industry ture in the form of new equity issues which took classifications) to form industry subsamples: place in 40 firm-years was also included in the holdings companies, energy and utilities, chemical models but are not correlated to board turnover. and materials (metal and non-ferro), electrical 23 Current and acid ratios were significant and electronics, consumer products and health but were not included in the model of Table 11.4 care/pharma. services (transport, leasing, 310 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER hotel, ...).The utility sector was subsequently Burkart, M., Gromb, D., Panunzi, F., 1997. deleted due to small sample size. Large shareholders, monitoring and the value 35 See also Meeus (1998) and Wymeersch of the firm. Quarterly Journal of Economics, (1998) for a discussion of the content of the 693–728. codes of good corporate governance. Cadbury, A., 1992. Report of the committee on the financial aspects of corporate governance, Gee & Co Ltd, London, December. REFERENCES Coughlan, A.T., Schmidt, R.M., 1985. Executive compensation, managerial turnover, and firm Aghion, Bolton, 1992. An incomplete contracts performance: An empirical investigation. approach to financial contracting. Review of Journal of Accounting and Economics 7, Economic Studies 59, 473–494. 42–66. Agrawal, A., Knoeber,C., 1996. Firm performance Daems, H., 1998. De paradox van het Belgisch and mechanisms to control agency problems kapitalisme, Lannoo, Brussels. between managers and shareholders. Journal Demsetz, H., 1983. The structure of ownership of Financial and Quantitative Analysis 31, and the theory of the firm. Journal of Law 377–397. and Economics 26, 375–390. Bacon, J., 1993. Corporate Boards and Demsetz, H., Lehn, K., 1985. The structure of Corporate Governance, published by The corporate ownership: Causes and conse- Conference Board in co-operation with Booz quences. Journal of Political Economy 93, Allen & Hamilton Inc. 1155–1177. Banerjee, S., Leleux, B., Vermaelen, T., 1997. Denis, D.J., Denis, D.K., 1993. Managerial Large shareholdings and corporate control: discretion, managerial structure, and An anlysis of stake purchases by French corporate performance: A study of leveraged holding companies. European Financial recapitalizations. Journal of Accounting and Management 3, 23–43. Economics 16, 209–236. Barca, F., Becht, M., 2000. Who Controls Denis, D.J., Denis, D.K., 1995. Performance Corporate Europe. Oxford University Press, changes following top management dis- Oxford. missals. Journal of Finance 50, 1029–1057. Barclay, M.J., Holderness, C.G., 1989. Private DeWulf, H. et al., 1998. Corporate Governance: benefits from control of public corporations. Het Belgische Perspectief, Intersentia Journal of Financial Economics 25, Rechtswetenschappen, Antwerpen-Groningen, 371–395. 245 p. Barclay, M.J., Holderness, C.G., 1991. Elliott, J. and Shaw, W. 1988. Write-offs as Negotiated block trades and corporate accounting procedures to manage percep- control. Journal of Finance 46, 861–878. tions. Journal of Accounting Research Baysinger, B.D., Butler, H.N., 1985. Corporate Supplement to vol. 26, 32–40. governance and the board of directors: Fama, E., Jensen, M., 1983. Separation of own- Performance effects of changes in board ership and control. Journal of Law and composition. Journal of Law, Economics and Economics 26, 301–325. Organization 1, 101–124. Franks, J., Mayer, C., 1998. Ownership and Berkovitch, W., Israel, R., Spiegel, Y., 1997. control of German corporations. Working Managerial compensation and capital paper, London Business School. structure, Working paper. Franks, J., Mayer,C., Renneboog, L., 1998.Who Bolton, P., von Thadden, E., 1998. Blocks, disciplines bad management? Discussion liquidity and corporate control. Journal of paper, CentER,Tilburg University. Finance 53(1), 1–25. Goergen, M., Renneboog, L., 2000a. Insider Bratton, W., McCahery, J., 1999. Comparative control by large investor groups and managerial corporate governance and the theory of the disciplining in listed Belgian companies. firm:The case against global cross reference, Managerial Finance, forthcoming. Working paper, Law faculty, Tilburg Goergen, M., Renneboog, L., 2000b. Prediction University. of ownership and control concentration OWNERSHIP,MANAGERIAL CONTROL AND GOVERNANCE 311 subsequent to initial public offerings in Rechtswetenschappen, Antwerpen-Groningen, Germany and the UK. In: Renneboog, L., 245 p. McCahery, J., Moerland, P., Raaijmakers, T. Michaely, R., Thaler, R., Womack, K., 1995. (Eds.), Convergence and Diversity of Price reactions to dividend initiations and Corporate Governance Regimes and Capital omissions: Overreaction or drift? Journal of Markets. Oxford University Press, Oxford. Finance 50(2), 573–608. Gilson, S.C., 1990. Bankruptcy, boards, banks, Moore, M., 1973. Management changes and and blockholders. Journal of Financial discretionary accounting decisions. Journal of Economics 27, 55–387. Accounting Research 11, 100–107. Grossman, S.J., Hart, O., 1980. Takeover bids, Morck, R., Shleifer, A., Vishny, R.W., 1989. the free-rider problem, and the theory of the Alternative mechanisms for corporate corporation. Bell Journal of Economics 11, control. American Economic Review 79, 42–64. 842–852. Grossman, S.J., Hart, O., 1988. One share-one Moses, D., 1987. Income smoothing and incen- vote and the market for corporate control. tives: Empirical tests using accounting Journal of Financial Economics 20, changes. The Accounting Review 62, 175–202. 358–377. Healey, P.,Palepu, K., 1988. Dividend initiations Murphy, K.J., Zimmerman, J.L., 1993. and omissions as earnings signals. Journal of Financial performance surrounding CEO Financial Economics 21, 149–176. turnover. Journal of Accounting and Hermalin, B.E., Weisbach, M.S., 1991. The Economics 16, 273–315. effects of board composition and direct Ofer, A., Siegel, D., 1987. Corporate financial incentives on firm performance. Financial policy, information and market expectations: Management 20, 101–112. An empirical investigation of dividends. Holderness, Sheehan, 1988.The role of majority Journal of Finance 42, 889–912. shareholders in publicly held corporations: An Pourciau, S., 1993. Earnings management and exploratory analysis. Journal of Financial nonroutine executive changes. Journal of Economics 20, 317–346. Accounting and Economics 16, 317–336. Jensen, M.C., Meckling, W.H., 1976. Theory of Renneboog, L., 1997. Shareholding concentration the firm: Managerial behaviour, agency costs and pyramidal ownership structures in and ownership structure. Journal of Financial Belgium. In: Balling, M., Hennessy, E., O’Brien, Economics 3, 305–360. R. (Eds.), Corporate Governance, Financial Joskow, P.L., Rose, M.L., 1994. CEO pay and Markets and Global Convergence. Kluwer firm performance: Dynamics, asymmetries, Academic Publishers, Dordrecht, December. and alternative performance measures. Shleifer, A., Vishny, R.W., 1986. Large share- Working paper Massachusetts Institute of holders and corporate control. Journal of Technology, Cambridge, MA. Political Economy 95, 461–488. Kaplan, S.N., 1994.Top executives, turnover and Shleifer, A., Vishny, R.W., 1997. A survey of firm performance in Germany. Journal of Law, corporate governance. Journal of Finance Economics and Organization 10, 142–159. 52(2), 737–783. Kaplan, S.N., Minton, B.A., 1994. Strong, J., Meyer, J., 1987. Asset writedowns: Appointments of outsiders to Japanese Managerial incentives and security returns. boards: Determinants and implications for Journal of Finance 20, 643–663. managers. Journal of Financial Economics Teoh, S., Welch, I., Wong, T., 1998. Earnings 36(2), 225–258. management and the underperformance of Kaplan, S.N., Reishus, D., 1990. Outside direc- seasoned equity offering. Journal of Financial torships and corporate performance. Journal Economics 50(1), 63–99. of Financial Economics 27, 389–410. Viénot, M., 1995. The Board of Directors of Meeus, D., 1998. De recente Belgische Listed Companies in France, published by aanbevelingen inzake corporate governance. Conseil National du Patronat Français and In: De Wulf, H. et al., Corporate Governance: the Association Française des Enterprises Het Belgische Perspectief, Intersentia Privées, Paris, July 10. 312 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Warner, J.B., Watts, R.L., Wruck, K.H., 1988. Williamson, O.E., 1983. Organizational form, Stock prices and top management changes. residual claimants, and corporate control. Journal of Financial Economics 20, Journal of Law and Economics 26, 351–380. 431–460. Wymeersch, E., 1994. Aspects of Corporate Weisbach, M., 1988. Outside directors and CEO Governance. Journal of Corporate turnover. Journal of Financial Economics 20, Governance 2, 138–149. 431–460. Wymeersch, E., 1998. De Belgische initiatieven White, H., 1980. A heteroscedasticity- inzake corporate governance. In: De Wulf, H. consistent covariance matrix estimator and et al., Corporate Governance: Het Belgische a direct test for heteroscedasticity. Perspectief, Intersentia Rechtswetens chap- Econometrica 48(2), 3–20. pen, Antwerpen-Groningen, 245 p. Chapter 12

Julian Franks, Colin Mayer and Luc Renneboog2 WHO DISCIPLINES MANAGEMENT IN POORLY PERFORMING COMPANIES?1

Source: Journal of Financial Intermediation, 10(3–4) (2001): 209–248. ABSTRACT Economic theory points to five parties disciplining management of poorly performing firms: holders of large share blocks, acquirers of new blocks, bidders in takeovers, nonexecutive directors, and investors during periods of financial distress. This paper reports the first comparative evaluation of the role of these different parties in disciplining management. We find that, in the United Kingdom, most parties, including holders of substantial share blocks, exert little disciplining and that some, for example, inside holders of share blocks and boards dominated by nonexecutive directors, actually impede it. Bidders replace a high proportion of management of companies acquired in takeovers but do not target poorly performing management. In contrast, during periods of financial constraints prompting distressed rights issues and capital restructuring, investors focus control on poorly performing companies.These results stand in contrast to the United States, where there is little evidence of a role for new equity issues but nonexecutive directors and acquirers of share blocks perform a disciplinary function. The different governance outcomes are attributed to differences in minority investor protection in two countries with supposedly similar common law systems. Journal of Economic Literature Classification Number: G3. © 2001 Academic Press Key Words: corporate governance; control; restructuring; board turnover; regulation.

1. INTRODUCTION may act on behalf of shareholders and replace management when they are HOW DO CAPITAL MARKETS DISCIPLINE THE thought to perform poorly. Finally, financial management of poorly performing firms? crises may trigger interventions by share- We attempt to answer this question in the holders when new equity is issued. context of the UK capital market by This paper provides the first comparative running a “horse race” between the five assessment of the degree of managerial principal competing parties suggested in disciplining provided by the five parties.The the literature. First, shareholders, and in assessment evaluates the relation between particular large shareholders, may inter- disciplining and the performance of firms. vene directly and replace management We examine the extent to which the parties when performance is poor. Second, man- provide significant disciplining of poorly agement replacement may follow the performing management and consider acquisition of a large block of shares. whether that disciplining is focused. Third, bidders may discipline the manage- We measure focus by whether disciplining ment of the acquired company. Fourth, is concentrated exclusively on poorly nonexecutive directors, i.e., outside directors, performing firms. We measure significance 314 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER by the extent to which different interventions these results stand in marked contrast to contribute to high board turnover in poorly what was described above for the United performing companies. A governance Kingdom. mechanism can be focused on poorly We believe that these differences are at performing companies but have an insignif- least in part a consequence of regulation. icant effect on overall board turnover in At first sight, this is surprising since the these firms, i.e., have a high level of Type 1 United Kingdom and United States are errors. A governance mechanism can be countries usually characterized as having significant in dismissing a large number of similar “common law” regulatory systems managers in poorly performing companies (see La Porta et al., 1998). Four examples but be unfocused in also dismissing a large will illustrate significant differences in number of other managers in well perform- regulation and how these lead to different ing firms, i.e., have a high level of Type 2 governance outcomes in the two countries. errors. First, the United Kingdom has a Takeover The evidence reported in Franks and Code that makes accumulation of control- Mayer (1996) illustrates this distinction. ling blocks expensive; the United States They report that, on average, target firms does not. Second, the United Kingdom has are not poorly performing companies. stronger minority protection laws making Bidders do not therefore provide a focused the acquisition of partial controlling blocks form of corporate control. Nevertheless, as well as takeovers expensive. For exam- they could still perform a significant ple, in the United Kingdom a transaction disciplinary function if they give rise to a between a large shareholder and its con- high level of managerial replacement in nected company must be undertaken at poorly performing firms. arms length and requires the consent of The horse race evaluates both the focus noncontrolling shareholders; in the United and significance of disciplining by different States, shareholder protection is limited to parties.The results are quite striking.We find seeking redress in the courts for unfairly that at least two parties—nonexecutive priced transactions. Third, in the United directors and directors with large share Kingdom all seasoned equity issues, above stakes—tend to entrench management by 5% of share capital, have to be in the form reducing board turnover in poorly performing of rights issues and rights requirements can firms. Neither existing holders nor new only be waived with considerable difficulty. purchasers of large share blocks exert much These requirements provide dispersed disciplining and, as noted above, bidders shareholders with significant control when impose high board turnover after takeovers firms need to raise new equity financing. In but in an unfocused way.It is only when there the United States, shareholders can and do is financial distress, requiring equity issues waive rights issue requirements for almost and capital restructuring, that disciplining is all seasoned offerings. Finally, in the United both significant and focused on the manage- States there are significant fiduciary ment of poorly performing firms. obligations on directors, breaches of which Some of the results reported for the create high management turnover United Kingdom are similar to those (Romano, 1991). There are few such recorded for the United States, most obligations in the United Kingdom where notably entrenchment by insiders. “actions to enforce the duties of directors However, others are quite different. In the of quoted companies have been almost United States, nonexecutives perform a nonexistent” (see Stapledon, 1996, pp. disciplinary function (see Hermalin and 13–14). Where the role of nonexecutives is Weisbach, 1991); active outside shareholders strengthened in the United Kingdom, discipline management when share blocks through, for example, the adoption of the change hands (see Bethel et al., 1998); and Cadbury Code (see Cadbury, 1992), it has there is no reported role for new share been shown that the disciplinary function of offerings in disciplining management. All boards increases (see Dahya et al., 2000). MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 315 In summary, we find that there are randomly selected from all companies differences in regulation, even within two quoted on the London Stock Exchange in common law countries, and that these 1988. We collected data on their perform- differences are associated with signifi- ance and board turnover over the period cantly different governance outcomes. The 1988 to 1993 and combined these with results of this paper therefore enrich the information on ownership, sales of share La Porta et al., (1998) view of regulation blocks, takeovers, board structure, and as an important influence on the operation capital structure. of capital markets. To be included in the sample, companies Section 2 of the paper describes the were required to have data on boards and methodology employed in this paper. ownership for at least 3 of the first 6 years Section 3 examines the relation between of the sample period to allow panel data board turnover and performance and the analyses to be performed. Companies role of each of the five parties in disciplin- delisted through takeovers or insolvencies ing poor management in the United for the first 3 years, between 1988 through Kingdom. Section 4 reports regressions of to 1990, were therefore excluded. In board turnover on performance: Section 4.1 addition, 7 of the original 250 companies reports the results for the total sample and were dropped because of lack of perform- Section 4.2 for poorly performing firms. ance data. Section 5 contrasts the UK results in this Data on the composition of the board of paper with those reported for the United directors were compiled for each year from States and shows how regulation may 1988 to 1993 from annual reports, account for these differences. Section 6 Datastream, the Financial Times, and summarizes the paper. Nexus databases. They include the names, tenure, and age of the CEO, Chairmen, and all directors, both executive and nonexecutive. 2. METHODOLOGY We measured annual executive turnover of the board from 1988 to 1993. Board This paper is concerned with identifying turnover is calculated by dividing the total who precipitates board restructuring in number of directors who leave the poorly performing firms. The literature on company by total board size. Executive and managerial disciplining points to five nonexecutive turnover is calculated in the parties: (i) Shleifer and Vishny (1986) same way, except that the denominator is show that large shareholdings mitigate free the number of executive and nonexecutive rider problems of corporate control; directors, respectively. CEO and Chairman (ii) Scharfstein (1988) models the way in turnover represents the proportion of which takeovers perform a disciplinary sample companies where the CEO and function; (iii) Burkart et al. (1997, 1998) Chairman, respectively, leave the company. argue that trades in share blocks may be All turnover figures are corrected for more cost effective than full takeovers; natural turnover. We distinguish between (iv) Fama (1980) and Fama and Jensen natural and forced turnover, classifying a (1983) describe how managerial labor resignation as “natural” if the director was markets and nonexecutive directors assist described as having left the board for in the governance of firms; and (v) Jensen reasons of retirement, death, or illness. (1986) and Aghion and Bolton (1992) Otherwise the resignation was classified as discuss the role that capital structure plays being forced. The normal retirement age is in reducing agency costs. between 62 and 65 but some voluntary This paper examines the role of all 5 retirement does occur before that; we took forms of interventions in disciplining 62 as the minimum retirement age and management. A sample of 250 companies, viewed any earlier retirement as forced.3 excluding financial institutions, real estate This reflects the difficulty of establishing companies, and insurance companies, was whether public announcements of resignations 316 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER result from forced retirements or reflect also examine performance measured the natural career progression of good relative to industry benchmarks, namely managers. Where a company was taken abnormal returns and return on equity over after 1990, board turnover was relative to industry averages. collected for two years after the acquisition Since the focus of this paper is on the to determine the survival rate of the disciplining of management of poorly pre-takeover directors of the target firm. performing companies, we investigated the In the regressions, we examine the relation between very poor performance relation of executive board and CEO and executive board turnover in greater turnover to five different measures of per- detail. Our sample of 243 companies has formance: abnormal share price returns, about 24 companies in the lowest decile in dividend cuts and omissions measured as a any one year, but across the entire sample dummy variable (equal to minus one where the number of different companies in the there is a cut or omission), after tax cash lowest decile totals 90.5 To expand the set flow margins (cash flows divided by total of poorly performing companies, we sales), after tax rates of return on book collected a second sample of 50 companies equity, and earnings losses measured as a in the lowest decile of abnormal share price dummy variable (equal to minus one where returns in any one of the three years from there is a loss). Abnormal share price 1988 to 1990. For the sample of all poorly returns were taken from the London Share performing companies, we used two further Price Database (LSPD).4 We employed measures of poor performance: earnings two measures of leverage, the ratio of losses combined with dividend cuts and pre-tax earnings to interest charges, and a omissions, and abnormal returns of less debt to book (and market) value of assets than minus 50% combined with earnings ratio. losses and dividends cuts and omissions. Different performance measures were In Section 3, we provide a univariate used because it is not always clear what analysis of the relation between board constitutes poor performance and because turnover and performance, and the parties governance may be more sensitive to one organizing interventions. We report either performance measure than another. Marsh individual year data for the whole sample (1992) examines 6000 UK dividend period, or we choose the sample of companies announcements over the period 1989 to from 1990 and aggregate data over three 1992. His evidence, which is consistent years from 1990 to 1992. with that in the United States, shows that In Section 4 we report the results of “dividend cuts are interpreted by the panel regressions of executive board market as powerful signals of bad news turnover on performance, ownership, and both about the current situation and about capital structure over the period 1988 to future prospects” (p. 50). Ball et al. 1993. We relate executive board turnover (1997) examine the discretion that man- to performance in the current year, and agers have in different countries to smooth with lags, to five classes of variables: earnings; for example, UK managers have little discretion while German managers (i) Ownership for the different cate- have considerable discretion and as a gories of investors described below, result, tend to use hidden reserves to (ii) Changes in share stakes of different smooth earnings and hide earnings losses. categories of investors, The three measures (abnormal returns, (iii) Takeovers, dividend cuts and omissions, and earnings (iv) Board structure: the proportion of losses) yield different incidence of poor nonexecutives on the board and sep- performance in our data set; for example, aration of the position of chairman 17.0% reported dividend cuts or omissions and CEO, on average each year, compared with (v) Capital structure and the incidence 10.4% that reported earnings losses. We of new equity issues. MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 317 In addition we include interactive terms its effect on board turnover of the worst between performance and the above five performing companies, in other words in categories.The results reported below refer having a high level of Type-1 errors. A to interactive terms with performance governance mechanism can be significant lagged one year; regressions using interac- in dismissing a large number of managers tive terms with contemporaneous perform- in poorly performing companies but ance were also performed. unfocused in dismissing a large number of We report executive board turnover other managers as well; i.e., it has a high panel regressions estimated using a Tobit level of Type-2 errors. regression to take account of fact that To rank the contribution of different there are frequently no changes to boards. governance mechanisms to board turnover, We also undertook OLS regressions with we report their economic as well as logistic transformations of the dependent statistical significance. Economic significance variable. Within (fixed effect) regressions is measured as the effect on board turnover were also performed and time dummies for of moving from the mean to the extreme individual years were used. Since high (upper or lowest) decile value of the board turnover in one year might lead to relevant independent variable. The eco- low turnover in a subsequent year,we inves- nomic significance of events, such as tigated the robustness of the results using a takeovers and new equity issues, are meas- cross-sectional OLS regression where the ured by their marginal impact on board dependent variable is accumulated board turnover,i.e., the coefficient on the relevant turnover for the year of poor performance dummy variable. and two subsequent years. The results reported below refer to the panel regressions that include time as well as cross- 3. DATA ON BOARD TURNOVER sectional effects. AND PERFORMANCE AND PARTIES In Section 4.1 we report the results for INITIATING INTERVENTIONS the complete randomly selected sample of firms. Although the potential size of our sample is 1458 firm years (number of com- This section investigates the relation panies 6 years), it is reduced to 1193 between board turnover, performance, firm years as a result of takeovers (180), and the five different parties initiating bankruptcies (10), and missing data (75), interventions. and if the independent variable is lagged then there is a further loss of 243 firm 3.1. Board turnover and years. In Section 4.2 we report regressions performance for a sample of poorly performing firms drawn from the lowest decile of abnormal Table 12.1 provides a snapshot of the returns in any one of the years 1988 to relation between board turnover and 1990. performance. Panel A of the table parti- The results in Section 4.1 provide a tions the sample into deciles of perform- measure of focus—To what extent are ance using abnormal returns. We choose different interventions focused exclusively the sample of companies from 1990 and on high executive board turnover in poorly aggregate data over three years from 1990 performing firms? The results in Section 4.2 to 1992. The period 1990 to 1992 is provide an indication of significance—To chosen because 1990 is the first year when what extent do different interventions con- the threshold for disclosing share stakes tribute significantly to high board turnover held by outsiders was reduced from 5 to in badly performing companies? A gover- 3%, and subsequent regressions demon- nance mechanism can be focused, in the strate that performance has an impact on sense of being well targeted on just poorly executive board turnover in the current and performing companies but insignificant in following two years. 318 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 243 differences test - test on decile - difference corrected for natural turnover). natural for corrected Board –1990 ice performance in 1990 for six deciles; 1990 for in ice performance year period 1990–1992. period year Board turnover 23% Average 1 vs 5 vs 10 1 executive members,executive respectively. CEO and t, death, by total board size. or illness) Executive test on t - t 13% to 5% 12% to 23% 19% 27% to 27% 45% to 45% UK Quoted Companies Annual Board Turnover in 1990–1992 Partitioned by Decile of Abnormal Share Price Performance Calculated in 1990 for a Sample of 1990 for Calculated in Performance Share Price of Abnormal by Decile 1990–1992 Partitioned in Annual Board Turnover This table reports the turnover of the total board, directors, reports table the turnover and nonexecutive This the three- for the executive the CEO and chairmen Annual board Dividend cuts Constant or differences in Earnings Positive on Sample sizePanel B: Board turnover over 1990–1992 partitioned by dividends changes and earnings, respectively during 1989 24 24 24 23 23 23 23 Chairmen 15.8% 6.9% 5.8% 7.2% 2.9% 5.9% 7.0% 0.862 1.007 CEOs 28.8% 19.2% 11.4% 11.6% 14.5% 10.4% 13.6% 1.815 1.953 Source: Annual reports, Database. Datastream and London Share Price panel B categorizes companies by dividend cuts and omissions, by dividend 1990. companies losses in B categorizes panel and earnings Chairmen 15.9% 8.4% 1.464 19.1% 5.1% 2.302 turnover data for the sample of 243 companies are categorized by performance: share pr by abnormal companies are categorized the sample of 243 companies A categorizes data for panel turnover Nonexecutive directorsNonexecutive 7.4% 1.7% 4.9% 4.2% 3.6% 4.8% 4.2% 0.758 0.511 CEOs 25.9% 7.0% 2.795 26.0% 11.2% 1.846 chairman turnover represents the proportion of sample companies where the CEO and chairman, where of sample companies represents the proportion turnover chairman respectively, leave the company ( Nonexecutive directorsNonexecutive 6.0% 5.8% 0.077 9.8% 3.7% 1.965 Executive directorsExecutive 21.1% 10.0% 9.9% 8.1% 8.0% 6.9% 9.2% 2.300 2.388 and nonexecutive director turnover is calculated in the same way, except that the denominator is the number of executive and non the same way, is calculated in of executive is the number turnover director that the denominator except and nonexecutive Total boardTotal 15.5% 7.0% 7.7% 6.8% 6.0% 6.4% 7.3% 2.122 2.085 is calculated by dividing the total number of directors who leave the company (excluding those leaving as a result of retiremen those leaving of directors who the total number leave the company (excluding is calculated by dividing Executive directorsExecutive 14.9% 7.0% 2.776 17.5% 8.0% 2.754 Panel A: Annual Board turnover over 1990–1992 partitioned by decile of abnormal share price performance in 1990 t Decile of abnormal returns Worst 1 2 3 5 9 Best 10 Table 12.1 Annual board turnover of: turnover of: board turnoverTotal 11.5% (average)Sample size and omissionsNote. increased dividends 6.2% 36 means 181 2.640 losses earnings 14.7% in means 6.4% 3.303 18 190 MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 319 Table 12.1 records that there is a high (vi) executive directors and their immediate level of board turnover in poorly performing family and trusts, and (vii) nonexecutive companies. It also shows that the relation directors and their immediate family and between board turnover and performance is trusts.8 We will refer to directors and their highly nonlinear. Annual board turnover is families as “insiders” and financial substantially higher in decile one than in institutions, industrial and commercial any of the other deciles, for example, companies, and other major shareholders 15.5% compared with 6.8 and 6.4% for as “outsiders.”9 The sizes of share stakes deciles five and ten, respectively. Executive held by government and real estate compa- board turnover is much higher than nonex- nies were collected but not reported ecutive turnover because nonexecutives because they are so small. perform both a monitoring and advisory The distinction between different outside function. CEO turnover is also much higher holdings is important because some may be in decile 1 (at almost 28.8%) than in other passive in the face of poor management deciles (for example, 11.6% in decile 5). performance while others are active. For Turnover of chairmen is relatively high in example, institutional shareholders are decile one, although the level is lower than often regarded as passive, and industrial for CEOs, reflecting the fact that some companies and individuals/families as chairmen are nonexecutive and perform a active. Corporate investors may have more monitoring role. When the sample was knowledge about the industry than other partitioned using abnormal returns accu- investors, and individuals and families may mulated over two years, 1989–1990, the have more incentive to intervene as relation between performance and turnover principals rather than agents. was very similar to that in panel A.6 This section reports the pattern of Panel B reports that companies with ownership for the sample companies. Panel A losses and companies with dividend cuts or of Table 12.2 records the largest individual omissions have more than twice the execu- share holding for all companies in each tive board turnover of better performing year from 1988 to 1993 with the average companies. Companies with dividend cuts for all years being 15.3%. The largest five have 3.7 times the CEO turnover of those shareholders accounted for between 29.7 with increasing or stable dividends, and and 36.7% of the company’s shares companies with earnings losses have 2.3 depending on the year. There is a large times those without. We therefore find that increase in reported blocks from 31.4 to there is a strong, nonlinear relation 41.0% between 1989 and 1990, which we between board turnover and performance. attribute to the change in the disclosure rule on block ownership, referred to earlier. Panel B of Table 12.2 reports that the 3.2. Ownership concentrations median size of the largest stake lies in the We collected data on the size of shareholdings range 5–15% for all individual years, but over the period 1988–1993. All directors’ there are a significant number of blocking holdings greater than 0.1% are included as minorities, defined as a stake of at least well as outside share stakes greater than 25%. For example, in 1988 almost 24% of 10 5% until 1989. From 1990, the statutory stakes are in excess of 25%. Panel C dis- disclosure threshold for outside sharehold- aggregates large shareholders by their type ers was reduced to 3%.7 and size of holding in 1991. Institutional Shareholdings were classified according investors hold the highest proportion to 7 categories: (i) banks, (ii) insurance (52.6%). Insiders, directors, and their companies, (iii) institutional shareholders families are the next most significant hold- including investment trusts, unit trusts, and ers. The difference between outsider and pension funds, (iv) industrial and commer- insider blocks is important because the cial companies, (v) families and individuals, latter may entrench poor management. not directly related to any director, Although not shown in the table, insider 320 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER b a vely ldings over the period 1988 to 1993. the period over ldings All reported large shareholdings ings by outsiders in excess of 5% are excess in by outsiders ings ports the size distribution of the largest ports distribution the size b 97.4% 9.2% 51.8% 19.2% 14.3% 2.9% ividuals 3.8% 0.0% 1.3% 0.8% 1.7% 0.0% Size and Category of the Largest Shareholder in the Sample of UK Quoted Companies, in of the Largest and Category Shareholder Size 1988 to 1993 Panel A shows the average largest shareholding, the average A shows Panel the sum of largest 5 shareholdings, and the sum of all reported large shareho c Panel C: Proportion of firms with the largest shareholder by type and size holding for 1991 1993 15.5% 58.1% 13.5% 11.6% 1.3% b 1989. 5 to 3% in from threshold the disclosure in the reduction 1990 reflects in increase The c are not included. and real estate companies by the government held sum is less than 100% because small shareholdings The 1992 10.6% 56.0% 17.4% 13.5% 2.4% a 239, were Sample sizes 241, 241, 240, 207, 1988 to 1993, each of the six years and 205 in respectively. 1991 10.0% (directors)Insiders 52.1% 28.0% 2.1% 19.6% 8.8% 7.9% 14.6% 7.1% 3.8% 2.1% shareholding by type of shareholder in 240 companies in 1991. in 240 companies in by type of shareholder shareholding than 0.1% are included; by directors greater All holdings sharehold of 3% thereafter. Source: excess reported 1988 and 1989 in in Annual reports. 1990 11.6% 44.2% 21.9% 18.2% 4.1% Panel A: Size of largest shareholding, sum ownership 5 share stakes and all reported large shareholdings respecti Panel B: Percentage of firms with largest shareholding by size holding Families, ind 1990199119921993 16.5%1989 15.6% 15.0% 13.7% 17.0% 39.8% 36.4% 36.7% 35.2% 30.4% 20.7% 41.0% 42.7% 40.7% 19.1% 33.5% 3.3% Panel B shows the percentage of companies with a large shareholding in five size ranges over the period 1988 to 1993. the period over ranges size C re five in Panel with a large shareholding of companies the percentage B shows Panel Table 12.2 Year19881989 Largest shareholding 15.3%Year 15.6% [0–5%] Largest 5 shareholdings Category of shareholder Total [5–15%] 29.7%All 30.7%Note. [0–5%] [15–25%] [5–15%] [15–25%] [25–50%] 30.6% 31.4% [25–50%] [50–100%] [50–100%] 1988 23.0% 34.7% 18.4% 20.9% 2.9% Industrial companiesIndustrial 13.0% 0.8% 3.8% 4.2% 3.8% 0.4% Institutional investors 52.6% 6.3% 37.9% 6.3% 1.7% 0.4% MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 321 holdings are roughly split two-thirds In contrast, board turnover is closely executive and one-third nonexecutive direc- related to performance (as noted above) tors. The size distributions of institutional but not to the size of firms: differences in investors and insider holdings are very board turnover are not economically large different. Insiders have a greater number of or statistically significant across the two blocking minority stakes than financial groups of firms, those below and those institutions, for example, 9.2% of their above the median capitalization. stakes are greater than 25% compared In summary, although the United with 2.1% for institutional investors. Kingdom is described as a relatively dis- Levels of concentration in the United persed capital market, coalitions of share- Kingdom are similar to those reported in holders can potentially exert significant the United States. Holderness and Sheehan voting power, insiders have substantial (1988) find that 13% of all publicly traded blocks, and there is a strong relation corporations and 5% of companies traded between concentration of ownership and on the NYSE and AMEX exchanges have a size, but not performance of firms. Since we single shareholder (family or another firm) have observed in the previous section a holding a majority of the shares. In our UK strong association between board turnover sample, the figure is 3%. Measuring and performance but not to the size of concentration by cumulating the largest firms, this suggests that concentrations of five holdings, Demsetz and Lehn (1985) ownership may not bear a close relation to report a mean of 24.8% in the United board turnover. We find confirmation for States compared with 33% in our UK this in the regression results in Section 4. sample. Denis and Denis (1995) record insider ownership of 11.7% for the United States, as against 11.8% for our 3.3. Takeovers and trades in share UK sample.11 blocks By Continental standards, these are low In this section we report the incidence of levels of concentrations. In Italy, 84% of full acquisition of firms and trades in share Italian companies have a single share- blocks.The rate of takeovers in the original holder owning majority stakes (see Bianco sample of 243 companies, for the period up et al., 1996). In Belgium, 93% of quoted to 1993, is 6%.12 If those same companies industrial companies have a single share- are tracked to 1997, the rate of acquisition holder who owns a block of at least 25% of increases to 13%. The rate of takeover for voting rights (Renneboog, 2000), and in the second sample, of poorly performing Germany, there is a single shareholder with firms, is substantially higher. Over the at least 25% of shares in 85% of large period 1988–1993, the rate of acquisition quoted companies (see Franks and Mayer, was 22%.13 If those same companies are 2000). tracked to 1997, the rate of acquisition Table 12.3 disaggregates the UK sample increases to 28%. This suggests that there by both size and performance measured by is a higher acquisition rate among the abnormal returns for three years worst performing companies but this only 1990–1992.There is little relation between occurs after considerable lag from the year concentration of ownership and perform- of poor performance.14 ance; for example, largest shareholdings Data on board turnover were collected in are similar in the worst and best perform- companies subject to takeover both before ing firms. However,concentration is related and after the acquisition. They suggest rel- to the size of equity capitalization; for atively low rates of turnover prior to the example, the sum of institutional shares is takeover, but very high turnover post- significantly greater in below-median- takeover.For example, during the two years capitalization than in above-median- pre-takeover, total annual board turnover capitalization firms, 31.3% compared with is 15.6% and annual CEO turnover is 20.5% (for the worst performing sample). 17.4%, whereas for the two years 322 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 0.847 0.267 1.364 tests on differences - rket capitalization in 1990. in capitalization rket chairman, respectively, leave the com- tions, companies, industrial and families g those leaving as a result of retirement, g those leaving e denominator is the number of executive and of executive is the number e denominator Smallest Belowmedian of market Large Above median of market t Companies, Capitalization of Equity Market by Size Partitioned rming CEO turnover 6.7% 14.8% Executive board turnoverExecutive 6.5% 7.7% Sum of insiders’ shares (%) 11.3% 9.9% 0.684 Largest shareholding (%)Largest shareholding 14.0% 13.0% 0.266 Sample size 23 8 CEO turnover 28.6% 28.0% 0.030 Executive board turnoverExecutive 17.5% 14.7% 0.382 Sum of insiders’ shares (%) 13.1% 6.4% 1.753 Average for 1990–1992 capitalization capitalizationSample size in means 15 27 Concentration of Ownership in of Ownership Concentration and Best Perfo Worst : Annual reports, Database. London Share Price . The table reports average ownership concentration over the period 1990 to 1992 for firms with below median and above median ma median and above median with below firms 1990 to 1992 for . the period over concentration ownership reports table average The abnormal returns 1990 returns abnormal shares (%) and family Sum of companies 3.5% 6.6% Source quintile of annualquintile shares (%) investor Sum of institutional 29.9% 12.5% 6.627 pany (corrected for natural turnover). natural for pany (corrected Best performing nonexecutive members,nonexecutive the CEO and where of sample companies represents the proportion respectively. turnover CEO and chairman death, or illness), the same way, is calculated in by total board size. that th turnover director except and nonexecutive Executive abnormal returns 1990 returns abnormal shares (%) and family Sum of companies 7.3% 5.9% 0.485 quintile of annualquintile shares (%) investor Sum of institutional 31.3% 20.5% 3.030 individuals, and insiders. of directors who the total number leave the company (excludin is calculated by dividing Board turnover Worst performingWorst (%) Largest shareholding 16.2% 11.9% 1.029 Ownership concentration is reported for the largest shareholding and the sum of all disclosed shares held by investment institu by investment shares held and the sum of all disclosed the largest is reported shareholding for concentration Ownership Table 12.3 Note MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 323 post takeover they are 88 and 94%, In Panel A of Table 12.4 we show the respectively.15 Moreover, there is little total number of purchases of share blocks difference in turnover between poorly per- by new shareholders in excess of 5% for forming companies and better performing the three-year period, 1991 to 1993.16 We companies, suggesting that takeovers pro- choose this subperiod because the disclo- vide relatively unfocused disciplining. For sure threshold changed to 3% in 1990. example, if we define poorly performing There are a total of 303 purchases of companies as those in the bottom decile of stakes greater than 5% and 82 greater abnormal share price performance in one than 10%; the latter represents an annual of the two years prior to takeover, total rate of 9% per year compared with 6.7% board turnover is identical at 88% post for the United States cited in Bethel et al.17 takeover in both samples of poorly Almost one-half of block sales greater than performing and better-performing firms. 10% are made by companies, families, and In the United States, Bethel et al. insiders, suggesting greater scope for more (1998) examine the relation between block active investing. In Panel B of the table we purchases of 5% or more and firm per- examine changes in the level of concentra- formance.They find that activist blockhold- tion by both existing and new shareholders. ers acquired stakes in highly diversified We accumulate share blocks by adding firms with poor profitability.They also find together individual purchases in each com- that the target firm’s profitability pany greater than 5% in any one year.The increased after the block purchase, and as panel reports that in 89 companies more a result, they conclude that this market than 10% of the equity was purchased by works as a market for corporate control. existing and new shareholders over the We investigate the size of this market in 3-year period. Panel B also shows that a share blocks in the United Kingdom and the single investor or a coalition of blockholders extent to which it is motivated by poor per- purchases an equity stake of 25% or more formance leading to the disciplining of in 19 companies.The acquisition of blocking management. We also distinguish between minority stakes takes place on average in active and passive blockholders. 3.1% of listed firms per year,similar to the

Table 12.4 Purchases of Share Blocks for 243 Companies for the Period 1991 to 1993

Panel A: Number of share blocks purchased by new shareholders exceeding 5% for the period 1991 to 1993 [5–10%] [10–25%] [25–50%] 50% Total Institutions 168 38 5 0 211 Companies 24 15 5 0 44 Families 19 6 2 0 27 Directors 10 11 0 0 21 Total 221 70 12 0 303 Panel B: Number of companies with increases in concentration greater than 5% over the period 1991 to 1993 Year [5–10%] [10–25%] [25–50%] 50% Total 1991 37 36 4 2 79 1992 16 25 5 2 48 1993 14 9 4 2 29 Total 67 70 13 6 156

Note. Panel A reports purchases of share blocks of between 5 and 10%, 10 and 25%, 25 and 50%, and more than 50% over the period 1991 to 1993 by type of purchaser. Panel B reports the number of companies with increases in concentration of more than 5% by new and existing shareholders over the period 1991 to 1993. Source: Annual reports. 324 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER annual rate of takeover activity in the in both sets of firms over the period, reflecting United Kingdom of 3–4%, where control the increasing emphasis on nonexecutives passes when a majority of shares are in corporate governance in the United acquired by the bidder, usually via a tender Kingdom.18 More strikingly, in those offer. companies where there is a change in CEO, Although not reported in the table, we there is a significant reduction in the examined the relation between purchases of proportion of companies with combined share blocks and performance, measured roles of CEO and chairman from 42.9%, over two years prior to, and the year of, the before 1990, to 14.3% after. Subsequent purchase. The number of share blocks is regressions suggest that separation of CEO virtually identical in the worst and best and chairman plays an important role in deciles of performance and similar to the disciplining of management of poorly average for the complete sample. performing firms. In summary, there is a high level of both In summary, there is little relation takeovers and trades in share blocks in the between the proportion of nonexecutives on United Kingdom.There is some evidence of the boards of firms and corporate perform- a relation between the incidence of ance suggesting little disciplinary role asso- takeovers and poor performance but not ciated with nonexecutive directors. between trades in blocks and performance. However, there is evidence that the role of This suggests that takeovers may be per- CEO and chairman is more frequently forming a disciplinary function but trades separated after changes in CEOs.We examine in blocks do not.We investigate this further these observations further in the regressions in Section 4. in Section 4.

3.4. Board structure 3.5. Capital structure Panel A of Table 12.5 reports the board Jensen (1989) suggests that creditors may structure of the sample of firms partitioned have greater incentives than shareholders by decile of abnormal share price perform- to monitor and change management in ance in 1990. There is evidence that the exchange for new loans or the restructuring proportion of companies in which the roles of existing loans. This suggests that board of CEO and chairman are combined is turnover may be particularly high where lower in the worst performing decile of poor performance is combined with high firms than in other deciles. However, in leverage (or low interest cover). other respects there is very little relation Alternatively, shareholders may be able to between board structure and performance; exert greater control over management for example, the proportion of nonexecutive where poor performance forces companies directors is almost identical in the best and to seek outside equity finance. In this worst performing companies. section we examine interventions by share- Panel B examines how the structure of holders in companies facing financial the board alters after a change in CEO.The constraints. purpose is to analyze the extent to which a An analysis of our entire sample of 243 change in CEO is used to strengthen corpo- companies shows that leverage increases as rate governance by altering the composi- performance declines. Although not shown tion of the board. It partitions the sample in a table, for the period 1990 to 1992, of firms into those where there was a there are significantly higher levels of change in CEO and those where there was capital leverage in the lowest decile of no change using data from 1990. It reports performance than in higher deciles—a the average board structure two years median of 39.3% in the lowest decile com- before and two years after 1990 for the pared with 34.9 and 23.9% in the 5th and two samples of firms.The proportion of non 10th deciles, respectively. Similarly, interest executive directors on the board increased coverage is significantly lower in decile 1 than MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 325 22.5% Average Best test on difference in - 0.032 0.643 2.449 n, the age and tenure of CEO 22.5%] as a change in CEO in 1990 and for CEO in in as a change eports of nonexecutives, the proportion [12.0, 5.0%] 13.12, 3.213 0.332 8.85%] [ 27.9, 27.9%] [ 45, 45% [ Decile 1 Decile 2 Decile 3 Decile 5 Decile 9 Decile 10 Worst Board Characteristics and Firm Performance for 243 Companies for Performance and Firm Board Characteristics Panel A shows board size, A shows on the board,Panel of nonexecutives the proportion CEO and Chairma with combined of companies the proportion –statistics of differences between companies with and without CEO turnover in 1990 in with and without CEO turnover between companies –statistics of differences Chairman tenure (years)Chairman Panel B: Board characteristics of companies with and without CEO turnover in 1990 two years prior post Companies 5.5 5.1 Sample Two years prior 5.9 6.5 Two years after 7.1 t 6.1 6.1 CEO tenure (years) age (years)Chairman 55.0 4.1 60.3 5.1 59.8 5.1 59.6 8.2 60.3 5.2 58.5 5.7 59.0 5.3 CEO age (years) 49.7 53.2 51.5 55.9 54.4 54.5 52.6 Combined CEO/ChairmanCombined 27.1% 43.1% 31.0% 42.0% 34.8% 28.4% 32.9% Companies Sample Two years prior Two years after t-test on difference in Chairman age (years)Chairman 121 57.8 59.1 1.419 Chairman age (years)Chairman 19 59.4 59.4 companies in which there was no change in CEO in 1990. CEO in in was no change there which in companies CEO age (years) 0.130 CEO age (years) 110 51.6 53.1 1.801 Total number of directors number Total of nonexec.Proportion dir. (%) 37.5% 8.7 37.3% 8.7 38.2% 8.4 42.4% 43.5% 9.6 40.7% 39.8% 10.9 8.4 9.3 Combined CEO/ChairmanCombined 0.062 CEO age (years) 19 51.9 52.6 0.329 combined CEO and chairman, and age of CEO and chairman averaged over two years either side of 1990 for companies in which there in CEO and chairman, companies combined of 1990 for side either two years over averaged and age of CEO chairman w Combined CEO/ChairmanCombined 28 CEO/ChairmanCombined 42.9% 180 42.2% 14.3% 38.9% and Chairman averaged annually over the period 1990 to 1992 for 6 deciles of abnormal share price performance in 1988. in performance share price of abnormal B r 6 deciles 1990 to 1992 for Panel the period over annually averaged and Chairman Proportion nonexecutive directors nonexecutive Proportion 0.068 0.305 Table 12.5 Average 1990–1992 with CEO turnover nonexec.Proportion dir. (%) 28 sizewithout CEO turnover 36.4% nonexec.Proportion dir. (%) (Average 1988–1989) 180 sizet 36.2% (Average 1988–1989) (Average 1991–1992) age (years)Chairman Note. 41.5% (Average 1991–1992) means 40.8% means 1.144 1.464 2.483 0.221 Panel A: Board charateristics by decile of abnormal share price performance in 1990 326 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER in decile 5: a median of 1.8 compared with issue and another five of a capital 4.0 in the worst performing decile.19 reconstruction or public recontracting of Table 12.6 shows that high leverage, existing debt. In one case the bank stated combined with poor performance, is related at an Extraordinary General Meeting to increased executive board turnover. that a renewal of loan facilities was Companies in the lowest decile of share conditional on a resolution to approve price performance and the lowest quartile the sale of assets. Since much of UK debt of interest coverage had significantly higher is in the form of private bank debt, the executive board and CEO turnover in the actual level of bank restructuring is year of poor performance and the two years much greater than that which is publicly subsequently than those in the highest quar- revealed. tile of interest coverage, 69.6 and 24.2%, It is clear from the descriptions in The respectively. Companies in the lowest decile Financial Times (FT) that the party initiat- of share price performance and the highest ing the boardroom changes is not necessar- quartile of capital leverage had higher exec- ily creditors. For example, in the departure utive board and CEO turnover than those in of the CEO of Burton the FT reported the lowest quartile of capital leverage “that he had not performed with sufficient (though only the latter was significant). vigor to impress [the board’s] non-executive The annual rate of new equity issues in directors” (November 30, 1990). In the sample of poorly performing firms used another company, Cookson’s, the FT stated in the subsequent regressions at 11.5% is that “the CEO/Chairman resigned after it almost identical to that of the total sample became clear that he had lost the confi- at 11.6%. This suggests an important role dence of the company’s own senior for new equity issues in distressed compa- executives.” (November 30, 1990). For nies. We investigated this further by under- Platon, “a series of boardroom changes taking press searches on 34 firms that were was foreshadowed when the company both below average performers and had detailed plans for a sterling open [equity] high levels of debt.The criteria for selection offer.” The chairman of Era resigned a were that firms both had interest coverage week after “a long and angry shareholders’ less than two and were in the bottom three meeting.” Finally, in the case of Caledonian deciles of performance in at least one year Newspapers, a large shareholder when during the period 1990–1993.20 In 28 approached about subscribing for new firms the CEO or chairman resigned, or equity responded that “they would put both resigned. Eighteen firms or about more money up, but if so, it was good-bye 54% of the sample raised new equity management.” finance. Of these, 15 were rights issues or An important question is why large insti- open offers, while the remaining three were tutional shareholders are active only when offered to new shareholders in the form of poorly performing firms make distressed placings.21 In three cases the offer took the rights issues. Senior management at the form of convertible preference shares, largest fund managers in the United otherwise it was for straight equity. Kingdom informed us that although they There was a substantial number of other might intervene where there was very poor ownership changes. In 24 companies, or performance, in the face of management 72% of the sample, there was at least one opposition, they were likely to avoid con- of the following: a new issue, a takeover, or frontation because they disliked the conse- the emergence of a large shareholder. In quent publicity and the costs of organizing some cases board changes coincided with other shareholders. However, it was a dif- one of these events, but in many cases cap- ferent story when the poorly performing ital or ownership changes preceded board company required new financing:“it comes changes by a matter of several months. to a crunch when companies raise addi- Debt restructuring is also important. tional finance” or “it all unpicks when a There were five cases of a public debt company needs money.” MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 327 *** 1.04 rtiles of interest coverage and capital coverage of interest rtiles cutive board turnover and CEO turnover board turnover cutive n at least one of the years in the period n at least in one of the years . 1 represents Quartile interest the lowest * 1.27 3.41 ** 0.51 1.94 2.49 ** *** Cumalative executiveturnover: Interest cover Cumalative CEO turnover: Interest cover turnover: Capital leverage Cumalative executive turnover: Capital leverage Cumalative CEO The Relation between Board Turnover and Leverage for Companies in the Lowest Decile of Performance Decile the Lowest in Companies for and Leverage Turnover between Board Relation The This table reports the cumulative turnover in poorly performing companies by quartile of interest coverage and capital leverage coverage of interest by quartile companies performing poorly in reports turnover table the cumulative This Significant differences in sample means at better than the 10% level. sample means at better in differences Significant Significant at better than 5% level. at better Significant Significant at better than 1% level. at better Significant * (Lowest) (Highest) Minus Minus ** -statistics on differences in means in -statistics on differences Quartile 1Quartile 2.25 t Quartile 4Quartile 32 34.3% 27.0% 33 24.2% 50.2% 31 44.7% 34.6% 31 22.6% 42.5% Quartile 3Quartile 34 45.4% 28.2% 34 55.9% 66.0% 35 45.0% 30.9% 36 47.2% 69.6% leverage. was calculated over the year of poor performance and two years subsequent. and two years of poor performance the year was calculated over based are then studied on qua turnover in Differences Quartile 2Quartile 25 44.6% 27.7% 27 59.3% 63.6% 25 39.5% 27.8% 27 59.3% 63.6% 1988–1993. For each of the poorly performing sample companies, exe 1988–1993. and the cumulative was identified performing of poor performance the year each of the poorly For cover and the highest book leverage. Poorly performing companies are defined as in the lowest decile of annual abnormal return i return of annual abnormal decile the lowest as in are defined book leverage. companies and the highest cover performing Poorly Quartile 2 Quartile 1Quartile 3.4 Source: based on annual reports, Own calculations Datastream, School’s Risk Measurement Service. London Business Table 12.6 Leverage 1Quartile Sample Mean 21 Stdev. 67.2% Sample 38.4% Mean 23 Stdev. 69.6% Sample 76.5%Note. Mean 22 Stdev. 56.4% Sample*** Mean 31.5% Stdev. 24 79.2% 72.1% Quartile 4 Quartile 328 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER In summary, the financial structure of the tables, comparisons were undertaken poorly performing companies is worse than with different estimation techniques that of other firms and there is a higher including industry and time dummies, OLS incidence of board turnover where leverage regressions, and fixed effect regressions.22 is very high (or interest cover is low) and Five different measures of performance are new finance is raised. In a significant reported: annual abnormal returns, indus- number of cases new financing includes try corrected annual abnormal returns, an equity. The regression results in the next earnings loss dummy, industry corrected section will shed further light on these return on equity, and a dummy for dividend observations. decreases and omissions.The best explanatory power is found in the earnings loss equation. Consistent with Ball et al. 3.6. Summary (1997), earnings losses may therefore The univariate analysis of this section be the most relevant signal of managerial reveals significant potential for coalition failure. Size, as measured by sales, was formation, a large market in acquisitions included as a control variable but is not and share block sales, and a high incidence significant in any of the regressions. of new finance raised by poorly performing To rank the contribution of different firms. However, we find little relation parties to board turnover, we report the between poor performance and concentra- economic as well as statistical significance tion of ownership, share block transactions, of the independent variables. These reflect takeovers, and the proportion of nonexecu- the effect on executive board turnover of tive directors on boards of firms. This sug- moving from their mean to their extreme gests that holders of large share blocks, (upper or lowest) decile value. For purchasers of share blocks, acquirers, and example, the highest decile ownership of nonexecutive directors do not perform a executives is 43.5% as against a mean strong disciplining function. In contrast, holding of 7.6%. The difference of 35.9% there is a large amount of new equity has been multiplied by the coefficient of raised by poorly performing companies, 0.3576 (line 9) to yield a marginal effect which is associated with board changes, on board turnover of moving from the indicating that shareholders intervene average to the highest decile executive 23 when new equity is raised. ownership of 12.8%.

Performance 4. REGRESSION RESULTS FOR Lines 3 to 5 of Table 12.7 show a strong BOARD TURNOVER ON GOVERNANCE negative relation between board turnover AND PERFORMANCE for four out of five measures of performance either concurrently or with lags (the This section reports the results of regres- exception being industry corrected return sions of executive board turnover on per- on equity). The economic effect of abnor- formance and the five sets of governance mal returns in the lowest decile is to raise variables described above over the period board turnover by 7.0% two years later. 1988 to 1993. Section 4.1 discusses the The economic effect of earnings losses is results for the total sample and Section 4.2 much larger in raising board turnover by for the worst performing companies. 24.6% over a three-year period. This suggests that board turnover is more sensitive 4.1. The total random sample to earnings losses than contemporaneous abnormal share price returns.24 It may be Table 12.7 records the results of a Tobit that management and shareholders regard panel regression on executive board earnings losses as a more serious sign of turnover; although not formally reported in managerial failure than abnormal returns. MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 329 We report below the regression results This is consistently observed across different describing the influence of existing share- estimation techniques. Purchases of share holders, changes in shareholdings, stakes are for the most part new holdings takeovers, leverage, and board structure on rather than increases in existing holdings. board turnover. The economic effect of increases in holdings by families and individuals (on average 0.7%) is to increase board Existing shareholders turnover by between 5.4 and 7.7%, depending on the performance measure If concentration of ownership overcomes a used. The economic effect of increase in free rider problem of corporate control, we holdings by executives (on average 0.7%) would expect there to be higher board is between 10.5 and 12.5% increases in turnover in poorly performing firms where board turnover. However, the interaction concentration of ownership is high. terms do not support the view that changes The signs of the coefficients on the interactive in family and executive holdings are terms between ownership and performance performing a disciplinary function (lines 23 should therefore be negative in lines 11 to and 24). We do not observe a relation of 15. In fact, there are few significant terms executive board turnover to industrial in lines 6–15, indicating that ownership companies shareholdings either on their concentration on its own or with interaction own (line 17) or interactively with per- terms does not play a significant role in disci- formance (line 22).There is therefore little plining management. An exception is holdings difference in impact of active and passive of executive directors, which are consistently shareholders on corporate governance.25 negatively related to board turnover for four performance measures. The economic effect of executive ownership (which has a mean Takeovers value of 7.6%) is to lower board turnover by between 12.2 and 15.1%, depending on the We included a dummy variable in the performance measure, but it is generally regression for whether the firm was taken unrelated to performance (line 14 in Table over and an interactive dummy variable 12.7). This suggests that large ownership with performance. The takeover variable gives executives modest protection against was significant in every regression but the outside intervention, irrespective of company interactive term with performance was sig- performance. This is robust to different nificant in none. The economic effect of estimation techniques. takeovers on board turnover is in the range 89.5 to 113.2% depending on the measure of performance used. This confirms the Increases in shareholdings result in Franks and Mayer (1996) that there is a very high level of board turnover Significant changes in shareholdings are associated with takeovers but it is unre- likely to give rise to changes in management lated to poor performance. In Section 3.3, irrespective of corporate performance. we observed that over an extended period However, if these changes are performing a of time, the incidence of takeovers is higher disciplinary function then there should be in poorly performing than other firms. This higher board turnover with poor perform- suggests that the disciplinary effect of ance and with changes in shareholdings. We takeovers may be delayed beyond the one- would therefore expect to observe positive or two-year lags in these regressions. coefficients on lines 16 to 20 and negative coefficients in lines 21 to 25. Table 12.7 shows a strong positive rela- Board structure tion between increases in share holdings by both families and executive directors, and If nonexecutive directors perform a corpo- executive board turnover (lines 18 and 19). rate governance function then we would 330 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 1) (dum 0.4200 0.00 0.2151 0.26 0.0532 0.59 5.8946 0.59 0.0470 0.82 10.4561 0.00 0.3671 0.00 0.4198 0.04 0.0124 0.20 0.0761 0.31 0.0085 0.53 0.0017 0.64 0.6443 0.06 0.0191 0.02 0.6004 0.21 1) corrected ROE omissions 0.0321 0.82 0.0475 0.74 0.0852 0.57 0.3433 0.00 0.3520 0.06 0.0206 0.82 0.0141 0.87 14.3359 0.00 28.3968 0.16 10.2785 0.04 0.0655 0.66 0.3397 0.00 0.2512 0.16 0.0278 0.34 0.0006 0.89 0.2432 0.65 0.1057 0.00 0.0010 0.85 0.1298 0.84 0.0015 0.67 0.2851 0.42 0.0026 0.40 0.1131 0.68 0.0043 0.19 0.8986 0.03 0.0006 0.80 Industry corrected Earnings losses Industry Dividend decreases/ performance T-1 performance * 0.1022 0.51 0.3576 0.00 0.2040 0.25 0.0019 0.98 0.0038 0.97 0.0367 0.18 0.00180.0007 0.55 0.88 0.0003 0.92 0.9214 0.02 0.0007 0.75 0.0603 0.79 0.1041 0.00 0.0016 0.60 0.0027 0.37 Ann. abn. Return ann. abn. Return (dummy Regressions of Executive Board Turnover on Governance and Performance for the Total Sample,Total 1988–1993 the for and Performance on Governance Turnover Board of Executive Regressions 9. directors executive 8. and indiv. families 7. cos industrial 0.0713 0.51 0.1010 0.33 0.0801 0.45 0.1179 0.27 0.1395 0.21 6. institutions 5. T Performance 4. T-1 Performance 0.0928 0.45 0.0536 0.68 3. T-2 Performance 2. Intercept 6.4890 0.16 4.1165 0.36 5.3199 0.23 1.0762 0.81 4.6022 0.33 1. Sample size 905 905 938 948 948 10. dir. non-executive : terms Interaction category per cum ownership Existing shareholders Existing 14. directors executive 15. dir. nonexecutive 13. and indiv. families 0.0004 0.94 Performance Performance 12. cos industrial Left censored 568 568 589 596 596 11. institutions Noncensored 337 337 349 352 352 Pooled Tobit regression: dependent variable is executive board turnover Table 12.7 1988–1993 Par. Est. p-value Par. Est. p-value Par. Est. p-value Par. Est. p-value Par. Est. p-value Independent variable MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 331 ) 0.1324 0.98 0.3294 0.09 0.3855 0.00 0.5732 0.58 0.9359 0.29 0.9204 0.13 0.0041 0.99 2.9039 0.77 Table 12.7 continued ( 0.3006 0.00 0.0031 0.84 0.6782 0.41 0.0048 0.49 0.0890 0.19 0.0574 0.50 0.0011 0.11 0.0151 0.85 0.3263 0.00 0.9735 0.33 1.4877 0.05 0.0535 0.85 0.3865 0.14 36.2614 0.00 performance T-1 performance 0.0026 0.06 0.0420 0.59 9.9548 0.35 0.0285 0.62 0.2614 0.29 2.8602 0.93 0.0152 0.94 0.3022 0.00 0.0090 0.64 1.0021 0.66 0.0030 0.69 0.0086 0.45 0.7496 0.66 0.0220 0.14 0.0126 0.15 0.0320 0.75 * 0.0026 0.05 0.0024 0.97 0.3300 0.16 0.3185 0.00 0.0113 0.56 0.0105 0.29 0.0072 0.35 33. Leverage 32. Separ. CEO-chair 31.% nonexec dir. 31.% nonexec 30. Merger/Acquis terms:Interaction board structure, leverage,T-1: takeovers, at and performance 113.2149 0.00 96.2547 0.00 98.4948 0.00 89.4849 0.00 96.7921 0.00 29. New equity issue 9.4970 0.00 9.9686 0.00 4.9607 0.12 6.6196 0.04 7.1537 0.03 28. Cap. Lev. (book) 0.0481 0.37 0.0440 0.40 0.0567 0.25 0.1656 0.00 0.0977 0.08 27. Separ. CEO-chair 1.9722 0.41 1.7487 0.45 0.4776 0.84 1.5695 0.50 0.8427 0.73 26.% nonexec dir. 26.% nonexec Interaction terms : terms Interaction category per shareholdings in increases Board structure, and takeovers leverage 25. dir non-executive 20. dir. nonexecutive 0.5908 0.28 0.6350 0.21 0.6241 0.13 0.4658 0.28 0.5673 0.21 24. directors executive 0.0171 0.14 0.0064 0.59 19. directors executive 1.3076 0.00 1.1639 0.00 1.1922 0.00 1.2396 0.00 1.3886 0.00 23. and indiv. families 18. and indiv. families 0.9933 0.01 1.1462 0.00 1.2302 0.00 1.4283 0.00 1.1242 0.00 22. cos industrial 17. cos industrial 0.1111 0.71 0.0650 0.82 21. institutions 0.0042 0.29 0.0051 0.23 0.4138 0.44 0.0016 0.67 0.6734 0.07 Increases in shareholding Increases in 16. institutions 0.1607 0.28 0.1774 0.22 0.2059 0.16 0.102934. Equity issues 0.48 0.2616 0.09 0.0299 0.75 332 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 1) 9.84E-01 (dum –1). of blocks Nine 5.1204 0.05 1 if there is a new issue and 1 if there tion, the period for and takeovers rior year performance. year eighth rior The al returns, losses (dummy, earnings ion in ownership disclosure from 5 to from disclosure ownership in ion 3% n of ownership by five classes of investors.The by five n of ownership , percentages. are all in and % of nonexecutives tes to the structure of the board (separation of tes to the structure of the board (separation rs.The concentrations in to increases fourth relates 4.5981 0.09 0.1227 0.53 27.4024 0.16 1) corrected ROE omissions 5.7543 0.03 6.8505 0.79 end decreases and omissions (dummy,end decreases and omissions cuts or omissions dividend ivid 3.6863 0.17 Industry corrected Earnings losses Industry Dividend decreases/ 6.8984 0.02 Ann. abn. Return ann. abn. Return (dummy 1 if there is separation,1 if there %nonexec. on the board), of nonexecutives directors is the proportion leverage, new equity issues Continued -test 0.001 0.0001 0.001 0.0001 0.001 1 if the firm is acquired. of the board,1 if the firm to the interaction seventh relates The structure, financial with p variables and takeover F –1), industry performance, average for on equity corrected return d The table records a Tobit regression of annual board turnover on performance, of annual board turnover regression ownership,Tobit shareholdings, in increases records a table The board composi Significant at better than the 10% level. at better Significant Significant at better than 5% level. at better Significant Significant at better than 1% level. at better Significant sq adj 0.1957 0.1962 0.2119 0.1512 0.1913 -value of -value over the period 1989–1990.The ninth relates to size of firm measured by log of total sales. of firm 1989–1990.The to size relates the period ninth variables,over Ownership capital leverage relates to a dummy variable which equals zero in the years 1988–1989 and 1 in the other years (1990–1993) reflecting the reduct (1990–1993) reflecting years the other 1988–1989 and 1 in the years in equals zero which to a dummy variable relates takeovers third relates to the interaction of concentration of ownership with prior year performance for the different classes of investo the different for performance year with prior of ownership of concentration to the interaction relates third CEO-Chairman of share holdings.The fifth relates to the interaction between these increases and performance in the prior year.The sixth rela year.The the prior in and performance between these increases to the interaction relates of share holdings.The fifth independent variables are reported.The first relates to performance with up to two-year lags.The with up to two-year are reported.The to concentratio second relates to performance variables independent relates first Performance loss 1988 to 1993. returns, annual abnormal include measures of performance The industry abnorm for both with and without a correction 37.Total sales (log)37.Total p 1.56E-07 5.79E-01 1.42E-07 6.14E-01 4.47E-08 8.72E-01 6.69E-08 8.20E-01 5.76E-09 36. Disclosure dum Source: based on annual reports, calculations own Datastream, School’s Risk Measurement Service. London Business Table 12.7 Pooled Tobit regression: dependent variable is executive board turnover 1988–199335. Merger/Acquis Par. Est. 0.9832R Pr>Chi 0.17Note. Par. Est. p-value 0.6005 Par. Est. 0.17 Pr>Chi Par. Est.* ** p-value*** Par. Est. Pr>Chi MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 333 expect to observe more board turnover in an increase in board turnover of between poorly performing companies with separate 3.9 and 6.9% depending on which per- chairmen and chief executives and with a formance measure is used. We also reran high proportion of nonexecutive directors our regressions using interest coverage. We (negative signs on the interactive terms in expect a high level of board turnover to be lines 31 and 32, respectively). If, however, related to low levels of interest coverage nonexecutives perform more of an advisory particularly when combined with poor per- than a monitoring role then we would formance. We find a strong relation expect nonexecutives and separate chairmen between turnover and interest coverage, to reduce executive board turnover including interactions with lagged perform- irrespective of performance (i.e., negative ance. The significance of leverage and signs in lines 26 and 27) and the interactive interest coverage is consistently observed terms in lines 31 and 32 to be insignifi- using different estimation techniques.26 cantly different from zero. What we There is a strong correlation between observe is more consistent with the latter new equity issues and board turnover (four than the former. For all five measures of significant positive coefficients in line 29 performance, the relation with proportion and the remaining one has a significant of nonexecutives (line 26) is negative and negative interactive coefficient with per- the interactive terms (lines 31 and 32) are formance). When new equity is issued then not significantly different from zero board turnover increases by between 6.6 (weakly so in the presence of dividend and 10.0%. The significant interaction cuts).The economic effect of an increase in between new equity issues and performance nonexecutives to the upper decile average in the earnings loss regression is consistent of 63.6% from a mean of 38.9% is with the case study observation of associated with a decline in board turnover Section 3.5 that corporate refinancing in of between 7.4% and 9.5%, depending the wake of poor performance provides an upon the performance measure. The opportunity to restructure the board. proportion of nonexecutives therefore has a substantial negative influence on board turnover. Overview The results to date suggest that neither Financial structure owners nor purchasers of share blocks perform a disciplinary function. The only If capital structure influences executive significant effect of large block holdings is board turnover, we expect a high level of associated with those in the hands of exec- executive board turnover to be related to utives and these are used to entrench high levels of capital leverage (a positive rather than discipline management. Share coefficient in line 28), and low levels of sales by families and executives are interest coverage combined with poor associated with board turnover but not performance (a negative coefficient in line interactively with performance. Takeovers 33). In addition, we examine whether board have a very large impact on board turnover turnover is associated with new equity but again not interactively with performance. issues (a positive coefficient in line 29 and Share block sales and takeovers therefore a negative coefficient in line 34 with an have a significant effect on board turnover interaction with performance).We find that but they are not focused on the worst- high leverage is significantly related to performing companies. Nonexecutives turnover in two of the five regressions, and appear to perform more of an advisory and it is negatively related to the interaction of supportive than a disciplinary role. In con- leverage with performance in two of the trast, financial variables (leverage and new remaining regressions.The economic effect equity issues taken together) are economi- of leverage increasing to the upper decile cally and significantly related to board level of 72.2% from a mean of 32.7% is turnover and performance.They are therefore 334 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER both significant and focused and are thus 4.4 and 6.8% increases in board turnover, far ahead in the horse race. respectively. Increases in institutional ownership are associated with significant 4.2. Poor performance decreases in board turnover, suggesting that institutions sell to other investors to accom- The results in the previous section were plish board replacements in poorly performing based upon all firms in the sample. In this companies. The economic effect of such section, we examine only poorly performing institutional sales is between 3.5 and 7.2% firms. We look at how board turnover is increases in board turnover for a reduction in related to the five governance mechanisms holdings from upper decile to average levels. in the presence of five definitions of poor performance. The first is annual abnormal Board structure returns of less than —50%. The second is incidence of earnings losses. The third is The proportion of nonexecutives on the dividend cuts and omissions. The other two board of poorly performing companies is definitions take even more extreme measures not significantly related to board turnover. of poor performance: earnings losses com- The number of nonexecutives does not bined with dividend cuts and omissions, and therefore affect managerial disciplining. abnormal returns of less than 50% However, separation of the roles of chair- combined with earnings losses and dividends man and chief executive is associated with cuts and omissions. Clearly, the sample sizes significantly higher board turnover— for the fourth and fifth definitions are between 8.5 and 12.0% more turnover. appreciably smaller than for the other three. Table 12.8 reports regression results for Financial structure our sample of poorly performing firms with the same variables as in Table 12.7, exclud- Capital leverage is associated with signifi- ing the interactive terms with performance. cantly higher board turnover. An increase from average to upper decile levels of lever- Existing shareholders age raises board turnover by between 2.5 and 3.9%. Firms with earnings losses that The results for the worst performing firms make rights issues have a 13.8% higher level are almost identical to those of the com- of board turnover than those that do not. plete sample. There is no evidence of share concentrations affecting turnover, except Takeovers for holdings by executive directors. Again there is strong evidence of entrenchment. Takeovers are a highly significant influence The economic effect of moving from on board turnover in the worst performing average to upper decile executive ownership companies and are associated with a 64.1 is a decrease of between 10.4 and 17.9% to 78.7% increase in executive board in executive board turnover in the worst turnover.This is, however,appreciably lower performing firms depending on what measure than the board turnover of firms in the of poor performance is used. complete sample that are targets of takeovers. Increases in shareholdings Summary As in the complete sample, increases in shareholdings by families and executive The results for the worst-performing directors are associated with significant companies are very similar to those for the increases in board turnover. The economic complete sample.There is evidence of man- effects of moving from average to upper agerial entrenchment through executive decile ownership by families and executive share ownership while block holdings in the directors are between 5.8 and 8.2%, and hands of other investors are not associated MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 335 with managerial disciplining. Purchases of perform poorly. However, there is some share blocks by families and executives, and evidence that a higher proportion of nonex- financial constraints are associated with ecutive directors is negatively related to increases in board turnover and takeovers CEO turnover, the same result as reported have an even more significant impact. above for board turnover using the While the previous tests established the complete sample. Both financial ratios— focus of different governance mechanisms, capital leverage and interest coverage28— these results shed light on significance. lead to increased CEO changes for two Takeovers emerge as being the most signifi- measures of performance. cant governance mechanism but are unfo- Overall, board structure is more impor- cused. Purchases of share blocks by families tant in the CEO than the board regressions and executives are also significant but unfo- with separation of the position of CEO and cused. Only financial constraints are both chairman leading to higher CEO turnover. focused and significant, in the case of new Boards are therefore instrumental in dis- equity finance in particular when poor per- missing CEOs in response to earnings formance is measured by earnings losses. losses or dividend cuts. To achieve wider board restructuring, investors require the leverage of external finance provided by 4.3. Other tests high debt levels. We performed several tests of the robustness of the results to alternative specification and Serial correlation tests different definitions of variables. If there is high annual board turnover in one year it may be followed by low board CEO in place of executive board turnover turnover, thus inducing serial correlation in We examined the effect of using CEO the panel data. To check for this we per- replacement in place of board turnover and formed a cross sectional regression where related CEO replacement in a logistic the dependent variable was the annual regression on the same performance gover- average board turnover for each company nance measures. We find that there is a in the three-year period including the year significant relation between CEO replace- of poor performance.The independent vari- ment and performance when companies ables were measured in the year of poor made losses, when dividends were cut or performance in one set of regressions and omitted, and when past abnormal returns with a lag of one year in another set of were negative. Ownership variables are not regressions. These cross-sectional regres- generally significant, except for holdings by sions were performed using data from firms executive directors, with the sign suggesting in two years, 1990 and 1991. We find that managerial entrenchment. For four the results are consistent with the panel performance measures, changes in share data, although the level of significance is holdings of nonexecutive directors are different in a number of cases. We find significant implying that greater ownership statistical support for the entrenchment by this group is related to higher CEO effect of insider holdings by directors, for replacement. This effect is independent of the relation between leverage (and interest performance, suggesting a nondisciplinary cover) and performance, rights issues, reason for replacement. and executive board turnover, and that The structure of the board is important nonexecutive directors support incumbent for all five measures of performance; in management. particular, the separation of the CEO and chairman leads to greater CEO replace- Exogeneity tests ment.27 This suggests that the presence of a nonexecutive chairman is of considerable An important assumption in the previous importance in governance when firms specification is that all the independent 336 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 2.6717 0.06 0.3431 0.30 0.4588 0.08 0.5507 0.48 0.0112 0.96 50% and losses Companies, 1988–1993 1.2174 0.34 0.6495 0.03 0.1778 0.58 0.5312 0.25 0.5870 0.00 0.6628 0.17 Earnings losses Abn. return rming 0.33900.4166 0.19 0.08 0.5000 0.00 0.7840 0.05 0.7006 0.01 0.1717 0.59 0.4011 0.02 0.5132 0.20 50% Earnings losses and omissions and omissions and dividend cuts 0.5499 0.03 0.0714 0.78 0.2886 0.06 0.5044 0.21 Ann. abn return Dividend cuts and dividend cuts < Regressions of Executive Board Turnover on Governance and Performance for the Worst Perfo Worst the for and Performance on Governance Turnover Board of Executive Regressions Nonexecutive dir.Nonexecutive 0.1054 0.87 0.1597 0.82 0.3288 0.69 Families and indiv.Families directorsExecutive 1.0760 0.4983 0.08 0.06 1.1109 0.4874 0.06 0.11 1.5230 0.7542 0.01 0.01 1.2999 0.7155 0.06 0.02 1.0070 0.4494 0.33 0.16 Industrial cosIndustrial 0.4391 0.25 0.4388 0.28 0.7214 0.03 0.5559 0.24 0.5183 0.39 Increases in shareholding Increases in Institutions Nonexecutive dir. Nonexecutive Executive directors Executive Families and indiv. Families Sample selection < Pooled Tobit regression: dependent variable is executive board turnover Industrial cosIndustrial 0.1302 0.46 0.0303 0.88 0.0297 0.86 0.0050 0.98 criterion Intercept shareholders Existing Institutions 14.2893 0.1201 0.11 19.6005 0.42 0.08 0.1422 14.9767 0.41 0.12 0.0405 24.6063 0.78 0.04 0.1210 37.1205 0.02 0.53 0.3211 0.19 Table 12.8 1988–1993Censored data Par. Est. p-val. Par. Est. 103 p-val. Par. Est. p-val. Par. Est. p-val. 71 Par. Est. p-val. 117 56 21 Independent variable Sample sizeNoncensored data 125 228 100 171 131 248 82 138 52 73 MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 337 1 if rming 1 if the firm 0.2975 0.22 13.9756 0.16 2.60E-06 0.65 anies from the worst from perfo anies tion in ownership disclosure from 5 to from disclosure ownership in tion ons. The table reports a Tobit regression ons.Tobit reports table a The leverage, are all in and % of nonexecutives e board (separation of CEO-Chairman e board (separation 0.0903 0.61 6.5226 0.45 end cuts and omissions, losses and divi- earnings 1.80E-08 0.98 ivid 1 if there is a new issue and takeovers 1 if there 4.3605 0.48 0.1074 0.45 50%, losses, earnings d 3.08E-07 0.55 5.3182 0.51 0.0625 0.70 5.4827 0.33 0.0870 0.54 The worst performing companies include those in the original random sample of 243 companies supplemented with a further 50 comp sample of 243 companies random the original those in include companies worst performing The Significant at better than 5% level. at better Significant Significant at better than 1% level. at better Significant Significant at better than the 10% level. at better Significant sq adj 15.4 17.6 20.8 20.9 11.9 -test 0.001 0.001 0.001 0.001 0.1 percentages. 3% over the period 1989–1990 is included as is the size of the firm measured by log of total sales. of the firm as is the size 1989–1990 is included the period variables,3% over Ownership capital is acquired. A dummy variable which equals zero in the years 1988–1989 and 1 in the other years (1990–1993) reflecting the reduc is acquired. (1990–1993) reflecting years the other 1988–1989 and 1 in the years in equals zero which A dummy variable there is separation,there %nonexec. on the board), of nonexecutives directors is the proportion leverage, new equity issues of annual board turnover on levels and increases in concentration of ownership by five classes of investors, by five of ownership concentration in and increases on levels the structureof annual board turnover of th F Separ. CEO-chair 8.4895 0.09 9.8290 0.09 2.3962 0.63 12.0251 0.06 5.4033 0.51 dend cuts, cuts and omissi losses and dividend with earnings of less than 50% together returns and annual abnormal and omissions Cap. (book) Leverage New equity issueMergers/acquisitionsDisclosure dummy 0.0751 78.7200 11.0971 0.03 0.00 0.12 0.0707 76.0784 13.6903 0.04 0.00 0.08 0.0982 76.5755 5.3056 0.00 0.00 0.49 0.0835 69.1113 11.7129 0.02 0.00 0.20 64.0700 0.0635 15.7946 0.00 0.07 0.18 Total sales (log)Total 1.19E-07 0.85 3.12E-08 0.96 % nonexec. dir. Source: based on annual reports, calculations own Datastream, School’s Risk Measurement Service. London Business Board structure, leverage, and takeovers R Note. * ** *** decile. critera: selected on one of five samples were The of less than returns annual abnormal 338 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER variables are exogenous.To investigate this United States. In Section 5.3, we describe question we lagged all independent variables how regulatory differences between the by one year in a panel OLS regression with United Kingdom and United States might fixed effects.We find that most of the results explain differences in outcomes between in the previous regressions are supported: two seemingly similar capital markets. board turnover increases with past poor performance and it decreases with high 5.1. Influence of UK regulation concentration of ownership when held by executive directors (entrenchment). Book leverage is positively correlated with execu- 5.1.1. Protection of minorities tive turnover, but interest coverage is not. A Minority and dispersed shareholders are Tobit with lags produced some significance protected in the United Kingdom by The on rights issues with board turnover. City Code on Takeovers and Mergers Demsetz and Lehn (1985) and (called “The Code”) and by UK company Himmelberg et al. (1999) argue that own- law.These create obstacles to building con- ership is endogenous to performance. We trolling stakes. For example, an outside investigated this by running reversed blockholder who owns 15% or more of the regressions of performance on ownership equity of a firm must make public their and changes in ownership, disaggregated by intentions of launching a takeover.Where a different classes of investor. We ran six stake of 30% or more has been acquired, regressions using abnormal returns, earn- there is a compulsory tender provision for ings losses, changes in earnings per share, all remaining shares and the tender price return on equity, cash flow margins, and must be at least that paid for any shares changes in dividends per share as depend- acquired over the previous 12 months. ent variables. No consistent relation was Purchases of share blocks in excess of 3% found between these measures of perform- together with the identity of the buyer must ance and either ownership or changes in be disclosed to the market. These rules are ownership. designed to establish “fair play” in takeovers and to reduce the potential for Summary predators to purchase stakes cheaply. However, they have the effect of raising We have investigated the robustness of the acquisition costs. results reported in the previous sections to Other rules affect incentives to acquire replacing board turnover with CEO less than 100% of the shares of a target. turnover and to tests of serial correlation The Stock Exchange lays down specific and exogeneity of the independent variables. rules concerning transactions between We have found that the conclusions of the controlling blockholders, who own more previous section are robust to these than 50% of shares, and related parties. changes and that there is some evidence These state that the firm “must be capable that board structure has more of an at all times of operating and making influence on CEO replacement than on decisions independently of any controlling executive board turnover. shareholder and all transactions and relationships in the future between the applicant and any controlling shareholder 5. INFLUENCE OF REGULATION must be at arm’s length and on a normal ON GOVERNANCE commercial basis.”29 A majority of the directors of the board of the subsidiary In Section 5.1 we discuss how UK regulation must be independent of the parent firm and may have affected the above results on minority shareholders have the right to be governance. In Section 5.2, we compare consulted about, and approve, transactions the results that we have found in the United with the parent firm (Sections 11.4 and Kingdom with those reported in the 11.5 of The Stock Exchange rules). The MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 339 effect of these rules is to increase the costs Olivier Company, shareholders controlling of partial stakes. They explain why almost 30% of the shares prevented a waiver of a all bids are made conditional on acceptance rights issue.33 Even where shareholders by 90% or more of target shareholders.The vote to drop pre-emption rights the dis- remainder can be purchased compulsorily at count of any new issue must not exceed the original bid price using a squeeze out 10% of the market price at the time of the rule under the 1948 Companies Act. In a issue’s announcement (paragraph 4.26, recent bid by Capital Shopping Centres Stock Exchange Rules, 1999). These rules (CSC) for 25% of shares of Liberty ensure that old shareholders cannot be International that they did not already own, diluted by new shareholders.34 They are the independent directors of Liberty advised reinforced by guidelines, authorized by the minority shareholders not to accept the National Association of Pension Funds and offer.They argued that “it does not give any the Association of British Insurers, limiting premium for full control of the company” companies to raising 5% of their share (The Financial Times, October 19, 2000). capital each year by any method apart from rights issues—and 7.5% in any 5.1.2. Fiduciary duties of directors rolling three-year period. Can regulation explain the results reported earlier about the role of nonexecutive 5.2. Comparison of UK and U.S. directors in poorly performing companies? empirical results One important characteristic of UK regula- In some respects, the determinants of board tion is the lack of fiduciary responsibilities turnover in the United Kingdom and United of directors. Stapledon (1996) finds that States are similar. We find a significant although directors in the United Kingdom negative relation between board turnover owe their companies “fiduciary duties of and performance, similar to results reported honesty and loyalty, and a duty of care and for the United States (see Weisbach, 1988; skill,” in practice “actions to enforce Warner et al., 1988; and Coughlan and the duties of directors of quoted companies Schmidt, 1985).There is strong evidence of have been almost non-existent” 30 entrenchment by insiders in both countries (pp. 13–14). Problems of mounting such in the form of a negative relation between actions may have been exacerbated by free board turnover and insider holdings (see rider problems, the difficulty of recovering Hermalin and Weisbach, 1991; and costs of the action from the firm, and the McConnell and Servaes, 1990, for results in illegality of contingent fees (Miller, 35,36 31 the United States). 1998). In the absence of a duty of care, We have found little evidence in our UK we would expect nonexecutive directors to sample of a relation between concentration perform an advisory role. of outside shareholdings and board turnover. The U.S. results are similar. For 5.1.3. Rights issues example, Holderness and Sheehan (1988) report that the identity of large outside Another significant mechanism in the horse block owners is important, and that firms race was equity issues. Section 89(1) of the with majority outside blockholders have Companies Act 1985 states that seasoned better performance than those with diffuse new equity issues by companies must be in ownership, but the differences are not sta- the form of rights issues.32 Section 95 of tistically significant. Denis and Kruse the same Companies Act describes the cir- (2000) find that the presence of large cumstances under which pre-emption rights blockholders, other than directors and their may be waived. It requires a super-majority families, does not have an impact on indus- vote by shareholders of 75% or more on try-adjusted operating performance and each and every occasion an equity issue is they find no evidence of ownership struc- to be made. In a recent case, involving the ture influencing nonroutine board turnover. 340 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER However, in other respects UK and U.S. where nonexecutive directors dominate the results differ significantly. There is more board. Also, Gilson (1990) and Kaplan and evidence in the United States than in the Reishus (1990) find that nonexecutive United Kingdom of disciplining associated directors of poorly performing companies with sales of share blocks. In the United lose reputation and are frequently unable States, Bethel et al. (1998) report that to find replacement positions. In the United purchases of share blocks by active Kingdom, we found no evidence of investors are targeted on poorly performing disciplining by nonexecutive directors; companies. Holderness and Sheehan indeed, the relation is negative between the (1988) find that when their majority proportion of nonexecutives and board blocks trade, there is substantial manage- turnover. Second, we find that capital ment turnover and stock prices increase. In structure and new equity financing are the United Kingdom, we have found little particularly significant influences on board evidence that changes in share blocks by turnover in the United Kingdom.We are not potentially active investors perform a disci- aware of any U.S. study reporting this plinary function. relation. The Table below summarizes There are two still more pronounced dif- evidence on the parties performing ferences between the United Kingdom and disciplining in the United Kingdom and United States. First, in the United States, United States. It shows the effect on Weisbach (1988) reports a closer relation disciplining of different parties’ interventions of CEO turnover to performance in firms in the two countries.

Effect on disciplining United Kingdom United States Negative Executive block holders Nonexecutive Executive block holders directors Neutral Other block holders Purchasers of Other block holders blocks Providers of new finance Positive Bidders Providers of new finance Some purchasers of blocks Bidders Nonexecutive directors

5.3. Regulatory differences between Code requiring full bids for companies to the United Kingdom and United States be made but there are much more extensive takeover defences in state legislation and 5.3.1. Protection of minorities company charters than in the United Kingdom (Miller, 1998).39, 40 We have already reported that protection The impediments to the exercise of control of minorities is extensive in the United by dominant shareholders in the United Kingdom. In terms of Goshen’s characteri- Kingdom and the more liberal view of zation (1998) of systems of minority takeovers encourage full acquisitions of protection, the United Kingdom has a companies rather than control through “property rule” which prevents any trans- partial share blocks.This may explain why, action from proceeding without the minor- in contrast to the United States, trades in ity owner’s consent. In contrast, the United share blocks in the United Kingdom do not States has a “liability rule” which allows involve active shareholders and are not transactions to be imposed on an unwilling disciplinary in nature. minority but ensures that the minority is adequately protected in objective market 5.3.2. Fiduciary duties of directors value terms.37, 38 Protection of investors, especially minorities, is primarily the While powers to enforce fiduciary responsi- concern of the courts. Until this year, there bilities on directors in the United Kingdom was no U.S. equivalent of the UK Takeover are weak, in the United States, directors MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 341 (both executive and nonexecutive) have a fiduciary responsibilities of nonexecutives, duty of care to shareholders and can be and rules relating to new equity issues.42 sued for failing to fulfil their fiduciary These differences in protection appear to responsibilities.41 This may explain why be related to the more active role of share nonexecutives play quite different roles in block purchasers and nonexecutive directors the United Kingdom and United States— in the United States and the more active an active governance function in the United role for providers of new finance in the States as against an advisory role in the United Kingdom. United Kingdom. There is evidence to support the view that litigation encourages boards to be active in the United States 6. CONCLUSIONS (see Millstein and MacAvoy, 1998). This is strengthened by the high proportion of The question posed at the beginning of the nonexecutive directors in the United paper was:Who initiates control changes in States—an average of 75% of the total poorly performing companies? Five parties board compared with only 33% in our UK were suggested: existing large sharehold- sample (see Kini et al., 2000). ers, purchasers of blocks, bidders in takeovers, nonexecutive directors, and 5.3.3. Rights issues shareholders supplying new equity finance. Coalitions of five shareholders can on As noted above, in the United Kingdom new average control more than 30% of shares equity issues generally take the form of in the United Kingdom. However, there is rights issues. In the United States, companies little evidence that they do. On the contrary, frequently obtain shareholders’ agreement the main source of block holder control to drop pre-emption rights. Brealey and comes from those in the hands of insiders Myers (1996) suggest that “the arguments and these are used to entrench rather than [by management] for dropping to discipline management. pre-emption rights do not make sense” An as yet undocumented characteristic (p. 405). Our results imply that managers of the UK capital market is an active mar- have incentives to drop pre-emption rights ket in share blocks. Markets in share blocks to allow equity issues to be made to new could be used to discipline poorly perform- shareholders at a discount to the ing management, and in the United States equilibrium price, thereby diluting existing there is some evidence that they do. But in shareholder wealth. The discount would be the United Kingdom they do not; instead, in exchange for implicit or explicit board turnover is primarily associated with agreements to new shareholders to leave full acquisitions in takeovers. existing management in place. The U.S. The role of boards in exercising gover- evidence provided by Loderer et al. (1991) nance is also weak as evidenced by the fact suggests that a significant minority of new that nonexecutive directors do not perform seasoned equity issues are made at a a disciplinary function. In this respect, the discount from the market price, although United Kingdom is quite different from the not necessarily below their equilibrium United States. In the United Kingdom, price. Stronger rights requirements in the ineffective implementation of fiduciary United Kingdom may therefore have responsibilities results in nonexecutive allowed investors to exert greater control directors regarding their role as being over management as a pre-condition for primarily advisory rather than disciplinary. the provision of new finance. If neither holders of share blocks nor In summary, while La Porta et al. boards discipline management in poorly (1998) suggest that the rights of share- performing companies in the United holders in the United Kingdom and United Kingdom, who does? We find that capital States are similar, we find significant structure is a significant determinant of differences in minority investor protection, board changes and high levels of leverage 342 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER and low interest coverage are associated restricted to the very worst performing with high levels of board turnover in poorly firms. The greater reliance on “unfocused” performing companies. At first sight, this takeovers may to some extent have com- suggests that creditor intervention is the pensated for this but, if regulation makes main source of corporate reorganization. this an expensive form of intervention, then However, evidence from 34 case studies there may still be inadequate restructuring and from the regression analyses revealed in the United Kingdom prior to the an important role for new equity issues in emergence of financial distress. board restructurings. Regulation appears to be a significant influence on this pattern of governance in NOTES the United Kingdom. Strong minority protection has discouraged partial accumula- 1 We are grateful for comments from partici- tion of share blocks in favor of full pants at conferences at the London School of acquisitions in takeovers. Weak fiduciary Economics, Fondazione ENI Enrico Mattei in obligations on directors have resulted in Milan, Wissenschaftszentrum Berlin fur nonexecutives playing more of an advisory Sozialforschung, the University of Amsterdam, than a disciplinary role. Rights issue the University of Venice, the Autonoma University of Barcelona, the conference of the requirements protect existing shareholders ESRC Network for Industrial Economics, against wealth transfers initiated by the the European Corporate Governance Network, management of poorly performing firms the European Finance Association Meeting in and allow outside shareholders to impose Milan, the CEPR European Summer board changes as a condition for the provi- Symposium in Financial Markets in Gerzensee, sion of new equity finance. the CEPR workshop on corporate finance in Thus, despite the frequent categorization Toulouse, the SUERF meeting in Budapest, the of UK regulation under the same “common TMR Meeting in Florence, the EFMA Meeting law” classification as the United States, in Lisbon, and the Annual Meetings of the there are significant differences—more American Finance Association. We are particularly grateful to our discussants Fabrizio Barca, minority investor protection in the United Alan Hughes, Alexander Ljungqvist, Marco Kingdom than in the United States, less Pagano, Sigurt Vitols, and Jozef Zechner. We fiduciary obligations on directors in the received useful suggestions from Margaret United Kingdom but stricter rights issue Bray, Luis Correia da Silva, Rafel Crespi, Jay requirements. These differences appear to Dahya, Carles Gispert, Marc Goergen, Michel be associated with more governance Habib, Uli Hege, Joe McCahery, Piet Moerland, through partial acquisitions of share blocks Anthony Neuberger, Kjell Nyborg, Enrico and by nonexecutive directors in the United Perotti, Christian Rydqvist, Henri Servaes, States, but less governance through the James Tomkins, David Webb, and Michael provision of new financing than in the Weisbach. We thank Matthew Beardmore-Gray and Huw Jones of Prudential for very helpful United Kingdom. Subtle differences in discussions. We are grateful to the referees of regulatory systems may therefore be our paper and to the editor of the Journal of associated with pronounced difference in Financial Intermediation for very helpful com- governance outcomes. ments. Julian Franks and Colin Mayer acknowl- What is the significance of these differ- edge financial support from the European ences for corporate performance? We have Union’s Training and Mobility of Researchers not attempted to answer this question Network (Contract FRMXCT960054) and Luc directly in this paper; however, it might be Renneboog is grateful for financial support argued that greater reliance on financial from the Netherlands Organization of Scientific constraints and less reliance on boards in Research. 2 To whom correspondence should be the United Kingdom leads to a greater addressed. concentration of disciplining on the worst- 3 Weisbach (1988) also assumes that any res- performing firms.This is consistent with the ignation over 62 is natural turnover,unless there observation that higher board turnover is is evidence of conflict. MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 343 4 This uses a Capital Asset Pricing Model confined to the Fortune 500 companies, with Bayesian updating and a thin trading whereas our sample is drawn from all quoted correction. Further details can be found in companies. LSPD, London Business School, 1997. 18 Hermalin and Weisbach (1998) predict 5 This is after adjusting for bankruptcies, that the “probability that independent directors acquisitions, and double counting of those are added to the board increases following poor companies appearing in the bottom decile in corporate performance.” more than one year. 19 Similar relations between interest coverage 6 Lai and Sudarsanam (1998) also report and other measures of performance are found. For declines in board turnover after performance example, median interest cover for companies with declines. dividend cuts is 0.6 compared with 4.5 for those 7 The disclosure threshold in the United with stable or increasing dividends. States is 5%. 20 A level of two is chosen because invest- 8 As well as direct (or beneficial) holdings, we ment grade companies “typically have coverage included all nonbeneficial holdings held by ratios exceeding two times interest expense” directors on behalf of families and charitable (Copeland et al., 1995, p. 178). trusts. Directors do not obtain cash flow bene- 21 Rights issues are offered to existing share- fits from these holdings but they have control holders and any rights not taken up may be sold rights. We also investigated nominee holdings for the benefit of the shareholder; in open offers, and found that in 95% of the cases, institu- rights not taken up may not be sold by the tional investors used the nominee registration to original holder and can be sold by the company reduce administrative costs.The nominee share- to other shareholders. holdings were classified according to category 22 We controlled for the change in the disclo- of shareholder using nominee accounts. sure threshold from 5 to 3% in 1989 by 9 Recent IPOs may particularly affect the including a dummy variable which equals 0 for pattern of ownership; however,the large majority 1988–1989 and 1 afterwards. We also verified of our companies, 71%, have been listed for at the robustness of our results, by including least eight years. dummy variables for the years and by running 10 Individual stakes of 30% or greater were the regression on subsamples which excluded almost always built up prior to the company’s the years 1989 and 1990. This did not signifi- IPO. In other cases they resulted from acquisi- cantly alter the results reported below. tions in which target shareholders did not tender 23 In the case of zero-one (or minus one) all their shares. dummy variables (earnings losses, dividend cuts 11 Bristow’s data (1995) show that for 3963 and omissions, takeovers, and new equity firms median insider ownership is 12.5% and issues), economic effects relate to the effect of for other much smaller samples the median rises switching from zero to one (or minus one). The to about 16%. tables with the means and percentiles of inde- 12 The quotation of 3 was suspended or pendent variables are available on request. cancelled, 1 company was taken private and the 24 We examined the effect of interactions equity of 3 other firms was converted into a between dividend cuts/omissions and negative different security. share price performance on board turnover by 13 A total of 14% of companies had quota- restricting the regression to the sample of tions suspended or cancelled. companies that incurred abnormal losses. The 14 The bankruptcy rate is low. For the original regression results were little affected. sample of 243 companies it is zero. For the 25 The results were rerun assuming coalitions additional sample of poorly performing compa- of shareholders, for example, all outside share- nies it is 4% for the period 1988–1993. It holders were grouped together, with executive increases to 6% if we follow the sample to 1997. and nonexecutives forming two other groups. 15 Martin and McConnell (1991) also report The results are similar to those in Tables 12.7 high board turnover post takeover, between 58 and 12.8.These results are available on request. and 64%, depending upon whether the takeover 26 Results remained unaltered when leverage was hostile or friendly. and interest cover were included with a lag of a 16 If we included increases in existing share- year, rejecting a reversed causation explanation holdings, the totals would increase by 22 for [5, for their significance. 10%], 15 for [10, 25%], and 3 for [25, 50%]. 27 Including a variable for the length of 17 The lower rate of block sales in the United tenure of the CEO eliminated the significance States may be due to their sample being of the separation variable. However, length of 344 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER tenure was only available for a subsample of of parents. Eisenberg (1976) states that “the firms (130 observations were lost) and has not checks on unfair dealing by the parent are few. therefore been shown in the table.There is likely In theory, of course, the fairness of the parent’s to be less CEO–chairman separation in compa- behaviour is subject to the check of judicial nies where CEOs have been in place for long review; but in practice such review is difficult periods of time and a nonexecutive chairman even where the courts have the will to engage in may be able to exert less influence where CEOs it, and they often lack the will.” are firmly entrenched. As a consequence, tenure 40 La Porta et al. (1997, 1999) measure anti- extinguishes the significance of separation. director rights in the United Kingdom and United 28 Interest coverage was substituted for States and find greater protection for minorities leverage in separate regressions. in the United States. However, they focus on the 29 See Sections 3.12 and 3.13 of Chapter 11 rights under commercial law and do not consider of the Stock Exchange rules. the influence of the nonstatutory, but highly effec- 30 See also Parkinson (1993). tive,Takeover Code and Stock Exchange rules. 31 In Germany there are particular lawsuits by 41 Clark (1986) describes the circumstances shareholders that must be funded by the company, under which shareholders can be sued in the for example, where shareholders object to being United States. For example, in Smith v. Van “squeezed out” when a blockholder has at least Gorkum, shareholders successfully sued direc- 95% of the target’s shares. The minority may tors for a breach of duty of care with respect to demand a court hearing at the company’s expense. a merger. However, Clark also notes the paucity 32 If shareholders fail to take up their rights, of such successful cases. the rights may be sold for the shareholder’s 42 An example in differing interpretations is benefit.These pre-emption rights are recognised Preemptive Right to New Issues.The evidence is in European Community law. clear that in the United Kingdom waiver of 33 Financial Times, May 29, 1998. rights issues is very difficult unless the company 34 The Exchange may relax these rules if the is in distress, whereas in the United States company is in severe financial difficulties. waivers are the norm. Nevertheless, La Porta et al. 35 Morck et al. (1988), and McConnell and (1997) give the two countries the same score. Servaes (1990) find that corporate performance as measured by Tobin’s Q initially rises with low levels of insider ownership (for example, up to 5% in the Morck et al. study) and then declines. REFERENCES 36 Our results are also consistent with the use of anti-takeover amendments, recorded by Aghion, P., and Bolton, P. (1992). An Borokhovich et al. (1997), to entrench manage- incomplete contracts approach to financial ment, and Stultz’s argument (1988) that anti- contracting, Rev. Econ. Stud. 59, 473–494. takeover amendments substitute for insider Atiyah, P., and Summers, R. (1987). “Form and ownership as an entrenchment mechanism. Substance in Anglo-American Law,” 37 The difference that is drawn here between Clarendon, Oxford. UK and U.S. minority protection conforms with Ball, R., Kothari, S. P., and Robin, A. (1997). the more general distinction which Atiyah and Summers (1987) and Posner (1996) draw “The Effect of Institutional Factors on between reliance on substantive reasoning under Properties of Accounting Earnings: U.S. law and formal reasoning in UK law. International Evidence,” working paper, 38 The United States does have the Williams University of Rochester, May. Act of 1968 which introduced rules on block Bethel, J. E., Liebeskind, J. P., and Opler, T. disclosure (10% later amended to 5%), a min- (1998). Block share purchases and corporate imum period for which a tender offer must be performance, J. Finance 53, 605–634. left open, and a provision explicitly allowing Bianco, M., Gola C., and Signorini, L. F.(1996). targets to sue bidding firms. Dealing with separation between ownership 39 Protection may even fall short of that and control: State, family, coalitions and provided by the application of a fair price rule. Gilson (1995) argues that in Sinclair Oil Corp v. pyramidal groups in Italian corporate gover- Levien business judgment rather than intrinsic nance, Fondazione Eni Enrico Mattei Note di fairness tests should have been applied. It is Lavoro: 05/96, February. debatable whether minority protection in the Borokhovich, K. A., Brunarski, K. R., and United States interferes with business judgments Parrino, R. (1997). CEO contracting and MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 345 anti-takeover amendments, J. Finance 52, Franks J., and Mayer, C. (2000).The ownership 1495–1517. and control of German corporations, Rev. Brealey, R., and Myers, S. (1996). “Principles Finan. Stud, in press. of Corporate Finance,” McGraw–Hill, Gilson, R. J. (1995). “The Law and Finance of New York. Corporate Acquisitions,” Foundation Press, Bristow, D. K. (1995).The time series and cross Westbury, New York. sectional properties of directors and officers’ Gilson, S. C. (1990). Bankruptcy, boards, banks stock holdings, mimeo, UCLA. and blockholders, J. Finan. Econ. 27, Burkart, M., Gromb, D., and Panunzi, F.(1997). 355–387. Large shareholders monitoring, and the Goshen, Z. (1998). Liability rule or property value of the firm, Quart. J. Econ. 112, rule: Voting in conflict of interests in 693–728. corporate law and minority’s protection, Burkart, M., Gromb, D., and Panunzi, F.(1998). mimeo. Why higher takeover premia protect minority Hermalin, B., and Weisbach, M. (1991). The shareholders, J. Polit. Econ. 196, 176–204. effects of board composition and direct Cadbury, A. (1992). “Report of the Committee incentives on firm performance, Finan. on the Financial Aspects of Corporate Manage. 20, 101–112. Governance,” Gee & Co, London. Hermalin, B., and Weisbach, M. (1998). Clark, R. C. (1986). “Corporate Law,” Little, Endogenously chosen boards of directors and Brown & Co., Boston. their monitoring of the CEO, Amer. Econ. Copeland, T., Koller, T., and Murrin, J. (1995). Rev. 88, 96–118. “Valuation: Measuring and Managing the Himmelberg, C. P., Hubbard, R. G., and Palia, D. Value of Companies,” Wiley, New York. (1999). Understanding the determinants of Coughlan, A. T., and Schmidt, R. M. (1985). managerial ownership and ownership, Executive compensation, managerial turnover J. Finan. Econ. 53, 353–384. and firm performance: An empirical investi- Holderness, C. G., and Sheehan, D. P. (1988). gation, J. Acc. Econ. 7, 42–46. The role of majority shareholders in publicly Dahya, J., McConnell, J., and Travlos, N. (2000). held corporations: An exploratory analysis, “The Cadbury Committee, Corporate J. Finan. Econ. 20, 317–346. Performance and Top Management Turnover,” Jensen, M. (1986). Agency costs of free cash working paper, Cardiff Business School. flow, corporate finance and takeovers, Amer. Demsetz, H., and Lehn, K. (1985).The structure Econ. Rev. 76, 323–339. of corporate ownership: Causes and conse- Jensen, M. (1989). The eclipse of the public quences, J. Polit. Econ. 93, 1155–1177. corporation, Harvard Bus. Rev. 61–74. Denis, D., and Denis, D. K. (1995). Performance Kaplan, S., and Reishus, D. (1990). Outside changes following top management turnover, directorships and corporate performance, J. Finance 50, 1029–1057. J. Finan. Econ. 27, 389–410. Denis, D., and Kruse,T. (2000). Managerial dis- Kini, O., Kracaw, W., and Mian, S. (2000). The cipline and corporate restructuring following nature of discipline by corporate takeovers in performance declines, J. Finan. Econ. 55, an active takeover market, mimeo, Georgia 391–424. State University. Eisenberg, M. (1976). “The Structure of the La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Corporation,” Little, Brown & Co, Boston. and Vishny, R. (1997). Legal determinants of Fama, E. (1980). Agency problems and the the- external finance, J. Finance 52, 1131–1150. ory of the firm, J. Polit. Econ. 88, 288–307. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Fama, E., and Jensen, M. (1983). Separation of and Vishny, R. (1998). Law and finance, ownership and control, J. Law Econ. 26, J. Polit. Econ. 106, 1113–1155. 301–325. La Porta, R., Lopez-de-Silanes, F. Shleifer, A., Franks, J., and Mayer, C. (1996). Hostile and Vishny, R. (1999). Ownership around the takeovers in the UK and the correction of world, J. Finance 54, 471–517. managerial failure, J. Finan. Econ. 40, Lai, J., and Sudarsanam, S. (1998). Corporate 163–181. restructuring in response to performance 346 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER decline: Impact of ownership, governance and Parkinson, J. (1993). “Corporate Power and lenders, Europ. Finance Rev. 1, 197–233. Responsibility: Issues in the Theory of Loderer, C., Sheehan, D. P., and Kadlec, G. Company Law,” Oxford Univ. Press, Oxford. (1991). The pricing of equity offerings, Posner, R. (1996). “Law and Legal Theory in J. Finan. Econ. 29, 35–58. the UK and USA,” Clarendon, Oxford. Marsh, P.(1992). Dividend announcements and Renneboog, L. (2000). Ownership, managerial stock price performance, unpublished manu- control and the governance of poorly per- script, London Business School. forming companies listed on the Brussels Martin, K., and McConnell, J. J. (1991). Stock Exchange, J. Banking Finance 24, Corporate performance, corporate takeovers, 1959–1995. and management turnover, J. Finance 46, Romano, R. (1991). The shareholder suit and 671–687. litigation without foundation, J. Law Econ., McConnell, J. J., and Servaes, H. (1990). Organ. 7, 55–87. Additional evidence on equity ownership and Scharfstein, D. (1988). The disciplinary role of corporate value, J. Finan. Econ. 27, takeovers, Rev. Econ. Stud. 55, 185–199. 595–612. Shleifer,A., and Vishny, R. (1986). Large share- Miller,G. (1998). Political structure and corpo- holders and corporate control, J. Polit. Econ. rate governance: Some points of contrast 94, 461–488. between the United States and England, in Stapledon, G. (1996). “Institutional “The Sloan Project on Corporate Shareholders and Corporate Governance,” Governance,” pp. 629–648. Corporate Clarendon, Oxford. Governance Today, Columbia Law School. Stulz, R. (1988). Managerial control of voting Millstein, I., and MacAvoy, P.(1998).The active rights: Financing policies and the market for board of directors and performance of the corporate control, J. Finan. Econ. 20, 25–54. large publicly traded corporation, Columbia Warner, J. B., Watts, R. L., and Wruck, K. H. Law Rev. 98, 1283–1321. (1988). Stock prices and top management Morck, R., Shleifer, A., and Vishny, R. (1988). changes, J. Finan. Econ. 20, 431–460. Management ownership and market valua- Weisbach, M. S. (1988). Outside directors tion: An empirical analysis, J. Finan. Econ. and CEO turnover, J. Finan. Econ. 20, 20, 293–316. 431–446. Chapter 13

Jay Dahya, John J. McConnell and Nickolaos Travlos* THE CADBURY COMMITTEE, CORPORATE PERFORMANCE, AND MANAGEMENT TURNOVER

Source: Journal of Finance, 57(1) (2002): 461–483.

ABSTRACT In 1992, the Cadbury Committee issued the Code of Best Practice which recommends that boards of U.K. corporations include at least three outside directors and that the positions of chairman and CEO be held by different individuals.The underlying presumption was that these recommendations would lead to improved board oversight. We empirically analyze the relationship between CEO turnover and corporate performance. CEO turnover increased following issuance of the Code; the negative relationship between CEO turnover and performance became stronger following the Code’s issuance; and the increase in sensitivity of turnover to performance was concentrated among firms that adopted the Code.

THE CADBURY COMMITTEE WAS APPOINTED The apparent reasoning underlying the by the Conservative Government of the Committee’s recommendations is that United Kingdom in May 1991 with a broad greater independence of a corporate board mandate to “ . . . address the financial improves the quality of board oversight. aspects of corporate governance” (Report As of 2001, the Code has not been of the Committee on the Financial Aspects enshrined into U.K. law and compliance of Corporate Governance, 1992, Section with its key provisions is entirely voluntary. 1.8). The Committee was chaired by Sir Nevertheless, the Code is not without Adrian Cadbury, CEO of the Cadbury con- “teeth.” First, the Cadbury Committee’s fectionery empire, and included other sen- report explicitly recognizes that legislation ior industry executives, finance specialists, would very likely follow if companies did and academics. In December 1992, the not comply with the guidelines of the Code Committee issued its report, the corner- (Report of the Committee on the Financial stone of which was The Code of Best Aspects of Corporate Governance, 1992, Practice, which presents the Committee’s Section 1.1). Second, the report has been recommendations on the structure and given further bite by the London Stock responsibilities of corporate boards of Exchange (LSE), which, since June 1993, directors.The two key recommendations of has required a statement from each listed the Code are that boards of publicly traded company that spells out whether the com- companies include at least three nonexecutive pany is in compliance with the Code and, if (i.e., outside) directors and that the positions not, requires an explanation as to why the of chief executive officer (CEO) and chair- company is not in compliance. man of the board (COB) of these companies To appreciate the significance of the be held by two different individuals.1 Cadbury Committee and its recommendations, 348 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER it is important to appreciate the environment who felt that it did not go far enough. The surrounding the establishment of the general unease of those who felt it went too Committee. First, the Committee was far can be summarized as a concern that appointed in the aftermath of the the delicate balance between shareholders “scandalous” collapse of several prominent and managers is better left to the forces of U.K. companies during the later 1980s and competition. A less generous interpretation early 1990s, including Ferranti, Colorol of this perspective, which was most Group, Pollypeck, Bank of Credit and frequently espoused by corporate managers, Commerce International, and Maxwell might be characterized as “leave us Communication. The broadsheet press alone—we know best.” popularly attributed these failures and others to weak governance systems, lax There is danger in an over emphasis on board oversight, and the vesting of control monitoring; on non-executive directors in the hands of a single top executive. independence...[and] on controls over decision making activities of companies. The Cadbury Committee was set up in (Green, 1994) response to a number of corporate scandals that cast doubt on the systems The general concern of those who thought for controlling the ways companies are that the report did not go far enough run.The downfall of powerful figures such centered on the “voluntary” nature of the as Asil Nadir or the late Robert Maxwell, Report’s recommendations. whose personal control over their companies was complete, raised fears about the The Committee’s recommendations concentration of power. (Self-regulation are steps in the right direction. seen as the way forward, 1992) But...[s]hareholders, investors and creditors will have been disappointed Second, historically, executive (i.e., inside) that just when the corporate failures of directors have heavily dominated U.K. recent years cried out for bold and imag- boards. For example, during 1988, for only inative legal reform, the body from which 21 companies of the Financial Times (FT) so much had been expected came up with 500 did outside directors comprise a a little tinkering and a voluntary code majority of the board and, when boards are (Cadbury Committee Draft Orders ranked according to the fraction of outside Mixed News for Shareholders, 1992). board members, outsiders comprised only 27 percent of the median board’s member- Against this background, this study empiri- ship (The Corporate Register, 1989). In cally investigates the impact of the key comparison, outsiders comprised a majority Cadbury recommendations on the quality of the board for 387 of the Fortune 500 of board oversight in U.K. firms over the companies. Furthermore, for the median period 1989 through 1996. board of the Fortune 500 companies, We begin our investigation with the outside directors comprised 81 percent of presumption that an important oversight the membership (Annual Corporate Proxy role of boards of directors is the hiring and Statements).With respect to the joint posi- firing of top corporate management. We tion of CEO and COB, the United Kingdom further presume that one indicator of effec- and United States historically are similar. tive board oversight is that the board For example, during 1988, a single individ- replaces ineffective or poorly performing ual jointly held the positions of CEO and top management. Finally, we presume that COB for 349 of the Fortune 500 and for corporate performance is a reliable proxy 328 of the FT 500. for the effectiveness of top management. At its issuance, the Cadbury Report was With those presumptions in place, we greeted with skepticism both by those who empirically investigate the relationship felt that it went too far and by those between top management turnover and CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 349 corporate performance before and after block shareholders, where a block the Cadbury Committee issued its recom- shareholder is defined as any shareholder mendations. owning greater than three percent of the We assemble a sample of 460 U.K. company’s stock. Such data are available industrial companies listed on the LSE as for 548 of the firms in the initial sample. of December 1988. For each company, we Stock price and accounting data are taken collect data on management turnover, from Datastream for the years 1985 board composition, and corporate perform- through 1988. If such data are not ance for up to seven years before and four available for the years 1985 through 1988, years after the issuance of the Cadbury the firm is dropped from the sample. Forty- Report. With these data, we determine that seven of the 548 firms were dropped the relationship between top management because of insufficient stock price data; turnover and corporate performance is 41 were dropped due to insufficient statistically significant both before and accounting data. The resulting sample after adoption of the Cadbury Committee’s contains 460 firms. These firms are then recommendations, that is, poorer perform- identified according to their Financial ance is associated with higher turnover. Times Industry Classification (FTIC). The Importantly, for our purposes, this relation- sample includes at least one firm from each ship is significantly stronger following of the 33 FTIC categories. adoption of the Committee’s recommenda- To keep the sample at 460 firms at all tions. Upon further exploration, the times, when a firm ceases to be listed, we increased sensitivity of turnover to per- search chronologically among newly listed formance is due to an increase in outside industrial firms until we identify the first board members among firms that complied firm with book value of assets within plus with the key provisions of the Code. or minus 20 percent of the book value of The next section describes our sample assets of the firm that ceased to be listed. selection procedure. Section II presents For this firm to be eligible for our sample, descriptive statistics for the sample. we require that data be available on Section III presents the results of our management identity, board composition, empirical analysis. We reserve our litera- share ownership, and financial perform- ture review until Section IV, in which we ance. Finally, we require that if the existing present our conclusions in the context of firm was (was not) in compliance with the prior related empirical studies. Code, the replacement firm must (must not) be in compliance. In this way, a replacement firm was identified for each I. SAMPLE SELECTION firm that ceased to be listed within at most four months of delisting. We continue this Our investigation focuses on top manage- procedure each year from December 1988 ment turnover during the eight-year inter- onward, replacing firms that are no longer val surrounding publication of the Cadbury listed on the LSE, through the end Report in December 1992 (i.e., December of 1996. 1988 through December 1996). To begin, For each firm in the sample, for each we randomly selected 650 out of a total of year, we collect the names of board 1,828 industrial firms on the Official List members, the number of outsiders, the of the LSE as of year-end 1988 (Stock number of shares held by the board and by Exchange Yearbook, 1988–1996). For institutions, and the number of block hold- each of the 650 firms for which data are ers from the Corporate Register. We take available in the Corporate Register for stock returns and accounting data from 1988, we determine the names of board Datastream. For new firms, accounting members, the outside directors, the total data for three years prior to LSE listing shares held by the board, the total shares are taken from filings with the LSE at the held by institutions, and the number of time of listing. The shares of some newly 350 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER listed firms traded elsewhere prior to In our tests, we employ both accounting entering the LSE Official List. For these earnings and stock returns to measure firms, stock price data are collected for up corporate performance. Specifically, as our to three years preceding their listing dates. measure of accounting earnings, we use For other firms, we use price data three-year average industry-adjusted return beginning with their entry onto the on assets (IAROA). For each firm in the Official List. sample and for each year,we calculate return To determine top management turnover, on assets (ROA) as earnings before depreci- we compare the names of top management ation, interest, and taxes (EBDIT) divided by from year to year over the time period beginning-of-the-year total assets. Then, for December 1988 through December 1996. each firm with the same FTIC as the sample For each company, we identify the top exec- firm, we calculate ROA in the same way. utive as the individual with the title of CEO Next, for each FTIC group for each year, we or Executive Chairman. In addition, we determine the median ROA. IAROA is calcu- identify other board members as members lated by subtracting the industry median of the top management team if the board ROA from the sample firm’s ROA for each of member is an employee of the firm and the three years prior to a turnover event.The holds the title of Chief of Operations or average of these three IAROAs is in our Managing Director. If the name of the top measure of accounting performance. executive changes between successive For measuring stock price performance, years, we classify that as turnover in the we use industry- and size-adjusted stock top executive. For other members of the top returns (ISARs), where ISARs are calcu- management team, if a name disappears lated by subtracting the daily stock returns from the top management list, that event of an industry- and size-matched portfolio is deemed to be a turnover in the top from the return of the sample firm management team excluding the top executive. beginning 36 calendar months prior to, and If the top executive exits the list of top ending 2 days prior to, the announcement management and is replaced by another of the management change. To construct member of the top team, that event is the industry- and size-matched portfolio, considered turnover in the top executive for each sample firm, all other firms with position, but not turnover in the top man- the same FTIC code are ranked from agement team.We do not count as turnover largest to smallest according to their the event in which the position of Executive equity market values. The firms are then Chairman is split into the positions of CEO divided into four size portfolios. The differ- and COB. (Henceforth, we refer to the top ences between the return on the stock in executive position as the CEO.) our sample and the equal-weighted average We further identify turnover as “forced” return of the industry- and size-matched by examining articles in the Extel Weekly portfolio are calculated. The sum of these News Summaries, the Financial Times, and differences is the ISAR for that firm. McCarthy’s News Information Service. Turnover is labeled forced when (a) a news article states that the executive was II. CHARACTERISTICS OF THE “fired”; (b) an article states that the exec- SAMPLE utive “resigned”; or (c) an article indicates that the company was experiencing poor To conduct our analysis, we split manage- performance. In addition, for criteria ment turnover along two dimensions. First, (b) and (c), the executive must be less than we split turnover events into a pre-Cadbury 60 years old and no other article can time period (1989 through 1992) and a indicate that the executive took a position post-Cadbury time period (1993 through elsewhere or cite health or death as the 1996). Descriptive data for these two reason for the executive’s departure. All samples are presented in the first set of other turnover is labeled “normal.” columns in Table 13.1. CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 351 * ** * Stock (2) (5.00) (50.7) (25.7) * * ** th the Cadbury * and postpublication - and postadoption of - and postadoption mple firms are classified mple firms 4 20.09 (20.44) 4 11.09 (10.44) 4 2.25 (2.43) 4 158.6 (58.7) 44 7.13 46.6 (7.00) (40.6) 4 3 (3) 1 5.49 12 1 19.88 (19.82) 1 9.79 (9.99) 1 2.23 (2.44) 1 150.1 126.1 y y y y y y y y y y y y y y 288 4 to 1 to 1 to 1 to 1 to 1 to 1 to 1 to 4 to 4 to 4 to 4 to 4 to 4 to Accounting information and share prices information Accounting y y y y y y y y y y y y y y N 22 Corporate Register. N 4). the in included companies industrial from sample is taken The -statistic and Wilcoxon statistic.Tests are comparisons of before and after Cadbury statistic.Tests and after of before are comparisons Wilcoxon -statistic and (48.8) 146.8 (45.3) t y * 150 6.69 (6.00) 4.53 (4.00) 48.6 (43.4) 17.9 (15.4) N Always in Never in * 1 through ** y (50.8) 148.8 (36.9) 1 and * ** 460 y N , on the LSE. and listed entering that leave the sample between 1989 and 1996 are replaced by firms Firms Full Sample Compliance Compliance Adopted Cadbury 4 through y 1993–1996 20.04 (19.05) 22.09 (19.51) 16.29 (16.22) 1993–1996 11.35 (10.83) 13.32 (13.01) 16.09 (15.99) 1993–1996 2.24 (2.34) 2.09 (1.96) 8.34 (9.27) 1993–19961993–1996 31993–1996 7.29 (3) 46.0 (7.00) (43.1) 7.41 3 48.5 (7.00) (45.8) (3.05) 5.02 21.5 1 (5.00) (20.9) (1) Years Mean (Median) Mean (Median) Mean (Median) Years Mean (Median) Corporate Register , the Financial, Ownership, 460 U.K. 1996 1989 through for and Board Characteristics the Period over Firms Industrial denote significance at the one and five percent levels, percent at the one and five denote significance respectively, both the for * and (£ million) 1993–1996 156.8 (56.7) 155.6 (53.2) 150.7 (49.2) Number of block holders of block Number 1989–1992 2 (2) 3 (2.04) 1 (1) Institutional ownership 1989–1992 21.39 (20.84) 22.55 (20.60) 16.37 (15.93) Board ownership 1989–1992 10.70 (10.96) 12.83 (12.60) 16.45 (15.93) the LSE on date closest to departure. the are from Management and board characteristics Exchange Yearbook the Cadbury recommendations ( the Cadbury recommendations Top executive ownership executive Top 1989–1992 2.23 (2.39) 2.22 (2.23) 9.93 (10.11) Descriptive statistics for a random sample of 460 publicly traded U.K. traded 1996.The 1989 through sample of 460 publicly statistics sa a random the period Descriptive for over firms industrial recommendations, periods, or (c) adopted Cadbury recommendations. two four-year pre- over (a) and (b) are analyzed in Sample firms periods,of the Cadbury 1996). Report 1992 and 1993 through (1989 through two four-year pre over (c) are analyzed in Sample firms Percent outside directorsPercent 1989–1992 35.3 Book value of assetsBook value 1989–1992 149.2 into three sets based on whether they were (a) always in compliance with the Cadbury recommendations, compliance in (a) always three sets based they were on whether into wi compliance in (b) never values. Board size 1989–1992 5.71 (5.00) Table 13.1 Datastream. are from ** 352 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER Second, we classify the observations significantly lower institutional ownership, according to whether the firm that and fewer outside block holders than the experienced the turnover was (or was not) other two sets. Apparently, firms with in compliance with the two key provisions greater “inside” ownership of shares are of the Code. This second classification less likely to adopt the Code. scheme gives rise to three sets of firms.The As regards board composition, for the first set includes 150 firms that were in full sample prior to Cadbury, 35.3 percent compliance with the Code for each year of directors are outsiders; after Cadbury, that the firm is in our sample (hereafter, this figure is 46.0 percent. Almost all of the “always-in-compliance” set). The this increase occurs in companies that second set includes 22 firms that were came into compliance with the Code. For never in compliance with the Code during this set, the fraction of outsiders increases any year in which the firm is in our sample from 26.1 percent before adoption to (hereafter, the “never-in-compliance” set). 46.6 percent afterward. For the always- The third set includes those firms that in-compliance set, the percentage of out- came into compliance with the Code during side directors prior to Cadbury (48.6 a year in which the firm is in our sample percent) is nearly identical to the percent- (hereafter,the “adopted-Cadbury” set; 288 age afterward (48.5 percent). Finally, most firms). Descriptive data for the first and of the increase in outside directors came second sets are split into pre- and about through an increase in board size as post-Cadbury time periods. These data are opposed to the replacement of inside direc- presented in the second and third sets of tors with outside directors. The median columns in Table 13.1. Descriptive data for board increases by two members, from five the third set of firms (i.e., the adopted- to seven for the full sample, and most of Cadbury set) are split into pre- and post- this increase occurs among the adopted- Cadbury adoption time periods (i.e., y 4 Cadbury set. through y 1 and y 1 through y 4, As regards the positions of CEO and where y equals the year in which the firm COB (not shown), not surprisingly, there is came into compliance with the Code). considerable variation before and after These data are presented in the fourth set Cadbury and across the various sets of of columns in Table 13.1. firms. For the full sample prior to Cadbury, The descriptive data include the mean the CEO is also the COB in 36.5 percent of and median of book value of assets, share the companies; after Cadbury, that fraction ownership by the CEO, share ownership by drops to 15.4 percent. Of course, most the board, share ownership by institutions, of this change is due to the set of companies number of block holders, board size, that became compliant with the Code. For and number of outside directors. In terms this set, prior to Cadbury, a single individ- of book value of assets, the three sets of ual held the position of CEO and COB in firms are remarkably similar before and 39 percent of the firms; after adoption of after Cadbury and to each other. (Other Cadbury, in none of these companies did a financial data [not shown] also exhibit little single individual hold both positions. variation across the three sets of firms.) A related question is when did firms In terms of share ownership, regardless become compliant with the key recommen- of the category of investor, the fraction of dations of the Code. At least some firms shares held by that category is essentially came into compliance every year through- unchanged from before to after Cadbury. out the interval 1989 through 1996, but Additionally, on this dimension, the always- the bulk of these firms, 202 out of 288, in-compliance set and the adopted-Cadbury became compliant after 1992. Of these set are similar to the full sample and to 202, 82 were in compliance with one or the each other.However,the never-in-compliance other of the two key Cadbury provisions set has significantly more ownership by the prior to becoming fully compliant. However, CEO, significantly greater board ownership, 160 were not in compliance with either CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 353 recommendation prior to simultaneously to 4.30 percent (p-value 0.04). Further adopting both provisions, and, again, most more, the increase in CEO turnover is of these occurred after 1992. concentrated in the adopted-Cadbury set of firms. For this set of firms, the rate of all CEO turnover increased from 7.24 percent III. MANAGEMENT TURNOVER to 8.87 percent (p-value 0.01), and the rate of forced CEO turnover nearly What our analysis shows thus far is that doubled, from 2.71 percent to 4.98 percent the informal arm-twisting associated with (p-value 0.01). For the always-in- the Cadbury recommendations appears to compliance set, the rate of CEO turnover is have had considerable impact on the size essentially unchanged from before to after and composition of boards of directors, and Cadbury. For the never-in-compliance set, on the number of firms in which one indi- the rate of turnover declined modestly from vidual holds the titles of CEO and COB. before to after Cadbury, but, given the Indeed, as of 1998, 96 of the FT 100 and small sample size, we are inclined not to 90 percent of all LSE firms were Cadbury- place much weight on this result. Thus, the compliant (The Corporate Register, 1998). increase in CEO turnover following The key questions to which we now turn Cadbury is primarily attributable to those are: What impact have these changes had firms that adopted the key provisions of the on top management turnover and on the Code of Best Practice. sensitivity of turnover to corporate CEO turnover data are consistent with performance? an argument that the Cadbury Committees’ recommendations increased the quality of board oversight.That is, turnover,especially A. Incidence and rate of top forced turnover, in the CEO position has management turnover increased and this increase is concentrated Table 13.2 shows the incidence and rates of in the set of firms that adopted the key CEO turnover for the full sample and the provisions of the Code of Best Practice. three subsets. As in Table 13.1, the data are Of course, it could be that the increased arrayed into pre- and post-Cadbury time management turnover that we document periods (1989 through 1992 and 1993 following Cadbury is random across firms. through 1996) and pre- and post-Cadbury The pertinent issue for our purposes is adoption time periods (y 4 through whether turnover is correlated with corpo- y 1 and y 1 through y 4). The inci- rate performance. That is, are the “right” dence of turnover is the number of managers being replaced? That is the key instances in which we identify a change in question to which we now turn. the CEO. The rate of turnover is the annualized rate calculated as the incidence of B. Relationship between top turnover divided by 460 firms divided by management turnover and corporate four years. The first two rows present data performance on all CEO turnover and the second two rows present data on forced CEO turnover. Table 13.3 presents a preliminary look at For the full sample, the incidence and the connection between forced CEO rate of CEO turnover increase significantly turnover and corporate performance, where from before to after issuance of the performance is measured as three-year Cadbury Report.The increase in turnover is average IAROA as described in Section I. due to an increase in what we have classi- For each calendar year, firms are ranked fied as forced turnover. For example, for the from lowest to highest on the basis of their full sample, the rate of all CEO turnover prior three-year average IAROA. For each increased from 6.48 percent to 7.71 percent year, observations are then sorted into (p-value 0.02), and the rate of forced quartiles with quartile one containing the CEO turnover increased from 3.10 percent 115 firms with the lowest IAROA and 354 OPRT OENNE UNDERPERFORMANCECORPORATE ANDMANAGEMENT GOVERNANCE, TURNOVER

Table 13.2 Incidence and Rates of CEO Turnover in 460 U.K. Industrial Firms, 1989 through 1996 CEO turnover for a random sample of 460 publicly traded U.K. industrial firms over the period 1989 through 1996. The sample firms are classified into three sets based on whether they were (a) always in compliance with the Cadbury recommendations, (b) never in compliance with the Cadbury recommendations, or (c) adopted Cadbury recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods, pre- and postpublication of the Cadbury Report (1989 through 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, pre- and postadoption of the Cadbury recommendations (y 4 through y 1 and y 1 through y 4). For each firm, the name of the CEO in the Corporate Register is compared from 1988 through 1996 to determine turnover.Turnover is classified as forced by examining news articles in the Extel Weekly News Summaries, the Financial Times, and McCarthy’s News Information Service.

Always in Never in Full Sample Compliance Compliance Adopted Cadbury N 460 N 150 N 22 N 288

Years Incidence Rate Incidence Rate Incidence Rate Years Incidence Rate

All CEO Turnover 1989–1992 119* 6.48* 35 5.44 4 4.55 y 4 to y 180** 7.24** 1993–1996 138 7.71 37 5.75 3 3.41 y 1 to y 4 98 8.87 Forced CEO Turnover 1989–1992 57* 3.10* 24 3.76 3 3.26 y 4 to y 130** 2.71** 1993–1996 79 4.30 20 3.14 1 1.09 y 1 to y 4 58 4.98

** and * denote significance at the one and five percent levels, respectively, for both the t-statistic and Wilcoxon statistic.Tests are comparisons of before and after Cadbury values. Table 13.3 Forced CEO Turnover in 460 U.K. Industrial Firms Grouped by Quartiles of Performance, 1989 through 1996 Forced CEO turnover for a random sample of 460 publicly traded nonfinancial U.K. firms grouped into quartiles based on IAROA in the two four-year periods during the interval 1989 through 1996. IAROA is calculated as earnings before interest, taxes and depreciation divided by the total book value of assets less the median performance of firms in the same FTIC grouping.Three years of IAROA are averaged.Turnover is classified as forced by examining news articles in the Extel Weekly News Summaries, the Financial Times, and McCarthy’s News Information Service. The sample firms are classified into three sets based on whether they were (a) always in compliance with the Cadbury recommendations, (b) never in compliance with the Cadbury recommendations, or (c) adopted Cadbury recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods, pre- and postpublication of the Cadbury Report (1989 through 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, pre- and postadoption of CORPORATE PERFORMANCE ANDMANAGEMENT TURNOVER the Cadbury recommendations (y 4 through y 1 and y 1 through y 4).

Interval

Quartile 1 Quartile 4 (Lowest IAROA) Quartile 2 Quartile 3 (Highest IAROA) Rate Rate Rate Rate Years Incidence (%) Incidence (%) Incidence (%) Incidence (%)

Full sample 1989–1992 33* 7.2* 15 3.3 9 2.0 0 0.0 1993–1996 47 10.2 25 5.4 7 1.5 0 0.0 (a) Always in compliance 1989–1992 15 10.0 6 4.0 3 2.0 0 0.0 1993–1996 15 10.0 5 3.3 0 0.0 0 0.0 (b) Never in compliance 1989–1992 2 9.1 1 4.5 0 0.0 0 0.0 1993–1996 1 4.5 0 0.0 0 0.0 0 0.0 (c) Adopted Cadbury y 4 to y 1 16** 5.5** 8 2.3* 6 2.1 0 0.0 y 1 to y 4 31 10.8 20 6.9 7 2.8 0 0.0

** and * denote significance at the one and five percent levels, respectively, for both the t-statistic and Wilcoxon statistic. Tests are comparisons of before and after Cadbury values. 355 356 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER quartile four containing the 115 firms with for forced turnover. We estimate separate the highest IAROA. regressions using three-year prior IAROAs For the full sample, both before and and three-year ISARs as our performance after Cadbury, the incidence and rate of measures. We include yearly observations forced CEO turnover increases as we move of four control variables: fraction of from the best to the poorest performing shares owned by directors, fraction of shares firms. Additionally, the data indicate that owned by institutions, number of block the increase in CEO turnover from before to shareholders, and log of total assets. after Cadbury that we document in Table 13.3 The results of our regressions are is due to an increase in turnover in the presented in Tables 13.4 and 13.5. In lowest two performance quartiles in the Table 13.4, the performance variable is adopted-Cadbury set of firms. For example, logIAROA. In Table 13.5, performance is for this set of firms, the rate of turnover in logISAR.2 Panel A of each table presents quartiles one and two increased by nearly regressions with all CEO turnover as the 100 percent, from 5.5 percent to 10.8 percent dependent variable and Panel B presents (p-value 0.01) and by almost 300 regressions with forced turnover as the percent, from 2.3 percent to 6.9 percent dependent variable. In total, we have 20 (p-value 0.04), respectively, from before regressions that have either logIAROA or to after adoption of Cadbury. In comparison, logISAR as an independent performance for the always-in-compliance set, in the variable. In each regression, the coefficient same bottom two quartiles, the rate of of the performance variable is negative turnover is essentially unchanged from and, with two exceptions, each has a before to after Cadbury. p-value of less than 0.05. Thus, CEO The data in Table 13.3 are representative turnover is significantly negatively corre- of the pattern of turnover (not shown) that lated with corporate performance: the emerges when we consider all CEO poorer the firm’s performance, the greater turnover and when we evaluate perform- the likelihood that the CEO will depart his ance based on ISARs. That is, turnover is position. (We also estimate regressions concentrated in the poorest performing separately for the pre- and post-Cadbury quartiles of firms, and the increase in time periods [not shown]. In every regres- turnover is concentrated in the adopted- sion, the coefficient of the performance Cadbury set of firms. variable is negative with a p-value less than 0.05. Thus, turnover is significantly C. Multivariate analysis of the negatively correlated with performance relationship between top both before and after Cadbury.) management turnover and Of the four control variables, only the corporate performance fraction of shares owned by directors regularly has a p-value less than 0.10. The The final questions, to which we now turn, coefficient of this variable is always are whether the relationship between negative, which indicates that, after con- turnover and performance is statistically trolling for performance, increased share significant and whether the sensitivity of ownership by the board reduces the likeli- turnover to performance is greater follow- hood that the CEO will depart his position. ing Cadbury.To answer those questions and We now turn to the effect of Cadbury on to control for other factors that may influ- CEO turnover and the effect of Cadbury ence managerial turnover,we estimate logit on the relationship between CEO turnover regressions with pooled time series, cross and corporate performance. The five section data. Initially, we estimate regres- regressions in each panel explore that ques- sions in which the dependent variable is 1 if tion from different perspectives. The first a firm experiences CEO turnover during a regression in each panel is estimated for calendar year and 0 otherwise.We estimate the full sample of firms and includes an separate regressions for all turnover and indicator variable (Dum for 1993–1996) CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 357 which takes a value of 0 for all observations compliance set. Thus, publication of the before January 1993 (the pre-Cadbury Code of Best Practice did not have an period) and a value of 1 for all observations impact, per se, on the rate of turnover after that date (the post-Cadbury period) among top U.K. executives; rather, the along with a performance variable, either effect was concentrated among those firms logIAROA or logISAR, and the four that altered their board structures to com- control variables. In each panel, in the first ply with the Code. This is not to say that the regression, the coefficient of the indicator rate of turnover among top executives in variable Dum for 1993–1996 is positive firms that were always-in-compliance was with p-values ranging from 0.04 to 0.11. “too low” either before or after Cadbury. Thus, even after controlling for corporate The data only show that the rate of performance, turnover is higher in the turnover for these firms did not change post-Cadbury period. However, as we between the pre- and post-Cadbury periods. observed in Table 13.4, increased turnover In comparison, the rate of turnover appears to be attributable to the set of firms increased significantly among firms that that came into compliance with the came into compliance with the Cadbury Cadbury Committees’ recommendations (the recommendations during the period of adopted-Cadbury set) as opposed to those this study. firms that were always in compliance. To determine whether the increase in To determine whether the Cadbury/ turnover is correlated with performance, we turnover relationship is due to a general estimate a regression with only the phenomenon affecting all firms or whether adopted-Cadbury set of firms that it is due specifically to a change in board includes the adopted Cadbury dummy structures traceable to the Cadbury (Dum-for-Adopt) and the adopted Cadbury recommendations, we next estimate the dummy interacted with our measures regressions separately for the always- of performance (either Dum-for-Adopt in-compliance set of firms and for the logIAROA or Dum-for-Adopt logISAR) adopted-Cadbury set.The only difference in along with our measures of performance the regressions is that for the adopted- (either logIAROA or logISAR) and our Cadbury set, the indicator variable four control variables. These are the key (Dum-for-Adopt) takes on a value of 0 in regressions of our analysis and are given as all years prior to the year in which the firm the fourth regression in each panel. came into compliance with the Code and a The coefficient of the interaction value of 1 for all subsequent years. These variable indicates whether the increase in are the second and third regressions in turnover among firms that adopted each panel. Cadbury is randomly distributed across For the always-in-compliance set, the those firms or is concentrated among the coefficient of the Cadbury dummy variable poorest performing firms. In each regres- (Dum for 1993–1996) is always positive, sion, the coefficient of the interaction vari- but the p-values range from 0.79 to 0.92. able is negative with p-values ranging from Thus, publication of the Cadbury Report 0.02 to 0.07. Additionally, the coefficient had a trivial impact, if any, on the rate of of the adopted Cadbury dummy (Dum-for- turnover among CEOs in firms that were Adopt) is reduced by 60 percent and now already in compliance with the key provi- has p-values ranging from 0.60 to 0.77. sions of the Code. For the adopted-Cadbury These results indicate that the increase in set, the coefficient of the indicator variable CEO turnover is not random; rather it is Dum-for-Adopt is always positive with (inversely) correlated with performance: p-values ranging from 0.06 to 0.08. After controlling for performance, the like- Additionally, the magnitude of the coeffi- lihood that the CEO will depart his position cient is at least four times the magnitude of is greater once a poorly performing the coefficient of the Cadbury dummy firm comes into compliance with the key (Dum for 1993–1996) for the always-in- provisions of the Code. The answer to the 358 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 4 y 288 ed Cadbury 0.052 (0.89) 0.039 (0.20) 0.064 (0.08) 0.566 (0.08) 0.062 (0.86) 0.331 (0.30) 3.019 (0.00) 2.583 (0.00) N ort (1989 through Extel Weekly News e through 1996.through IAROA commendations ( commendations 288 0.739 (0.02) 0.038 (0.96) 3.228 (0.00) 2.799 (0.00) N -values are in parentheses. are in -values P 288 0.593 (0.06) 0.148 (0.66) 0.112 (0.72) 3.180 (0.00) 2.570 (0.00) equals one for the period following the adoption of the key the adoption following the period equals one for N multiplied by logIAROA. multiplied The sample firms are classified into three sets based they were on whether into are classified sample firms The Dum-for-adopt the total book value of assets less the median performance of firms in the same in of firms performance of assets less the median the total book value 150 1.859 (0.10) 1.849 (0.09) N Dum-for-adopt ivided by 460 0.457 (0.11) 0.055 (0.92) 2.034 (0.02) 1.866 (0.08) TotalSample Always in Compliance Cadbury Adopted Cadbury Adopted Cadbury Adopted McCarthy’s News Information Service. 4). Datastream.The are from equals one when turnover and share prices dependent variable information Accounting Panel A: Logit Regressions of All CEO Turnover on Log IAROA and Cadbury Status A:Turnover Panel of All CEO Regressions Logit y , and logIAROA equals one for the period 1993 through 1996. 1993 through the period equals one for 1 through logIAROA logIAROA y Financial Times logIAROA 1 and Logit Regressions of CEO Turnover on IAROA and Status of Cadbury Compliance, 1996 1989 through Turnover of CEO Regressions Logit , the y Dum for 1993–1996 Board size Board size Dum for single CEO/COB Dum for single CEO/COB Prop outsiders Prop outsiders Dum for 1993–1996 Dum-for-adopt Dum-for-adopt Log IAROA occurs. through Board variables 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, 1996).1992 and 1993 through two four-year of the Cadbury pre- and postadoption re over (c) are analyzed in Sample firms Cadbury variable FTIC grouping.Three years of IAROA are averaged. CEO turnover is classified as normal or forced by examining news articles th in FTIC grouping.Three of IAROA are averaged. by examining years or forced as normal is classified CEO turnover Summaries recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods,recommendations. two four-year of the Cadbury Rep over pre- and postpublication (a) and (b) are analyzed in Sample firms (a) always in compliance with the Cadbury recommendations, compliance in (a) always with the Cadbury recommendations, compliance in (b) never or (c) adopt Performance variable Performance is calculated as earnings before interest, before is calculated as earnings taxes, d and depreciation Intercept CEO turnover for a random sample of 460 publicly traded U.K. industrial firms in two four-year periods during the interval 1989 the interval during U.K. traded periods sample of 460 publicly a random for two four-year in CEO turnover firms industrial Table 13.4 of the Cadbury Report.Therecommendations dummy is interactive Variable N CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 359 501.58 603.03 0.105 (0.12) 0.762 (0.12) 0.189 (0.03) 0.852 (0.08) 0.058 (0.08) 0.103 (0.69) 0.031 (0.25) 0.618 (0.07) 0.053 (0.87) 0.364 (0.30) 3.921 (0.00) 2.493 (0.00) 588.65 499.20 0.170 (0.04) 0.854 (0.08) 0.659 (0.06) 0.129 (0.68) 4.659 (0.00) 0.139 (0.05) 2.819 (0.00) 0.844 (0.09) 569.29 454.11 0.166 (0.04) 0.820 (0.10) 0.631 (0.07) 0.164 (0.61) 0.132 (0.68) 4.882 (0.00) 0.142 (0.05) 2.995 (0.00) 0.812 (0.08) 485.07 387.66 0.129 (0.06) 1.114 (0.05) 2.293 (0.00) 0.122 (0.06) 1.887 (0.10) 1.092 (0.05) 621.87 572.89 0.131 (0.05) 1.260 (0.10)0.051 (0.48) 1.039 (0.22) 0.076 (0.38) 1.140 (0.15) 0.044 (0.46) 1.176 (0.15) 0.049 (0.45) 1.144 (0.15) 0.043 (0.46) 1.190 (0.01) 0.531 (0.08) 0.151 (0.79) 2.932 (0.00) 0.159 (0.02) 1.294 (0.08)0.039 (0.60) 1.027 (0.21) 0.045 (0.48) 0.985 (0.28) 0.028 (0.72) 1.032 (0.21) 0.031 (0.68) 0.597 (0.65) 0.044 (0.46) 1.745 (0.16) 0.984 (0.04) Panel B: Logit Regressions of Forced CEO Turnover on Log IAROA and Cadbury Status Turnover B: CEO Panel of Forced Regressions Logit logIAROA logIAROA logIAROA logIAROA Institutional ownership Block holders Log assets Board share ownership Board size Board size Dum for single CEO/COB Dum for single CEO/COB Prop outsiders Prop outsiders Dum-for-adopt Dum for 1993–1996 Dum-for-adopt Log IAROA Institutional share ownership Block holders Log assets Board share ownership Log-likelihood Observations 3,680 1,200 2,304 2,304 2,304 Chi-square 86.45 (0.00) 37.10 (0.00) 60.84 (0.00) 70.36 (0.00) 70.93 (0.00) Control variables: Control Board variables Log-likelihood Cadbury variables Observations 3,680 1,200 2,304 2,304 2,304 Intercept variable Performance Control variables Control Chi-square 89.35 (0.00) 53.58 (0.00) 88.66 (0.00) 87.69 (0.00) 88.21 (0.00) 360 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER 4 y 288 ted Cadbury 0.037 (0.37) 0.050 (0.10) 0.055 (0.85) 0.064 (0.82) 0.573 (0.08) 0.144 (0.60) 0.262 (0.43) 0.019 (0.00) 2.495 (0.00) N ort (1989 through Extel Weekly News commendations ( commendations in the in 989 through 1996.989 through ISARs prior to, prior 2 and ending days 288 0.680 (0.05) 0.023 (0.00) 2.612 (0.00) N -values are in parentheses. are in -values P 288 0.572 (0.08) 0.119 (0.77) 0.090 (0.85) 0.022 (0.00) 2.639 (0.00) equals one for the period following the adoption of the key the adoption following the period equals one for N multiplied by logISAR. multiplied The sample firms are classified into three sets based they were on whether into are classified sample firms The Dum-for-adopt 150 0.012 (0.03) 2.925 (0.00) N Dum-for-adopt 460 0.519 (0.09) 0.066 (0.87) 0.019 (0.00) 3.194 (0.00) TotalSample Always in Compliance Cadbury Adopted Cadbury Adopted Cadbury Adopted McCarthy’s News Information Service. 4). Accounting information and share prices come from Datastream.The dependent variable equals one when Datastream.The turnover 4). come from dependent and share prices variable information Accounting Panel A: Logit Regressions of All CEO Turnover on Log ISAR and Cadbury Status A:Turnover Panel of All CEO Regressions Logit y , and logISAR equals one for the period 1993 through 1996. 1993 through the period equals one for 1 through logISAR logISAR y Financial Times logISAR 1 and Logit Regressions of CEO Turnover on ISAR and Status of Cadbury Compliance, 1996 1989 through Turnover of CEO Regressions Logit , the y Dum for 1993–1996 Board size Board size Dum for single CEO/COB Dum for single CEO/COB Prop outsiders Prop outsiders Dum-for-adopt Dum for 1993–1996 Dum-for-adopt LogISAR through occurs. Board variables 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, 1996).1992 and 1993 through two four-year of the Cadbury pre- and postadoption re over (c) are analyzed in Sample firms Cadbury variables recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods,recommendations. two four-year of the Cadbury Rep over pre- and postpublication (a) and (b) are analyzed in Sample firms Performance variable Performance (a) always in compliance with the Cadbury recommendations, compliance in (a) always with the Cadbury recommendations, compliance in (b) never and (c) adop Summaries are industry- and size-adjusted cumulative excess stock returns computed using daily stock returns beginning 36 calendar months beginning stock returns daily computed stock using returns excess are industry- cumulative and size-adjusted news articles by examining change. or forced as normal to the announcement of top executive is classified prior CEO turnover Intercept CEO turnover for a random sample of 460 publicly traded U.K. industrial firms in the two four-year periods during the interval 1 the interval during periods U.K. traded sample of 460 publicly the two four-year a random in for firms industrial CEO turnover Table 13.5 of the Cadbury Report.Therecommendations dummy is interactive Variable N CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 361 588.82 590.73 0.049 (0.31) 0.763 (0.13) 0.045 (0.10) 0.042 (0.34) 0.039 (0.93) 0.564 (0.08) 0.060 (0.83) 0.134 (0.65) 0.272 (0.39) 0.039 (0.00) 4.291 (0.00) 0.087 (0.15) 0.850 (0.07) 584.07 584.02 0.061 (0.27) 0.840 (0.07) 0.590 (0.07) 0.040 (0.00) 0.085 (0.16) 4.624 (0.00) 0.810 (0.08) 581.41 509.66 0.066 (0.26) 0.807 (0.09) 0.538 (0.08) 0.227 (0.60) 0.030 (0.92) 0.049 (0.00) 0.077 (0.17) 4.538 (0.00) 0.804 (0.08) 555.36 491.39 0.142 (0.03) 0.925 (0.05) 0.009 (0.07) 0.126 (0.04) 3.023 (0.00) 1.190 (0.02) 629.65 598.19 0.119 (0.07) 1.040 (0.21)0.078 (0.25) 1.100 (0.19) 0.101 (0.07) 0.638 (0.58) 0.044 (0.48) 0.635 (0.58) 0.048 (0.47) 0.567 (0.67) 0.043 (0.47) 0.921 (0.05) 0.598 (0.04) 0.050 (0.86) 0.030 (0.00) 0.087 (0.15) 1.187 (0.13)0.059 (0.39) 1.062 (0.16) 0.072 (0.29) 0.817 (0.40) 0.039 (0.51) 0.837 (0.34) 0.031 (0.68) 0.936 (0.22) 0.039 (0.51) 4.892 (0.00) 0.930 (0.04) Panel B: Logit Regression of Forced CEO Turnover on Log ISAR and Cadbury Status Turnover B: CEO Panel of Forced Regression Logit logISAR logISAR logISAR logISAR Institutional share ownership Block holders Log assets Board share ownership Board size Board size Dum for single CEO/COB Dum for single CEO/COB Prop outsiders Prop outsiders Dum for 1993–1996 Dum-for-adopt Dum-for-adopt LogISAR Institutional share ownership Block holders Log assets Board share ownership Log-likelihood Observations 3,680 1,200 2,304 2,304 2,304 Control variables Control Board variables Chi-square 86.69 (0.00) 49.76 (0.00) 75.93 (0.00) 80.80 (0.00) 81.40 (0.00) Cadbury variables Log-likelihood Observations 3,680 1,200 2,304 2,304 2,304 Intercept variable Performance Control variables Control Chi-square 117.41 (0.00) 48.37 (0.00) 50.21 (0.00) 50.46 (0.00) 50.61 (0.00) 362 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER question of whether the “right” managers p-values that range from 0.07 to 0.08. In are leaving the firms appears to be yes, contrast, in none of the regressions does assuming, of course, that our measures of the coefficient of the dummy for the performance properly identify the right CEO/COB or the coefficient of the interac- managers. tion of this variable with our measures of Thus far, we have employed an indicator performance begin to approach statistical variable to capture the key provisions of the significance (p-values range from 0.69 Code of Best Practice. A further question to 0.93). is: Which of the key provisions is responsi- Apparently, the increased sensitivity of ble for the increased sensitivity of turnover turnover to corporate performance for the to corporate performance? To address that adopted-Cadbury set of firms (and the question, we estimate a final regression contemporaneous loss in significance of with the adopted-Cadbury set of firms in the interaction of Dum-for-Adopt with which we include annual observations on performance) is attributable to the increase the fraction of outside directors (Prop in the fraction of outside directors. Outsiders), an interaction between the Splitting the responsibilities of the CEO fraction of outsiders and our measures of and COB between two individuals appears corporate performance (either Prop to have had no effect on the rate of CEO Outsiders logIAROA or Prop Outsiders turnover. logISAR), an indicator variable to identify observations in which the positions of CEO and COB are held by a single indi- D. Spurious correlation? vidual (equal to 1) or by two individuals A question that may arise is whether the (equal to 0) and an interaction between correlation between management turnover this indicator variable and our measures of and corporate performance interacted with corporate performance (either Dum for Cadbury compliance is spurious. More Single CEO/COB logIAROA or Dum for specifically, is it possible that both Single CEO/COB logISAR).These variables turnover and Cadbury compliance are are designed to capture the changes caused by poor performance, perhaps brought about by the Code of Best because poorly performing firms adopt Practice. Because adoption of the Code led Cadbury to placate shareholders and, to a general increase in board size, we also concurrently, dismiss top managers—a include the number of directors and an change in management that would have interaction between the number of direc- occurred even in the absence of Cadbury? tors and our measure of performance. Several analyses seem to indicate that this These regressions, which also include a is not the case. performance measure, Dum-for-Adopt, and First, by construction, for the adopted- the four control variables, are shown as the Cadbury set of firms, all post-adoption fifth regression in each panel. CEO turnover follows compliance with the According to the regressions, when the Code. This occurs because we use year-end board composition and CEO/COB variables data to determine whether a firm is in com- are included, the coefficients of the interaction pliance. Only after the year-end in which of the Dum-for-Adopt and our measures of the firm becomes compliant with the Code performance are not significant (p-values do we consider turnover to be postadoption. range from 0.60 to 0.96). Additionally, the Thus, all post-Cadbury CEO turnover is coefficient of the fraction of outsiders on postadoption. Related to this point, most the board is positive, albeit not significant, postadoption turnover does not follow in each regression (p-values range from closely after Cadbury compliance. For 0.30 to 0.43). More interestingly, the example, for the 58 instances of postadoption coefficients of the interaction between the forced CEO turnover, 18 occur within fraction of outsiders and our measures 12 months after the year-end of adoption, of performance are always negative, with 23 occur in months 13 through 24, and CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 363 17 occur in months 25 through 36. Thus, F. Corporate performance and forced CEO turnover is not clustered in turnover in the top team the months immediately following adoption. The same is true for all CEO As noted at the outset, we focus our turnover. Second, corporate performance discussion on turnover in the CEO position. prior to adoption for those 288 firms that However, we also gathered turnover data became Cadbury-compliant is not poor. for the entire top team of managers. For For example, over the three years prior to the top team of managers, excluding the adoption, both the mean IAROA and the CEO, we conduct each of the same analyses mean ISAR are positive: They are as undertaken for the CEO. In general, the 0.057 and 0.039, respectively, but results for the top team (excluding the neither is statistically significantly CEO) are similar to, albeit weaker than, different from zero (p-values 0.24 and those for the CEO. For example, the regres- 0.40). Thus, it is not just poorly perform- sions reported in Tables 13.4 and 13.5 for ing firms that adopt Cadbury. Third, even CEO turnover are also estimated for for the set of 57 firms that came into turnover in the top management team compliance and then experienced (excluding the CEO). The signs of the forced CEO turnover, the three-year coefficients for these regressions (not preadoption mean IAROA and ISAR are shown) are identical to those of Tables positive (0.027 and 0.014), but 13.4 and 13.5; however,the p-values of the not significantly different from zero variables are not significant at traditional (p-values 0.55 and 0.69). levels. For example, the sign of the Cadbury In sum, adoption of Cadbury and CEO 1993 through 1996 dummy variable is turnover are not simultaneous, adoption of positive with p-values that range from 0.16 Cadbury is not concentrated among poorly to 0.20. Similarly, the sign on the performing firms, and firms that adopt Dum-for-Adopt variable is also positive in Cadbury and have CEO turnover are each regression, but has p-values that not performing poorly prior to adoption. range from 0.17 to 0.24. The coefficient These analyses argue against spurious for the interaction of Dum-for-Adopt and correlation. our measures of performance in the same regression is always negative with p-values that range from 0.15 to 0.26.4 In short, E. How much additional turnover? the regressions for turnover in the top management team (excluding the CEO) are To give some indication of the economic consistent with those of turnover in the significance of the statistical relationship CEO, but the levels of statistical signifi- we document, we use the last regression in cance are weaker. Panel A and Panel B of Table 13.5 to calculate the implied increase in the instances of total CEO turnover and forced CEO turnover for the adopted-Cadbury set IV. COMMENTARY AND of firms during years y – 4 through y – 1. CONCLUSIONS The predicted instances of total CEO turnover are 95 and the predicted We initiated this study with a degree of instances of forced CEO turnover are skepticism. Given the potential bite associ- 54. These compare with actual total ated with the recommendations of the turnover of 80 and actual forced turnover Cadbury Committee, we are not surprised of 30. Thus, the regressions imply all CEO to observe a significant increase in board turnover would be 20 percent higher and sizes, a significant increase in the number forced turnover would be 80 percent and fraction of outside board members, and higher had these firms been in compliance a significant reduction in the number and with the Code over the four years prior to fraction of firms with a single individual as adoption.3 CEO and COB. Further, because of prior 364 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER studies on the relationship between sensitivity of turnover to performance is corporate performance and CEO turnover, due to an increase in outside board mem- we also are not surprised to find a bers. These results are consistent with, and significant (negative) correlation between support, the argument that the Cadbury corporate performance and top manage- recommendations have improved the qual- ment turnover both before and after ity of board oversight in the United Cadbury (Coughlan and Schmidt (1985), Kingdom. However, a caveat is in order: Warner, Watts, and Wruck (1988), Increased management turnover and Weisbach (1988), Gilson (1989), Martin increased sensitivity of turnover to our and McConnell (1991), Murphy and measures of performance do not necessar- Zimmerman (1993), Kaplan (1994), Kang ily mean an improvement in corporate and Shivdasani (1995), Franks and Mayer performance. As observed by Bhagat and (1996), Huson, Parrino, and Starks Black (1999), prior research on board (1998), Mikkelson and Partch (1997), and composition and corporate performance Denis and Sarin (1999)). We were, how- generally appears to show that board ever, skeptical as to whether the observed composition does affect the way in which changes in board composition would lead boards accomplish discrete tasks, such as to changes in corporate decision making or hiring and firing top management, responding to a change in the relationship between to hostile takeovers, setting CEO compen- corporate performance and top management sation and so forth (Klein and Rosenfeld turnover. (1988), Kaplan and Reishus (1990), Part of our skepticism may stem from Rosenstein and Wyatt (1990), Byrd and the mixed results of prior studies on board Hickman (1992), Shivdasani (1993), composition and management turnover.For Denis and Denis (1995), Kini, Kracaw, and example, for 367 publicly traded U.S. Mian (1995), Agrawal and Knoeber companies, Weisbach (1988) determines (1996), Cotter, Shivdasani, and Zenner that CEO turnover is more highly negatively (1997), Hermalin and Weisbach (1998), correlated with performance in firms with and Bhagat and Black (2000)). However, outsider-dominated boards. Contrarily, for such studies generally show less (or no) 270 publicly traded Japanese companies, connection between board composition and Kang and Shivdasani (1995) find that the corporate profitability. Our study analyzes sensitivity of CEO turnover to performance the effect of the Cadbury recommendations is unrelated to the fraction of outside direc- on a discrete board task. In a subsequent tors. Finally, Franks, Mayer and Renneboog study, we intend to investigate whether the (2000) examine CEO turnover for a sample Cadbury recommendations have influenced of poorly performing U.K. firms for the corporate performance more generally. period 1988 through 1993.They are unable to draw definitive conclusions as to whether or not CEO turnover is more NOTES sensitive to performance when the board comprises more outside directors. * Dahya and McConnell are from Krannert The other part of our skepticism largely Graduate School of Management, Purdue stems from our general expectation that, University, and Travlos is from Athens prior to Cadbury, market forces were likely Laboratory of Business Administration and to have propelled boards toward efficient Cardiff Business School, Cardiff University.This structures. Thus, we are surprised to paper has benefited from the helpful comments observe a significant increase in manage- and suggestions of George Benston, David ment turnover following Cadbury adoption, Denis, Diane Denis, Julian Franks, Paul Marsh, to find an increase in the sensitivity of Robert Parrino, Ronan Powell, David Power, Raghu Rau, Andrew Stark, Jason Xiao, and management turnover to corporate seminar participants at Emory University, performance following Cadbury adoption, Purdue University, the University of Iowa, the and, especially, to find that the increase in University of Oklahoma, the EFMA (Paris) and CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 365 the FMA (Orlando). McConnell acknowledges The Corporate Register, 1989 to 1996, 1998 financial support from the Center for (Hemmington Scott Publications, London). International Business Education and Research Cotter, James F., Anil Shivdasani, and Marc (CIBER) at Purdue University. Dahya acknowl- Zenner, 1997, Do independent directors edges financial support received from the enhance target shareholder wealth during Nuffield Foundation. Travlos acknowledges tender offers? Journal of Financial financial support received from the Kitty Kyriacopoulos Chair in Finance. Economics 43, 195–218. 1 The report also recommended: (a) full disclo- Coughlan, Anne T., and Ronald M. Schmidt, sure of the pay of the chairman and the highest 1985, Executive compensation, management paid director; (b) shareholders’ approval on turnover, and firm performance, Journal of executive directors contracts exceeding three Accounting and Economics 7, 43–66. years; (c) executive directors pay be set by a Denis, David J., and Diane K. Denis, 1995, board subcommittee composed primarily of Performance changes following top manage- outsiders; and (d) directors establish a subcom- ment dismissals, Journal of Finance 50, mittee of the board, comprised mainly of outside 1029–1057. directors, to report on the effectiveness of the Denis, David J., and Atulya Sarin, 1999, company’s system of internal control. 2 We also estimated the regressions with Ownership and board structures in publicly market model excess returns and CAPM excess traded corporations, Journal of Financial returns as our measure of performance. The Economics 52, 187–224. p-values of the coefficients are essentially Franks, Julian, and Colin Mayer, 1996, Hostile unchanged. takeovers and the correction of managerial 3 As a benchmark, we calculated the implied failure, Journal of Financial Economics 40, instances of total CEO turnover during years 163–181. y 1 through y 4 to be 96 versus actual Franks, Julian, Colin Mayer, and Luc turnover of 98 and forced turnover to be 56 Renneboog, 2000, Who disciplines managers versus actual forced turnover of 58. in poorly performing companies? Working 4 The results of our analyses of the top team excluding the CEO are available from the authors. paper, London Business School. Gilson, Stuart, 1989, Management turnover and financial distress, Journal of Financial Economics 25, 241–262. REFERENCES Green, Sir Owen, 1994, Pall Mall Lecture on U.K. Corporate Governance, February 24. Agrawal, Anup, and Charles R. Knoeber, 1996, Hermalin, Benjamin E., and Michael S. Firm performance and mechanisms to control Weisbach, 1998, Endogenously chosen agency problems between managers and boards of directors and their monitoring of shareholders, Journal of Financial and the CEO, American Economic Review 88, Quantitative Analysis 31, 377–397. 96–118. Bhagat, Sanjai, and Bernard S. Black, 1999, Huson, Mark R., Robert Parrino, and Laura T. The uncertain relationship between board Starks, 1998, Internal monitoring composition and firm performance, Business mechanisms and CEO turnover: A long term Lawyer 54, 921–964. perspective, Unpublished manuscript, Bhagat, Sanjai, and Bernard S. Black, 2000, University of Alberta. Board independence and long-term perform- Kang, Jun-Koo, and Anil Shivdasani, 1995, Firm ance, Working paper, University of Colorado performance, corporate governance, and top and Columbia Law School. executive turnover in Japan, Journal of Byrd, John W., and Kent A. Hickman, 1992, Do Financial Economics 38, 29–58. outside directors monitor managers? Kaplan, Steven N., 1994, Top executives, Evidence from tender offer bids, Journal of turnover and firm performance in Germany, Financial Economics 32, 195–221. Journal of Law, Economics and Organization Cadbury committee draft orders mixed news 10, 142–159. for shareholders, 1992, Financial Times, Kaplan, Steven N., and David Reishus, 1990, June 2, p. 16. Outside directorships and corporate 366 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER performance, Journal of Financial Economics Report on the Financial Aspects of Corporate 27, 389–410. Governance (With the Code of Best Kini, Omesh, William A. Kracaw, and Shehzad Practice), 1992 (Gee and Co. Ltd., London). Mian, 1995, Corporate takeovers, firm Rosenstein, Stuart, and Jeffrey G. Wyatt, 1990, performance, and board composition, Journal Outside directors, board independence, and of Corporate Finance 1, 383–412. shareholder wealth, Journal of Financial Klein, April, and James Rosenfeld, 1988, Economics 26, 175–191. Targeted share repurchases and top Self-regulation seen as the way forward, 1992, management changes, Journal of Financial The Financial Times, May 28, p. 17. Economics 20, 493–506. Shivdasani, Anil, 1993, Board composition, Martin, Kenneth J., and John J. McConnell, ownership structure, and hostile takeovers, 1991, Corporate performance, corporate Journal of Accounting and Economics 16, takeovers, and top management turnover, 167–198. Journal of Finance 46, 671–687. Stock Exchange Yearbook, 1989–1996 Mikkelson, Wayne, and Megan Partch, 1997, (London Stock Exchange, London). The decline of takeovers and disciplinary Warner, Jerold B., Ross L. Watts, and Karen H. management turnover, Journal of Financial Wruck, 1988, Stock prices and top manage- Economics 44, 205–228. ment changes, Journal of Financial Murphy, Kevin J., and Jerold L. Zimmerman, Economics 20, 461–492. 1993, Financial performance surrounding Weisbach, Michael S., 1988, Outside directors CEO turnover, Journal of Accounting and and CEO turnover, Journal of Financial Economics 16, 273–315. Economics 20, 431–460. Part 4 Directors’ remuneration

INTRODUCTION

NE OF THE IMPORTANT TOOLS IN THE ARMOURY of the board of directors Oto align the interests of shareholders and managers is the executive compensation contract for the CEO and other executive managers. In theory, a properly written contract should achieve such alignment by making level of compensation conditional upon performance. However, in practice, there are problems in both contract writing and enforcement and in specifying the appropriate performance measures and the time horizon for the award of compensation. Many scholars (e.g. Bebchuk and Fried, 2004) have argued that entrenched top managers capture the pay setting process and manipulate the boards into awarding them excessive pay.They argue that such managers often get rewarded for failure rather than success in delivering performance and shareholder value.Thus, far from being a solution to the agency problem, executive compensation may itself be a manifestation of an agency problem and reflect failure of the governance arrangements in firms. The sensitivity of pay to performance is therefore an important test of governance effectiveness. Executive compensation arrangements may vary from country to country depending on governmental and public attitudes to ‘high’ managerial remuneration. The papers in this part deal with many of these issues and provide empirical evidence. Martin Conyon and Kevin Murphy (Ch.14) document differences in CEO pay and incentives in the United States and the United Kingdom for 1997. After controlling for size, sector and other firm and executive characteristics, they find that CEOs in the US earn 45 percent higher cash compensation and 190 percent higher total compensation. The calculated effective ownership percentage in the US implies that the median CEO receives 1.48 percent of any increase in shareholder wealth compared to 0.25 percent in the UK.The differences can be largely attributed to greater share option awards in the US arising from institutional and cultural differences between the two countries. In the next paper,Wayne Guay (Ch.15) deals with the problem of appropriate measures of compensation incentives for performance. An important determinant of performance is the level of risk taken by managers in their corporate decisions. It is in the interest of shareholders that managers are incentivised to accept high risk if that avoids the problem of under-investment in risky projects by risk adverse managers. On the other hand, excessive risk incentive can lead to value destroying, speculative investments.To control risk-related incentive problems, equity holders are expected to manage both the convexity and slope of the relation between firm performance and managers’ wealth. The author finds stock options, but not common stockholdings, significantly increase the sensitivity of CEOs’ 368 DIRECTORS’ REMUNERATION wealth to equity risk. Cross-sectionally, this sensitivity is positively related to firms’ investment opportunities. This result is consistent with managers receiving incentives to invest in risky projects when the potential loss from under-investment in valuable risk- increasing projects is greatest. Firms’ stock-return volatility is positively related to the convexity provided to managers, suggesting convex incentive schemes influence investing and financing decisions.This paper has pioneered the literature on not only how different compensation components influence corporate risk taking and performance but also how to measure risk taking and performance incentives. An interesting manifestation of the managerial capture of the pay setting process referred to above is when executive stock options are repriced. Following the stock market crash and the dotcom collapse of the late 1990s, stock options with exercise prices set in relation to pre-crash levels lost their incentive character since they had fallen too ‘deep in the water’. In response to this development many firms re-set the exercise price at much lower levels. This retrospective ‘manipulation’ of compensation incentive was, in the eyes of many scholars and observers (e.g. Bebchuk and Fried, 2004), a case of top managers’ ‘having their cake and eating it too’ or enjoying a ‘heads, I win and tails, you lose’ gamble. Chidambaran and Prabhala (Ch.16) test the management capture view and examine firms that reprice their executive stock options and find little evidence that repricing reflects managerial entrenchment or ineffective governance. Repricing grants are economically significant, but there is little else unusual about compensation in repricing firms. Repricers tend to be smaller,younger,rapidly growing firms that experience a deep, sudden shock to growth and profitability. They are also more concentrated in the technology, trade, and service sectors and have smaller boards of directors. Repricers have abnormally high CEO turnover rates, which is inconsistent with the entrenchment hypothesis. Over 40 percent of repricers exclude the CEOs’ options when they reprice. Julie Ann Elston and Lawrence G. Goldberg (Ch.17) focus on executive compensation arrangements in Germany with a different corporate governance system from that in many other countries including the US, UK and France. The presence of workers on German supervisory boards under the co-determination system also means that the pay setting process has a different political dynamic in Germany than in countries without such representation. However, with the growth of international mergers like DaimlerChrysler, which dramatically illustrated the compensation differentials between US and German top managers, interest in executive compensation practices in different countries, particularly in Germany, has increased. Using unique data sources for Germany, the authors find that, similar to US firms, German firms also have agency problems caused by the separation of ownership from control, with ownership dispersion leading to higher compensation. Compensation is also positively related to firm size but there is evidence that bank influence has a negative impact on compensation. Large shareholdings also restrain managerial compensation suggesting greater monitoring by block shareholders and bank creditors. Does stock ownership by managers necessarily align shareholder and managerial interests and lead to better corporate performance and shareholder value creation? John Core and David Larcker (Ch.18) provide evidence by examining a sample of US firms that adopt ‘target ownership plans’, under which managers are required to own a minimum amount of stock.They find that, prior to plan adoption, such firms exhibit low managerial equity ownership and low stock price performance. Managerial equity ownership increases significantly in the two years following plan adoption.The authors also observe that excess DIRECTORS’ REMUNERATION 369 accounting returns and stock returns are higher after the plan is adopted. Thus, for the sample of firms, the required increases in the level of managerial equity ownership result in improvements in firm performance. This study highlights the importance of executive stock ownership as an effective corporate governance tool.

REFERENCE

Bebchuk L. and Fried J. (2004), Pay without performance, Cambridge, Mass: Harvard University Press.

Chapter 14

Martin J. Conyon and Kevin J. Murphy* THE PRINCE AND THE PAUPER? CEO PAY IN THE UNITED STATES AND UNITED KINGDOM

Source: The Economic Journal, 110 (2000): 640–671.

ABSTRACT We document differences in CEO pay and incentives in the United States and the United Kingdom for 1997. After controlling for size, sector and other firm and executive characteristics, CEOs in the US earn 45% higher cash compensation and 190% higher total compensation. The calculated effective ownership percentage in the US implies that the median CEO receives 1.48% of any increase in shareholder wealth compared to 0.25% in the UK.The differences, can be largely attributed to greater share option awards in the US arising from institutional and cultural differences between the two countries.

CORPORATE GOVERNANCE PRACTICES IN THE Although CEO pay levels in the United United Kingdom received increased attention Kingdom have grown in recent years, they in the 1990s, culminating in influential remain far behind pay levels enjoyed by reports issued by the Cadbury (1992), CEOs in the United States. The interna- Greenbury (1995) and Hampel (1998) com- tional pay gap is especially pronounced mittees. Among other recommendations, the after including gains realised from exercis- reports outlined a best-practice framework ing share options. Chief executives in the for setting executive pay, and significantly 500 largest UK companies in aggregate expanded disclosure rules for UK executive made £330 million (or £660,000 each) in compensation. The Greenbury and Hampel 1997, including £74 million from exercis- reports were, in part, a response to a ing options. In contrast, the top 500 US growing controversy over chief executive CEOs made in aggregate £3.2 billion (or officer (CEO) pay levels triggered when £6.3 million each), including £2.0 billion executives in several recently privatised from option exercises.1 Indeed, Disney’s electric utilities exercised share options Michael Eisner, dubbed by pay-critic Graef worth millions of pounds. However, in spite Crystal (1991) as the ‘Prince of Pay’, of these reports, CEO pay levels rose more exercised options worth £348 million in than 18% in 1997 alone, even as public- December 1997, thus single-handedly out- sector workers were being asked to accept earning the aggregate paycheques of the raises of less than 3% (Buckingham and top 500 CEOs in the United Kingdom. Cowe, 1998 a, b).The continuing controversy, British Sky Broadcasting’s Sam Chisolm, the coupled with enhanced data availability highest-paid UK executive, is a mere pauper through the new disclosure requirements, by American standards: his £6.8 million has sparked considerable academic interest pay package would only rank as the 97th in UK executive pay practices. highest among US chief executives. 372 DIRECTORS’ REMUNERATION These anecdotal comparisons, while driven compensation from 1989–97, and on the by option gains in the robust US stock prevalence of stock option plans from market, hint at important differences in 1979–97. We show that, although cash CEO pay practices in the United Kingdom compensation has been growing at about and the United States. The purpose of this the same rate in both countries since the article is to provide a comprehensive mid-1990s, the prevalence of option plans comparison of pay practices in the two has been growing in the United States while countries, and to generate stylised facts to declining in the United Kingdom. stimulate future research. Existing inter- Section 3 explores stock-based financial national comparisons of pay practices have incentives for CEOs in the two countries. typically relied on non-comparable survey CEO wealth is linked directly to company data, or have focused on narrow definitions share-price performance through their of compensation that usually exclude the share holdings, their option holdings, and grant-date value of share options.2 In through shares awarded through long-term contrast, our results are based on compa- incentive plans (LTIPs). We show that rable and complete measures of CEO pay these various holdings can be aggregated to and stock-based incentives, utilising detailed form an ‘effective’ ownership percentage (or, data made available through enhanced dis- following Jensen and Murphy (1990 a,b), closure requirements in the United States the ‘Pay-Performance Sensitivity’). Similar and (more recently) the United Kingdom. to our analysis of CEO pay levels, we Our use of micro data allows us to analyse compute the effective ownership percentage differences in compensation and incentives for each CEO and analyse how it varies while controlling for factors such as com- across countries and with different firm pany size, industry, human capital, growth and executive characteristics. We find that opportunities, and performance. American CEOs, on average, own much We begin in Section 1 by offering an larger fractions of their firms’ stock than introduction to executive compensation, do British CEOs. For example, the median aimed at academic economists new to the holding for US CEOs is 0.29%, while the area. In this section, we identify available median holding for UK CEOs is only 0.05%. data sources for CEO pay in the two The median effective ownership (including countries, describe our sample, and discuss options and LTIP grants) is more disparate: how to measure and value share options and 1.48% for US CEOs vs. only 0.25% for other pay components.3 Section 2 analyses UK CEOs. the level and structure of executive compen- Section 3 also explores how CEO cash sation, based primarily on data from 510 UK compensation varies with share-price and 1,666 US corporations. We show that, performance. We show that the elasticity while company size is an important deter- of cash compensation to performance is minant of pay in both countries, the rewards higher in the United States than in the for scale are more pronounced in the United Kingdom for all industries, although United States than in the United Kingdom. the difference is only statistically significant We document that share option grants, for the financial services industry. In valued at grant date, comprise a fairly addition, we examine the relation between small percentage of total pay for the CEO turnover and company performance. typical British CEO, but are much more We find that CEO turnover is negatively important for American CEOs. In addition, correlated with shareholder returns in both we document that American CEOs indeed the United Kingdom and United States, out-earn their British counterpart, earning indicating that CEOs in both countries are 45% more in cash pay and 190% more in more likely to lose their jobs following poor total pay, even after controlling for size, performance. The cross-country difference industry, growth opportunities, CEO human in the turnover-performance relationship is capital, and other observable characteristics. not statistically significant. Also, Section 2 offers some cross-country Section 4 considers a variety of explana- time-series evidence on the levels of cash tions for the observed Anglo-American CEO PAY IN THE US AND UK 373 differences in compensation and incentives, in the United Kingdom, and the proxy including agency theory, taxes, and culture. statement in the United States.6 Our objective is not to reconcile completely Although we offer some longitudinal the differences in pay practices, but rather to comparisons, our analysis is based primarily identify potential explanations as opportu- on 1997 fiscal-year data, since this was the nities for future research. We argue that first year covered by the new UK disclosure traditional agency-theoretic considerations requirements.The UK data analysed in this offer little insight in explaining the differ- paper are drawn directly from the annual ences, unless US and UK executives differ reports for the 510 largest companies systematically in their ability, productivity, (ranked by market capitalisation). The or risk aversion. We document that, while pay and ownership data are matched to personal income tax rates and rules in Datastream data on company size, industry, the United States and United Kingdom are and performance.Together,these companies generally quite similar,there are differences account for virtually all (98%) of the in corporate tax rules that may explain at market capitalisation of the entire UK stock least some of the observed differences in market. The fiscal 1997 US compensation compensation and incentives. Finally, we and company data are extracted from consider a variety of economic, political, Standard and Poor’s (S&P’s) Compustat’s and cultural factors that help explain why ‘ExecuComp’ database, which includes share option compensation has increased proxy-statement data for 1,666 top execu- dramatically in the United States, but not in tives in the S&P 500, the S&P Mid-Cap the United Kingdom. Section 5 summarises 400, the S&P Small-Cap 600, and other our results, and explores implications supplemental S&P indices. For each country of our results to broader multi-country and company, we identified the CEO (or comparisons of international pay practices. most senior executive officer), and collected information on share ownership, current and prior option grants, salaries, annual 1. AN INTRODUCTION TO bonuses, benefits, and LTIP cash and EXECUTIVE COMPENSATION share awards. The largest US companies are considerably larger than the largest UK companies, 1.1. Data sources and our combination of large, mid, and The United States, United Kingdom, and small-cap US companies is meant to pro- Canada are currently the only countries vide a distribution of US firms similar to that require detailed disclosure on the the UK distribution. Nonetheless, our sam- compensation practices for individual top ple of US firms remains somewhat larger, corporate executives.4 Disclosure rules with mean (median) 1997 market capitali- for US executives were standardised and sation of £3.4 billion (£790 million) in the expanded in 1992 by the US Securities and United States, compared to £2.2 billion Exchange Commission (SEC), and require (£480 million) in the United Kingdom. Our details on share ownership, share options, results below include company size controls and all components of compensation for to adjust for systematic differences in size. the top five corporate executives. Disclosure rules in the United Kingdom were significantly expanded in recent years following 1.2. Measuring and valuing the the Greenbury (1995) and Hampel (1998) components of pay reports, and require disclosure of data com- Compensation arrangements in both the parable to those available for US executives United Kingdom and the United States (including previously unavailable details on contain the same basic components. CEOs share option grants and holdings).5 Although in both countries receive base salaries and there are a variety of ‘secondary sources’ are eligible to receive annual bonuses paid for compensation data in the two countries, based on accounting performance. CEOs in the primary data source is the annual report both countries also typically receive share 374 DIRECTORS’ REMUNERATION options, which are rights to purchase shares of the prior 48 monthly observations on of stock at a pre-specified ‘exercise’ price cash dividend per share.8 for a pre-specified term. CEOs also often In spite of its prevalence in both practice participate in long-term incentive plans and academia, there are many drawbacks (LTIPs).In the United Kingdom, LTIPs are to using the Black-Scholes formula for typically grants of shares of stock that calculating the value of an executive share become ‘vested’ (i.e., ownership is trans- option. First, the Black-Scholes value is, ferred to the CEO) only upon attainment of at best, a measure of the company’s certain performance objectives. LTIPs in opportunity cost of granting the option, the United States take two primary forms: and will typically overstate the value to (1) ‘restricted stock’ grants that vest with the executive-recipient (Hall and Murphy, the passage of time (but not with perform- 2000). Second, executive share options are ance criteria); and (2) multi-year bonus subject to forfeiture if the executive leaves plans typically based on rolling-average the firm prior to vesting; this probability three or five-year cumulative accounting of forfeiture reduces the cost of granting performance. the option and thus implies that the Black- We define total compensation as the sum Scholes formula overstates option values. of base salary, annual bonus, LTIP awards, Third, the Black-Scholes formula assumes and share options valued at grant date. that options can only be exercised at the We measure LTIP share grants at the face expiration date, but executive options can value of the shares on the grant date, and be exercised immediately upon vesting, impose 20% discounts for performance- which typically occurs relatively early in the contingent UK grants. LTIP cash awards option’s term.9 Finally, following recom- are valued as the amount actually paid mendations in the Greenbury (1995) report, during the fiscal year. In valuing share share options granted in the United options, we follow the approach used by Kingdom typically vest only upon attain- both practitioners and academic researchers ment of some performance criteria, by measuring the grant-date expected value often based on earnings-per-share growth. using the Black and Scholes (1973) formula, Although the existence of performance adjusted for continuously paid dividends: criteria will naturally reduce the company’s cost of granting an option, the expected Option Value Pe ln(1 d)TN(z) discount is fairly modest, because the criteria are seldom binding.10 Moreover, to the XeLn(1r)TN(z T), (1) extent that the performance criteria are where P is the grant-date share price, X is correlated with share prices, the criteria the exercise price, T is the time remaining will be binding only when the intrinsic value until expiration, d is the annualised dividend of an unrestricted share option is low. yield, is the stock-price volatility, r is the Subject to these caveats, we present Black- risk-free discount rate, N( ) is the cumulative Scholes values of stock options in our normal distribution function, and analyses below.

ln(PX) [ln(1 r) ln(1 d) 22]T z T 2. THE LEVEL AND COMPOSITION (2) OF CEO PAY

We measure the risk-free rates for the two 2.1. Summary statistics countries as the average yield on 7-year UK and US Treasury bills.7 Volatilities are Table 14.1 provides summary statistics for defined as the standard deviation of monthly the level and composition of fiscal 1997 continuously compounded returns over the CEO pay, by company size, and industry, prior 48 months, multiplied by 12 . and country.Total pay is defined as the sum Dividend yields are computed as the average of salaries, bonuses, benefits, other cash CEO PAY IN THE US AND UK 375 pay, grant-date values of share options, and with 1997 revenues in excess of £1,500 grant-date value of LTIP shares (dis- million is £811,000, far larger than the counted, where appropriate, for perform- £287,000 median pay for companies with ance contingencies). Dollar-denominated revenues below £200 million. Median and data for each US executive are converted average total pay is somewhat less in to UK pounds using the average exchange utilities than in other industries. The rate during the company’s fiscal year; this bottom panel shows that American CEOs rate varied between 1.61 $/£ and 1.65 $/£ earn substantially more than their British during our sample period. counterparts, for every size and industry As reported in the top panel of Table 14.1, group. The average total compensation for the average total compensation for the 510 the 1,666 US CEOs is £3.6 million, or UK CEOs is £589,000, while the median 500% more than the average pay for UK pay is £414,000. Total pay increases with executives. Similarly, the US median pay of firm size: the median pay for companies £1.5 million is 260% more than the

Table 14.1 Summary statistics for 1997 CEO total compensation, by company size and industry

Total pay Average composition of total pay (%)

Sample Average Median Base Annual Option LTIP Other Group firms (£000s) (£000s) salary bonus grant shares pay

United Kingdom All companies 510 589 414 59 18 10 9 5 By firm sales (millions) Less than £200 152 452 287 64 17 10 4 5 £200 to £500 119 403 335 61 19 8 6 6 £500 to £1,500 116 601 507 54 20 10 12 4 Above £1,500 123 927 811 55 16 10 15 4 By industry Mining/manufacturing 217 564 436 59 17 9 9 5 Financial services 84 559 411 60 22 6 7 4 Utilities 19 448 382 58 15 6 14 8 Other 190 645 397 58 17 11 8 5 United States All companies 1,666 3,565 1,508 29 17 42 4 8 By firm sales (millions) Less than £200 339 1,166 686 38 14 43 1 4 £200 to £500 379 1,833 926 36 18 36 3 7 £500 to £1,500 458 3,038 1,604 28 18 40 5 9 Above £1,500 490 7,056 3,552 20 17 48 5 10 By industry Mining/manufacturing 842 3,388 1,540 28 17 43 3 8 Financial services 198 6,277 2,787 19 20 47 5 8 Utilities 120 1,333 707 43 15 23 6 13 Other 506 3,326 1,438 32 16 43 3 6

Note: UK data from the largest companies in fiscal 1997, ranked by market capitalisation. US data include firms in the S&P 500, the S&P MidCap 400, the S&P SmallCap 600, and companies in S&P supplemental indices. Revenues for financial firms defined as net interest income (banks) and total income (insurance companies).Total compensation defined as the sum of salaries, bonuses, benefits, share options (valued on date of grant using the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for performance-contingent awards), and other compensation. US dollar-denominated data are converted to UK pounds using the average $/£ exchange rate during the fiscal year. 376 DIRECTORS’ REMUNERATION median UK pay. The US premium is grants comprise a much larger percentage especially pronounced for large firms (42%). The divergence between UK and (where the median US CEOs earns 340% US pay practices is, again, especially pro- more) and financial firms (where the nounced for companies with revenues average US CEOs earns 580% more). exceeding £1,500 million. Within this The right-hand portion of Table 14.1 group, salaries account for more than half describes the average composition of CEO of pay for UK CEOs, but account for only pay in the two countries. On average, CEOs one fifth of pay for US CEOs. Similarly, in the United Kingdom receive 59% of share option grants account for nearly their total pay in the form of base salaries, 50% of pay for CEOs in large US firms, 18% in bonuses, 10% in share options but only account for 10% of pay in large (valued at grant-date), and 9% in LTIP British firms. shares (valued at grant-date, with a 20% Table 14.2 compares base salaries and discount for performance contingencies). In the prevalence of contingent-pay practices contrast, base salaries comprise a much in the two countries. The median United smaller percentage of total pay for US States base salary of £317,000 is more executives (only 29%), while share option than 30% higher than the median United

Table 14.2 Summary statistics for components of CEO pay, by company size and industry

Annual bonus Value of option Value of LTIP Base received grant shares salary median % with Median (£) % with Median (£) % with Median (£) Group (£000s) bonus (for £0) grants (for £0) grants (for £0)

United Kingdom All Companies 240 81 91 50 69 32 161 By firm sales (millions) Less than £200 175 76 59 40 70 14 58 £200 to £500 200 82 69 44 45 25 94 £500 to £1,500 264 87 104 59 73 44 148 Above £1,500 410 80 146 59 108 50 294 By industry Mining/manufacturing 254 84 90 53 67 34 165 Financial services 222 83 115 45 50 29 132 Utilities 240 84 86 47 43 58 120 Other 221 77 83 48 110 29 177 United States All companies 317 83 270 72 1,142 19 325 By firm sales (millions) Less than £200 195 74 99 69 592 6 107 £200 to £500 256 82 180 64 787 15 180 £500 to £1,500 335 85 299 71 1,053 22 315 Above £1,500 487 89 518 80 2,505 30 579 By industry Mining/manufacturing 318 85 280 74 1,111 18 318 Financial services 395 92 437 81 1,713 31 611 Utilities 291 81 157 53 392 25 177 Other 301 78 239 68 1,232 15 336

Note: Median data for bonuses, options, and LTIP represent the median value of award/grant (in £000s) for the subsample of CEOs actually receiving awards/grants during the 1997 fiscal year. Revenues for financial firms defined as net interest income (banks) and total income (insurance companies). Share options (valued on date of grant using the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for performance-contingent awards), and other compensation. US dollar-denominated data are converted to UK pounds using the average $/£ exchange rate during the fiscal year. CEO PAY IN THE US AND UK 377 Kingdom salary of £240,000; median the value of the grant (for those receiving salaries are higher in the United States for grants) is higher in the US. each size and industry group.The percentage of CEOs receiving bonuses is roughly the 2.2. Time-series comparisons same in the two countries (81% in the UK compared to 83% in the US). However, Although longitudinal comparisons of total conditional on receiving a bonus, US compensation are not possible because of bonuses are much higher: the median bonus UK disclosure requirements prior to 1997, paid in the United States of £270,000 is time-series data on cash compensation triple the median bonus paid in the United (salaries and bonuses) are available for Kingdom. American CEOs are more likely the sample companies. Figure 14.1 shows the to receive option grants than their UK median cash compensation received from counterparts (72% in the United States 1989–97 for UK CEOs and US CEOs in the versus only 50% in the United Kingdom). S&P 500, and from 1992–97 for US CEOs In addition, they are likely to receive much in the MidCap 400 and SmallCap 600.The larger grants: the median option grant in figure includes data from firms in our 1997 the United States (for CEOs receiving sample of 510 UK and 1,666 US firms.11 options) of £1,142,000 is nearly twenty As shown in the figure, the median cash pay times the median grant value for UK CEOs for UK CEOs has grown from £158,000 in (who receive only £69,000). Finally, 1989 to £340,000 in 1997 (representing Table 14.2 shows that 32% of UK CEOs a 10% average annual growth). In the receive LTIP share grants, while only 19% United States, median pay among S&P of US CEOs receive similar grants. However, 500 CEOs has grown 6.4% annually, from although fewer US executives receive LTIP £574,000 in 1989 to £945,000 in 1997. share grants (primarily restricted stock), Since 1992, UK CEO cash compensation

1,000 US S&P 500 900

800

700

600 US Mid Cap

500

400 US Small Cap

300 Cash compensation (£000s) 200 UK Top 500

100

0 1989 1990 1991 1992 1993 1994 1995 1996 1997

Figure 14.1 Median Cash Compensation of US and UK CEOs, 1989–97. Note: Cash compensation includes salaries and bonuses. UK data from the largest 510 companies in fiscal 1997, ranked by market capitalisation. US data include firms in the firms in the 1997 S&P 500, the S&P MidCap 400, and S&P SmallCap 600. US dollar-denominated data are converted to UK pounds using the average $/£ exchange rate during each calender year. 378 DIRECTORS’ REMUNERATION has increased on average 11.2% per year, in Figure 14.2.The UK data are from Main compared to growth rates of 10.7%, (1999), and are based on data provided 11.9%, and 10.2% for CEOs in the S&P by a large compensation consulting firm. 500, MidCap, and SmallCap companies, The US data for 1979–96 are from respectively. the Conference Board ‘Top Executive The results in Figure 14.1 suggest that cash Compensation’ reports, which cover pre- compensation in UK firms is roughly ‘on dominately S&P 500 companies.13 The par’ with median pay among CEOs in Small- data for ‘US Small and MidCap’ firms are Cap US firms, but still lies significantly extracted from ExecuComp, and defined below pay in larger US firms.12 Moreover, as the fraction of MidCap and SmallCap as documented by Hall and Liebman companies in which the top five executives (1998) and Murphy (1999), the value of hold any options during the year. share options granted to US CEOs has As reported by Main (1999) and grown much faster than their cash compen- replicated in Figure 14.2, option grants in sation since the early 1990s. Indeed, as the United Kingdom grew dramatically in indicated in Table 14.1, share options have popularity from the mid-1980s to the emerged as the single largest component of early-1990s. In particular, in 1978 only compensation for CEOs in the largest US 10% of UK companies offered options to companies. Although comparable longitudi- their top executives, by 1983 over 30% of nal data on share-option grants in the companies offered options, and by 1986 United Kingdom are not systematically nearly 100% offered options. However, the available, time-series data on the preva- use of share options in the UK fell substan- lence of option plans among UK and US tially in the mid-1990s; by 1997 only 68% companies are available and are reported of companies offered options to their top

100

90 United States

US Small & 70 MidCap firms

60 United Kingdom

20 Percentage of firms with option plans 10

0 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97

Figure 14.2 Prevalence of Stock Option Plans, 1979–97. Note: UK data from Main (1999). US data 1979–96 from the Conference Board ‘Top Executive Compensation’ reports (various issues). US data for 1997 include percentage of S&P 500 companies that have executives holdings options (these data closely track the Conference Board data from 1992–96). US Small and MidCap data include percentage of companies in the S&P MidCap 400 and SmallCap 600 that have executives holding options. CEO PAY IN THE US AND UK 379 executives. In contrast to the UK experience, as the elasticity of cash compensation Figure 14.2 shows that the prevalence of to company revenues. Rosen (1992) sum- option plans in the US has increased rather marises academic research covering a than decreased in recent years. In particu- variety of industries and a variety of time lar, the percentage of large firms with periods in both the United States and the option plans grew from 82% in 1992 to United Kingdom. Even though there is 97% in 1997. The growth in prevalence is some variation in wage-size elasticities, his more pronounced in smaller firms: in 1992 general conclusion is that the ‘relative only 55% of all small and midcap firms uniformity [of estimates] across firms, offered options to its top executives; 96% of industries, countries, and periods of time is the companies had plans in place by 1997. notable and puzzling because the technology Figure 14.2 shows the percentage of that sustains control and scale should vary companies that have option plans, but does across these disparate units of comparison.’ not provide information on the magnitude Table 14.3 replicates these earlier findings of the grants to executives. Although UK for 1997 CEO pay data, with pay-size elas- option data prior to 1997 are not publicly ticities estimated from the regression: available, Hemmington Scott estimate the total number of option shares held by ln (1997 CEO Pay) ln(1997 Sales). the CEO in both 1991 and 1997. For (3) 395 UK companies with share option data Consistent with prior findings (see Murphy, in both years the Hemmington Scott reports 1999), the estimated elasticities for US cash that the mean CEO option holdings (meas- compensation (salary plus bonus) are not ured as a percentage of common equity) significantly different from ≈ 0.3 for all fell from 0.26% in 1991 to 0.22% in 1997; companies or for each of the four industry median holdings over the same time period groups. The US elasticities for total com- increased slightly from 0.09% to 0.11%. pensation (including salaries, bonuses, In contrast, mean option holdings among benefits, share options, LTIP-shares, and US CEOs in 1,128 Execu-Comp firms with other compensation) are somewhat higher share option data in both 1997 and 1992 than the elasticities for cash compensation, (the first year such shares were publicly reflecting primarily that option grants are disclosed) grew from 0.87% in 1992 to increasingly larger in bigger firms. 1.1% in 1997; median holdings grew from Interestingly, the estimated pay-size 0.16% to 0.64% over the same time period. elasticities for UK firms in Table 14.3, Overall, Fig 14.1 and 14.2 and our while positive and statistically significant results on option shareholdings suggest for all industries, are uniformly and signifi- that UK and US pay levels are diverging cantly smaller for UK firms than for US rather than converging in recent years. firms. For example, an elasticity of cash While CEO cash compensation is growing pay to sales of 0.316 in the United States at roughly the same rate in both countries, and 0.197 in the United Kingdom indicates the use of share options has increased in that doubling size increases cash pay the United States while declining in the by 32% in the United States but only by United Kingdom. In Section 5 below, we 20% in the United Kindgom. Similarly, analyse the political and economic factors doubling size increases total pay by 41% in contributing to the divergence of option- the United States, but only by 22% in the granting practices in the two countries. United Kingdom.14 Table 14.4 reports coefficients of regres- 2.3. Regression results sions estimating the US pay premium after controlling for company size and industry. The best-documented empirical finding in We use revenues as our measure of company the executive compensation literature is the size, and classify companies into four broad consistency of the relation between CEO industry groups; our results are not changed pay and company size, typically measured using alternative size definitions (assets, 380 DIRECTORS’ REMUNERATION

Table 14.3 Estimated elasticities of CEO compensation with respect to firm revenues

United States United Kingdom

Industry Salary bonus Total pay Salary bonus Total pay

All companies 0.316 [0.24] 0.413 [0.28] 0.197* [0.31] 0.217* [0.26] Mining & manufacturing 0.294 [0.20] 0.423 [0.35] 0.189* [0.31] 0.157* [0.23] Financial services 0.342 [0.42] 0.465 [0.34] 0.177* [0.26] 0.205* [0.30] Utilities 0.334 [0.36] 0.459 [0.30] 0.209 [0.17] 0.314 [0.32] Other 0.329 [0.28] 0.366 [0.20] 0.270 [0.45] 0.296 [0.33]

Note: Elasticities computed from regressions of ln(Pay) on ln(Sales) for 510 UK and 1,666 US companies. R2 in brackets. Revenues for financial firms defined as net interest income (banks) and total income (insurance companies). * indicates that the estimated elasticity in the United Kingdom is significantly different from the US elasticity at the 1% level. market capitalisation) or more narrowly chairman. Combining the CEO and chairman defined industries.The dependent variable in position is expected to increase pay for columns (1) and (2) is ln(Salary Bonus), two reasons. First, the combination is a while the dependent variable in columns proxy for added responsibility and/or (3) and (4) is ln(Total Pay).The coefficient ability of the incumbent CEO. Second, the on the US dummy in column (1) of 0.3861 combination is a proxy for the CEO’s influ- indicates that, after controlling for size and ence over the board and the remuneration industry, CEOs in the United States earn committee. In the United Kingdom, such approximately 47% more than their British combinations are the exception: only 18% counterparts.15 Similarly, the US dummy in of the sample companies have CEOs who column (3) of 1.142 indicates that total also serve as board chairmen. However, in expected pay is 215% higher in the United the United States such combinations are States after controlling for size and industry. the rule: nearly two-thirds (66%) of the Existing empirical studies of executive sample have combined CEO/chairmen. compensation have consistently documented Columns (2) and (4) of Table 14.4 that company size and industry are the two report coefficients for cash and total most important factors determining levels compensation.The control for risk is insignif- of CEO pay. However, researchers routinely icant in the cash compensation regression, include additional controls for risk (because but significant for total compensation, CEOs will demand higher expected pay while the control for investment options is levels if compensation is risky), investment insignificant in both regressions.16 The opportunities (because firms with growth human capital controls are significant, and options need better managers), and human suggest that CEO age-earnings profiles for capital.The regressions in columns (2) and cash (total) compensation turn downward (4) of Table 14.4 include as additional after age 55 (age 43).17 The coefficients explanatory variables the volatility of on the CEO/chairman dummy variables are shareholder returns (as a proxy for risk), positive and significant, indicating that the the book-to-market ratio (the inverse of a combination increases cash and total com- proxy for investment opportunities), and pensation by 11% and 19%, respectively. CEO age and age-squared (traditional Including the control for CEO/chairman human capital variables to allow concave combinations reduces the estimated US age-earnings profiles). In addition, the premium slightly, from 47% to 45% for regressions include a dummy variable set salary and bonus, and from 215% to to unity if the CEO and board chairman 190% for total compensation. position are combined, and zero if the Several stylised facts emerge from corporation has a separate non-executive the results of Tables 14.1–14.4 and CEO PAY IN THE US AND UK 381 4.0) 2.6) 7.8) ted data are f grant using f grant ompensation. 0.0009 ( 0.1248 ( 0.7146 ( 2.1) 9.3) 0.0956 ( 0.7891 ( 3.1) — 0.2) — 0.0006 (0.1) 0.3) — 0.8702 (4.0) 1.5) 5.0) 0.0006 ( 0.0007 ( 0.0609 ( 0.0625 ( 0.3902 ( 2.9) 5.4) 0.1130 ( 0.3848 ( —— 0.0663 (3.0) — 0.0779 (3.3) — — Dependent variable 0.306 0.315 0.437 0.474 (1) (2) (3) (4) (Dummy) — 0.1038 (2.5) — 0.1771 (3.7) (Dummy) Explanatory regressions for 1997 CEO compensation for Explanatory regressions (Dummy) 0.3861 (9.5) 0.3747 (7.7) 1.142 (23.9) 1.072 (18.8) (Dummy) 0.3236 (6.1) 0.3278 (5.7) 0.3553 (5.7) 0.4083 (6.1) ) 0.2873 (26.9) 0.2693 (20.8) 0.3689 (29.3) 0.3883 (25.7) (Dummy) : as the sum of salaries, t-statistics defined parentheses.Total in compensation bonuses, benefits, on date o (valued share options 2 Sales Sample Size 2,165 1,805 2,165 1,805 R US Firm CEO Age CEO Age-Squared CEO/Chairman Combined Book to Market Ratio Volatility of Shareholder Returns Other Industry the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for performance-contingent awards), performance-contingent for at 80% of face value the Black-Scholes formula), (valued and other stock grants LTIP-related c year. the fiscal during rate $/£ exchange the average to UK pounds using converted Book to market defined as the book value of tangible equity divided by the end-of-year market capitalisation. market by the end-of-year equity divided of tangible dollar-denomina as the book value US defined Book to market Finance Utility Interceptln( 4.023 (52.6) 2.293 (3.7) 3.839 (42.6) 1.435 (2.0) Table 14.4 Independent variable ln (Salary and Bonus)Note ln (Total Pay) 382 DIRECTORS’ REMUNERATION Figs 14.1–14.2. First, American chief shares held by the CEO declines with executives are paid significantly more company size in the United Kingdom, but than British chief executives, even after increases with company size in the United controlling for company size, industry, and States. In the largest firms, US CEOs hold other firm and executive characteristics. on average shares worth £145 million Second, although CEO pay increases with (median £5 million), while UK CEOs hold company size in both countries, the pay- shares worth only £3.4 million (median size gradient is significantly higher in the £330,000). United States than in the United Kingdom. Since agency costs arise when agents Third, the US pay premium can be traced receive less than 100% of the value of largely, but not entirely, to the prevalence output, the CEO’s share of ownership is a and magnitude of share option grants, natural measure of the severity of the which have increased in the United States agency problem. As reported in Table 14.5, since the mid-1990s, while declining in the the average and median shareholdings for United Kingdom. UK CEOs (expressed as a percentage of outstanding shares) are 2.13% and 0.05%, respectively, significantly smaller than the 3. THE RELATION BETWEEN average and median holdings among US CEO PAY AND CORPORATE CEOs of 3.10% and 0.29%, respectively. PERFORMANCE In the largest firms, US CEOs hold on average 1.61% of their company’s shares Empirical investigations of executive (median 0.09%), while UK CEOs hold only incentives have typically focused on the 0.21% (median 0.01%). In fact, whether relation between the CEO’s wealth and measured in pounds or percentages, or the wealth of the company’s shareholders. at averages or medians, share ownership CEO wealth is directly related to shareholder among US CEOs is substantially higher wealth through the CEO’s holdings of shares, than ownership among UK CEOs for each share options, and LTIP shares. In addition, size and industry group in Table 14.5. CEO wealth can be indirectly related to Consequently, share ownership mitigates shareholder wealth through accounting- more of the agency problem in the United based bonuses (reflecting the correlation States than in the United Kingdom. between accounting returns and share-price Holdings of unexercised share options performance) and through year-to-year also provide a direct link between CEO and adjustments in salary levels, target bonuses, shareholder wealth, because the value of and option and LTIP grant sizes. Finally, the options held increases with increases CEO wealth (and reputation) is affected by in the share price. Table 14.5 shows that performance if CEOs lose their jobs when the average US executive holds options their firms are performing poorly. to purchase 1.18% of the company’s outstanding shares (median 0.72%), while 3.1. The direct relation between the average UK executive holds options to CEO and shareholder wealth purchase only 0.24% (median 0.11%) of his company’s shares. Again, pair-wise The most obvious linkage between CEO and comparisons from the top and bottom shareholder wealth comes from the CEO’s panels of Table 14.5 reveal that holdings of holdings of company shares. As reported in previously granted but unexercised share Table 14.5, the share ownership for UK options are substantially higher in the CEOs is worth an average of £7 million, sub- United States than in the United Kingdom stantially smaller than the average holdings for every size and industry group. among US CEOs of over £60 million. The Finally, holdings of unvested LTIP shares ownership distribution is significantly provide an additional direct link between skewed: the median holdings for UK and US CEO and shareholder wealth. First, the value executives are £460,000 and £3.3 million, of the underlying shares naturally increases respectively. Interestingly, the value of penny-for-penny with increases in the share CEO PAY IN THE US AND UK 383

Table 14.5 Summary statistics for stock-based CEO incentives, by company size and industry

Share holdings Share holdings Option holdings Pay-performance (£ millions) (% of common) (% of common) sensitivity (%)

Group Average Median Average Median Average Median Average Median

United Kingdom All companies 7.01 0.46 2.13 0.05 0.24 0.11 2.33 0.25 By firm sales (millions) Less than £200 9.86 1.41 4.38 0.63 0.38 0.21 4.72 1.09 £200 to £500 9.50 0.70 2.55 0.14 0.24 0.14 2.75 0.42 £500 to £1,500 4.55 0.13 0.76 0.02 0.19 0.12 0.91 0.16 Above £1,500 3.40 0.33 0.21 0.01 0.10 0.04 0.31 0.05 By industry Mining/ 6.01 0.26 1.91 0.04 0.24 0.14 2.11 0.23 manufacturing Financial services 5.95 0.84 1.98 0.07 0.19 0.05 2.14 0.31 Utilities 0.16 0.11 0.01 0.00 0.03 0.02 0.05 0.02 Other 9.31 0.71 2.64 0.10 0.28 0.11 2.89 0.37 United States All companies 60.37 3.26 3.10 0.29 1.18 0.72 4.18 1.48 By firm sales (millions) Less than £200 16.63 2.07 5.32 0.96 1.84 1.37 6.98 3.65 £200 to £500 23.84 2.93 3.94 0.58 1.39 0.94 5.20 2.05 £500 to £1,500 32.25 2.64 2.36 0.25 1.12 0.70 3.43 1.26 Above £1,500 145.26 4.96 1.61 0.09 0.62 0.40 2.17 0.56 By industry Mining/ 33.77 3.14 2.78 0.31 1.22 0.76 3.87 1.53 manufacturing Financial services 127.46 7.58 2.25 0.31 0.98 0.52 3.17 1.01 Utilities 6.19 0.58 0.25 0.05 0.35 0.12 0.53 0.16 Other 91.34 4.22 4.63 0.38 1.40 0.95 5.96 2.01

Shares held as Options held as Options LTIP shares as LTIP % of firm shares % of firm shares Delta % of firm shares Delta where 0 Option Delta 1 is the share-weighted-average slope of the Black-Scholes function at the year-end stock price, for options outstanding at the fiscal year end, and (LTIP Delta) 1. price. Second, there is an additional link to various ownership percentages. However, a the extent that the shares vest only upon share option provides the same incentives meeting share-based performance criteria. as a share of stock only if the option is deep The average US executive holds LTIP shares ‘in the money’ (that is, the share price is far (primarily restricted stock) on 0.09% of the in excess of the exercise price). Therefore, company’s outstanding shares, while the each share option should count somewhat average UK executive holds LTIP shares on less than one share of stock when adding 0.03%; the median LTIP shareholdings is the holdings to form an aggregate measure zero in both countries. of CEO incentives. In constructing our One way to aggregate the various com- aggregate measure of CEO incentives, we ponents of the direct link between CEO and weight each option by the ‘Option Delta’, shareholder wealth is to simply sum the which ranges from near zero (for deep 384 DIRECTORS’ REMUNERATION out-of-the-money options) to near one (for American-style restricted shares (which deep in-the-money options on non-dividend vest with the passage of time, without a paying stock).18 The option delta is a well- performance contingency) have an LTIP known concept from option pricing theory, delta of one, because a £1 increase in and equals the ‘slope’ of the Black-Scholes the share price results in a £1 increase function (that is, the change in the Black- in the LTIP share. Calculating LTIP deltas Scholes value for an incremental change in for performance-contingent grants is more the share price). Formally, complicated. The delta is near zero when there is little chance of achieving threshold Option Delta eln(1 d)T N(z), (4) performance, near one when the LTIP shares are highly likely to become vested, where d, T, and z are as defined in (1). and above one in the range where small Option deltas for ten-year ‘at the money’ changes in share prices have a large effect options are ≈0.9 for options on non-divi- on the likelihood of vesting. As a simplifi- dend paying shares, and ≈0.6 for options on cation, we assume that LTIP deltas for shares with a 3% dividend yield. all LTIP shares are one, independent of Calculating the option delta for each performance-vesting contingencies. option held at the end of the fiscal year Our aggregate measure of direct incen- requires exercise price and expiration-term tives is the CEO’s ‘effective’ ownership information for each outstanding option percentage, computed as the following delta- grant. The information required to calcu- weighted average: late option deltas for 1997 grants is publicly disclosed (and hence available) for Shares held as companies in both the United Kingdom %of firm shares and United States. The required data are not uniformly available, however, for grants Options held as Options made prior to 1997, but unexercised at the %of firm shares Delta end of the 1997 fiscal year. In particular, although most UK companies provide LTIP shares as LTIP (5) detailed data on each grant held at the %of firm shares Delta fiscal year closing, some UK companies and all US companies only provide data on Our measure of effective CEO ownership is (1) the number of share options held, and essentially the ‘Pay-Performance (2) the average exercise price for unexer- Sensitivity’ introduced by Jensen and cised options granted in prior years.19 Murphy (1990a).The difference is that we In cases where complete data were not have measured the effective ownership available, we calculated deltas for an percentage, while Jensen and Murphy aggregated prior grant, with an exercise measured the change in CEO wealth per price equal to the average exercise price $1,000 change in shareholder wealth, and a term assumed to be five years.20 which equals the effective ownership LTIP shares are also not equivalent to percentage multiplied by ten. Also, our unrestricted shares owned, because they effective CEO ownership statistic yields a may be forfeited if certain employment and distribution of CEO incentives. That is we performance objectives are not achieved. calculate directly the incentive (pay- Analogous to our treatment of stock performance term) for each CEO separately. options, we weight each LTIP share by an This differs from much of the prior compen- ‘LTIP Delta’, which is a measure of the sation literature which relies on regression change in value of each LTIP share for an techniques to derive a single average incremental change in the share price. We estimate of the pay-performance link.21 assume that LTIP recipients will remain The final two columns of Table 14.5 show employed long enough for all time-related the average and median pay-performance restrictions to lapse. Under this assumption, sensitivity for CEOs grouped by country, CEO PAY IN THE US AND UK 385 8.9) 4.4) 0.4) 15.1) s net interest 0.320 ( 0.557 ( 0.037 ( 0.320 ( 7.0) 1.9) 20.6) 0.946 ( 0.188 ( 0.414 ( 5.0) — 3.6) 1.3) 9.5) 0.967 ( 2.533 ( 0.645 ( 1.103 ( 4.3) 0.7) 11.0) 2.724 ( 0.324 ( 1.038 ( Dependent variable: Stock-based pay-performance sensitivity — 0.009 (5.5) — 0.003 (9.8) ——— 2.303 (1.4) 0.018 (0.4) — — 4.384 (14.3) 0.003 (3.9) 0.092 0.157 0.080 0.102 (1) (2) (3) (4) (Dummy) — 2.898 (7.9) — 0.394 (5.8) (Dummy) 1.764 (5.2) 2.026 (5.4) 0.304 (4.2) 0.360 (5.2) Explanatory regressions for stock-based pay-performance sensitivity stock-based pay-performance for Explanatory regressions (Dummy) 2.467 (6.9) 0.963 (2.2) 1.225 (16.1) 0.652 (8.2) ) (Dummy) (Dummy) : parentheses. in regressions) t-statistics OLS) and asymptotic t-statistics median a (for (for defined firms financial Sales for 2 Sales Sample Size 2,174 1,812 2,174 1,812 R CEO Age-Squared US Firm CEO Age Volatility of Shareholder Returns Other Industry Book to Market Ratio CEO/Chairman Combined Utility Finance income (banks) and total income (insurance companies). (insurance (banks) and total income income Sensitivity. of the Pay-Performance our definition for table on the prior See the footnote In ( Intercept 8.181 (12.1) 32.191 (5.8) 3.052 (21.3) 9.545 (9.3) Table 14.6 Independent variable OLS regressionsNote Median regressions 386 DIRECTORS’ REMUNERATION company size, and industry.The average and sensitivities are substantially higher in the median pay-performance sensitivities for US United States, even after including our CEOs are 4.18% and 1.48%, respectively, control variables. significantly higher that the average (2.33%) and median (0.25%) sensitivities for UK 3.2. The indirect relation between CEOs. In the largest firms, the median US CEO and shareholder wealth CEO has a sensitivity, or effective ownership, of 0.56%, more than ten times the effective In addition to the direct relation through ownership of their British counterparts. ownership, LTIP shares, and options, Table 14.6 presents coefficients from CEO wealth is indirectly related to OLS and median regressions where the company stock-price performance through dependent variable is the stock-based pay- performance-based bonuses, raises, and performance sensitivity from (5). The coef- LTIP and option grant sizes. The CEO pay ficient on the US dummy in column (1) of literature has yet to reach a consensus on 2.467 indicates that the average effective the appropriate methodologies and metrics ownership is 2.5 percentage points higher in to use in evaluating the indirect relation the United States than the United between CEO pay and company stock-price Kingdom, after controlling for company performance.22 However,the most common size and industry.The regression in column approach has involved estimating some (2) includes controls for risk, investment variant of the following first-difference opportunities, and human capital intro- regression: duced in Table 14.4: the volatility of returns, the book-to-market ratio, CEO age ln(Salary Bonus)it and age-squared, and a dummy variable ln(Shareholder Value)it, (6) indicating that the CEO and board chairman position are combined. The volatility where the change in shareholder value, and book-to-market variables are insignifi- ln (Shareholder Value)t ignores share cant, while the age variables are significant issues or repurchases and therefore equals and indicate that pay-performance sensitivi- the continuously accrued rate of return on ties are monotonically increasing after age common stock, rt.The estimated coefficient 52. The coefficient of the CEO/chairman is the elasticity of cash compensation with dummy variable is positive and significant, respect to shareholder value (or, following indicating that pay-performance sensitivi- Rosen (1992), the ‘semi-elasticity’ of pay ties are higher for combined CEO/chairmen. with respect to the rate of return).23 So, although the level of pay is higher when Table 14.7 reports coefficients from the posts are combined (and corporate estimating (6) with US interactions. The governance reformers typically argue for pay-performance elasticity for UK CEOs in these posts to be separated for this reason) column (1) is 0.1213, indicating that an the link between pay and performance additional ten percentage point shareholder (effective ownership percentage) is also return corresponds to an additional 1.2% higher. Including the control for CEO/ pay raise. The pay-performance elasticity chairman combinations reduces the coeffi- for US CEOs in column (1) is 0.27 cient on the US dummy variable to 0.9632, (≈0.1213 0.1488), or more than double but the coefficient remains significantly the elasticity for UK CEOs. The difference positive. Columns (3) and (4) present in elasticities is statistically significant at coefficients from median regressions. The the 10% level. The remaining columns in book to market and volatility measures are Table 14.7 report estimated elasticities for now significant. Importantly, the US dummy the industrial groups. The estimated US variables in both columns (3) and (4) are elasticity is higher than the corresponding positive and significant (although quantita- UK elasticity for all industries, although the tively smaller than in the mean regressions), difference is only statistically significant for indicating that median pay-performance the financial services industry. CEO PAY IN THE US AND UK 387 0.5) 0.0350 ( 1.4) 0.6) 0.1348 (1.3) tion. the in firms data include US 0.6182 ( 0.6359 ( 2.8) 1.0) 0.3273 ( 0.1809 ( 2.8) Bonus) 0.0883 ( ln (Salary 2.8) 0.0861 ( Dependent variable: 0.038 0.050 0.107 0.065 0.026 (1) (2) (3) (4) (5) ) 0.1213 (2.0) 0.1569 (2.2) ) 0.1558 (1.9) 0.1268 (1.5) 0.7567 (2.9) 1.062 (1.0) 0.1101 (0.7) US Dummy CEO pay-performance elasticities for salary and bonus, for elasticities CEO pay-performance by industry ( US Firm ) Shareholder Return : asymptotic t-statistics parentheses. White in 1997, fiscal in the largest 510 companies UK data from capitalisa by market ranked 2 R Sample Size 2,069 1,009 273 135 652 Return ( (Dummy) S&P 500, 400, the S&P MidCap and the S&P SmallCap 600. Datastream (UK) and Compustat from (US). return Shareholder ln(1 Intercept 0.1448 (7.1) 0.1552 (6.6) 0.2298 (3.4) 0.5689 (1.3) 0.1158 (3.0) Table 14.7 Independent variable All industries Mining & manufacturingNote Financial services Utilities Other industries 388 DIRECTORS’ REMUNERATION 0.4) 2.2) 5.8) tistics in 0.12 ( 1.81 ( 1.63 ( 1.4) 1.8) 1.0) S&P 500, 400, the S&P MidCap the 1.15 ( 0.78 ( 12.54 ( 1.0) 0.1) 3.9) 0.78 ( 0.20 ( 2.34 ( 1.6) 1.7) 7.8) 0.40 ( 1.56 ( 1.70 ( 1.4) 2.8) 10.7) 0.26 ( 1.65 ( 1.71 ( Dependent variable: CEO Turnover in 1996 0.030 0.037 0.023 0.034 0.042 (1) (2) (3) (4) (5) ) 1 if CEO is in last year of office in 1996. in of office last year 1 if CEO is in is 0.119. Mean of the dependent variable 1996. is measured in return Shareholder z-sta Dummy) 0.43 (0.7) 0.29 (0.3) 2.66 (1.1) 11.37 (1.6) 0.50 (0.5) US CEO turnover-performance regressions, by industry CEO turnover-performance ( US Firm 2 ) Shareholder Return : Dependent variable Return ( Sample Size 1,942 984 241 135 582 (Dummy) ln(1 S&P SmallCap 600, supplemental S&P indices. in and companies Datastream (UK) and Compustat from (US). return Shareholder Intercept Table 14.8 Independent variable All industriesPseudo R Mining & manufacturingNote Financial services Utilities Other industries parentheses. 1997, fiscal in the largest 510 companies UK data from capitalisation. by market the in ranked firms data include US CEO PAY IN THE US AND UK 389 3.3. CEO turnover and corporate (iii) we are not controlling for corporate performance governance variables, including the composition of the board. We leave a comprehensive Another important potential source of investigation to future research. managerial incentives is the threat of being fired for poor performance. Several researchers, beginning with Coughlan and Schmidt (1985), Warner et al. (1988), and 4. DISCUSSION Weisbach (1988) have documented an inverse relation between CEO turnover and Consistent with our objective of generating shareholder returns for US companies; stylised facts to stimulate future research, Conyon (1998) has shown that the inverse our paper to this point has been primarily relation holds for UK data as well, and descriptive. That is, although we have Kaplan (1994 a, b) reports similar findings documented that American CEOs are for Japan and Germany. higher paid and have better stock-based Table 14.8 reports the coefficient incentives than British CEOs, we have estimates from logistic regressions of CEO offered little in the way of theory or turnover on shareholder returns by industry. conjecture to help explain these interesting The dependent variable is equal to one if differences. In this section, we will briefly the CEO is in his last full year of office in discuss a variety of factors that may fiscal year 1996.The independent variables explain the crosscountry differences. As include 1996 fiscal-year performance, a noted earlier, our objective is not to US dummy, and the US dummy interacted reconcile completely the differences in with the performance variable to identify pay practices, but rather to identify crosscountry differences in the turnover- potential explanations, leaving formal performance relation. The coefficient on hypothesis development and testing to the performance variable is negative in all future research. regressions (and significant in all but financial services), indicating that CEOs in 4.1. Agency-theoretic the United Kingdom are indeed more likely considerations to depart following poor performance. The interaction variable is positive but The traditional principal-agent model insignificant in all regressions, suggesting highlights the trade-off between risk and that there are no systematic US-UK dif- incentives. Increasing the pay-performance ferences in the CEO turnover-performance sensitivity imposes more risk on CEOs, who relation in our data.24 in turn demand higher levels of expected Our conclusion that the turnover- compensation to compensate for the performance relation in the United Kingdom additional risk. Therefore, the facts that is not significantly different from that in US CEOs are better paid and have stronger the United States is robust to alternative incentives than UK CEOs may both reflect definitions of performance (using various equilibrium incentive contracts. lagged performance measures, and allowing To illustrate, suppose that firm value is performance-size interactions). In addition, given by x e ‡, where e is executive our conclusion is consistent with Kaplan effort, and ‡ is (normally distributed) (1994a,b), who finds no systematic differ- uncontrollable noise, ‡ ≈ N(0, 2). Moreover, ences in turnover-performance relations in suppose that managerial contracts take the the United States vs. Japan or the United simple linear form (x) s bx,where States vs. Germany. However, we view our s is a fixed salary and b is the sharing results here as preliminary because (i) they rate (or ‘pay-performance sensitivity’). are based on only a single year of turnover Assuming that the executive has exponential data; (ii) we are not controlling for utility, U(x) er[Wc(e)], where r is the turnover related to normal retirement; and executive’s absolute risk aversion and c(e) 390 DIRECTORS’ REMUNERATION is the convex disutility of effort, the optimal may explain divergences in pay between the sharing rate is given by:25 United States and United Kingdom. Similarly, American CEOs may have more 1 b . (7) decision rights and influence over corporate 1 r2c results than do British executives. Unfortunately, traditional agency theory Equation (7) implies that the optimal pay- gives little guidance on why the career performance sensitivity b will equal 1 when paths, production functions, or hierarchical output is certain (2 0) or executives structures should vary across international are risk-neutral (r 0). Incentives will be boundaries. weaker for more risk-averse executives (∂b/∂r 0), and will also be weaker the greater the uncontrollable noise in firm value (∂b/∂2 0). Moreover, expected 4.2. Taxes compensation E (w) s bE (x) will As emphasised by Miller and Scholes increase monotonically with b to compensate (1982) and Abowd and Bognanno (1995), for both the increased risk imposed, and corporate and personal tax regimes affect the increased effort induced, by higher the optimal structure of executive compensa- pay-performance sensitivities. tion contracts. Overall, the personal income Equation (7) suggests a finite number of tax rules and rates affecting executive pay factors that might explain higher pay and are quite similar in the United States and incentives among US executives. First, United Kingdom. Cash compensation and American CEOs may be less risk averse or gains from ‘unapproved’ (or ‘nonqualified’) have steeper marginal costs of effort than share options are taxed as personal income their British counterparts, but to our in both countries, at comparable rates.27 knowledge there is no theory or empirical ‘Approved’ share options, granted infre- work suggesting such international differ- quently to top executives because of ences in risk-aversion coefficients. Second, institutional restrictions, are taxed at UK performance might be measured with capital gains rates in both countries. In substantially more noise than in the United recent years, the capital gains rate in the States, leading to lower pay-performance UK has increased from 30% to 40%, while sensitivities and lower expected levels of falling from 28% to 20% in the US. pay. However,we find no evidence that cash Apart from the difference in capital flows or shareholder returns are systemati- gains tax rates, which is both relatively cally more variable in the United Kingdom recent and fairly benign (given the paucity than in the United States. of approved share grants), there are no Overall, the traditional principal-agent major differences in the personal tax regimes model encapsulated in (7) does not offer in the United Kingdom and the United States promising explanations for the difference in related to executive compensation. However, pay levels and incentives in the two countries. there are two significant differences in Extensions of the model to incorporate the extent to which compensation is differences in both ability and in the deductible from corporate profits as an marginal productivity of CEO effort might ordinary business expense. First, the help reconcile the data, but only given the exercise-date spread between the market additional assumptions that executives and exercise price on unapproved options is are more able and more productive in the treated as a deductible compensation United States.26 For example, Granick expense in the United States, but not in (1972) shows that US managers have the United Kingdom. Second, under rules greater exposure to different functions and passed in 1993, the United States limits operating divisions and that such diversity the deductibility of ‘non-performance- is rewarded at the top of the hierarchy.The based’ compensation (including salaries, complexity of the filtering and training restricted stock, and discretionary bonuses) processes, and differences in each country, to $1 million.28 Thus, UK rules allow CEO PAY IN THE US AND UK 391 deductions for cash compensation but and opted against requiring an accounting not for exercised options, while US rules change, adopting instead enhanced footnote allow deductions for exercised options but disclosure of granting practices.The October limit deductions for cash compensation. As 1995 FASB retreat ensured the continued discussed in the next sub-section, we trend in option awards, since the grants believe that these differences have played remained invisible from an accounting at least some role in the relative prevalence perspective, but fully deductible from a of share options in US pay packages. tax perspective. The S&P500 Index (a broad measure of 4.3. The rise (US) and US stock-market performance) increased fall (UK) of share options by 300% in the 1990s; the UK FTSE Index increased by only 150% during the The fact that US CEOs are better paid with same period. Proponents of large share stronger pay-performance sensitivities is option grants have pointed to the US stock- largely attributable to the relative preva- market performance as evidence that lence and magnitude of share options in the options provide incentives for executives to US. As documented by Hall and Liebman increase shareholder wealth. But, the robust (1998) and Murphy (1999) and suggested stock market has also been a contributing by Figure 14.2, the importance of share factor to the growing demand for option options in US compensation packages is a compensation among US executives. The relatively recent phenomenon, reflecting current cohort of executives has not expe- an explosion in option granting practices rienced a major market downturn, and the since the mid-1980s. Over the same period, overwhelming majority of US share options many UK companies have rejected share issued since 1980 have been exercised well option plans in favour of performance in-the-money. As a result of these past share plans such as LTIPs (Main, 1999). successes, both executives and employees Understanding the factors that have led have embraced share options as the quickest, to the rise in share options in the United and (by their perception) the surest, route States, and the fall of share options in the to obtaining substantial wealth. United Kingdom, is critical in understand- Economic, political and cultural factors ing the differences in compensation and have also shaped option granting policies in incentives in the two countries. the United Kingdom. In contrast to the The increased popularity of share option United States, where these factors have grants for US executives can be traced, in encouraged the use of share option com- large part, to a combination of economic, pensation, a variety of recent statutory and political, and cultural factors. Shareholder non-statutory arrangements in the United groups and academics in the early 1990s Kingdom have discouraged share option called for more stock-based compensation in grants. Options in the United Kingdom CEO pay packages, citing the lack of mean- became controversial in 1995, after execu- ingful rewards and penalties in the typical tives in several recently privatised electric package. The deductibility limitations on utilities exercised options worth millions non-performance-related pay introduced by of pounds. The influential Greenbury report the Clinton Administration in 1993 further (1995) encouraged companies to replace fuelled the popularity of stock options. The their option plans with LTIP share plans trend was nearly reversed in the mid-1990s, which ‘may be as effective, or more so, however, when the Financial Accounting than improved share option schemes in Standards Board (FASB) announced plans linking rewards to performance’ Greenbury to institute a grant-date accounting charge (1995, paragraph 6.32). In response to for companies granting share options. the Greenbury report and the ongoing Ultimately, FASB yielded to pressure from controversy, the government tightened the the business community, accounting firms, restrictions on approved option awards, and the hi-tech (‘Silicon Valley’) sector, reducing the amount that could be awarded 392 DIRECTORS’ REMUNERATION (expressed as the aggregate exercise price) perhaps it is natural that the United States from the greater of £100,000 or four times reacts to claims of excessive CEO pay cash emoluments to only £30,000. by increasing the link between pay and An important non-statutory consideration performance, thus exacerbating income for UK companies is the codes of conduct inequality, while the United Kingdom reacts issued by institutional investors and their through wage compression and reducing representatives. Although not legally binding, the pay-performance link. policy guidelines from the Association of The divergence in top wages in the UK British Insurers (ABI) and other investor vs. US is not, of course, limited to top groups are highly influential and closely executives. The best doctors, engineers, followed.The ABI guidelines (1994, 1995), professors, athletes, lawyers, investment for example, effectively constrain the bankers, and entertainers all earn substan- issuing of share options—approved and tially more in the United States than in the unapproved—to four times cash compensa- United Kingdom. The US wage premiums tion.29 More recently, the Pension Investment for ‘superstars’ in all occupations persist Research Consultants (PIRC, 1998) called in spite of the similarities in language, on companies to toughen performance culture, tax regimes, and institutions.These targets for executive option schemes and premiums may be evidence that the US other incentive plans. market for superstars (Rosen, 1981) is In sum, executive pay has been highly more competitive than the UK market, controversial in both the United States and with ‘winner-take-all’ rents flowing to the the United Kingdom, and has generated a producers (Frank and Cook, 1995). Thus, variety of responses from the government perhaps the CEO-pay differentials should and tax authorities, the accounting boards, be examined in the context of broader the media, and institutional investors. competitive and culture factors. Although the root causes of the controversy in the two countries are basically similar, the responses have been markedly different: 5. CONCLUSION while the net effect of these factors has been to greatly increase option grants in The recent wave of international corporate the United States, the net effect has mergers, such as Daimler-Chrysler and discouraged the use of option grants in the British Petroleum-Amoco, has sparked United Kingdom in favour of LTIPs. academic and practitioner interest in understanding international differences in executive 4.4. Culture pay policies. International comparisons are inherently difficult, because of fundamental We have documented that US CEOs have differences in data availability, tax regimes, higher pay and stock-based incentives than corporate governance and the organisation CEOs in the United Kingdom, and shown of business. For a variety of reasons, the that these results are driven, in large part, United States and United Kingdom offer a by the recent divergence in share option natural laboratory to examine in detail practices. Although it is difficult for us to differences in compensation and incentives explain why a similar controversy over CEO practices. First, the United Kingdom and pay has led to increased option grants in United States (along with Canada) are the United States but decreased grants currently the only countries that require in the United Kingdom, it is tempting to detailed disclosure on the compensation attribute at least part of the divergence practices for individual top corporate to cultural differences between the two executives. Second, the United Kingdom countries. The United States, as a society, and United States share a common has historically been more tolerant of language and have similar capital markets income inequality, especially if the inequality and underlying economies. Third, the is driven by differences in effort, talent, or United States and United Kingdom employ entrepreneurial risk taking. In this light, similar corporate governance structures, CEO PAY IN THE US AND UK 393 especially when compared to those in Japan, deductibility rules—which encourage option Germany, and other economic powers. compensation in the United States while These important similarities, in a sense, discouraging option compensation in the stack the deck against finding significant United Kingdom—help explain the observed differences in pay practices. differences in pay structures. Ultimately, Nonetheless, several stylised facts have however, the differences largely reflect emerged from our comprehensive compari- subtle political and cultural differences in son of CEO incentives and pay practices in the two countries. In the US, the controversy the United Kingdom and the United States. over CEO pay has led to tighter links Expected pay levels, after controlling for between executive pay and performance company size and industry, are significantly (primarily through an explosion in option higher in the United States than in the grants), exacerbating wage inequality given United Kingdom. Although base salaries the robust US stock market. In the United are modestly higher in the United States, Kingdom, the pay controversy has led to and annual bonuses substantially higher, statutory and non-statutory policies that the driving force behind the US premium discourage large share option grants, less- is the prevalence and sheer magnitude of ening the pay-performance link and leading share option grants to US executives. to a relatively compressed wage structure. The divergence between UK and US pay The differences in pay practices between practices is especially pronounced in large the United States and the United Kingdom firms and financial firms. (or, more generally among any countries) The link between CEO wealth and share- would have few consequences on supply holder wealth is also much stronger in the and demand relations if the managerial United States than in the United Kingdom. labour markets were truly isolated within Jensen and Murphy (1990 a) argued that country boundaries. But, large companies US CEOs have insufficient incentives to are increasingly multinational, and must increase shareholder wealth, but incentives continuously deal with expatriate compen- among British CEOs pale compared to sation and horizontal wage inequities their American counterparts. CEOs in the caused by heterogeneous local labour US hold more shares of stock (measured market conditions. Moreover, top-level both in value and as a percentage of out- managers are increasingly mobile and will standing shares), hold more share options, naturally flow to higher-paid markets: 62 of and hold at least as many LTIP shares as the top 800 CEOs in the United States are do UK CEOs. Overall, the pay-performance foreign born, 30 are originally European.30 sensitivity is substantially higher in the Understanding the magnitude, causes, and United States than the United Kingdom, for consequences of international differences every size and industry group. Moreover, in executive pay practices has important the indirect relation between cash compen- implications for our understanding of the sation and stock-price performance is also globalisation of world commerce. more strongly positive in the United States than the United Kingdom. In short, CEOs in the United States have more incentives NOTES to improve shareholder wealth than do CEOs in the United Kingdom. * We would like to thank John Core, Harry The differences between pay and incentive DeAngelo, Linda DeAngelo, Wayne Guay, Brian practices in the United States and United Hall, Steve Machin, Brian Main, Simon Peck Kingdom are somewhat surprising, given and Andrew Samwick and an anonymous the similarities in corporate governance referee for comments on this paper and its predecessors. Advice from practitioner-colleagues at and the managerial labour markets. The PriceWaterhouseCoopers (especially Stephen differences are especially surprising since Clements) is appreciated. We would like to executive pay has been highly controversial thank Graham Sadler for excellent research in both countries, for similar underlying assistance in assembling the UK data. Financial reasons. We believe that corporate tax support provided by the Economic and Social 394 DIRECTORS’ REMUNERATION Research Council (Award number R000237246) Compustat ExecuComp data, and the and the USC Marshall School is gratefully less-comprehensive surveys published by Forbes, acknowledged. Business Week, and the Wall Street Journal. 1 The US dollar-denominated data in this 7 Options typically expire after either 5 or paper are converted to UK pounds by the 10 years; our results are not sensitive to contemporaneous exchange rate, which whether we use the 5, 7, or 10-year risk-free ranged from 1.61$/£ to 1.65$/£ over the 1997 rates. The US risk-free rate is estimated at fiscal year. 6.3% for all fiscal 1997 data.The risk-free rate 2 Abowd and Bognanno (1995) use survey is set to 6.0% for UK firms with fiscal closings data from compensation consultants to show prior to July 31, 1997, and 7.0% for firms with that CEO pay differs internationally in both later closings. level and structure but little attempt is made to 8 Dividend yields above 5% are ‘trimmed’ to ensure commensurability between economies 5%, and volatilities are trimmed to lie in the and control for scale, etc. Conyon and range 20% to 60%. We imposed these Schwalbach (1999) show differences in constraints because abnormal historical divi- executive cash compensation across European dend yields and volatilities are poor predictors economies after controlling for company size of yields and volatilities over the term of the and job position. Main et al. (1994) compare option. These constraints do not, however, cash compensation in 1990 between the United change any of our qualitative results. States and United Kingdom but exclude any 9 Early exercise has ambiguous implications comparison of the structure of pay. for the cost of granting options. On one hand, 3 Murphy (1999) provides a comprehensive the right to exercise early increases the amount review of the executive compensation literature, an outside investor would pay for the option, with an emphasis on US data. Recent UK and hence increases the option’s cost. On the research includes Main et al. (1996), Cosh and other hand, risk-averse undiversified executives Hughes (1997), Conyon (1998), Conyon and tend to exercise much earlier than would Peck (1998b), and Ezzamel and Watson a rational outside investor, and these early (1998). With the exception of Main et al., these exercise decisions reduce the company’s cost of studies exclude share options and focus only on granting options (Carpenter, 1998). cash compensation. 10 The most common criterion requires 4 Other countries, including Japan and earnings-per-share growth exceeding 2% in any Germany, require disclosures of the aggregate three years of the options’ term. The Pension amount paid to the group of top directors or Investment Research Consultants (PIRC, 1998) executives, but do not identify individuals. argue that most requirements are non-demanding Disclosure in Canada was mandated in 1993 by as their index of large companies has achieved the Ontario Securities Regulation, and covers all real EPS growth rates in excess of 3%. publicly traded companies in the province of Similarly, we calculated in our data set of UK Ontario (including all companies on the Toronto companies that average (median) real earnings Stock Exchange). See Zhou (1999) for an per share growth between 1992 and 1997 was extensive analysis of CEO pay and incentives 11.04% (13.32%). Even the company at the in Canada. 20th percentile achieved average real EPS 5 Although the US reporting requirements growth over the period of 2%. are generally more stringent than the UK 11 The full time-series is not available for all requirements, data on prior share option grants of our 1997 sample firms, and the medians in are more detailed in the United Kingdom. Figure 14.1 are therefore based on smaller sam- 6 Secondary sources in the United Kingdom ples in earlier years. The results are unchanged include Datastream and PriceWaterhouse- when we restrict the analysis to only firms with Coopers’ Corporate Register published by complete time-series data. Hemmington Scott.These sources typically ignore 12 The firms in the Small-Cap sample are share options, combine other pay components considerably smaller than the firms in the into a single number, and focus on the highest- UK sample, with mean (median) 1997 market paid executive rather than the CEO. Information capitalisation of only £344 million (£247 million) about options can be obtained from the Register in the US Small-Cap sample compared to of Directors’ Interests. In practice it can be £2.2 billion (£484 million) in the UK sample. difficult to access this source (see Main et al., 13 The Conference Board discontinued their 1996, p. 1632). Secondary sources in the plan-prevalence data series in 1997. For this United States include the comprehensive year, we use ExecuComp data on the fraction of CEO PAY IN THE US AND UK 395 S&P 500 companies in which the top five details there is some risk that the abundance of executives hold any share options.This definition information will mask rather than highlight of prevalence closely tracks the Conference the nature and scale of option schemes’, and Board’s survey responses for the 1992–6 period allowed less-than-complete option information where both sources of prevalence data are when remuneration committees are ‘satisfied available. that this will not result in failure to disclose 14 Guy (1999) documents empirical evidence information of material importance’. Greenbury that the annually estimated CEO pay-size elas- (1995, p. 29). ticity varies positively with annual median pay. 20 About 80% of UK annual reports provided This is consistent with our results since median full information (i.e. the exercise price and US pay is higher than in the United Kingdom. expiration date) on every prior grant of options 15 Calculated as e0.3861 1 ≈ 0.47. still outstanding. In the remaining 20% of cases 16 However,in unreported regressions on only an exercise price is almost always given although UK data, cash compensation is positively and it is weighted exercise price of all options held significantly related to volatility, and negatively (see Conyon and Sadler, 1999). The expiration and significantly related to book-to-market date is given in only 50% of cases and is ratios. Since our primary objective here is to typically the expiration date of the longest-dated identify cross-country differences in pay levels, option. US proxy statements provide information we leave further examination of cross-country on the number and intrinsic value of options differences in slope coefficients to future held at the end of the fiscal year (based on the research. fiscal year-end stock price, P). The number (N) 17 For example, 0.0663 Age 0.0006 and intrinsic value (Y) of previously granted Age2 reaches a maximum at age 0.0663/ options is calculated by subtracting new grants (2 0.0006) 55.25. from total outstanding options, and adjusting 18 The percentage option holdings multiplied the year-end intrinsic value of the new grants by the option delta is a measure of the change from the total intrinsic value. We treat the in CEO option-related wealth corresponding to a previously granted options as a single prior grant change in shareholder wealth. More formally, with exercise price X, where N(P X) Y,or suppose that the CEO holds N share options, X P (Y/N). and suppose that shareholder wealth increases 21 See, for example, Conyon and Peck (1998 a). by £1. If there are S total shares outstanding, 22 See Murphy (1999) for a discussion and the share price P will increase by P £1/S, comparison of the various approaches. and the value of the CEO’s options will increase 23 Most of the UK literature has been by NP(∂V/∂P), where V is the Black-Scholes concerned with estimating this type of equation value of each option. Substituting for P, the and ignoring the importance of the direct rela- CEO’s share of the value increase is given by tion between CEO pay and shareholder wealth (N/S) (∆V/∂P), or the CEO’s options held as a (as reviewed in Conyon et al., 1995). Moreover, fraction of total shares outstanding multiplied Gregg et al. (1993) show that the ‘implicit’ link by the ‘slope’ of the Black-Scholes valuation. between CEO pay and shareholder returns For examples of this approach which yield a became de-coupled in the 1990s. In fact, this distribution of CEO incentives, see Jensen and may simply be attributable to the restructuring Murphy (1990b),Yermack (1995), and Murphy of CEO pay packages and the importance of (1999). Hall and Murphy (2000) offer a modified non-cash forms of compensation. approach to measure the pay-for-performance 24 Estimating logit regressions for each incentives of risk-averse undiversified executives. country separately yielded negative and signifi- An alternative approach, adopted by Jensen cant correlations between CEO turnover on and Murphy (1990a) for the United States and shareholder returns in both the US and UK data. Main, et al. (1996) for the United Kingdom, 25 For similar derivations of the optimal pay- involves estimating the option pay-performance performance sharing rate, see Lazear and Rosen sensitivity as the coefficient from a regression of (1981), Holmstrom and Milgrom (1991), the change in option value on the change in Gibbons and Murphy (1992), and Milgrom and shareholder wealth. The procedure is similar to Roberts (1992). how we identify the ‘indirect’ relation between 26 Half of us are unwilling to concede to such pay and performance below. assumptions. 19 Although the Greenbury report generally 27 The highest marginal tax rate is approxi- required detailed disclosure of prior grants, it mately 40% in both countries. The top UK rate recognised that ‘in the disclosure of share option affects incomes above £28,000, while the top US 396 DIRECTORS’ REMUNERATION bracket is indexed for inflation and is currently Conyon, M. J. and Peck, S. I. (1998b). ‘Board about £175,000 for married taxpayers. control, remuneration committees and top 28 Research on the effects of the deductibility management compensation.’ Academy limitation, which apply only to the top five Management Journal, vol. 41, pp. 146–57. executives in publicly traded corporations, Conyon, M. J. and Sadler, G. V. (1999). has concluded that the rule has had a modest ‘CEO compensation, option incentives and influence on both the level and structure of US pay plans (Perry and Zenner, 1999; Rose and information disclosure.’ Warwick University Wolfram, 1997). mimeograph. 29 Conyon and Sadler (1999) and Main Conyon, M. J. and Schwalbach, J. (1999). (1999) consider the implications of this rule. ‘Corporate governance, executive pay and 30 Based on data from Forbes’ Annual Survey performance in Europe.’ In (J. Carpenter and of Executive Compensation, May 19, 1997. D. Yermack, eds.) Executive Compensation and Shareholder Value: Theory and REFERENCES Evidence, pp. 13–34. Dordrecht: Kluwer Academic Press. Conyon, M. J., Gregg, P.and Machin, S. (1995). Abowd, J. and Bognanno, M. (1995). ‘Taking care of business: executive compensa- ‘International differences in executive and tion in the UK.’ ECONOMIC JOURNAL, vol. 105 managerial compensation.’ In (R. Freeman (May), pp. 704–15. and L. Katz, eds.) Differences and Changes in Cosh, A. and Hughes, A. (1997).‘Executive remu- Wage Structures, pp. 67–103. Chicago: neration, executive dismissals and institutional University of Chicago Press. shareholdings.’ International Journal of Association of British Insurers (1995) ‘Share Industrial Organization, vol. 15, pp. 469–92. options and profit sharing incentive schemes.’ Coughlan, A. T. and Schmidt, R. M. (1985). ABI, London. ‘Executive compensation, managerial turnover Association of British Insurers (1994) ‘Long and firm performance.’ Journal of Accounting term remuneration for senior executives.’ and Economics, vol. 7, pp. 43–66. ABI, London. Crystal, G. (1991). In Search of Excess: The Black, F.and Scholes, M. (1973).‘The pricing of Overcompensation of American Executives. options and corporate liabilities.’ Journal of New York: Norton & Company. Political Economy, vol. 81, pp. 637–59. Ezzamel, M. and Watson, R. (1998). ‘Market Buckingham, L. and Cowe, R. (1998a). comparison earnings and the bidding-up of ‘Boardroom pay 98: Pleas for restraint fail to executive cash compensation: evidence from curb executive excess.’ The Guardian, 22 July. the United Kingdom.’ Academy Management Buckingham, L. and Cowe, R. (1998b). ‘The Journal, vol. 41, pp. 221–31. boardroom bonanza: exclusive survey reveals Frank, R. H. and Cook, P. J. (1995). Winner- top directors’ pay soaring by 18%.’ The Take-All Society. New York: Free Press. Guardian, 22 July. Gibbons, R. and Murphy, K. J. (1992). ‘Optimal Cadbury, A. (1992).‘Code of best practice.’ Report incentive contracting in the presence of from the Committee on Financial Aspects of career concerns: theory and evidence.’ Corporate Governance. London: Gee Publishing. Journal of Political Economy, vol. 100, Carpenter,J. (1998).‘The exercise and valuation pp. 468–505. of executive stock options.’ Journal of Granick, D. (1972). ‘Managerial Comparisons Financial Economics, vol. 48(2), pp. 127–58. of Four Developed Countries: France, Britain, Conyon, M. J. (1998). ‘Directors’ pay and United States and Russia.’ Cambridge MA: turnover: an application to a sample of large MIT Press. UK firms.’ Oxford Bulletin of Economics and Greenbury, R. (1995). ‘Directors’ Remuneration: Statistics, vol. 60, pp. 485–507. Report of a study group chaired by Sir Conyon, M. J. and Peck, S. I. (1998a). ‘Recent Richard Greenbury.’ London: Gee Publishing. developments in UK corporate governance.’ Gregg, P., Machin, S. and Szymanski, S. (1993). In (T. Buxton, P. Chapman and P. Temple, ‘The disappearing relationship between eds.) Britain’s Economic Performance. directors’ pay and corporate performance.’ London: Routledge. CEO PAY IN THE US AND UK 397 British Journal of Industrial Relations, Main, B. G., O’Reilly, C. and Crystal, G. (1994). vol. 31, pp. 1–10. ‘Over here and over there: a comparison of Guy, F.(1999).‘Earnings distribution, corporate top executive pay in the UK and the USA.’ governance and CEO pay.’ ESRC Centre for Contributions to Labour Studies, vol. 4, Business Research, University of Cambridge. pp. 115–27. Working Paper No. 126. Milgrom, P.and Roberts, J. (1992). Economics, Hall, B. J. and Leibman, J. B. (1998). Organizations, and Management. New ‘Are CEOs really paid like bureaucrats?’ Jersey: Prentice Hall. Quarterly Journal of Economics, vol. 113 (3), Miller, M. and Scholes, M. (1982). ‘Executive pp. 653–91. compensation, taxes and incentives.’ In (W. Hall, B. J. and Murphy, K. J. (2000). ‘Optimal Sharpe and C. Cootner, eds.) Financial exercise prices for executive stock options.’ Economics: Essays in Honor of Paul American Economic Review, Volume 20, Cootner, pp. 179–201. Englewood Cliffs NJ: No 2, May, pp. 209–214. Prentice Hall. Hampel, R. (1998). ‘Committee on Corporate Murphy, K. J. (1999).‘Executive compensation.’ Governance: Final Report.’ London: Gee In (O. Ashenfelter and D. Card, eds.) Publishing. Handbook of Labor Economics, vol. 3. Holmstrom, B. and Milgrom, P. (1991). Amsterdam: North Holland. ‘Multi-task principal-agent analyses: incentive PIRC (1998). Executive Share Schemes: contracts, asset ownership and job design.’ Trends in 1998. London, Pension Investment Journal of Law, Economics and Organization, Research Consultants. vol. 7, pp. 24–52. Perry, T. and Zenner, M. (1999). ‘Pay for per- Jensen, M. and Murphy, K J. (1990a). formance? Government regulation and the ‘Performance pay and top-management structure of compensation contracts.’ incentives.’ Journal of Political Economy, University of North Carolina mimeograph. vol. 98(2), pp. 225–64. PriceWaterhouseCoopers Corporate Register, Jensen, M. and Murphy, K. J. (1990b). ‘CEO various years, published Hemmington Scott. incentives: it’s not how much you pay, but Rose, N. and Wolfram, C. (1997). ‘Regulating how.’ Harvard Business Review, vol. 68 CEO pay: assessing the impact of the (May/June), pp. 138–49. tax-deductibility cap on executive compensa- Kaplan, S. (1994a). ‘Top executive rewards and tion.’ MIT. firm performance: a comparison of Japan and Rosen, S. (1981).‘The economics of superstars.’ the United States.’ Journal of Political American Economic Review, vol. 71 (5), Economy, vol. 102(3), pp. 510–46. pp. 845–58. Kaplan, S. (1994b). ‘Top executives, turnover, Rosen, S. (1992).‘Contracts and the market for and firm performance in Germany.’ Journal of executives.’ In (L. Werin and H. Wijkander, Law, Economics and Organization, vol. 10(1), eds.) Contract Economics, pp. 181–211. pp. 142–59. Oxford: Blackwell. Lazear, E. and Rosen, S. (1981). ‘Rank order Warner, J., Watts, R. and Wruck, K. (1988). tournaments as optimum labour contracts.’ ‘Stock prices and top management changes.’ Journal of Political Economy, vol. 89, Journal of Financial Economics, vol. 20, pp. 841–64. pp. 461–92. Main, B. G. (1999). ‘The rise and fall of execu- Weisbach, M. (1988). ‘Outside directors and tive share options in Britain.’ In (J. Carpenter CEO turnover.’ Journal of Financial and D.Yermack, eds.) Executive Compensation Economics, vol. 20, pp. 431–60. and Shareholder Value: Theory and Evidence, Yermack, D. (1995). ‘Do corporations award pp. 83–113. CEO stock options effectively?’ Journal of Dordrecht, Kluwer Academic Press. Financial Economics, vol. 39, pp. 237–69. Main, B. G., Bruce, A. and Buck, T. (1996). Zhou, X. (1999). ‘CEO pay, firm size and ‘Total board remuneration and company corporate performance: evidence from performance.’ ECONOMIC JOURNAL, vol. 106, Canada.’ Canadian Journal of Economics, pp. 1627–44. (forthcoming). Chapter 15

Wayne R. Guay* THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK: AN ANALYSIS OF THE MAGNITUDE AND DETERMINANTS

Source: Journal of Financial Economics, 53(1) (1999): 43–71.

ABSTRACT To control risk-related incentive problems, equity holders are expected to manage both the convexity and slope of the relation between firm performance and managers’ wealth. I find stock options, but not common stockholdings, significantly increase the sensitivity of CEOs’ wealth to equity risk. Cross-sectionally, this sensitivity is positively related to firms’ investment opportunities. This result is consistent with managers receiving incentives to invest in risky projects when the potential loss from underinvestment in valuable risk-increasing projects is greatest. Firms’ stock-return volatility is positively related to the convexity provided to managers, suggesting convex incentive schemes influence investing and financing decisions.

1. INTRODUCTION The convexity, or curvature, of the wealth-performance relation refers to the JENSEN AND MECKLING (1976) ILLUSTRATE sensitivity of managers’ wealth to the volatil- that, to reduce agency conflicts with ity of equity value. To date, no empirical managers, shareholders are expected to evidence exists on the importance of tie managers’ wealth to firm, or stock convexity in the design of executives’ incen- price, performance. By using compensation tives. This study quantifies the impact of policy to manage the slope of the relation equity risk, or stock-return volatility, on the between managers’ wealth and stock price, value of stock options and common stock shareholders can induce managers to held by corporate CEOs, and provides take actions that increase equity value. evidence on cross-sectional determinants of Managing this slope, however, is not suffi- convexity in executives’ incentive schemes. cient to control agency conflicts arising Smith and Stulz (1985) show that when between stockholders and managers. As is managers’ wealth is dependent upon firm well-recognized in studies by Jensen and performance, risk-aversion can cause Meckling (1976), Haugen and Senbet managers to pass up risk-increasing, (1981), and Smith and Stulz (1985), the positive net-present-value projects. They convexity of the relation between stock illustrate how shareholders can reduce price and managers’ wealth, in addition this risk-related agency problem by using to the slope, must be managed to induce stock options or common stock to struc- managers to make optimal investment and ture managers’ wealth as a convex function financing decisions. of firm performance. Since risk-related THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 399 investment problems are expected to be options, but not common stockholdings, greatest for firms with substantial invest- play an economically significant role in ment opportunities, the magnitude of increasing the convexity of the relation convexity in executives’ wealth-performance between managers’ wealth and stock price. relation is predicted to be positively related I measure convexity as the change in the to the proportion of assets that are growth value of managers’ stock options and options. Smith and Watts (1992) also stockholdings for a given change in stock- hypothesize that growth options are a return volatility. The median change in the determinant of executives’ incentives. value of CEOs’ option portfolios for a 10 However, their argument focuses primarily percentage point change in the standard on the slope of the wealth–performance deviation of stock returns is approximately relation. Theory suggests that, in addition $300,000, with an interquartile range of to influencing the slope, the investment $425,000.1 Convexity provided by common opportunity set is also a determinant of stock, on the other hand, is several orders the convexity in the wealth–performance of magnitude lower than that of stock relation. options, and is of little economic impor- To illustrate differences between the slope tance for most CEOs in the sample. The and convexity of the wealth–performance median change in the value of CEOs’ relation, consider the options and common common stockholdings for a 10 percentage stock held by the following two CEOs. As of point change in return volatility is only December 31, 1993, the CEO of Conrail $22, with an interquartile range of $2400. Inc. held 94,400 shares of stock, worth Since most firms are financially healthy, $6.3 million, and 102,500 stock options, common stock, when viewed as an option worth approximately $3.9 million. At that on the firm’s asset value, is generally so time, the CEO of GTE Corp. held 61,100 deep ‘in the money’ that the payoff to shares of stock worth $2.1 million and shareholders is effectively a linear function 539,900 stock options worth approximately of firm value. $4.3 million. Using a Black and Scholes I also find that the convexity in CEOs’ (1973) option-pricing framework (its incentive schemes is positively related to application in this study is discussed in the proportion of a firm’s assets that are Section 4), and parameter values as of growth options. Thus, firms appear to December 31, 1993, the securities held by provide managers with incentives to invest each of the CEOs would increase in value in risky projects when the potential loss by about $600,000 for a 5% increase in from forgoing valuable risk-increasing proj- their firm’s stock price. That is, the slope ects is greatest. Finally, equity risk is shown of the wealth–performance relation is to be positively related to the convexity approximately the same for both CEOs provided to CEOs, suggesting that managers’ on this date. However, the convexity in investment and financing decisions are their wealth–performance relation differs influenced by their risk-taking incentives. considerably. The GTE CEO’s securities Section 2 briefly highlights the relation would increase in value by about $505,000 between convexity, managers’ preferences for a 5 percentage point increase in the toward firm risk, and risk-related agency annualized standard deviation of GTE’s costs.The data are summarized in Section 3. stock returns. This figure compares to a Section 4 describes the procedure used to $55,000 increase in the value of the estimate the slope and convexity of the Conrail CEO’s securities for the same wealth–performance relation, and provides increase in stock-return volatility.Therefore, descriptive findings. The cross-sectional the GTE CEO appears to have significantly relation between firms’ investment oppor- greater risk-taking incentives than the CEO tunities and the convexity provided to of Conrail Inc. managers is explored in Section 5. Section 6 Using compensation data for 278 corpo- examines whether firms’ stock-return rate CEOs, I present evidence that stock volatility is related to the convexity of 400 DIRECTORS’ REMUNERATION managers’ wealth–performance relation. I can also contribute to the wealth effect when conclude with Section 7. their payoffs are concave or convex functions of firm performance (see Section 5.2 for further discussion). 2. MANAGERS’ PREFERENCES The presence of convex, or option-like, TOWARD FIRM RISK AND RISK- payoffs in the structure of executives’ RELATED AGENCY COSTS incentives is well-recognized by both academics and practitioners. For example, The relation between firm risk and managers’ Jensen and Meckling (1976) point out that incentives is well-developed in the literature. equity holders in a levered firm essentially When a manager holds common stock and hold a European call option to buy the firm stock options, a dependence exists between at an exercise price equal to the face value his wealth and the firm’s stock-price of debt. As such, the value of common performance (see Jensen and Meckling, stock increases with the volatility of the 1976; Jensen and Murphy, 1990). This firm’s cash flows. Haugen and Senbet dependence is commonly referred to as the (1981) analyze the incentive effects of wealth–performance relation. Since stock convex payoffs from employee stock price varies over time, the payoffs to this options. Several empirical studies, such as incentive scheme are uncertain, and risk is Agrawal and Mandelker (1987), DeFusco imposed on the manager. In Pratt (1964), et al. (1990), and Tufano (1996),have also a risk-averse manager is shown to be examined the convexity implications of indifferent between a risky payoff and the managers’ stock-based wealth. However, following payoff with certainty: none of these studies has quantified the magnitude of this convexity for common Certainty equivalent E(wealth)risk stock or stock options, or explored its premium. (1) determinants. The following excerpt from the Committee Report on Executive Differentiating this expression with respect Compensation in BellSouth’s 1993 proxy to firm risk yields the following: statement (p. 11) illustrates that corporate compensation committees also recognize ∂CE/∂ ∂E(wealth)/∂ the need to structure compensation plans ∂(risk premium)/∂. (2) to give executives appropriate risktaking incentives: Smith and Stulz (1985) and Lambert et al. (1991) illustrate how this expression BellSouth’s long term [incentive] program partitions the effect of firm risk on man- is intended to focus the executive group agers’ preferences into two components. on the achievement of corporate goals .... The first component, ∂E(wealth)/∂, Recognizing the dynamic convergence of which I will refer to as the wealth effect, industries such as telecommunications, represents the change in expected wealth entertainment, and cable which will experienced by a manager when firm risk continue throughout the next decade, the changes. When the payoffs of an incentive Committee wants to motivate BellSouth scheme are a linear function of firm executives to take the risks necessary to performance, the wealth effect is zero, secure a strong foothold for BellSouth since changes in the distribution of firm in this extremely competitive new performance do not affect managers’ marketplace. expected wealth. Expected wealth is an increasing function of risk when managers The second component in Eq. (2), ∂(risk own securities with convex payoffs, such premium)/∂, which I will refer to as the as stock options or common stock, that risk-aversion effect, captures the influence of increase in value as firm risk increases. risk-aversion on managers’ utility. If man- Bonus plans and other incentive schemes agers are risk-averse and poorly diversified THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 401 with respect to firm-specific wealth, an a better understanding of the magnitude increase in firm risk decreases managers’ and cross-sectional variation in the wealth expected utility. The magnitude of the effect.2 However, I do not ignore the risk- risk-aversion effect is expected to depend aversion effect in the empirical tests. Since upon the degree of diversification in a securities that induce convexity in the manager’s portfolio of wealth, the level of wealth-performance relation, such as stock a manager’s wealth, and manager-specific options and common stock, are also likely risk-aversion parameters. to influence managers’ aversion toward firm A manager’s overall preference toward risk, the wealth effect and risk-aversion firm risk will depend upon the relative effect are expected to be correlated. magnitudes of the wealth effect and the Section 5 addresses the implications of this risk-aversion effect. If the risk-aversion correlation and how the research design is effect dominates, the manager will prefer structured to control for these effects. to decrease firm risk. This condition can give rise to risk-related agency problems. When shareholders hold well-diversified 3. DATA AND DESCRIPTIVE portfolios, they would like managers to STATISTICS invest in all positive net-present-value projects, irrespective of the risk associated I compile compensation and stock-based with those projects. However, due to a lack wealth data for 278 corporate CEOs as of of diversification, risk-averse managers may December 31, 1993. To obtain the sample, choose to forgo some positive net-present- I first rank the largest 1000 firms included value projects that would increase firm in the Compustat database by market value risk. Smith and Stulz (1985) and Milgrom as of December 31, 1988. By allowing a and Roberts (1992) argue that by making five-year period between the ranking date adjustments to the slope and convexity and the compensation measurement period, of the wealth-performance relation, share- I reduce the potential bias of including only holders can reduce the likelihood that successful firms in the sample, and increase managers pass up valuable risky projects. the probability of finding cross-sectional Holding the slope constant, greater convexity variation in the firms’ financial characteris- in the wealth-performance relation is tics. However, because firms are required expected to shrink the gap between the to have survived from 1988 until 1993, a risk-aversion effect and the wealth effect. survival bias could be present, as some Though, in principle, the wealth effect could firms are likely to have been eliminated due dominate the risk-aversion effect, Lambert to bankruptcy or being acquired. From et al. (1991) find that, for most plausible this initial sample, 500 firms are chosen incentive schemes, managers are likely to for inclusion in the final sample using a remain averse to firm risk. uniform selection method where every other Unlike the wealth effect, which can be firm is selected. estimated with option pricing techniques I delete firms without a December-end and readily available data about managers’ fiscal year from the sample. Firms are also stock options and common stockholdings, removed if data are not available from the a measure of the risk-aversion effect Center for Research in Security Prices requires manager-specific data that is (CRSP) or Compustat. These selection much more difficult, if not impossible, to criteria result in a sample of 315 firms. For obtain. For example, the risk-aversion 20 of these firms, either the time series of effect is expected to be a function of a proxy statements is incomplete, or sufficient manager’s total wealth, the degree of proxy data are not available to construct diversification in a manager’s portfolio the CEO’s option portfolio. To check for of wealth, and a manager’s utility function. errors in the procedure used to construct Given these measurement problems, this option portfolios, I compute the cash value study focuses primarily on developing each CEO would obtain if all options in the 402 DIRECTORS’ REMUNERATION constructed portfolios were exercised on combined with the time series of options December 31, 1993, and compare this granted, to construct an estimate of the amount to the cash value of the CEOs’ composition of each CEO’s option portfolio options if exercised immediately. This cash as of December 31, 1993. Detailed infor- value disclosure is required for all named mation on option exercises is not readily executive officers. CEOs are excluded from available from proxy statements prior to the sample if these two cash value measures 1992. I assume that, unless otherwise differ by more than 25% and $200,000, noted, options with the shortest remaining resulting in the removal of 14 CEOs. time-to-maturity are exercised first. Since Finally, CEOs are excluded if they own this exercise strategy is expected to simu- more than one-third of their corporation’s late the true exercise strategies with error, common stock, since it is doubtful that it is important to remove CEOs with these managers’ compensation schemes are constructed option portfolios that differ designed for contracting purposes. This significantly from the option portfolio criteria reduces the sample by three CEOs. characteristics disclosed in the 1993 proxy The results are not sensitive to using a statements, as described in Section 3. lower ownership threshold, such as five or Table 15.1 indicates that common stock ten percent stockholdings.The final sample and options are significant components of contains 278 CEOs. CEOs’ incentive schemes. Using the Black and Scholes (1973) model, the median stock option portfolio has a value of $1.78 3.1. Summary of compensation million. Furthermore, the substantial and stock-based wealth data standard deviation and interquartile range For each CEO, I compile data on common (not reported) suggest there is considerable stock, restricted stock, and outstanding variation in the extent to which firms use stock options held as of December 31, options to provide managers with incentives. 1993, as well as salary and bonus received The median value of CEOs’ common stock- during 1993. With the exception of the holdings is $2.74 million, which is about composition of the CEOs’ option portfolios, 50% larger than the median value of this data is readily available from the firms’ option holdings. Though restricted stock is 1993 proxy statements.To reduce the influ- held by less than half of the CEOs, nearly ence of a few CEOs that hold extremely 25% of the CEOs hold restricted stock large quantities of common stock, the upper- valued at $1 million or more. The median most percentile of aggregate stockholdings value of total stock-based wealth, including and sensitivities of these stockholdings to stock options, common stockholdings, and stock price and equity risk are set equal to restricted stock, is $6.79 million, and the stockholdings and sensitivities of the several times larger than the median CEOs at the 99th percentiles. annual cash compensation of $866,700. I construct the CEO’s option portfolios using the past time-series of proxy statements. For each CEO, I collect data on the 4. ESTIMATING CONVEXITY number of options granted in each fiscal IN THE WEALTH–PERFORMANCE period, the average exercise price per RELATION option granted, and the actual or maximum allowable time-to-maturity for the options As discussed above, the focus of this study granted. In approximately 5% of the fiscal is not on the slope of CEOs’ wealth– periods, it was necessary to average option performance relation, but instead on the grants over a multi-year period. Beginning convexity of this relation. I measure the in 1992, proxy statements report the total convexity contributed by a stock option or number of stock options held by each share of common stock as the change in named executive officer. I use the number the security’s value for a 0.01 change in the of options held at the end of 1993, annualized standard deviation of stock THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 403

Table 15.1 Summary statistics for CEOs’ stock option portfolios, common stockholdings, and cash compensation

The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat, ranked by market value of equity on December 31, 1988. The CEOs’ salary and bonus are for the fiscal year ending December 31, 1993. The number of options held by the CEOs and their stock- based wealth are as of December 31, 1993. Option values are based on the Black-Scholes formula for valuing European call options, as modified to account for dividend payouts by Merton (1973). Option value is calculated as [SedTN(Z) XerTN(Z T(1/2))], where Z is equal to [In(S/X) T(r d /2)]/T(1/2). The parameters in the Black-Scholes model are set as follows: S price of the underlying stock at December 31st, 1993, E exercise price of the option, annualized volatility, estimated as standard deviation of daily logarithmic stock returns over the last 120 trading days in 1993, multiplied by 252(1/2), r ln(1 risk-free interest rate), where the risk-free interest rate is the yield, as of December 31, 1993, on a U.S.Treasury strip with the same time to maturity as the remaining life of the stock option, T remaining time to maturity of the option in years, as of December 31, 1993, and d ln(1 expected dividend rate), where the expected dividend rate is set equal to the dividends paid during 1993 divided by the year-end stock price.Total stock-based wealth is the sum of the value of the CEO’s options, stockholdings, and restricted stock.

Compensation component Standard (in $ millions) Mean deviation Minimum Median Maximum

Salary bonus 1.10 0.93 0.00 0.87 8.87 Number of options held (000s) 257.89 329.69 0.00 170.55 2500.00 Value of option portfolio 4.23 6.83 0.00 1.78 42.53 Value of common stockholdings 24.23 88.75 0.00 2.74 600.90 Value of restricted stock 0.97 2.46 0.00 0.00 25.17 Total stock-based wealth 29.43 89.80 0.04 6.79 600.90

returns. The contribution of stock options accommodate these differences is not and common stock to the convexity in the straightforward. First, there is no clear wealth-performance relation can vary method of determining the value of an widely across CEOs, and is a function of employee stock option from the employee’s firms’ financial characteristics, the quanti- perspective, as opposed to the firm’s per- ties of each security held, and the specific spective. Second, since there is considerable parameters that underlie the stock-based variation in the remaining time-to-maturity components, such as the exercise prices and for the options held by the CEOs, it is times to maturity of the options in a man- difficult to estimate the expected time until ager’s portfolio. these options will be exercised. I estimate the incentive effects of I estimate the sensitivity of an option employee stock options using the Black- portfolio’s value to equity risk as follows. Scholes formula for valuing European call First, for each option in a CEO’s portfolio, options, as modified to account for dividend I compute the Black-Scholes partial payouts by Merton (1973). The details of derivative of option value with respect to a this procedure are presented in Section 0.01 change in the annualized standard A.1. of the appendix. Huddart (1994) and deviation of stock returns. Next, the partial Cuny and Jorion (1995) point out that the derivatives are weighted by the number of optimal exercise policy is likely to deviate options in the portfolio. The procedure and from the assumptions underlying the Black- parameters used to compute the Black- Scholes framework, due to managers’ Scholes partial derivatives are described in risk-aversion and the non-transferability Section A.1. of the appendix. of employee stock options.3 However, To check the sensitivity of the reported adjusting the Black-Scholes model to results to alternative option valuation 404 DIRECTORS’ REMUNERATION techniques, all tests are reproduced using the sum of book debt plus market value of methods suggested in Hemmer et al. (1994) equity is 0.99. and in Statement of Financial Accounting Together with the other parameters, Standards No. 123, ‘Accounting for stock- I use the implied firm value to estimate the based compensation’ (FASB, 1995). To Black–Scholes partial derivative of each implement the Hemmer et al. and FASB firm’s stock price with respect to volatility. techniques, I assume that the expected time Note that because common stock is an to exercise for all options held is equal to option on firm value, its value is sensitive to 60% of the remaining time-to-maturity, the volatility of firm value. In contrast, a though the cross-sectional results are not stock option’s value is sensitive to stock- sensitive to varying this percentage. With return volatility.To make the two convexity respect to the Hemmer et al. method, I also measures comparable, I compute the assume that all options have a three-year change in common stock value with respect vesting period at the grant date.The option to a 0.01 change in the annualized standard values and convexity measures using these deviation of stock-returns. Since firm value procedures are generally about 10–25% is the sum of debt value plus equity value, lower than those using the Black-Scholes portfolio theory suggests that the volatility formula. However,all cross-sectional results of firm value can be estimated using meas- are qualitatively unchanged. ures of debt volatility and equity volatility, As illustrated by Black and Scholes and an assumption about the correlation (1973) and Smith (1976), the payoff to a between debt value and equity value. For share of common stock in a levered firm simplicity, I set this correlation equal to can also be viewed as a call option, where one, though the results are not sensitive to the option’s underlying asset is the value of this assumption. The sensitivity of common the firm and the exercise price is the face stock value to equity volatility is then value of the firm’s liabilities. Unlike stock estimated, at the margin, by holding debt options, the Black-Scholes parameters for volatility constant while stock-return common stock are not all readily available. volatility is allowed to vary by 0.01 (see However, an estimate of the value of the Section A.2. for further details). Finally, option component is readily observable in I estimate the aggregate sensitivity of a the stock price. Knowledge of this option CEO’s common stock to equity risk as the value reduces, by one, the number of Black– partial derivative of stock price with Scholes parameters that must be estimated. respect to stock-return volatility, multiplied I estimate parameters for the exercise price, by the number of shares of stock held by time to maturity, interest rate, dividend yield, the CEO. and volatility of firm value as described in To gain some insight for the source of Section A.2. of the appendix. Firm value, convexity inherent in common stock, consider the sum of market value of equity and how the value of common stock varies as a market value of debt, and the underlying function of firm value. In an unlevered asset in this option pricing application, is left firm, the payoff to common stock is a linear as the free parameter. Using the Black– function of firm value, and the value Scholes model, I compute the implied of common stock increases by $1 for every per-share value of each sample firm. As $1 increase in firm value. For a levered reported in Table 15.6 of the appendix, the firm, an increase in firm value will not implied per-share firm values are slightly accrue entirely to common stockholders, smaller than the sum of per-share book but will instead be shared between the value of debt plus stock price for nearly all debtholders and shareholders, assuming the firms. This relation is not surprising given probability of financial distress is greater that the value of an option is strictly than zero. Confirming this notion, a $1 greater than the price of the underlying change in firm value translates into a asset less the exercise price.The correlation median change in common stock value between the implied firm value and the of only $0.97, with $0.03 accruing to THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 405 the firms’ debtholders (see Table 15.6 in considerable error. The median aggregate the appendix). sensitivity of CEOs’ wealth to equity risk due to options is $29,893. These sensitivi- 4.1. Descriptive statistics on ties range from a minimum of $0 for the convexity in the wealth–performance 50 CEOs that do not hold options, up to relation a maximum of $347,256 for the CEO holding the option portfolio most sensitive Table 15.2 presents the characteristics to equity risk. of the stock option portfolios held by the The sensitivity of common stock to CEOs in the sample. Panel A indicates that equity risk is much smaller than that of the characteristics of the CEOs’ options options. The median change in per-share vary considerably. For each CEO, I compute stock price for a 0.01 change in stock- the mean price-to-strike ratio and mean return volatility is $0.00005, compared time-to-maturity of the options in their to $0.156 per option. This result is not portfolio. The price-to-strike ratio is the surprising, given that the median price- stock price divided by the option’s exercise to-strike ratio for a share of common price. The mean price-to-strike ratio stock is 1.9, which is much larger than the indicates, on average, the extent to which price-to-strike ratio of most CEOs’ stock the CEO’s options are in the money. The options (see Table 15.6 of the appendix). value of an option with a price-to-strike Even when multiplied by the quantity of ratio much greater than one increases stock held, the total sensitivity of common almost linearly with stock price, and as stock value to equity risk is very small for such, will be quite insensitive to changes in most CEOs, with a median sensitivity of $2. stock-return volatility. By comparison, an As a result, for most firms, the aggregate “at the money” option has a price-to-strike sensitivity of CEOs’ stock-based wealth to ratio of one, and will be more sensitive to equity risk due to options plus stock is equity risk. The mean price-to-strike ratio driven primarily by stock options. However, has a median value of 1.3, and ranges from the sensitivity of stock value to equity risk about 0.4 to 9.0.The mean time-to-maturity is significant for a small number of CEOs. is the number of years remaining to expira- For approximately 3% of the CEOs, the tion, on average, for a CEO’s options. The change in stock value for a 0.01 change in mean time-to-maturity has a median value stock-return volatility exceeds $25,000. of 7.2 years, and ranges from 1.5 to 16.5 Thus, while the sensitivity of stock value to years. This variation suggests that per- equity risk can potentially provide managers option incentive effects are expected to with incentives to shift wealth from bond- vary widely across CEOs. holders, this effect is expected to be Estimates of the sensitivity of CEOs’ extremely isolated, and relevant only for options and stock to equity risk are pre- firms experiencing severe financial distress. sented in Panel B of Table 15.2. In the first To interpret the economic significance of row, the sensitivity of option value to equity these sensitivities, one must consider both risk is averaged over the options in each the sensitivity of a CEO’s wealth to equity CEO’s portfolio. For the 228 CEOs with risk and his ability to alter the risk of the options, the median change in value, per firm. I estimate the latter component for option, for a 0.01 change in the standard each sample firm by first computing the deviation of stock returns is $0.156, and difference between 1993 stock-return ranges from $0.001 to $0.748.The consid- volatility and the 1993 volatility of a con- erable variation in these per-share statistics trol sample obtained from the CRSP tapes, indicates that the incentive effects of stock and matched on two-digit SIC code and options are heavily dependent on security- market value of equity. Each sample firm is specific parameters. Measures that ignore then assumed to experience a change in this variation are likely to estimate the equity volatility that brings it in line with its incentive effects of these instruments with industry and size-matched control portfolio. 406 DIRECTORS’ REMUNERATION option. iation of the iation sitivity of an sitivity dend-adjusted o-maturity for o-maturity ompustat, ranked sensitivities weighted sensitivities option portfolio. option Time-to-maturity is the Time-to-maturity h to equity risk averaged h to equity risk ercise price of an the ex end-adjusted Black–Scholes value of end-adjusted Black–Scholes value ivided by ivid Standard 1.50 0.867.18 1.68 0.38 1.50 1.07 6.25 1.30 7.23 1.67 8.17 9.00 16.53 0.167 0.105 0.001 0.097 0.156 0.212 0.748 45,967 51,641 82 14,108 29,893 56,590 347,256 a a options ($)options 40,557 50,777 0 6489 22,664 55,770 347,292 a a The sensitivity of CEOs’ sensitivity The and stock price wealth to equity risk a options ($) options price ($) price stock option’s value for a 1% change in stock price is the sensitivity of wealth to stock price averaged over all options in a CEO’s in all options over of wealth to stock averaged price stock is the sensitivity in price a 1% change for value option’s stock options or common stockholdings for a 1% change in the value of the firm’s stock (see the appendix). price of the firm’s the value sen in a 1% change average for The or common stockholdings stock options by the number of securities held. of securities by the number the d in stock is the change in price of wealth to a 1% change Sensitivity over all options in a CEO’s option portfolio. Total sensitivity of wealth to equity risk aggregates the per option and per-share option the per aggregates of wealth to equity risk sensitivity portfolio. a CEO’s option in Total all options over firm’s stock returns (see the appendix).The average sensitivity of an option’s value to equity risk is the sensitivity of wealt is the sensitivity to equity risk value of an option’s stock (see the appendix).The returns sensitivity firm’s average Black–Scholes value of a CEO’s stock option portfolio and common stockholdings for a 0.01 change in the annualized standard dev the annualized in a 0.01 change for and common stockholdings portfolio of a CEO’s stock option Black–Scholes value all options in a CEO’s option portfolio. a CEO’s option in all options the divi and stock in is the change due to options of wealth to equity risk Sensitivity remaining time to expiration for an option as of December 31, as of December an option for to expiration time of the time-t 1993. average remaining is the simple mean time-to-maturity The The mean price-to-strike ratio is the simple average of the price-to-strike ratios for all options in a CEO’s option portfolio. a CEO’s option in all options for ratios of the price-to-strike average is the simple ratio mean price-to-strike The Panel C: stock in to a 1% change value of an option’s sensitivity Average 0.272 0.173 0.006 0.147 0.231 0.366 1.160 Total sensitivity of wealth to equity risk: sensitivity Total stock Total sensitivity of wealth to equity risk: sensitivity Total stock ($) 2857 12,654 0 0 2 240 98,232 Mean time-to-maturity per option per Mean time-to-maturity Panel B: ($) to equity risk value of an option’s sensitivity Average Total sensitivity of wealth to equity risk: sensitivity Total ($) options by market value of equity on December 31, of equity on December value by market 1988. 1993 stock d is the December price ratio Price-to-strike Panel A: ratio Mean price-to-strike Per-share sensitivity of stock price to equity risk: of stock price sensitivity Per-share ($) 0.005 0.016 0.000 0.000 0.000 0.002 0.225 Total sensitivity of wealth to equity risk: sensitivity Total options – all CEOs ($) 37,700 49,982 0 4614 20,915 48,164 347,256 The data consist of option portfolios and common stockholdings for 278 CEOs selected uniformly from the 1000 largest firms on C the 1000 largest from firms 278 CEOs selected uniformly for and common stockholdings portfolios data consist of option The Table 15.2 Option portfolio characteristic ($) stock price in to a 1% change of stock price sensitivity Per-share Mean deviation 0.393 Minimum 0.247 Q1 0.029 Median 0.248a stock Excludes options. 50 CEOs that do not hold 0.336 Q3 Maximum 0.490 2.548 Total sensitivity of wealth to a 1% change in stock price: in of wealth to a 1% change sensitivity Total ($) options 72,169 104,844 0 3997 36,407 96,250 687,350 Total sensitivity of wealth to a 1% change in stock price: in of wealth to a 1% change sensitivity Total 324,164 904,534 353 38,784 90,685 222,387 6,300,594 Total sensitivity of wealth to a 1% change in stock price: in of wealth to a 1% change sensitivity Total stock ($) 251,995 886,860 0 10,571 37,307 108,341 6,008,957 THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 407 The mean change in CEOs’ stock-based up risky, positive NPV projects.This problem wealth from this change in volatility would is likely to be most severe in firms with be $324,000, with a standard deviation of substantial investment opportunities (see $562,000. This mean change in wealth is Milgrom and Roberts, 1992, Chapter 13, about 20% larger, at $391,000, when for a discussion of potential risk-related considering only CEOs with option port- agency costs in firms with valuable growth folios.The magnitude of these wealth effects opportunities).4 As such, the expected suggests that the convexity generated by loss from valuable projects bypassed by stock options is potentially large enough to managers is hypothesized to be positively influence managers’ behavior. related to the proportion of assets that The maintained hypothesis in this paper represent growth options. By providing is that firms add convexity to a manager’s managers with incentive schemes that have incentive scheme to encourage investment convex payoffs, equity holders can reduce in valuable risk-increasing projects, that is, these risk-related agency costs.The following to help overcome the risk-aversion effect. hypothesis follows directly: Convexity in the However, note that even in the absence of relation between managers’ wealth and stock convexity in the wealth–performance rela- price is positively related to the proportion tion, managers have incentives to undertake of assets that are growth options. positive NPV projects when the slope of the I use three proxies to capture variation in wealth–performance relation is greater than firms’ investment opportunities. These are: zero. Therefore, it is difficult to precisely (i) the book-to-market ratio, (ii) expenditures determine how much convexity,at the margin, on research and development, scaled by is necessary to induce managers to invest in market value of assets, and (iii) a measure of risk-increasing, positive NPV projects. investment expenditures defined as the sum To illustrate the differences between of capital expenditures plus acquisitions over the slope and convexity of the wealth– the most recent three years, divided by performance relation, Panel C of Table 15.2 market value of assets. Though these vari- reports descriptive statistics on the sensi- ables are expected to contain information tivity of CEOs’ wealth to stock price. On about investment opportunities, each has average, a stock option changes in value by unique limitations as a measure of this $0.27 for a 1% change in stock price, unobservable underlying construct.Therefore, compared to $0.39 for a share of common as in Baber et al. (1996) and Gaver and stock. Thus, in contrast to the convexity Gaver (1993), I employ common factor induced by each of these securities, a share analysis to construct a single variable that of stock increases the slope of the wealth– captures variation common to these observ- performance relation by about 40% more able proxies. Separate results are presented than a stock option does, on average. On an using the common factor as an alternative aggregate basis, both stock options and to the above three proxies. common stock make a substantial contribu- Table 15.3 summarizes the proxies for tion to the wealth–performance slope. The investment opportunities and the financial median change in option portfolio value for characteristics of the 278 sample firms. In a 1% change in stock price is $36,400, general, the sample firms are large, with compared to a median change of $37,300 median market value of assets equal to for common stockholdings. $5.4 billion. The range in values for book- to-market, R&D, and investment expenditures is quite large, suggesting that there is 5. CONVEXITY AND THE considerable variation in investment oppor- INVESTMENT OPPORTUNITY SET tunities across the sample firms.The factor score for investment opportunities has a As illustrated in Section 2, risk-related mean of zero, by construction, and varies agency problems can cause managers to pass from 0.85 to 2.27. 408 DIRECTORS’ REMUNERATION

Table 15.3 Summary firm characteristics

The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat, ranked by market value of equity on December 31, 1988. All financial variables are computed for fiscal year ending December 31, 1993. Book-to-market ratio is the book value of assets divided by the sum of book value of liabilities and market value of equity. R&D expenditures is R&D expense divided by the market value of assets (in %). Investment expenditures is the sum of capital expenditures plus acquisitions over the period 1991 through 1993, divided by market value of assets. Factor score is obtained using common factor analysis on the variables Book-to-market, R&D expenditures, and Investment expenditures. Market value of assets is book value of debt plus market value of equity. Leverage is calculated as [(book value of assets – book value of equity)/ market value of equity].

Firm characteristics Mean Standard Minimum Median Maximum deviation

Book-to-market ratio 0.74 0.21 0.24 0.78 1.20 R&D expenditures 0.80 1.78 0.00 0.00 13.80 Investment expenditures 0.13 0.10 0.01 0.12 0.71 Factor score (investment 0.00 0.54 0.85 0.05 2.27 opportunities) Market value of 16.59 34.15 0.20 5.42 315.21 assets($billion) Leverage 2.74 4.13 0.06 1.10 34.28

5.1. Empirical results: convexity sensitivity of CEOs’ wealth to stock price is and the investment included in all regressions.The market value opportunity set of assets is also included in the regressions to control for a relation between firm size To explore the relation between convexity and both the probability of having a formal and the investment opportunity set, I regress incentive compensation plan, such as a the sensitivity of CEOs’ wealth to equity stock option plan, and the level of compen- risk on the proxies for growth options sation (see Smith and Watts, 1992; Gaver described in the previous section. I use two and Gaver,1993). All t-statistics are calcu- alternative measures of the sensitivity of lated using heteroskedasticity-consistent CEOs’ wealth to equity risk: (i) the sensi- standard errors. tivity of CEOs’ option portfolios to equity The regression results are reported in risk, and (ii) the combined sensitivity of Table 15.4. The negative coefficient on stock options and common stockholdings book-to-market and positive coefficients to equity risk. on R&D and investment expenditures are Though stock options and common consistent with the hypothesis that the stockholdings add convexity to the relation investment opportunity set is positively between managers’ wealth and stock price, related to convexity in the relation between both securities also increase the slope of CEOs’ wealth and stock price. In Columns (2) this relation. Smith and Watts (1992) argue and (5), book-to-market ratio, R&D that because the management of investment expenditures, and investment expenditures opportunities is difficult to monitor, firms are replaced with the factor variable with greater investment opportunities are that captures variation common to these expected to tie managers’ wealth more observable proxies for investment opportu- closely to firm performance.5 To control for nities. In both columns, the coefficient on a relation between the wealth– performance this factor is positive and significant at the slope and investment opportunities, the 1% level. THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 409 2.28) 1.70) 0.87) ( (3.44) (3.08) ( (6.84) a a a b a 0.41 ( 0.01 market value of value market 80.80 4.64) (4.92) 17.12 (6.01) 9.61 ( a a a f liabilities end-adjusted Black-Scholes value ssets/market vlaue of assets).ssets/market R&D onus for 1993 (in $ millions). 1993 (in onus for Age ands). of wealth to equity Sensitivity 103.04 ivid 3.40) 3.76) Log(sensitivity of wealth to stock price) Log(sensitivity Log(market value of assets) value Log(market 4 ( ( (2.51) (6.01) 16.26 1 c a a a a c 55.79 77.29 Sensitivity of wealth to equity risk: options 2.08) 0.00 (0.74) 0.00 (0.85) 2.03) 1.57) (3.96)(3.54)( 17.10 (6.11) 26.44 12.79 ( a a a a a (Cash compensation) 3 0.70 ( 0.01 c 71.05 (Investment expenditures) (Investment 3 b 5.18) ( (5.32)(6.44) 19.46 11.00 (R&D) a a a 2 end-adjusted Black-Scholes value of a CEO’s stock option portfolio and common stockhold- portfolio of a CEO’s stockend-adjusted option Black-Scholes value b ivid 112.37 tock $ thousands). (in returns of a is (book value ratio Book-to-market end-adjusted Black-Scholes value of a CEO’s stock option portfolio and common stockholdings for a 0.01 for and common stockholdings portfolio of a CEO’s stockend-adjusted option Black-Scholes value 4.61) error. 3.67) 0.02) 0.00 (0.18) ( ivid a (6.59) 17.13 alue of assets (in %). Investment expenditures is the sum of capital expenditures plus acquisitions over the period %).alue of assets over (in plus acquisitions is the sum of capital expenditures Investment expenditures (Book-to-market) (2.44) 3.29 a 1 Age a (Sensitivity of wealth to stock price) (Sensitivity b 2 5 c c 82.62 ( 63.72 0.00 ( 22.47 20.76 31.77 20.12 19.25 32.59 Sensitivity of wealth to equity risk: options (000’s) & stock (000’s) a ) 27.76 ) 48.29 (1.63) 46.81 (1.54) ) 3.10 ) 18.19 ) 10.96 ) ) ) ) 7.06 (1.48) 6.71 (1.44) the market value of assets at December 31, of assets at December value the market 1993. Book-to- on the variables common factor analysis using score is obtained Factor ivided by market v Predicted sign (1) (2) (3) (4) (5) (6) -statistics parentheses are computed in using standard errors. heteroskedasticity-consistent White’s T ivided by R&D expense d R&D expense The relation between investment opportunities and the sensitivity of CEOs’ and the sensitivity opportunities between investment relation The wealth to equity risk 2 R Sensitivity of wealth to equity risk Sensitivity opportunities) stock price to stock price equity]. the d in is the change of wealth to stock price Sensitivity ings for a 1% change in the value of the firm’s stock price (in $ thousands). (in stock price of the firm’s the value in a 1% change for is the sum of salaryings plus cash b Cash compensation is the CEO’s age in years. is the CEO’s age in market ratio,market R&D expenditures, expenditures. o and Investment of [book value logarithm of assets] is the natural value Log[market 1991 through 1993,1991 through d Intercept (?) Factor score (investmentFactor ( expenditures is expenditures of a CEO’s stock option portfolio for a 0.01 change in the in a 0.01 change for portfolio $ thous of a CEO’s stock (in option stock returns of the firm’s standard deviation annualized the d and stock in is the change options from risk Table 15.4 Adjusted a at the 5% level. Statistical significance b at the 10% level. Statistical significance All variables are measured as of December 31, are measured as of December All variables 1993. the d in is the change options from of wealth to equity risk Sensitivity change in the annualized standard deviation of the firm’s standard deviation the annualized in change s Investment expendituresInvestment ( R&D expenditures ( Log (sensitivity of wealth Log (sensitivity ( log(market value of assets) value log(market ( Age ( Book-to-market ratioBook-to-market ( of wealth to Sensitivity ( Cash compensation ( 410 DIRECTORS’ REMUNERATION To interpret the magnitude of the factor the greater the expected risk-aversion effect. score coefficient, consider two CEOs with Though logarithm is the reported functional factor scores that differ by the standard form of this variable, the regressions are deviation of the sample firms’ factor not sensitive to using other concave func- scores.The coefficient on the factor score in tional forms, such as the square root. Since Column (5) implies that the compensation the same stock-based securities drive both schemes for these two CEOs are expected the sensitivity of CEOs’ wealth to equity to be set so that a ten percentage point risk and to equity value, a positive coeffi- rise in stock-return volatility increases the cient is predicted. stock-based wealth of the CEO managing When these variables are included, as the high growth options firm by about shown in Columns (3) and (6), the coeffi- $143,000 more than the CEO managing cient on log[sensitivity of wealth to stock the low growth option firm. price] is significantly positive, indicating that As noted in Section 2, the sensitivity of securities that increase the risk-aversion CEOs’ wealth to equity risk, or the wealth effect also increase the wealth effect. effect, is expected to measure the impact of However, the coefficient on the investment risk on managers’ utility with error.Missing opportunities factor remains significantly from this estimate is the risk-aversion positive in both columns, providing further effect, or the reduction in utility that support for the hypothesis that firms risk-averse, poorly diversified managers provide managers with incentives to invest experience when the volatility of their in risky projects when the potential wealth increases. To examine whether this loss from underinvestment in valuable omitted variable is influencing the results, I risk-increasing projects is greatest. include proxies for the risk-aversion effect Yermack (1995) concludes that invest- in the regressions. ment opportunities are not an important The dollar value of 1993 cash compen- determinant for grants of CEO stock sation, salary plus bonus, is included to options. However, the sensitivity of wealth control for the level of CEOs’ outside to stock price is used as the dependent wealth. The greater the cash compensation variable in his tests.The results in Table 15.4 that can be invested outside the firm, the suggest that firms are expected to consider better diversified the CEO is likely to be, the sensitivity of wealth to equity risk, in and the lower the expected risk-aversion addition to sensitivity of wealth to stock effect. CEO age is included to control for price, when granting stock options. manager-specific variation in diversification of wealth and degree of risk-aversion. 5.2. Specification checks The results are robust to the following alternative treatments of the age variable: Corporate hedging theory identifies circum- (i) including dummy variables for various age stances where firm value can be increased categories, (ii) interacting age with the other through reductions in risk (e.g., Myers, independent variables, and (iii) re-estimating 1977; Smith and Stulz, 1985; Smith and the regressions on subsamples of the CEOs Mayers, 1987, 1990; Froot et al., 1993). formed by age.The signs of the coefficients All else being equal, firms with strong on cash compensation and age are difficult incentives to hedge are expected to provide to predict since it is unclear how the wealth managers with weaker incentives to increase and risk-aversion effects are likely to be firm risk. To examine the sensitivity of correlated for these variables.The log of the the reported results to these factors, sensitivity of CEOs’ wealth to stock price is I include measures of financial leverage, also added as a proxy for the CEOs’ degree operating performance, cash flow volatility, of diversification. The more sensitive the and tax-loss-carryforwards in the Table 15.4 CEO’s wealth is to firm performance, the less regressions as proxies for firms’ incentives well-diversified the CEO is likely to be, and to hedge. The coefficients and significance THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 411 levels on the growth option variables (not 6. THE RELATION BETWEEN reported) are qualitatively similar to those STOCK-RETURN VOLATILITY AND reported previously.The coefficient on lever- THE SENSITIVITY OF CEOS’ age is in the predicted direction, negative, WEALTH TO EQUITY RISK and is significantly different from zero. The coefficients on the other hedging variables 6 The analysis in Section 2 suggests that are not significantly different from zero. managers are more willing to invest in Yermack (1995) and Dechow et al. risk-increasing projects as the convexity of (1996) hypothesize that liquidity constraints payoffs in the relation between their wealth can induce firms to use stock options in lieu and stock price increases. This relation of cash compensation. If firms with sub- between convexity and investment choice stantial growth options tend to be cash also underlies the hypothesis tested in constrained, the inferences drawn in this Section 5, that the sensitivity of CEOs’ section could be spurious. The results wealth to equity risk is directly related to reported in Table 15.4 are robust to includ- firms’ investment opportunities. Therefore, ing free cash flow, measured as cash flow it is of interest to examine whether firms’ from operations less capital expenditures, stock-return volatility, as a measure of the and financial leverage, as proxies for cash riskiness of a firm’s projects, is positively constraints. related to the sensitivity of managers’ The preceding analysis focuses exclu- wealth to equity risk. sively on the convexity of payoffs in CEOs’ To explore the relation between equity risk stock-based wealth. However, it is possible and the convexity of payoffs to managers, that firms make adjustments to other I regress contemporaneous stock-return forms of compensation, such as salary and volatility on the sensitivity of CEOs’ wealth bonuses, to offset changes in stock-based to equity risk.7 The dependent variable, wealth that CEOs’ experience when equity stock-return volatility, is computed over risk changes. To explore this possibility, 240 trading days, from 120 days before I examine the relation between changes in through 120 days after the compensation the CEOs’ salary and bonuses from 1993 to measurement date of December 31, 1994, and both changes in firms’ stock- 1993. The results are not sensitive to return volatility and estimates of realized alternatively measuring volatility over changes in CEOs’ stock-based wealth due 120 or 240-day periods starting at the to changes in equity risk. To estimate the compensation measurement date. To CEOs’ realized change in stock-based control for other determinants of equity wealth due to equity risk, the sensitivity risk, the log of market value of assets of the CEOs’ wealth to equity risk at and financial leverage are included in December 31, 1993, is multiplied by the all regressions. All t-statistics are calcu- change in stock-return volatility from 1993 lated using heteroskedasticity-consistent to 1994.The change in stock-return volatil- standard errors. ity is computed as the standard deviation Column (1) of Table 15.5 indicates that of daily stock returns over the last 120 firms’ stock-return volatility is positively trading days in 1994 minus the standard related to the sensitivity of CEOs’ wealth to deviation of stock returns over the last equity risk. However, given the findings in 120 trading days in 1993. Note that this Table 15.4, and the likelihood of a positive method measures the true changes in correlation between investment opportuni- wealth during 1994 with error due to new ties and equity risk, there is a concern that grants/exercises of options, changes in these inferences are spurious. When the stockholdings, and changes in the Black/ factor score for investment opportunities is Scholes parameters used to estimate the added to the regression, the coefficient on 1993 sensitivities. I find no significant the sensitivity of wealth to equity risk is correlations between these variables. about 25% lower,but remains significantly 412 DIRECTORS’ REMUNERATION

Table 15.5 The relation between stock-return volatility and the sensitivity of CEOs’ wealth to equity risk

Stock-return volatility a b1(Sensitivity of wealth to equity risk) c1(market value of assets) c2(Leverage) c3(Factor score for investment opportunities) error. Stock-return volatility is the annualized standard deviation of daily logarithmic stock returns (in %) over 240 trading days, starting with the last 120 trading days in 1993 and ending with the first 120 trading days in 1994, multiplied by 252(1/2). Sensitivity of wealth to equity risk from options and stock is the change in the dividend-adjusted Black-Scholes value of a CEO’s stock option portfolio and common stockholdings (in $ thousands) for a 0.01 change in the annualized standard deviation of the firm’s stock returns. Adjusted sensitivity of wealth to equity risk due to options and stock removes the influence of firm-specific stock-return volatility by using the sample mean standard deviation of stock returns in all Black-Scholes computations, instead of using the equity risk specific to each firm. Log[market value of assets] is the natural logarithm of [book value of liabilities market value of equity] at December-end 1993. Leverage is [(Book value of assets – book value of equity)/market value of equity]. Factor score is obtained using common factor analysis on the variables Book-to-market ratio, R&D expenditures, and Investment expenditures as described in Section 5.

Annualized standard deviation of stock returns (%)

(1) (2) (3) (4)

Intercept 55.37a 51.84a 53.91a 50.07a (10.47) (10.97) (9.68) (10.21) Sensitivity of wealth to equity risk: 0.047a 0.035a options stock (4.69) (3.90) Adjusted sensitivity of wealth to equity risk: 0.026a 0.016b options stock (2.97) (1.88) Control variables: log(market value of assets) 3.85a 3.51a 3.57a 3.21a (6.03) (6.20) (5.33) (5.41) Leverage 0.67a 1.04a 0.62a 1.02a (3.83) (4.53) (3.44) (4.31) Factor score: investment opportunities 6.20a 6.63a (5.12) (5.28) Adjusted R2 19.63 27.88 15.91 25.48

T-statistics are in parentheses and are computed using White’s heteroskedasticity-consistent standard errors. a Statistical significance at the 5% level. b Statistical significance at the 10% level.

positive. The regression results are robust the Black-Scholes partial derivatives used to including the book-to-market ratio, R&D to estimate the sensitivity of CEOs’ wealth expenditures, and investment expenditures to equity risk. To address this concern, in place of the factor score as proxies for I recompute all Black–Scholes partial investment opportunities. derivatives using the sample mean standard Since the standard deviation of stock deviation of stock returns, instead of the returns is an input in the Black-Scholes volatility specific to each firm. Though model, it is possible that the regression this adjusted measure of sensitivity of coefficients are influenced by a mechanical wealth to equity risk is not expected to fully relation between stock-return volatility and reflect the incentive effects of managers’ THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 413 compensation schemes, it is free from is potentially large enough to influence a mechanical relation with firm-specific investing behavior. In a sample of 278 stock- return volatility. As indicated in corporate CEOs, I find stock options play an Columns (3) and (4) of Table 15.5, the economically significant role in increasing coefficients on this adjusted measure of the convexity of the relation between convexity are about 50% smaller than in managers’ wealth and stock price. The the previous regressions. However, they magnitude of the convexity provided by remain significantly positive. common stock is much lower than that These findings support the hypothesis that of stock options, and of little economic firms’ stock-return volatility is positively importance for most CEOs in the sample. related to the convexity of payoffs in In cross-sectional tests, after controlling managers’ incentive schemes. The coeffi- for the slope of the wealth-performance cient on the sensitivity of CEOs’ wealth to relation, convexity is positively related to equity risk in Column (2) of Table 15.5 proxies for the importance of growth indicates that a difference of one sample options in firms’ assets. This finding sup- standard deviation in the sensitivities of ports the hypothesis that firms provide two CEOs’ wealth to equity risk is expected managers with incentives to invest in risky to be associated with a two percentage projects when the potential loss from point difference in the annualized standard underinvestment in valuable risk-increasing deviation of their firms’ stock returns. For projects is greatest. Finally, consistent with a firm with the sample median standard managers making investment and financing deviation of stock returns, this difference decisions in accordance with their risk- amounts to nearly 10% of the firm’s taking incentives, I find that firms’ equity equity risk. risk is positively related to the convexity provided to CEOs. This study makes two important contri- 7. CONCLUSION butions to the literature. First, I emphasize that the incentive effects of stock-based The determinants of executive compensation compensation encompass more than simply practices, and in particular the relation encouraging managers to increase stock between managers’ wealth and firm per- price. Specifically, I consider firms’ use of formance, is a topic of importance to both stock-based compensation to manage the academics and practitioners. Though the sensitivity of managers’ wealth to firm risk, slope of this relation has been examined in and find evidence consistent with firms considerable detail, no study has quantified using stock options to control risk-related or explored the determinants of curvature, agency problems. Second, I stress the or convexity, in this relation. I argue that to importance of considering an executive’s effectively control agency conflicts between complete portfolio of stock options and stockholders and managers, shareholders common stock when analyzing firms’ com- are expected to manage the convexity, in pensation practices. Yermack (1995) addition to the slope, of the relation observes that the absence of complete data between firm performance and managers’ on managers’ option portfolios can hinder wealth. Since convexity in an incentive researchers’ ability to find a strong link scheme generates a positive relation between between extant agency and financial con- a manager’s wealth and firm risk, compen- tracting theory, and firms’ use of stock sation components with convex payoffs, options. Since managers’ incentives derive such as stock options and common stock, from both newly awarded and previously can induce risk-averse managers to invest issued options and stock, firms are expected in valuable risk-increasing projects that to consider the incentive effects of out- they may otherwise forgo. standing options and stock when choosing My evidence indicates that the convexity the size and characteristics of the current of payoffs in managers’ stock-based wealth year’s options grant. 414 DIRECTORS’ REMUNERATION APPENDIX A days in 1993, multiplied by 252(1/2). There were 252 trading days in 1993. A.1. Estimating the sensitivity r ln(1 risk-free interest rate), where of stock option portfolios to the risk-free interest rate is the yield, stock-return volatility on December 31, 1993, on a U.S. Treasury strip with the same time to Estimates of the sensitivity of stock options maturity as the remaining life of the and common stock to equity risk at stock option. December 31, 1993, are based on the T remaining time to maturity of the Black-Scholes formula for valuing European option, in years, as of December 31, call options, as modified to account for 1993. I use the grant date and dividend payouts by Merton (1973), as duration of the option when granted follows: to compute the remaining time to maturity. If the grant date is unavail- Option value [SedTN(Z) able, I set it equal to July 1st in the XerTN(Z T(1/2))], year the option is issued. If the duration of the option is not specified, I set it equal to ten years at the grant where date, since over 90% of newly issued Z is the [ln(S/X) T(r d 2/2)]/ options in the sample have a ten-year T(1/2), N the cumulative probability duration. function for the normal distribution, S d ln(1 expected dividend rate), where the price of the underlying stock, X the the expected dividend rate is the per- exercise price of the option, the expected share dividends paid during 1993 stock-return volatility over the life of the divided by the year-end stock price. option, r the risk-free interest rate, T the time to maturity of the option in years, and d the expected dividend rate over the life of the option. A.2. Estimating the sensitivity The partial derivative with respect to of common stock value to stock-return volatility is defined as stock-return volatility I estimate the per-share sensitivity of ∂(option value)/∂(stock volatility) common stock value to equity risk at dT (1/2) e N(Z)ST , December 31, 1993, in two steps: i) Compute the implied, per-share firm where N is the normal density function. value using the Black–Scholes model. The parameters of the Black–Scholes ii) Use the implied firm value to estimate model are estimated for stock options the Black–Scholes partial derivative of as follows: stock price with respect to a 0.01 change S price of the underlying stock on in the annualized standard deviation of December 31, 1993. stock returns.8 X exercise price of the option. The Descriptive statistics for the parameters exercise price could not be obtained used to estimate the sensitivity of common for approximately 2% of the options. stock to stock-return volatility appear in For these options, the exercise price Table 15.6. These parameters are defined is set equal to the simple average of as follows: the stock prices prevailing at the Option value per-share price of common beginning and end of the year in which stock the option was granted. X the per-share book value of debt, annualized volatility, estimated as the estimated as the book value of total standard deviation of daily logarithmic liabilities divided by common shares stock returns over the last 120 trading outstanding. THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 415

Table 15.6 Description of parameters used in Black–Scholes computations for common stock

The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat, ranked by market value of equity on December 31, 1988. All data are for 1993. Per-share stock price is the stock price on December 31. 1993. Per-share book value of debt is the book value of liabilities as of December 31, 1993. Per-share market value of assets is per-share stock price plus per-share book value of debt. Implied per-share market value of assets is computed using the Black–Scholes model (see Section A.2 of the appendix).The standard deviation of equity returns is the standard deviation of daily logarithmic stock returns over the last 120 trading days in 1993, multiplied by 252(1/2). The standard deviation of debt is the standard deviation of monthly logarithmic returns on the Merrill Lynch corporate bond index that matches the firm’s S&P senior debt rating, multiplied by 12(1/2).The standard deviation of bond index returns is estimated over the five-year period ending December 1993.The standard deviation of equity and debt returns are used to compute the estimated standard deviation of returns on firm value () (see Section A.2 of the appendix). The risk-free interest rate is the yield, as of December 31st, 1993, on a U.S. Treasury strip with the same time to maturity as the weighted average maturity of the firm’s liabilities. The weighted average maturity of liabilities is estimated as described in Section A.2 of the appendix. Dividend yield is dividends per share paid during 1993, divided by the implied per-share market value of assets. Price-to-strike ratio is the implied per-share market value/per-share book value of debt. Sensitivity of stock price to a $1 change in firm value is the change in the dividend-adjusted Black-Scholes value of common stock for a $1 change in the value of the firm (see Section A.2 of the appendix).

Firm characteristics Minimum Median Maximum

Per-share stock price ($) 2.88 33.56 254.75 Per-share book value of debt ($) 0.84 38.90 1632.73 Per-share market value of assets ($) 9.78 76.75 1680.35 Implied per-share market value of assets ($) 9.51 75.04 1617.96 Standard deviation of equity returns (%) 11.82 22.22 62.06 Standard deviation of debt returns (%) 3.48 4.15 12.24 Est. std. dev. of returns on firm value (%) 3.89 10.70 52.20 Risk-free interest rate (%) 3.22 4.33 5.57 Weighted average maturity of liabilities (yrs.) 0.50 2.57 7.60 Dividend yield (%) 0.00 1.11 5.06 Price-to-strike ratio 0.82 1.88 18.56 Sensitivity of stock price to a $1 change in firm value ($) 0.68 0.97 1.00

annualized volatility of firm value, Merrill Lynch corporate bond index estimated as the annualized standard that matches the firm’s S&P senior deviation of the rate of return on a debt rating, multiplied by 12(1/2).The portfolio that includes the firm’s debt standard deviation of bond index and equity. Portfolio theory implies returns is estimated over the five-year that the variance of firm value is equal period ending December 1993. The 2 2 2 2 to X debt debt X equity equity correlation between equity and debt 2XdebtXequity Cov (debtequity). Xdebt returns is set equal to one, which and Xequity are the weights on equity assumes that shocks to firm value and debt in the firm’s capital structure. affect equity and debt values similarly. equity is estimated as the standard Though this assumption is not expected deviation of daily logarithmic stock to hold empirically (e.g., the value of returns over the last 120 trading days debt may be much more dependent (1/2) in 1993, multiplied by 252 . debt is upon interest rates than the value of estimated as the standard deviation equity), the results are not sensitive to of monthly logarithmic returns on the varying this correlation. 416 DIRECTORS’ REMUNERATION T the weighted average maturity of the Mikkelson (the referee), Cathy Schrand, Clifford firm’s liabilities estimated using Smith, Philip Stocken, Jerry Warner,Ross Watts, Compustat data on corporate liabili- Jerry Zimmerman, workshop participants at ties maturing in less than 1 year, Baruch College, University of Chicago, Columbia 2 years, 3 years, 4 years, 5 years, and University, Northwestern University, and The Wharton School, and especially S.P.Kothari for more than five years. When the firm helpful comments. I gratefully acknowledge has outstanding debt with different financial support from the American times to maturity, common stock is Compensation Association. Earlier drafts of this technically a compound option. That work were titled ‘Compensation, Convexity, and is, when a portion of the debt matures, the Incentives to Manage Risk’. the stockholders have the option to 1 A 10 percentage point change represents pay off that portion of debt and the standard deviation of the sample firms’ purchase an option to buy the firm for stock-return volatility, adjusted for industry and the remaining book value of debt. To firm size. Stock-return volatility is measured approximate the sensitivity of wealth as the annualized standard deviation of daily stock returns. to firm risk for this compound option, 2 Ofek and Yermack (1997) provide evidence I make the simplifying assumption that that managers can influence the magnitude of the firm has a single debt obligation their own stock-based wealth and degree of with time to maturity equal to the diversification. This suggests a less important weighted average time to maturity of role for the risk-aversion effect. However, to the the firm’s debt. I assume a maturity extent that convexity in managers’ incentive of ten years for the portion of debt schemes comes largely from non-portable maturing in more than five years. For employee stock options, managers are less likely most banks, utilities, and insurance to be able to influence the wealth effect. firms, maturity data for years two and 3 Cuny and Jorion (1995) and Statement of Financial Accounting Standards No. 123, over is unavailable. I assume these ‘Accounting for stock-based compensation’, firms have an average maturity on address valuation issues resulting from vesting long-term debt of 7.5 years, though the and portability restrictions. The difference results are not sensitive to alternative between Cuny and Jorion’s ‘corrected’ option maturity assumptions. value and Black/Scholes’ value is smallest when r ln(1 risk-free interest rate), where options do not have vesting restrictions. Since the risk-free interest rate is the yield, most of the options valued in this study are not as of December 31, 1993, on a newly granted options, measurement error due U.S.Treasury strip with the same time to vesting issues is not likely to be a serious to maturity as the weighted average problem. Failure to consider portability restrictions is likely to lead to upward biased estimates maturity of the firm’s liabilities. of option values and convexity. However, it is d ln(1 expected dividend rate on firm much more difficult to determine how portability value), where the expected dividend restrictions affect cross-sectional variation in rate is set equal to dividends paid option values and convexity. during 1993 divided by the implied 4 Note that this risk-related agency problem market value of the firm. Note that is somewhat different than the well-known the dividend rate on firm value is not underinvestment problem described by Myers known ex ante. It is obtained in step (i) (1977). He demonstrates that when fixed claims when the implied total market value of are present in the capital structure, equity holders the firm is computed. may forgo positive net-present-value projects if the gains accrue primarily to fixed claim holders. The risk-related agency problem described here does not require debt in the capital structure, but instead derives from risk-averse managers NOTES that are poorly diversified with respect to their firm-specific wealth. * I thank Ray Ball, Michael Barclay, John 5 Gaver and Gaver (1993) and Baber et al. Core, Paul Fischer, Dan Gode, Jarrad Harford, (1996) provide further empirical support for Christian Leuz, Jon Lewellen, John Long,Wayne this hypothesis. Yermack (1995) finds that the THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 417 investment opportunity set does not help in REFERENCES explaining the wealth-performance slope when examining new grants of stock options. Agrawal, A., Mandelker, G., 1987. Managerial 6 In addition to the determinants posited here, incentives and corporate investment and Jensen and Meckling (1976) explain how fixed claims in the capital structure can create incen- financing decisions. Journal of Finance 42, tives for equity holders to transfer wealth from 823–837. bondholders by increasing the firm’s risk. This Baber, W., Janakiraman, S., Kang, S., 1996. incentive to shift wealth from bondholders may Investment opportunities and the structure of increase the desire of equity holders to motivate executive compensation. Journal of Accounting managers to invest in risky projects. However, and Economics 21, 297–318. this argument is only relevant ex-post with Black, F., Scholes, M., 1973. The pricing of respect to the issuance of debt. Prior to issuing options and corporate liabilities. Journal of debt, equity holders, as the residual claimants, Political Economy 81, 637–659. are expected to structure incentive schemes Core, J., Schrand, C., 1998. The effect of that discourage wealth transfer from fixed claim holders. accounting–based debt covenants on equity 7 Theories of risk management hypothesize valuation. Journal of Accounting and that the benefits from hedging are increasing in Economics, forthcoming. cashflow volatility. Therefore, other things Cuny, C., Jorion, P., 1995. Valuing executive equal, managers of firms with highly volatile stock options with a departure decision. cashflows are expected to be provided with Journal of Accounting and Economics 20, weaker incentives to increase equity risk than 193–205. managers’ of firms with steady cashflows. Since Dechow, P., Hutton, A., Sloan, R., 1996. cashflow volatility and stock-return volatility Economic consequences of accounting for are positively correlated, the alignment of stock-based compensation. Journal of managers’ incentives with optimal hedging strategies can induce a bias against finding the Accounting Research 34 (Suppl.), 1–20. positive relation hypothesized between equity DeFusco, R., Johnson, R., Zorn, T., 1990. The risk and the sensitivity of managers’ wealth effect of executive stock option plans on to equity risk. Though this issue clouds stockholders and bondholders. Journal of the interpretation of the results if the null Finance 45, 617–627. hypothesis cannot be rejected, it does not cre- Financial Accounting Standards Board, 1995. ate interpretive difficulties when the null Accounting for stock-based compensation. hypothesis can be rejected. SFAS No. 123, Norwalk, CT. 8 In contrast to stock options, where the price Froot, K., Scharfstein, D., Stein, J., 1993. Risk of the underlying asset is allowed to vary freely management: coordinating corporate invest- above or below the exercise price during the option’s life, the option value of common stock ment and financing policies. Journal of may be reduced by the existence of bond Finance 48, 1629–1648. covenants that generally do not allow the value Gaver, J., Gaver, K., 1993. Additional evidence of the firm to drop below a certain point without on the association between the investment triggering technical default and forced renegoti- opportunity set and corporate financing, ation. To incorporate this feature into the dividend, and compensation policies. Black–Scholes computations requires detailed Journal of Accounting and Economics 16, assumptions about the constraints placed upon 125–160. equity holders by firms’ creditors. Since there is Haugen, R., Senbet, L., 1981. Resolving the no theoretical or empirical guidance in making agency problems of external capital through these assumptions, I use the standard Black– Scholes model to compute the sensitivity of options. Journal of Finance 36, 629–648. common stock value to firm risk. Failure to Hemmer, T., Matsunaga, S., Shevlin, T., 1994. incorporate this feature results in an upward- Estimating the “fair value” of employee stock biased estimate of the option component and options with expected early exercise. convexity contributed by common stock (see Core Accounting Horizons 8, 23–42. and Schrand (1998) for a detailed discussion Huddart, S., 1994. Employee stock options. of the impact of covenants on the option Journal of Accounting and Economics 18, component of common stock). 207–231. 418 DIRECTORS’ REMUNERATION Jensen, M., Meckling, W., 1976. Theory of the Myers, S., 1977. The determinants of corporate firm: managerial behavior, agency costs, and borrowing. Journal of Financial Economics 5, ownership structure. Journal of Financial 147–175. Economics 3, 305–360. Ofek, E., Yermack, D., 1997. Taking stock: does Jensen, M., Murphy, K., 1990. Performance pay equity-based compensation increase managers’ and top-management incentives. Journal of ownership. Unpublished working paper. New Political Economy 98, 225–264. York University. Lambert, R., Larcker, D., Verrecchia, R., 1991. Pratt, J., 1964. Risk aversion in the small and in Portfolio considerations in valuing executive the large. Econometrica (January), 122–136. compensation. Journal of Accounting Research Smith, C., 1976. Option pricing. Journal of 29, 129–149. Financial Economics 3, 3–51. Mayers, D., Smith, C., 1987. Corporate Smith, C., Stulz, R., 1985. The determinants of insurance and the underinvestment prob- firm’s hedging policies. Journal of Financial lem. Journal of Risk and Insurance 54, and Quantitative Analysis 20, 391–405. 45–54. Smith, C., Watts, R., 1992. The investment Mayers, D., Smith, C., 1990. On the corporate opportunity set and corporate financing, demand for insurance: evidence from the dividends, and compensation policies. Journal reinsurance market. Journal of Business 63, of Financial Economics 32, 263–292. 19–40. Tufano, P., 1996. Who manages risk? An Merton, R., 1973. Theory of rational option empirical examination of risk management pricing. Bell Journal of Economics and practices in the gold mining industry. Journal Management Science 4, 141–183. of Finance 51, 1097–1137. Milgrom, P., Roberts, J., 1992. Economics, Yermack, D., 1995. Do corporations award CEO Organization, and Management. Prentice-Hall stock options effectively? Journal of Financial Inc., Englewood Cliffs, NJ. Economics 39, 237–269. Chapter 16

N. K. Chidambaran* and Nagpurnanand R. Prabhala EXECUTIVE STOCK OPTION REPRICING, INTERNAL GOVERNANCE MECHANISMS, AND MANAGEMENT TURNOVER†

Source: Journal of Financial Economics, 69(1) (2003): 53–89.

ABSTRACT

We examine firms that reprice their executive stock options and find little evidence that repricing reflects managerial entrenchment or ineffective governance. Repricing grants are economically significant, but there is little else unusual about compensation in repricing firms. Repricers tend to be smaller, younger, rapidly growing firms that experience a deep, sudden shock to growth and profitability.They are also more concentrated in the technology, trade, and service sectors and have smaller boards of directors. Repricers have abnormally high CEO turnover rates, which is inconsistent with the entrenchment hypothesis. Over 40% of repricers exclude the CEO’s options when they reprice.

1. INTRODUCTION The parameters of executive stock options (ESOs) are fixed at the time of the STOCK OPTIONS HAVE BECOME AN increasingly grant. Most ESOs have three-to-four-year important component of executive compen- vesting schedule, final maturities of five to sation, and are used by most Fortune 500 ten years, and strike prices that are almost firms to compensate their senior executives. always equal to the company’s share price In an extensive review, Murphy (1998, p. 21) on the grant date.The parameters of option states that “The most pronounced trend in grants are, however,sometimes reset before executive compensation in the 1980s and the options expire. The most common 1990s has been the explosion of stock instance of such resetting is the “repricing” option grants, which on a Black–Scholes of executive stock options, in which firms basis now constitute the single largest lower the strike prices of ESOs following a component of executive pay.” The most decline in a firm’s share price. The new frequently cited explanation for granting strikes are often 30–40% lower than the options is their incentive effect. Options old strike prices, and the strike reset is often create a direct link between management accompanied by an extension of the option compensation and shareholder wealth and maturity. The net effect is to transform thereby align the interests of a firm’s options that are worth little into options with managers with those of its shareholders. considerable value to the repriced executive. 420 DIRECTORS’ REMUNERATION Repricing is formally executed either by repriced. This method is unpopular because canceling the old options and replacing it requires that repricing costs be explicitly them with new grants at more favorable expensed in income statements and not terms, or by simply rewriting the terms of merely reported in footnotes. In October the existing option contracts.The Financial 1999, the FASB ruled that repricings Accounting Standards Board (FASB) rec- conducted after December 15, 1998 should ognizes both types of transactions as being be accounted for using the variable method equivalent. (Carter and Lynch, 2000). ESO repricing has received considerable Proponents of repricing offer two major attention in the popular press. For instance, explanations for repricing. One argument a LEXIS–NEXIS search for “option is that deep-out-of-the-money options no repricing” reveals over 50 published stories longer provide any meaningful incentives to from January to December 1999. Most executives (Hall and Murphy, 2000), so it is articles harshly criticize repricing. Critics necessary to revise the strike price down- appear to be perturbed by two aspects of ward. A second argument, often cited by repricing. First, repricing typically follows high-technology companies, is that repricing a period of poor stock price performance is needed to retain key executives. Executives and decline in firm value. Thus, it seems to will leave if options are so underwater that reward managers for underperformance; in they are no longer a material part of the fact, it has the perverse effect of increasing compensation contract. This explanation is managers’ wealth when they ought perhaps especially prominent in Silicon Valley com- to be fired. Repricing is, therefore, seen as panies such as Symantec Corporation and a signal that the managers of the firm are Apple Computers, who indicate that they entrenched and shareholders are unable to have repriced their executive stock options replace them. Additionally, repricing seems to retain talented executives. to undermine the role of options as a link The academic literature on repricing is between management and shareholder somewhat sparse. A major impediment to wealth. With repricing, executives profit empirical research has been the lack of both when stock prices increase (when adequate disclosure on repricing. Starting options become in-the-money) and when in 1992, the SEC has required firms that stock prices fall precipitously (when out- reprice executive stock options to disclose of-the-money options are repriced). in their annual financial statements, all Active institutional investors have been instances of repricing over the preceding particularly strident in their criticism of ten fiscal years. Gilson and Vetsuypens ESO repricing. These investors argue that (1993) report instances of repricing in a repricing is an example of managerial small sample of financially distressed firms entrenchment, and that it illustrates the that file for bankruptcy. Saly (1994) and inability of existing governance mechanisms Acharya, John, and Sundaram (2000) to curb self-serving behavior by managers. provide theoretical models of the incentive Some institutions suggest that repricing effects of repricing. Brenner, Sundaram, should not be permitted without prior and Yermach (2000) develop models for shareholder approval. These attempts at valuing options subject to repricing redrafting company charters have met with features, by analogy to barrier options, and mixed results.1 The Financial Accounting like Chance, Kumar, and Todd (2000), find Standards Board also appears to have that size is about the only variable that taken a dim view of repricing. On August 12, explains the repricing decision. The typical 1998, the FASB decided to reconsider stock repricer is small relative to the population option accounting policies in light of the of the S&P 500 firm. repricing phenomenon. The FASB argues In this paper,we analyze over 200 repric- that repriced options cannot be regarded as ing announcements between 1992 and 1997 being “fixed” and that firms must use the reported in Standard & Poor’s executive variable accounting method if options are compensation database (EXECUCOMP). EXECUTIVE STOCK OPTION REPRICING 421 We present new evidence on cross-sectional drops is that the control firms are never determinants of repricing and on top explicitly identified in the data. We use sta- management turnover in repricing firms. In tistical methods to account for imperfectly doing so, we shed new light on the relevance observed control samples. Our approach of the agency and poor governance exploits the information in variables such explanations underlying much of the public as past returns or option grant prices that criticism of repricing. We begin by charac- point to plausible control firms, recognizes terizing the negative shock that leads that the instruments are imperfect, and to repricing. Repricers experience negative produces a statistical comparison of returns over the two years before the repricers with (never observed) control repricing fiscal year. The return shock is firms. We further implement the method- accompanied by a steep decline in prof- ology by conditioning directly on the itability and sales growth that is not “underwater” nature of options. reversed in the next two years.The evidence Second, our analysis contributes to the provides a clear picture of the typical literature by introducing several new find- repricer: it is a young and rapidly growing ings on the cross-sectional determinants of firm that experiences an abrupt and firms’ repricing decisions. While previous relatively permanent shock to growth work (Brenner et al., 2000; Chance et al., and profitability. 2000) finds that firm size is about the only We analyze repricing in the context of variable that explains why firms reprice, we overall executive compensation, focusing on uncover evidence that other industry and top management. We examine whether firm-specific variables matter. Younger repricings are economically significant firms, rapidly growing firms, firms belonging compensation events by analyzing the to the technology, trade, and service sectors, levels of compensation in repricers and the and firms with smaller boards are more compensation changes that accompany likely to reprice. We also find no evidence repricing. We find that repricing grants are that repricers have longer-serving managers, economically significant in relation to both more diffuse stock ownership, or especially overall pay levels and previous-year option low institutional ownership. Overall, the grants, and they also significantly exceed evidence provides little support for the notion option grants in matched peers of repricers. that repricing primarily manifests agency However,compensation adjustments accom- problems or ineffective governance of firms. panying repricing tend to be small, and We also develop evidence on top compensation levels in repricers are similar management turnover in repricers. One to those in matched peers. Thus, repricings criticism of repricing is that it reflects are themselves economically significant the inability of shareholders to fire poorly compensation events, but there is little else performing managers because these man- unusual about the compensation levels or agers are entrenched. The entrenchment changes in compensation that accompany hypothesis predicts that top management repricing. turnover rates in repricers are abnormally We then investigate cross-sectional low. However, we find the opposite result: determinants of a firm’s repricing decision. repricers have abnormally high CEO Given a shock to firm value, some firms attrition rates. There is little evidence that choose to reprice their executive stock repricers are unable to effect changes in options while others do not. What industry, management, as an entrenchment hypothesis financial, and governance characteristics might suggest. differentiate the two sets of firms? Our The last part of our paper documents cross-sectional analysis adds to the and analyzes an interesting heterogeneity literature in two ways. One contribution in executives included in firms’ repricing is methodological. A key difficulty in com- announcements. In over 40% of all paring repricers to control firms that do repricing announcements, the CEO is not not reprice despite similar stock price included in the list of executives repriced. 422 DIRECTORS’ REMUNERATION We find a pronounced imbalance in a repricing that took place after the fiscal compensation structures in these “non- year but before the annual proxy filing as CEO” repricers. Specifically, the CEO has an event pertaining to the proxy’s fiscal fewer options and more direct shareholdings, year. We discard seven cases because the while non-CEO executives have relatively repricing is caused by a merger, spinoff, or more options than shares in these firms. regrant of options in a division instead of Thus, a negative return shock in non-CEO options in a parent company. In four cases, repricers weakens incentives of non- the repricing does not relate to named CEO executives far more than those of executives of the firm but to lower employ- the CEO. Here, repricing seems to play the ees. In one case, option strike prices are role of mitigating the intra-management actually increased to the current market incentive imbalances created by negative price and the maturity is extended, perhaps return shocks. motivated by tax considerations. In another We proceed as follows. Section 2 case, the executive bought options at the describes the repricing sample. We charac- Black–Scholes value from the company. terize the industry and time trends in Our final repricer sample has 213 repricing and the nature of the return and instances of repricing.The sample is larger operating shocks that precede repricing. than that used by Brenner et al., who Section 3 describes repricing grants in the analyze 133 repricings in EXECUCOMP context of overall compensation policy. between 1992 and 1995. Our sample size Section 4 conducts a cross-sectional analysis also significantly exceeds that used in of firms’ decisions to reprice. Section 5 Chance et al. (2000), who analyze 53 analyzes CEO and non-CEO repricing. announcements between 1985 and 1994 Section 6 offers conclusions. identified through a LEXIS–NEXIS search. For each year in which a firm reprices executive stock options, we identify the 2. DATA executives whose options are repriced and, in particular, note whether the CEO’s Our repricing sample is based on Standard & options are repriced. To identify whether a Poor’s, 1998 EXECUCOMP database, repriced executive is a CEO or not, we which contains information for 51,555 match executive names to CEO titles deter- executives for 1,836 firms for fiscal year mined as described below, and also verify the 1992 to fiscal year 1997. In this six-year data manually in the proxy. We also count period, EXECUCOMP identifies a total of the number of executives for whom stock 864 executives who are repriced, which options are repriced for each repricing event. translates into 240 separate firm-level For each firm in EXECUCOMP, we repricing announcements. Following Brenner obtain dates when an executive assumes et al. (2000), we read the proxy state- the CEO title and when the executive ments, 10-K filings, or annual reports to leaves the CEO position. This allows us to cross-verify the repricing event. For some identify the CEO in charge at the beginning years, repricing is disclosed only in the of each year. When the dates are not filings in subsequent years. In general, reported, we use EXECUCOMP’s classifi- repricing-related reporting has become cation of CEO for that fiscal year. This both more uniform and more comprehensive procedure identifies CEOs for 10,042 in recent years. firm-years. For each CEO, we obtain the From the initial sample of 240 instances compensation package and the number of of repricing in EXECUCOMP, we discard options and shares held. We also obtain the 27 cases; 14 of these contain reporting aggregate number of options and shares errors. EXECUCOMP sometimes classifies held by all other executives named in the an ESO repricing that occurred in a prior proxy. If the CEO of a firm changes during year as belonging to the current year. a fiscal year, we classify the firm-year as The database, for example, misclassifies having a turnover event. Some observations EXECUTIVE STOCK OPTION REPRICING 423 for the 1997 fiscal year are missing from of 55 events in fiscal 1996, and 36 events the 1998 EXECUCOMP data.We fill these in fiscal 1997. The number of executives in from the 1999 EXECUCOMP database. repriced is relatively stable at a median For the entire sample of firms, we obtain value of four. Table 16.1 also shows that return data from the Center for Research in repricings not involving the CEO have Security Prices (CRSP) and cross-sectional become more common over time, starting data on firm characteristics from the from only one non-CEO repricing in 1992 annual COMPUSTAT tapes. We obtain to 22 in 1996. data on the number of members serving on Table 16.2 presents the number of repric- a company’s board of directors from ing announcements classified by industry. Investor Responsibility Research Center We classify all firms in EXECUCOMP into (IRRC). Board-related observations are 18 industry categories. Our primary classi- available for about 75% of the sample of fication is according to two-digit SIC codes, 11,016 firm-years. Additionally, we obtain but in some instances, we use four-digit SIC two measures of the fragmentation of firm codes, as this is more informative about the ownership reported in the SPECTRUM types of companies that engage in repricing. Disclosure database: institutional owner Unlike Brenner et al. and Chance et al., ship and aggregate shareholdings of 5% we find evidence of industry patterns in block-holders of the firm. Both variables repricing. Firms in the computers and elec- are available for slightly more than two- tronic parts, software and high-technology, thirds of the sample. We merge the return, and biotechnology industries account for firm characteristics, and governance data 80 of 213, or about 37.5%, of all repricing with the repricing data set. events. Other industry segments with some concentration of repricers are the trade 2.1. The repricing sample: industry (wholesale and retail) and service segments. and time trends The technology, trade, and service sectors together account for over two-thirds of all Table 16.1 presents an annual breakdown repricing announcements. On the other of the repricers in our sample and the hand, repricing is rare in the heavy indus- number of executives repriced at each tries such as utilities and mining. announcement. Repricing seems to have The more detailed industry classifications become more popular over time. There are appear to be helpful in identifying industry 15 repricing events in fiscal 1992, a peak patterns in repricing. Part of the gain in

Table 16.1 Repricing announcements by year

Table 16.1 presents annual distribution of all repricing events reported in EXECUCOMP.Column 1 shows the total number of repricing events. Columns 2 and 3 show the number of repricing events when the CEO is repriced and the number of repricing events when the CEO is not repriced, respectively. Column 4 shows the median (mean) number of repriced executives named in the proxy statements of the firm.

Number of

Year Repricers CEO repricers Non-CEO repricers Named executives

1992 15 14 1 4 (3.36) 1993 31 21 10 4 (3.59) 1994 33 17 16 4 (3.24) 1995 43 24 19 3 (4.02) 1996 55 33 22 4 (3.91) 1997 36 18 18 4 (3.87) TOTAL 213 127 86 4 (3.74) 424 DIRECTORS’ REMUNERATION Table 16.2 Repricing announcements by industry

Table 16.2 reports the distribution of repricing and non-repricing firms across industry segments. We classify firms into 18 industry segments based on SIC codes, as detailed in the table.

Industry Full Number of % CEO Non-CEO number Industry name sample repricers Repricing repricers repricers

1 Agriculture & food 330 6 1.81 5 1 100, 200, 2000–2090 2 Mining 1000–1090, 1400 126 0 0 0 0 3 Construction 1500–1700 120 2 0 1 1 4 Oil & petroleum 444 4 0.90 4 0 1300–1389, 2900–2990 5 Small scale manufact. 576 10 1.74 6 4 2100–2690, 2830–2832, 2837–2839 6 Chemicals/related manufact. 1,392 6 0.43 3 3 2800–2899, 3000–3569 7 Industrial manufact. 942 19 2.02 12 7 3680–3990 8 Computers & electronic parts 1,008 40 3.97 21 19 3570–3679 9 Printing & publishing 234 1 0.43 1 0 2700–2799 10 Transportation 312 0 0 0 0 4000–4790 11 Telecommunication 270 2 0.74 1 1 4800–4899 12 Utilities 822 5 0.60 3 2 4900–4999 13 Wholesale 372 11 2.96 9 2 5000–5190 14 Retail 756 20 2.65 10 10 5200–5799, 5900–5990 15 Services 1,002 43 4.29 24 19 5800–5820, 7000–7363, 7389–9999 16 Financials 1,476 4 0.20 4 0 6000–6999 17 Software & technology 414 21 5.08 12 9 7370–7377 18 Biotech 366 19 5.46 12 7 2833–2836

power comes from working with four-digit from non-software firms in the services SIC codes rather than two-digit SIC codes. sector.The detailed classifications, however, This allows us, for instance, to separate out do not eliminate all problems. For instance, biotech companies within the small manu- firms in the retail sector concentrating on facturing sector and software companies software and computers might be better EXECUTIVE STOCK OPTION REPRICING 425 classified as computer-related industries, starting from month 24 but not before. but these firms are classified as retailers Wilcoxon z (p) values for differences and share SIC codes all the way down between repricers and non-repricers for the to four-digit levels with other retailers not periods [24, 18], [18, 12], [12, in computer-related businesses. Without 6], and [6, 0] are 5.41, 10.12, adhoc choices on our part, we cannot 11.68, and 6.59, respectively, all signifi- reclassify these firms. cant at 1%, while Wilcoxon z (p) values for differences in periods [30, 24] and [36, 30] are 0.96 (p 0.34) and 2 2.2. The repricing sample: 1.21 (p 0.23), both insignificant. pre-repricing returns Dissimilarities in returns between repricers and non-repricers evidently surface about Repricing firms usually experience a two years prior to the repricing fiscal significant decline in stock prices over year-end and persist through the year of some period of time prior to repricing. For repricing. Negative returns are dispersed instance, Brenner et al. find that three-year over the two-year period even for the returns prior to repricing are significantly repricing sample, so the data provide less negative. We characterize this price drop in guidance on the magnitude of negative six-month intervals over the three years returns or their timing within the two-year before the repricing fiscal year-end. These period to qualify a non-repricer as a valid findings help define control firms that control firm. We choose firms with returns experience large negative price drops yet do of less than 15% in months [24, 6] not reprice their ESOs. Table 16.3 reports or 30% in [6, 0] as our control sample. the distributional features of six-month The former group of firms experience a buy-and-hold returns covering three years relatively permanent shock to prices, going back from time zero for repricers and while the latter group represents firms with all non-repricers, where year zero is the large negative shocks in a recent period. repricing fiscal year-end for repricers and We recognize that the definition is adhoc so every fiscal year-end for every firm is a the control sample is probably imperfect time zero for non-repricers. with errors of both inclusion and exclusion. Repricers experience lower returns than (We address this issue through econometric non-repricers in all six-month periods methods in Section 3.)

Table 16.3 Prior returns of repricers

Table 16.3 presents data on the distribution of stock returns for repricing and non-repricing firms. A repricer is a firm on the EXECUCOMP database that makes a repricing announcement in a fiscal year, and a non-repricer is a firm on the EXECUCOMP database that does not make such an announcement. Each row reports buy-and-hold returns (in percent) for several six-month periods prior to the fiscal year-end of the repricing year,which is month zero. For each period, we report the first quartile, median, third quartile, and the mean of the returns distribution.

Repricers All other firms

Period 1st Q Med 3rd Q Mean 1st Q Med 3rd Q Mean

[6, 0] 27.72 6.14 19.37 1.59 5.39 8.07 23.53 11.17 [12, 6] 36.1 16.67 5.61 11.71 3.13 10.59 25.71 13.86 [18, 12] 33.33 14.75 5.69 9.31 7.63 5.79 20.98 9.20 [24, 18] 20.6 1.18 26.37 8.31 2.5 11.14 27.79 16.03 [30, 24] 22.48 7.73 31.61 15.61 11.11 3.09 19 6.75 [36, 30] 6.35 13.46 43.17 21.94 3.37 11.47 28.98 16.54 426 DIRECTORS’ REMUNERATION 2.3. The repricing sample: EBITDA ratio for all firms in the same operating performance industry group and fiscal year. As recommended by Barber and Lyon, our analysis Panels A and B of Table 16.4 present data focuses on median rather than mean on the operating performance of repricers abnormal performance, as mean ratios can from year 2 through year 2, where year be distorted by outliers. zero is the repricing fiscal year-end for Repricers perform at industry levels in repricers and every fiscal year end for every year 2, when the median abnormal firm is a year zero for non-repricers. Our EBITDA ratio equals 0.52%, not signif- metric for judging operating performance is icantly different from zero; the Wilcoxon the “EBITDA” ratio, i.e., earnings before z(p) 0.44 (0.65). However,the operat- interest, taxes, and depreciation in year t ing performance of repricers tails off rather (COMPUSTAT item # 13) divided by the sharply in the next two years.The abnormal average book value of assets (COMPUSTAT EBITDA ratio in year zero is negative item # 6) in years t and t 1 (Barber and (median 6.71%) and significant at Lyon, 1996). Following Barber and Lyon, 1% (Wilcoxon z(p) 7.77 (0.00)) and we compute the abnormal EBITDA ratio the performance decline is not reversed in as the raw number less the median the next two years. The decline in repricer

Table 16.4 Operating performance of repricers, non-repricers and control firms

Table 16.4 reports data on operating performance of three groups of firms. Repricers are firms on the EXECUCOMP database that make a repricing announcement during a fiscal year.Non-repricers are firms that do not make such an announcement during the fiscal year. Control firms are firms with a return not exceeding 15% in months 24 through 6 or 6 through zero, where zero denotes the fiscal year-end month. The columns in Table 16.4 report characteristics for the fiscal year and a window of two years preceding and following the fiscal year. For each group of firms, we report EBITDA, defined as earnings before interest, taxes, and depreciation (COMPUSTAT item # 13) as a percentage of the average book value of assets (COMPUSTAT item # 6) in the fiscal year and the year preceding, abnormal EBITDA, which equals EBITDA minus industry median EBITDA, sales, percentage sales growth, and industry- adjusted sales growth (sales growth minus industry median sales growth). In each case, we report the median (mean) of the relevant characteristic.

Year 2 Year 1 Repricing year Year 1 Year 2

A: EBITDA Repricers 15.79 (15.16) 13.02 (11.35) 9.58 (7.53) 11.31 (7.79) 11.35 (7.70) Non-repricers 14.45 (14.78) 14.53 (14.84) 14.74 (15.05) 14.67 (14.93) 14.68 (14.85) Control firms 15.42 (14.67) 12.47 (11.76) 11.41 (9.81) 11.67 (9.88) 12.78 (11.08) B: Abnormal EBITDA Repricers 0.58 (0.91) 3.49 (4.82) 6.71 (8.53) 5.20 (8.43) 4.88 (7.90) Non-repricers 0.00 (0.56) 0.00 (0.58) 0.02 (0.57) 0.01 (0.46) 0.01 (0.32) Control firms 0.20 (0.77) 2.29 (3.63) 3.60 (5.64) 3.29 (5.54) 2.20 (4.37) C: Sales Repricers 222 (639) 276 (723) 320 (717) 334 (709) 355 (796) Non-repricers 652 (2627) 708 (2768) 803 (2994) 923 (3265) 984 (3417) Control firms 380 (1475) 417 (1544) 462 (1652) 500 (1814) 540 (1954) D: Sales growth Repricers 28.16 (47.12) 22.22 (53.7) 3.61 (13.26) 5.85 (13.73) 8.06 (27.28) Non-repricers 8.93 (17.30) 9.13 (17.43) 10.07 (18.71) 10.15 (18.15) 10.39 (17.52) Control firms 14.77 (33.91) 9.38 (20.06) 6.57 (16.24) 6.87 (15.04) 8.36 (16.53) E: Industry-adjusted sales growth Repricers 13.78 (34.43) 7.60 (40.89) 7.26 (0.82) 5.14 (2.09) 1.11 (15.84) Non-repricers 0.07 (7.40) 0.00 (7.35) 0.02 (7.86) 0.00 (7.38) 0.00 (6.51) Control firms 4.17 (22.31) 1.17 (8.76) 4.33 (4.38) 3.5 (3.29) 2.48 (5.07) EXECUTIVE STOCK OPTION REPRICING 427 operating performance is economically economically significant compensation event, significant, amounting to a decline of over it is necessary to understand repricing in the $6 in pre-tax cashflows for every $100 of context of overall executive compensation the book value of assets, or a quarter to a as well as changes in other compensation third of the pre-repricing EBITDA ratio. variables such as salaries, bonuses, or new Patterns for control firms are similar, option grants. In this section, we charac- though the troughs are not as pronounced. terize the levels of compensation and the To obtain additional insight into the adjustments in other pay components when operating shocks that cause repricing, we an executive is repriced, focusing on the top examine the pattern of sales and sales management level. growth surrounding the fiscal year of We begin by determining whether the repricing. Panels C, D, and E of Table 16.4 CEO in charge at the beginning of the repric- report sales, sales growth, and industry- ing fiscal year is included in the repricing adjusted sales growth for repricers, control announcement for each repricing firm. For firms, and all non-repricers, respectively. this purpose, we match the list of CEO Repricers are rapidly growing in year 2. titles with the list of repriced executives Their median raw and industry-adjusted that EXECUCOMP reports. We cross- sales growth are 28% and 13.78%, verify the repricing information with the respectively, and both are significant at actual proxy statements and/or annual 10- 1%. However, sales growth drops dramati- K filings. Our sample has 127 repriced cally in the next two years. The year zero CEOs. We first report the level of compen- sales growth rate is only 3.6% and the sation and its structure for these repriced industry-adjusted growth rate is 7.82%, CEOs and then analyze compensation vari- indicating that growth in repricers essentially able changes in the fiscal year of repricing. stalls and is well below industry levels. We also examine CEO pay changes in Sales growth remains mired in single digits the subsample of firms that reprice other after the repricing year, well below the executives but not the CEO. historically high growth of repricers. Patterns for control firms are similar except that the troughs are again less steep than 3.1. Exante compensation level 3 for repricers. We start by analyzing compensation levels The empirical evidence provides a clear in the fiscal year prior to repricing. Focusing picture of the typical repricing event. A on the prior year ensures that our analysis is firm that initiates repricing is likely to have exante. Thus, we only incorporate informa- enjoyed rapid, above-industry growth rates tion available to the board and compensation and industry-level profitability two years committee at the time of the repricing and before repricing.The growth rate of repricers do not confound exante compensation with experiences a steep and somewhat abrupt adjustments that occur in the repricing drop to well below industry levels in the fiscal year.Panel A in Table 16.5 reports the repricing year. Repricers never regain their total dollar compensation (field TDC1 in historical profitability levels or growth the EXECUCOMP database) and its three rates, which suggests that the shock that major components, i.e., salary, bonus, and precipitates repricing is lasting. option grants, for all repriced CEOs. The three components account for a median of 87% of total CEO pay in our sample. 3. REPRICING IN THE CONTEXT Median previous-year salary, bonus, OF COMPENSATION POLICY4 option grants, and total pay for the repriced CEOs are $375,000, $71,030, $247,400, The decision to reprice an executive’s stock and $941,700, respectively. To benchmark options is only one part of the overall these results, the corresponding median compensation decision concerning the compensation variables across all firms in executive.To assess whether repricing is an EXECUCOMP are $450,000, $234,600, 428 DIRECTORS’ REMUNERATION

Table 16.5 Compensation level, structure, and changes

Table 16.5 reports compensation levels and changes for a sample of 126 CEOs whose stock options are repriced between 1992 and 1997 and corresponding compensation variables for firms matched by size decile, industry, and fiscal year. Panel A reports salary, bonus, the Black–Scholes value of option grants, and total dollar compensation in the fiscal year prior to repricing. Panel B reports the dollar changes in salary and bonus between the repricing and prior year. Panel C reports the Black–Scholes value of (a) the total repricing-year grants, (b) fresh replacement options granted in lieu of the options canceled in repricing, and (c) new option grants. Panel C also gives the change in the Black–Scholes value of the new option grants relative to the previous-year grants. Columns 1 and 2 report data for repricers and matched firms, respectively, while Column 3 gives Wilcoxon z (p) values for testing differences between the two. For each characteristic, we report the median (mean).

Variable Repriced CEO Matched firm Wilcoxon z (p)

Panel A: Compensation levels ($000) Salary 375.00 (417.70) 369.50 (385.00) 0.30 (0.77) Bonus 71.03 (192.70) 159.20 (180.20) 3.42 (0.00) Option grants 247.40 (1356.00) 151.00 (323.10) 1.67 (0.09) Total compensation 941.70 (2028.00) 916.50 (1164.00) 0.49 (0.62) Panel B: Salary & bonus changes ($000) Salary 11.85 (10.15) 18.34 (19.82) 2.00 (0.05) Bonus 0.00 (16.68) 7.60 (24.97) 2.11 (0.04) Panel C: Options in repricing year ($000) Total grants 1222.00 (2931.00) 218.70 (345.70) 9.21 (0.00) Repriced options 870.10 (2307.00) — — New grants 123.10 (612.20) 218.70 (345.70) 1.61 (0.10) option grants 17.56 (240.6) 73.50 (67.71) 3.90 (0.00)

$181,900, and $1,197,000, respectively. three sources of systematic variation in Three of the four compensation variables— compensation: time period, firm size, and salary, bonus, and total compensation are industry sector. lower in repricers at significance levels Time: Figure 16.2, which plots the of 1% or better, while the fourth (option median annual compensation variables grants) is higher but only marginally signif- across all firms in EXECUCOMP, illus- icant at 10%. This suggests that options trates the time trend in compensation lev- are a relatively more important part of pay els. Median prior-year CEO salaries, in repricers. Indeed, options represent a bonuses, and option grants across all firms median (mean) proportion of 31.8% in EXECUCOMP increase from $420,000, (36.5%) of total compensation in repricers $87,000, and $144,000 in fiscal 1993 to versus 17.5% (22.8%) for all other firm- $500,000, $300,300, and $352,000, years and the difference is significant respectively, in fiscal 1997, while total pay at 1% (Wilcoxon z 3.96) (Figure 16.1). rises from $1 million to $1.6 million over The straight differences between repricers the same period. and non-repricers could well reflect Size: Executive compensation tends to compensation patterns peculiar to repricers. be lower in small firms (see, e.g., the recent However, the compensation differences work of Himmelberg and Hubbard, 2000). could also reflect systematic variation due Figure 16.3, which plots median compensa- to the characteristics of repricers, such tion for ten size deciles (based on book as the industries in which repricers are value of assets) of firms, illustrates this concentrated or perhaps time periods in strong relation. Median total dollar CEO which repricing is more prevalent. We find pay, for instance, amounts to $3.3 million EXECUTIVE STOCK OPTION REPRICING 429

NO SIGNIFICANT STOCK PRICE DECLINE 1–p1

OPTION TERMS SET FIRM CHOOSES NOT p1*(1–p2) TO REPRICE p1 1–p2 SIGNIFICANT STOCK PRICE DECLINE p2

REPRICE p1*p2

Figure 16.1 Model of the repricing decision as a sequential two-stage process, where p1 is the probability that firms experience a significant stock price decline and p2 is the probability that firms choose to reprice conditional on such a decline. Non-repricers either do not experience a significant price decline [probability (1 p1)] or choose not to reprice despite such a price decline [probability (p1 *(1 p2))].

Compensation ($000) by Year

1800 1600 1400 1200 1000 800 600 400 200 0 1993 1994 1995 1996 1997

Salary Bonus Options Total

Figure 16.2 Median annual values by year for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP. 430 DIRECTORS’ REMUNERATION

Compensation ($000) By Size Decile

3500

3000

2500

2000

1500

1000

500

0 Decile 1 Decile 2 Decile 3 Decile 4 Decile 5 Decile 6 Decile 7 Decile 8 Decile 9 Decile 10

Salary Bonus Options Total

Figure 16.3 Median annual values by size decile for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP. Firms are classified into ten size deciles (based on the book value of assets) of firms. Decile 1 and Decile 10 consist of the smallest and the largest firms, respectively.

Compensation ($000) By Industry

1200

1000

800

600

400

200

0 Trade Services Technology Other

Salary Bonus Options Total

Figure 16.4 Median annual values by industry for salary, bonus, option grants, and total compensation, across all firms in EXECUCOMP. Firms are classified into four industries: trade, services, technology, and other. for the top-decile firms versus $503,800 Industry: The data in Section 2.1 and for the lowest-decile firms. Similar trends Section 3 show that repricers tend to con- are also apparent in all components of centrate in specific industries, particularly CEO compensation. trade, services, and technology. Figure 16.4 EXECUTIVE STOCK OPTION REPRICING 431 depicts total compensation and its elements remain somewhat lower for repriced CEOs in these three sectors and corresponding (median $71, 030) than in matched compensation variables in all other indus- firms (median $159, 200). This finding tries taken together. Salary, bonus, and is not surprising, following Core and Guay total compensation are lower in the (2001) and Yermack (1995), who argue technology, services, and trade sectors that firms facing high cash needs and with compared to other sectors, and the differ- costly external finance reduce their use of ences are significant at 1% or better cash compensation. Repricers are small (z and p values not reported here). and rapidly growing firms that have experi- Additionally, option grants are significantly enced severe, permanent operating shocks, more dominant forms of compensation in the and plausibly need to conserve resources to technology sector, where options constitute recoup. Finally, repriced CEOs have about a median of 31% of pay versus 14–20% the same levels of overall compensation for other industry sectors. as in firms matched by industry, size, and The mechanics of adjusting for size, fiscal year. industry, and time are as follows. We divide all firms in EXECUCOMP, year by year, 3.2. Compensation changes into ten size deciles formed on the basis of the book value of assets (from COMPUS- In this section, we examine adjustments in TAT) at the beginning of the fiscal year.We compensation variables that tend to also classify firms as belonging to technol- accompany repricing. Panel B in Table 16.5 ogy, trade, services, or other industry reports median and mean salary and bonus sectors. For each firm, we then compute changes for repriced CEOs in the fiscal matched peer salary, bonus, option grant, year of repricing and corresponding statis- and total compensation as the median of tics for industry-, size-, and year-matched the relevant CEO compensation variable for firms. The median salary change accompa- all firms in the same size decile, industry nying repricing is $11,850 versus $18,340 group, and fiscal year.Panel A in Table 16.5 for matched non-repricing peers, and a reports compensation variables for similar difference exists for the means. repriced CEOs and matched peers. It also The difference is economically small but reports Wilcoxon z (p) values for testing significant at 5% (Wilcoxon z (p) differences between repriced CEOs and 2.00 (0.05)). The median bonus change their matched peers. is zero versus $7,600 for matched peers. The matching procedure does attenuate Again, the differences are economically some of the differences between repricers small but statistically significant at 5% and all non-repricers. Median total dollar (Wilcoxon z (p) 2.11 (0.04)). The compensation (TDC) of repriced CEOs low levels of bonuses and their apparent ($941,700) is closer to the TDC of the downward rigidity suggest that repriced peer-matched sample ($916,500) than CEOs have bonuses that amount to floor that of all non-repricers ($1,197,000) and amounts and are probably not a signifi- TDC differences are no longer significant. cant component of discretionary annual Likewise, salaries (median $375, 000) performance-linked incentive pay. are closer to and not significantly different Dealing with the third element of com- from peer-matched salaries (median pensation, option grants, is more intricate. $369, 500), while they are significantly Option grants in the repricing year consist lower than salaries in all non-repricers of two distinct components: (a) options (median $450, 000). Previous-year stock granted in lieu of the canceled (repriced) grants (median $247, 400) tend to be options; and (b) additional new grants over somewhat higher than those in matched and above the repriced options. The two firms (median $151,000), but as before, components must be disentangled to assess the difference is only significant at 10% the magnitude of the repricing grant and the (Wilcoxon z (p) 1.67 (0.09)). Bonuses size of new grants that accompany repricing. 432 DIRECTORS’ REMUNERATION This breakup, however,faces two obstacles. $123,100. This is somewhat lower than First, option grants reported in proxy matched firm grants, as seen in Table 16.5, statements (and EXECUCOMP) do not and also lower than previous-year grants separately identify the number of repriced (median $247, 400), although the dif- options. A second issue is the lack of ferences are only significant at about 10%. uniformity in reporting practices. Some The median ratio of current-year to previous- firms seem to regard repricing as just an year grants is 67% and is not significantly alteration of existing options.These firms do different from 100% (Wilcoxon z (p) not include repriced options in the repricing- 1.38 (0.19)). Thus, firms appear to year options grant. Other firms view repricing maintain or reduce the number of new as a cancelation of old options and regrant options granted in the fiscal year of of fresh replacement options. These firms repricing. The value of repriced options do include the replacement options as (median $870, 100) significantly exceeds part of the repricing-year option grants. the value of the previous-year grant Thus, straight grant numbers reported (median $247, 400) with Wilcoxon in EXECUCOMP and proxies are not z (p) 4.36 (0.00). Likewise, the ratio of comparable across firms. the number of repriced options to previous- We read proxies or annual reports of year grants has a median of 182% repricers to obtain the number of repriced (mean 284%) and significantly exceeds options and also use these proxies to ascer- 100% (Wilcoxon z (p) 5.11 (0.00)). tain whether repriced options are included The evidence suggests that repricing is in in the grants table. Sufficient information is fact a significant addition rather than an available for 109 out of 127 cases, mostly alternative to options granted in the normal in footnotes to the option grants table. In course by firms. about 75% of the sample, repricing-year A related question of some interest is grants include repriced options. Here, we whether large option grants are associated deflate the total repricing-year grant by with poorly performing firms in general, in the number of repriced options included which case we should not view repricing as in the grant to obtain new option grants. In a special event. To address this issue, we other cases (about two dozen), repricing- examine option grants for the control firms year option grants explicitly exclude repriced with negative returns used in Section 2 and options. Here, new options directly equal Table 16.3. The median (mean) current- grant figures reported in EXECUCOMP and year grants for this sample equal $247,600 we obtain the number of repriced options ($807,600). These are slightly below the separately by reading proxy statements. previous-year grants for the same firms Panel C in Table 16.5 reports the value (median (mean) equal to $264,300 of total grants, repriced options, and new ($889,400)), and also significantly below options, and the change in option grant rel- the value of options repriced.5 Repricing is ative to the previous year’s option grant. therefore special; it is not the case that The median Black–Scholes value of the total poorly performing firms that do not reprice repricing-year grant, which incorporates instead give large option grants in lieu of both repriced options and new grants, is repricing. $1,222,000.This far exceeds both matched An alternative to repricing is to issue grant values (median $218,700) and new options but leave the old, out-of-the- previous-year grants (median $247, 400). money options untouched. Firms might be Our findings are consistent with those of reluctant to use this alternative because the Brenner et al., who also find that the total number of options are capped at EXECUCOMP reported Black–Scholes levels pre-authorized by shareholders. It can value of repricing-year grants is unusually be difficult for firms to make large grants high for repriced executives. similar to those involved in ESO repricing Turning to the components of the total without seeking specific shareholder approval repricing-year grants, the median value of for increasing the cap, whereas repricing the new grants accompanying repricing is preserves decision-making at the board level EXECUTIVE STOCK OPTION REPRICING 433 without having to get shareholder approval.6 and by introducing new variables into the Leaving old options untouched also has the cross-sectional analysis. The expanded disadvantage that it could adversely affect analysis permits, in particular, a more the future diluted EPS of the firm. This detailed investigation of alternative expla- alternative can be more attractive, however, nations for repricing, particularly the following the October 1999 FASB ruling agency explanation that underlies much of that companies must expense costs of the public criticism of repricing. repricing because repriced options are Section 4.1 discusses univariate compar- deemed to have “variable” strike prices. isons of repricers with non-repricers and Taxes are unlikely to play a role in this with control firms that experience negative choice, as tax events in ESOs are triggered returns but do not reprice. Section 4.2 only at eventual option exercise. reports estimates of multivariate specifica- Finally, the above analysis focuses on tions in which repricers are compared compensation changes of repriced CEOs. to control firms, accounting for the unob- However, several firms reprice other servability of control firms. Section 4.3 members of the top management team but discusses an alternate definition of control not the CEO. We check whether CEOs in firms that conditions directly on the these firms receive backdoor compensation underwater nature of outstanding options. through unusual salary, bonus, or grant Section 4.4 offers a summary of the results changes. The data do not support this in relation to alternative explanations for hypothesis. For non-CEO repricers, the ESO repricing. median (mean) change in CEO salary, bonus, and option grants is $12,290 ($13,600), $0 ($72, 370), and 4.1. Univariate comparisons $22,270 ( $1, 050, 000), respectively, Table 16.6 presents cross-sectional data versus $17,570 ($20,340), $8,000 for repricers, all non-repricers, and control ($16,460), and $57,690 ($232,600), firms. Column 1 reports the data for respectively, for matched peers, with repricers, Column 2 reports data for all Wilcoxon z (p) values for differences equal non repricers, and Column 3 reports data to 1.66 (0.10), 3.52 (0.00), and for control firms. Column 4 reports the 0.36 (0.72), respectively. There is little nonparametric Wilcoxon (or Mann– evidence that firms give backdoor compen- Whitney) z-statistics and p-values for sation through other means as a substitute testing differences between repricers and for a conventional strike reset repricing. the control sample. The variables in Table In sum, compensation levels in repricers 16.6 are organized into four categories: match those in peers, and changes in com- (i) CEO turnover in repricers; (ii) three pensation variables accompanying repricing proxies for firm size, a focus of much are economically small.The repricing grant, previous work; (iii) data on several other however,is large and comparable to annual firm and governance characteristics includ- total compensation levels.Thus, the repricing ing firm age, ownership, and board size; grant itself is significant to the repriced and (iv) data on executive equity and executive but there is little else unusual in option holdings. the accompanying compensation changes.

4.1.1. CEO turnover 4. CROSS-SECTIONAL One of the main criticisms of repricing is CHARACTERISTICS OF that it reflects the agency problems of REPRICING FIRMS managerial entrenchment in firms. According to this argument, entrenched This section analyzes factors that explain under-performing managers are able to firms’ repricing decisions. We expand the reward themselves via repricing instead repricing literature through new methodology of being fired by shareholders for poor 434 DIRECTORS’ REMUNERATION

Table 16.6 Cross-sectional characteristics of repricers

Table 16.6 reports the number of CEO turnovers and several firm-specific characteristics of three groups of firms. Repricers are firms on the EXECUCOMP database that make a repricing announcement during a fiscal year. Non-repricers do not make such an announcement during the fiscal year. Control firms are firms with returns not exceeding 15% in months 24 through 6 or 6 through zero, where zero denotes the fiscal year-end month. The last column reports z and p values for Wilcoxon tests that compare characteristics of repricers with those of control firms.The firm-specific characteristics are the annual sales, book value of assets, market value of equity, shares and executive stock options held by all executives named in the firm’s proxy statement, the firm’s age (defined as the time in months since listing), the tenure of the CEO (in years), the aggregate institutional ownership and aggregate ownership of 5% block-holders in a firm, and the number of members in a firm’s board of directors. For each characteristic, we report the median (mean).

Repricing All Control Wilcoxon Variable firms non-repricers firms z (p-value)

# of Firms 213 10,072 1471 # CEO turnovers Year [0] 41 (19.25%) 634 (6.29%) 151 (10.26%) 2.19 (0.03) Years [0, 1] 55 (25.82%) 1,266 (12.57%) 268 (18.21%) 2.21 (0.03) Firm size Sales 276 (723) 708 (2,768) 417 (1,544) 3.71 (0.00) Total assets 309 (647) 814 (5,809) 383 (1,829) 3.20 (0.00) Equity value 332 (647) 718 (2,830) 353 (1,360) 1.85 (0.06) Other characteristics Firm age (months) 83 (112) 237 (214) 126 (196) 6.14 (0.00) CEO tenure 6.74 (7.83) 5.89 (8.17) 6.56 (8.13) 0.20 (0.83) Institutional ownership 49.05 (47.82) 50.76 (49.51) 47.34 (46.29) 0.79 (0.43) 5% block owners 26.12 (28.73) 20.56 (25.37) 23.69 (27.87) 0.32 (0.75) Board size 7 (7.41) 9 (9.84) 8 (8.60) 5.74 (0.00) Equity and option holdings Shares (all executives) 461 (2,625) 707 (7,308) 437 (3,855) 0.16 (0.87) Options (all executives) 444 (874) 691 (1,800) 385 (956) 2.38 (0.02) performance. A directly testable implication repricing announcements that occur towards of this entrenchment explanation for repric- fiscal year-ends. By either metric, repricers ing and the related contention by Chance show surprisingly high CEO turnover rates. et al. (p. 129) that “firms with greater About 19% of repricers experience a agency costs are more likely to reprice” is CEO turnover in the year of repricing and that repricers should have abnormally low 25% of repricers experience turnover in top management turnover.We contribute to the repricing year or the next. Both the this debate by presenting evidence on rates one-and two-year attrition rates are signif- of top management turnover in repricers. icantly higher than those for non-repricers Row 1 of Table 16.6 reports the number and for control firms. The turnover data of firms that experience a CEO change in in Table 16.6 provide little support for the the year of repricing, while Row 2 reports a notion that ESO repricers are especially two-year turnover rate, i.e., the number of reluctant to undertake a CEO change, as CEO changes in the year of the repricing an entrenchment hypothesis might suggest. or the next year. The two-year window is On the contrary, firms that announce useful to account for a potential lag a stock option repricing are also more between the time a CEO change is initiated likely to experience an accompanying and its actual implementation, and for CEO turnover. EXECUTIVE STOCK OPTION REPRICING 435 Besides providing a test of the Other explanations for small firm repricing entrenchment-agency explanation for do not rely on agency hypotheses. Small repricing, the turnover findings also have firms tend to be younger firms with less interesting implications for contingent- well developed lines of management, so claims style valuation models of executive the costs of replacing managers could be stock options (see, e.g., Carpenter, 1998). higher for these firms. Renegotiation of ESO valuation models attempt to incorpo- compensation contracts might also be rate features of executive stock options easier in small firms because these firms that are not present in standard option val- are organizationally less complex, which uation models. One stream of this literature makes for faster decision-making. accounts for the repricing feature of execu- It is not clear which of the above tive stock options (e.g., Brenner et al.). A explanations accounts for the size effect, different set of papers deals with turnover but the data certainly suggest that we should and its effect on ESO valuation (e.g., Cuny control for size in the multivariate analysis. and Jorion, 1995). Our evidence suggests Following Brenner et al., we use sales as a that ESO valuation models should incorpo- proxy for firm size. As the empirical distri- rate both repricing and turnover features bution of firm size is skewed, we use the simultaneously, because conditions that natural logarithm of firm size in regression lead firms to reprice also appear to specifications. increase the probability of CEO turnover. Valuation models that ignore the probability 4.1.3. Firm age of higher turnover while incorporating the repricing feature can overstate the value of We measure the age of the firm as the the reset feature in executive stock options. number of months since the first price for the firm is reported in the CRSP tapes. 4.1.2. Firm size Younger firms are less likely to have well developed lines of succession, so manage- We report three proxies for firm size— ment turnover is likely to be costly for the book value of all assets of a firm, the young firms. If executive retention moti- firm’s sales, and the market value of a vates repricing, young firms might be more firm’s equity. We measure all variables as likely to reprice.The evidence in Table 16.6 of the beginning of the fiscal year. By all is consistent with this explanation. Repricers three metrics of size, repricers are smaller tend to be younger, with a median time than non-repricers and control firms. For since listing of about seven years versus instance, the median sales of all non- about 19 years for all non-repricers and repricers are about 2.0–2.5 times the 11 years for control firms. median repricer sales, as in Brenner et al. It is difficult to unambiguously interpret 4.1.4. CEO tenure why small firms are more likely to reprice. One possibility is that repricing reflects Long-serving CEOs are perhaps more agency problems. Small firms perhaps face entrenched and have more pliable boards less scrutiny from the financial community of directors in place. If an entrenchment- and institutional ownership, which could expropriation hypothesis explains repricing, curb rent-seeking behavior,is perhaps lower repricers should have high CEO tenure.The in small firms. On the other hand, agency data in Table 16.6 provide little support problems are more pronounced in larger, for such a hypothesis. CEO tenure is mature firms that have more diffuse own- not statistically different between ership and greater separation between repricers and control firms, or between ownership and control. A finding that small repricers and the universe of non-repricers. firms reprice more often is actually incon- In each case, the median CEO tenure is sistent with the idea that repricing reflects about seven years.This result is also invari- shareholder–manager agency problems. ant to alternate definitions of long-serving 436 DIRECTORS’ REMUNERATION CEOs, such as CEOs who have served for at with greater market valuation of firms, least five years. suggesting that smaller boards are more shareholder value oriented. If repricing reflects shareholder–manager agency prob- 4.1.5. Institutional and block lems, repricers should have greater board ownership sizes, to the extent that board size is symptomatic of the existence of such We examine the aggregate level of institu- agency problems. tional ownership in a firm. Firms with low The last row in Table 16.6 reports institutional ownership are perhaps subject median and mean board sizes of repricers, to less intense monitoring, so the balance of non-repricers, and control firms. We find power between shareholders and managers little evidence that repricers have larger may tilt more towards management in these boards: in fact, the evidence suggests firms. If poor monitoring explains repricing, exactly the opposite. The median board in repricing will be more concentrated in firms repricers has seven members compared to with low institutional holdings. We find no nine members for all non-repricers and evidence to support this contention. Repricers eight for control firms. The differences in have median institutional ownership of board size are significant at 1%. To the about 49.05%, which is slightly but not extent that small boards reflect greater significantly higher than the 47.34% orientation towards shareholder value institutional ownership in control maximization, repricing is unlikely to be an firms. Institutional ownership is somewhat outcome of ineffective governance. higher in the universe of all non-repricers, consistent with higher institutional involvement in large firms. 4.1.7. Equity and option holdings In addition to institutional ownership, of executives we also obtain one other proxy for ownership fragmentation available in the CD Firms in which executives have large option SPECTRUM database—the aggregate holdings as compared to direct equity shareholdings of all 5% block-holders in a ownership are probably more likely to company. This variable is inversely related reprice. One explanation is that negative to the fragmentation in a firm’s ownership. return shocks cause greater misalignment in If shareholder expropriation explains managerial incentives when executives have repricing, firms with more diffuse and more options rather than straight equity. fragmented ownership should be more Another explanation is that, whatever be likely to reprice. We find little support for the reason for ESO repricing, it is worth this hypothesis. The median aggregate the effort only when option holdings are block holding in repricers (26.12%) is not sufficiently large. significantly different from that in control Table 16.6 reports the number of options firms (23.69%). and shares held by all named executives appearing in company proxy statements. We normalize all share prices to $25 so 4.1.6. Board size firms with beginning-of-year share prices greater than $25 have the number of share The number of members in a firm’s board and options scaled up, while those with of directors can be viewed in two ways in share prices below $25 are scaled down, to relation to repricing. One is that smaller make share and option holdings more com- boards offer greater flexibility and speed in parable across firms.The data in Table 16.6 renegotiating contracts with managers. show no significant difference between the This suggests that smaller boards should be shares held by all executives in repricing more likely to reprice. A different perspective firms from those held by all executives in of board size is offered by Yermack (1996), control firms. The data do indicate that who finds that small boards are associated repricers have more options outstanding EXECUTIVE STOCK OPTION REPRICING 437 relative to control firms. Therefore, stock median repricer market value but control options are not lower when executive share- firms are only 5% larger than repricers. holdings are high, as in Yermack (1995). Absent control firm comparisons, it is hard Options seem to be important elements of to say whether repricer versus non-repricer compensation and incentives in repricers differences reflects characteristics peculiar regardless of executive shareholdings in to repricers or those generally associated these firms. with firms that experience large price drops. Are repricers the types of firms in which one might expect greater use of options in 4.2. Multivariate tests executive compensation? The characteristics of repricers in Table 16.6 fit the profile of In this section, we report evidence from firms in which we expect options to be multivariate tests to simultaneously assess used more extensively. Repricers are small, the role of industry- and firm-specific young, and rapidly growing firms that have factors in differentiating repricers from high internal demands for cash, and non- firms that experience negative return cash compensation (such as stock options) shocks but elect not to reprice. is more likely to be used in such firms. Implementing such a comparison is Further, since repricers are young firms, afflicted by an identification problem. executives are less likely to have exercised Figure 16.1 illustrates this issue. While we their option grants. An industry effect in can observe firms that make a repricing repricers—specifically, their concentration announcement, non-repricers could either in the technology industry—also explains be firms that never experience significant the more widespread use of options. In the price declines or control firms that experi- 1990s, technology firms simultaneously ence such declines but choose not to witnessed rapid growth and a short supply reprice. We could identify the latter by of management talent, and used options specifying that control firms should have widely as “golden handcuffs” to facilitate experienced past returns below a negative management retention since options almost return cutoff, say r*, as in Table 16.6. always vest over periods of three to four However, such control samples would years. For instance, Auspex Systems Inc. probably be imperfect because the cutoff writes in its proxy statement dated return and the period over which it is November 20, 1997 that “...competition defined are essentially arbitrary choices. for qualified employees in the technology The control sample could also require more market is extremely intense, and, due to the complex definitions involving multiple rapid growth of many successful companies return cutoffs over several prior periods. in this sector, such competition is increas- Finally, including some firms as controls ing...” and “...the Board of Directors while altogether ignoring others ignores believes that having a stock [option] plan variation within a selected control sample. in place is vital to retaining, motivating, and For instance, a firm with a return of rewarding employees, executives and con- 40% is probably more likely to belong to sultants, by providing them with long-term the control sample than a firm with a equity participation in the company relating “borderline” return of 15%. However, directly to the financial performance and this differential is not reflected in the long-term growth of the company.” standard control sample procedure, whereby The evidence developed thus far emphasize both firms receive equal weights as members two main issues. First, there are several of the selected control sample. systematic differences between repricers and We use statistical methods based on both the universe of all non-repricers “partial observability” probit models and control firms.The evidence also under- (e.g., Abowd and Farber, 1982) to address lines the value of comparisons with control unobservability of control samples. Roughly firms. For instance, the median market speaking, the technique compares repricers value of all non-repricers is 216% of to a “weighted average” of non-repricers, 438 DIRECTORS’ REMUNERATION with higher weights assigned to firms that weights, while returns in [24, 18] and are more likely to be appropriate control [6, 0] are less important in precipitating firms, i.e., firms with more negative histories a repricing decision. Weights on returns in of past returns. The weights, however, need periods prior to t 24 are not signifi- not be specified by the researcher; the cant, so we do not include these in the final technique estimates the weighting function specifications. endogenously as part of the likelihood The second-stage estimates identify function and produces a statistical what types of firms are more likely to comparisons of repricers with the (never respond to negative return shocks by observed) control firms. repricing their executive stock options. In Formally, the procedure takes as input the baseline specification, Model 1, we two sets of variables. In the first stage, the include four industry dummies consisting input is a vector of instruments, say x1i of a technology group (our industry groups (e.g., past six-month returns), that help 8, 17, and 18), services (group 15), trade determine whether firm i belongs to the (groups 13 and 14), and industrial appropriate control sample of firms that manufacturing (group 7). To the industry have negative returns but choose not to dummies, we add firm size, proxied by the reprice. In the second stage, the input is a natural logarithm of sales. We also include second set of variables, say x2i (such firm age (our proxy being the time since as firm size, age, board size, etc.) that first listing on the CRSP tapes), which is potentially differentiate repricers from negatively related to management turnover non-repricers, conditional on a negative costs. Finally, we add the number of options shock.The likelihood function for the model held by all executives listed in EXECUCOMP can be written as for the firm, as a proxy for the demand for repricing. The baseline specification L(x1, x2) Repricers[p1(x1)*p2(x2)] estimates confirm that repricing concen- Non-Repricers[1 p1(x1)*p2(x2)] (1) trates in technology, services, trade, and manufacturing relative to all other sectors The likelihood function is derived by not- such as construction and utilities. Dummies ing that repricers have received a negative for technology, services, and trade are return shock large enough to precipitate con- 2.5–3.5 times the dummy for manufactur- sideration of repricing (probability p1) and ing, indicating a greater preponderance of have chosen to reprice (probability p2) in repricing in the first three sectors relative response to this shock, to give a compound to the latter. As in Brenner et al., firm size probability of p1 *p2, the first term in is important: small firms are more likely to Eq. (1). Firms that have not announced a reprice conditional on a negative return repricing either do not receive a large shock. Firm age is also significant and negative return shock (probability 1 p1), remains so across all specifications. The or receive such a shock but choose not to number of options is positively related to reprice (probability P1 *(1 p2)). The total the conditional probability of repricing. probability of these two possibilities is Models 2–4 sequentially add other 1 p1 p1 *(1 p2) 1 p1 *p2.The explanatory variables to the baseline speci- likelihood function follows. fication. Model 2 adds the change in the Table 16.7 provides estimates of several industry-adjusted EBITDA ratio (similar two-stage models fitted to the repricing results obtain with differences in unad- firms. The first-stage estimates determine justed EBITDA ratios) from year 2 to what types of firms are likely to be in the the year of repricing, and the change in control sample. As expected, the procedure sales growth from year 2 through the weights returns in six-month periods repricing year. As indicated in Table 16.4, between t 24 and t 0 negatively. profitability and sales growth shocks are Returns in the time interval [12, 6] and more severe for repricers compared to [18, 12] have the largest (absolute) control firms, and we test this proposition EXECUTIVE STOCK OPTION REPRICING 439

Table 16.7 Multivariate analysis of repricing firms

Table 16.7 reports estimates of a two-stage probit model fitted to firms on the 1998 EXECUCOMP database, to compare repricers with firms that face similar return shocks but choose not to reprice. The dependent variable is one if a firm announces a repricing in a fiscal year and zero otherwise. Independent variables for the first stage include six- month returns over four periods beginning in month 24, where month zero denotes the fiscal year-end. Independent variables for the second stage are dummies for membership in the technology, services, trade, and manufacturing sectors, the natural logarithm of annual sales, the natural logarithm of the number of options held by all named executives, firm age (years since first listing), the change in the EBITDA ratio and annual sales growth over the last two fiscal years, and the number of members in a firm’s board of directors. Numbers in parentheses denote t-statistics.

Variables Model 1 Model 2 Model 3 Model 4

First stage Intercept 1.28 (12.86) 1.15 (11.19) 1.33 (14.11) 1.23 (10.84) Return [24, 18] 0.38 (4.36) 0.37 (3.91) 0.41 (4.17) 0.44 (10.84) Return [18, 12] 0.97 (13.37) 0.88 (11.72) 1.14 (9.31) 1.08 (9.10) Return [12, 6] 1.33 (14.10) 1.24 (11.44) 1.43 (9.86) 1.36 (8.41) Return [6,0] 0.49 (4.51) 0.40 (3.47) 0.49 (3.14) 0.42 (3.14) Second stage Intercept 1.36 (2.13) 1.34 (2.23) 1.62 (1.77) 1.28 (1.46) Technology 1.71 (4.14) 1.49 (4.46) 1.88 (3.30) 1.69 (3.58) Services 1.45 (3.85) 1.23 (3.70) 2.15 (3.72) 1.55 (3.22) Trade 1.45 (4.51) 1.42 (4.66) 1.58 (3.59) 1.44 (3.59) Manufacturing 0.45 (1.97) 0.44 (1.86) 0.61 (1.76) 0.58 (1.79) Log (sales) 0.34 (3.61) 0.40 (4.17) 0.14 (1.14) 0.21 (1.67) Firm age (years) 0.05 (5.06) 0.05 (4.82) 0.06 (4.26) 0.05 (4.21) Log (#options) 0.07 (2.36) 0.06 (2.32) 0.05 (1.14) 0.04 (1.10) EBITDA 2.25 (1.98) 1.22 (0.71) Sales growth 0.64 (2.67) 0.72 (2.05) Board size 0.16 (2.91) 0.10 (1.94) Model significance 2 (d.f.) 434 (12) 491 (13) 392 (13) 394 (15)

in a multivariate context. Both the change in significance in the multivariate specification. profitability and the change in sales growth The major drops in coefficient magnitudes are significant, indicating that repricers are are for firm size and the change in the likely to have experienced steeper declines EBITDA ratio. Neither variable is significant in profitability and a greater shock to sales in the full specification even at 10%. The growth compared to control firms. Model 3 coefficient for firm size is sensitive to adds in the board size to the baseline spec- inclusion of board size: individually, when ification. Board size appears to subsume the only one of the two is included as a regres- explanatory power of firm size in explaining sor, it matters. However, when both firm firms’ repricing responses to negative return size and board size are included jointly, shocks.The coefficient for firm size falls in firm size loses significance. In the multi- statistical significance, and also drops by variate specification, coefficients for both about half. the manufacturing industry dummy and Model 4 reports the complete multivariate sales growth do not drop by much, though specification. Some variables drop in significance levels are reduced. Finally, the 440 DIRECTORS’ REMUNERATION industry dummy variables for technology, provides additional justification for the services, and trade continue to be significant, 40% out-of-moneyness underlying the as does firm age. Variables other than firm Appendix A computations.8 size, notably industry membership and firm We reestimate the two-step probit speci- age, do matter in explaining the conditional fication of Section 4.2, using the strike probability of repricing given a negative price ratio instead of past returns as the return shock. first-step instrument pointing towards plausible control firms. The results are 4.3. Control sample based on qualitatively similar to those in Table 16.7 approximate moneyness7 and are not reported here. As expected, we find that the strike price ratio has a nega- Section 4.2 uses prior stock returns to tive coefficient and is significant at better identify control firms that do not reprice than 1%. This indicates that firms with despite negative return shocks that plausi- lower strike price ratios, i.e., those with bly cause their options to go underwater. more out of money options, are more likely This section attempts to further refine to be at the node at which a repricing deci- control sample comparisons by conditioning sion must be made. With regard to the sec- more directly on the underwater nature of ond-step estimates, the significant changes outstanding ESOs. are in firm size and the change in the The key obstacle in assessing the money- EBITDA ratio in the full specification. ness of outstanding ESOs is that disclosures While firm size altogether loses its mar- do not report data on strike prices of ginal (10%) significance, the change in the outstanding options. This is, of course, a EBITDA ratio now becomes significant at well known issue afflicting all empirical better than 1%. (The results are available research on executive stock options. upon request.) However, an approximation is possible based on option grant prices reported in 4.4. Summary of cross-sectional proxies (see also Carter and Lynch, 2001). results Since options are normally granted at- the-money, the difference between strikes Our cross-sectional analysis provides an of current grants and lagged grants is a extensive investigation into factors that plausible proxy for the moneyness of explain repricing. We contribute several previously granted options. new results to the repricing literature by Accordingly, for each firm-year we examining a richer slate of variables. Two compute the weighted average of strike aspects of our results deserve special com- prices of option grants, with weights for a ment. First, we show that several variables given strike price proportional to the other than firm size explain why firms number of options granted at that price. reprice. Our results also provide a more From Table 16.3, return shocks to repricers thorough evaluation of agency-based criti- are spread over 24 months prior to the cisms of repricing, and they cast some repricing fiscal year-end. Hence, the ratio of doubt on the notion that repricing primarily the current-year strike to the two-year-back manifests shareholder–manager agency lagged strike is a rough measure of the problems or ineffective governance of moneyness of previously granted options. firms. Specifically, For our sample of repricers, median and mean strike price ratios equal 56.27% • Repricers are young and rapidly and 69.20%, respectively, suggesting that growing firms concentrated in the the median repricer has options that are technology, services, and trade sectors about 40% out-of-the-money.This accords (together accounting for over 70% well with the 40% figure reported in of all repricings). These are not both Brenner et al. and Chance et al., and traditionally viewed as the types of EXECUTIVE STOCK OPTION REPRICING 441 firms in which shareholder – manager “factors beyond the manager’s control” agency problems are especially because systematic factors constitute significant. background risk that makes it harder to fine • Institutional ownership is not lower tune incentives in repriced options. They in repricers nor is the ownership argue on this basis that repricing of repricers more fragmented. is unlikely to be associated with systematic or Repricers have significantly smaller industry-wide shocks to firms. Consistently, boards, while large boards are more we find that negative returns leading to likely symptomatic of ineffective repricing are associated with firm-specific governance or non-value maximiz- operating shocks to growth and EBITDA. ing behavior (e.g., Yermack, 1996). In the same vein, Acharya et al. also argue • Repricers experience abnormally that repricing should be more prevalent in high CEO attrition rates. This is firms where managerial input exerts more inconsistent with the entrenchment influence on future return distributions. view of repricing, which holds that We also find empirical support for this shareholders are unable to fire implication. Repricers tend to be young, executives in repricers despite their rapidly growing firms. Managers probably displeasure with management’s have more influence over the return distri- poor performance, because managers butions of such firms rather than of older are entrenched. firms in mature businesses.We also find that the technology dummy is positively related These findings suggest that it is premature, to repricing. Close to 40% of all repricings and probably incorrect, to conclude as do are concentrated in this sector, which Chance et al. (p. 129), based on a more together with the services and trade sectors limited analysis of an earlier sample, that accounts for over 70% of all repricings.The “firms with greater agency costs are more technology and service sectors, and perhaps likely to reprice.” to a lesser extent the trade sector, are While the cross-sectional analysis comes plausibly industries in which managerial down somewhat against the agency- input, as opposed to other competitive expropriation explanation for repricing, advantages in the product market, is key in it is also useful to consider whether it influencing future returns of firms. supports a view of repricing as optimal recontracting by shareholders with managers. Devising a test for optimality is 5. TOP MANAGEMENT REPRICING difficult because contract theory provides AND TURNOVER few concrete guidelines for such tests and theoretical work on repricing is relatively Our earlier analysis compares the sample sparse. The analysis in Acharya et al. of repricers with firms that do not reprice. provides some initial pointers. We explore In this section, we examine variation within whether our results, particularly some of the sample of firms that reprice their the newer ones, are consistent with their ESOs. We document that while the CEO is optimal recontracting framework. often included among the executives repriced According to Acharya et al., the key by a firm, several firms reprice other exec- consideration in determining optimality of utives but not the CEO.The phenomenon of repricing is the impact of managerial “non-CEO repricing” is an interesting facet control on the returns distribution of the of the data not noted in previous literature. firm. Repricing is more likely to be optimal Section 5.1 reports characteristics of when managers have more control over CEO repricers and non-CEO repricers and future return distributions. They suggest also examines these firms relative to two testable implications. First, they refute control firms, evaluating whether either of the notion that repricing is initiated due to the two subsamples displays support for 442 DIRECTORS’ REMUNERATION the agency, governance, or entrenchment included in the repricing announcement.We explanations for repricing. Section 5.2 cross-verify the repricing information with analyzes CEO turnover in non-CEO and the actual proxy statements in the repricing CEO repricers. year where available, and otherwise verify it in ten-year repricing tables provided 5.1. Do repricers reprice in subsequent proxy statements or annual their CEOs? 10-K filings. If the CEO is repriced, we define the repricing event as a “CEO We begin our analysis of CEO and non- repricing,” and if not, we call it a “non- CEO repricing by identifying the CEO in CEO repricing.” charge at the beginning of the repricing Table 16.8 reports the number of CEO fiscal year. For each repricing announce- and non-CEO repricings. Non-CEO repric- ment, EXECUCOMP reports names of ings constitute 86 out of 213, or about 40% repriced executives. We match this list of all repricings. Table 16.8 also reports of repriced executives with the list of CEO cross-sectional characteristics of both sets titles to determine whether the CEO in charge of firms. All CEO-related data pertain to the at the beginning of the repricing fiscal year is CEO in charge at the beginning of the

Table 16.8 Characteristics of CEO repricers and non-CEO repricers

Table 16.8 reports data on turnovers and several characteristics of firms in the 1998 EXECUCOMP database that announce a repricing between 1992 and 1997. Column 1 reports data for firms that reprice their CEO, while Column 2 gives data for firms that reprice executives other than CEOs. Column 3 reports z and p values for Wilcoxon tests that compare the two sets of firms. Characteristics reported here include annual sales, book value of assets, market value of equity, shares and options held by the CEO and other executives in the proxy statement, the difference between the options (as a fraction of option plus equity) of the CEO and other executives, the firm’s age (months since first listing), CEO tenure in years, the aggregate ownership of institutions and 5% block-holders, and the number of members in a firm’s board of directors. For each characteristic, we report the median (mean).

Non-CEO Wilcoxon z Variable CEO repricers repricers (p-value)

Number of firms 127 86 # CEO turnovers Year [0] 15 (11.81%) 26 (30.23%) 2.35 (0.02) Years [0, 1] 24 (18.89%) 31 (36.05%) 2.59 (0.01) Firm size Sales 291 (680) 265 (789) 0.07 (0.94) Total assets 341 (582) 271 (744) 0.84 (0.40) Market value of equity 296 (672) 363 (611) 0.78 (0.43) Other characteristics Firm age (months) 87 (121) 78 (97) 2.02 (0.04) CEO tenure 6.32 (7.63) 7.48 (8.14) 0.47 (0.64) Institutional owners 53.58 (49.29) 45.63 (45.37) 1.10 (0.27) 5% block owners 23.01 (25.89) 30.41 (33.51) 1.79 (0.07) Board size 7 (7.54) 7 (7.23) 1.08 (0.28) Equity and option holdings CEO options 131 (311) 81 (209) 2.61 (0.01) CEO shares 70 (341) 278 (908) 2.79 (0.01) NonCEO executive options 366 (661) 267 (508) 1.57 (0.12) NonCEO executive shares 77 (2,911) 74 (9,635) 0.22 (0.83) % Options (nonCEO) % Options (CEO) 4.78 (8.40) 21.99 (27.27) 2.80 (0.00) EXECUTIVE STOCK OPTION REPRICING 443 repricing fiscal year, with respect to whom Institutional ownership is actually higher in firms must make repricing and turnover CEO repricers, though not significantly so, decisions. Columns 1 and 2 give characteris- while block ownership is comparable to con- tics of CEO repricers and non-CEO repricers, trol firms. There is no compelling evidence respectively. Column 3 reports Wilcoxon of poor governance in either sample. z and p values for testing differences The data do, however, show pronounced in characteristics of CEO repricers and asymmetries in the structure of the equity non-CEO repricers. As in Section 4, the holdings of CEOs relative to other execu- cross-sectional data in Table 16.8 are tives in CEO versus non-CEO repricers. In organized into four categories: CEO turnover, non-CEO repricers, CEOs have more straight firm size, other firm characteristics, and equity compared to options, while other equity and options holdings. executives have relatively more options The CEO turnover data emphasize that compared to equity. In CEO repricers, the on average, both CEO repricers and non- tilt of equity versus options for CEOs CEO repricers are not particularly averse compared to other executives is not as to initiating CEO changes. In neither CEO pronounced. The asymmetry in CEO and repricers nor non-CEO repricers do we find other executives’ equity holdings relative evidence of abnormally low turnover rates. to options is captured by the variable For non-CEO repricers, the one- and two- “% Options (non-CEO) – % Options year attrition rates are 30% and 36%, (CEO)” in Table 16.8. To compute this respectively.The one- and two-year attrition variable, we first calculate the aggregate rates for CEO repricers are 11.81% and number of options held by executives other 18.89%, respectively.The one- and two-year than the CEO, and express this as a per- turnover rates for these firms significantly centage of the total number of options and exceed turnover rates for all non-repricers, shares they hold. We then compute the which are 6.29% and 12.57%, respectively same measure for the CEO, and subtract (from Table 16.6). For both samples, CEO the option percentage for the CEO from the turnover rates are also at least as high option percentage for non-CEO executives. as the rates for control firms, which are As Table 16.8 shows, the difference in the 10.26% and 18.21%, respectively. There option percentage is positive for both CEO is little evidence that turnover rates in repricers and non-CEO repricers, which either sample are abnormally low, as an indicates that executives have greater pro- entrenchment hypothesis would suggest. portions of their equitylinked compensation Other cross-sectional characteristics of in options compared to CEOs. However,the CEO repricers and non-CEO repricers also variable is significantly larger for non-CEO reveal no evidence of expropriation or repricers (median 21.99%) than for poor governance.We find no significant dif- CEO repricers (median 4.78%). This ferences in size, as measured by sales, indicates that for the sample of non-CEO assets, or market value of equity, between repricers, executives other than the CEO CEO repricers and non-CEO repricers. CEO have more options relative to straight repricers tend to be older (87 months) than equity holdings, while CEO repricers have non-CEO repricers (78 months), but both more equity relative to options. sets of firms are much younger than the The imbalance in options relative to equity control group of firms (126 months, from between the CEO and other executives Table 16.6). We also detect no statistically provides one possible explanation for the significant differences in CEO tenure: CEO incidence of non-CEO repricing. A price repricers have a median tenure of 6.32 years decline has a differential impact on the versus 7.48 years for non-CEO repricers and value of options and the value of shares. 6.56 years for control firms. Board sizes in Further, when the price decline is such both samples are roughly equal, with a that the options end up out-of-the-money, median board size of seven members; executives may place very little value on both types of firms have small boards. them (Hall and Murphy, 2000). Therefore, 444 DIRECTORS’ REMUNERATION if executives have more options compared that exclude the CEO from repricing are to equity, a negative return shock misaligns more likely to experience CEO turnover. incentives for non-CEO executives more Table 16.9 reports probit estimates in than it does for the CEO. The CEO’s incen- which the incidence of CEO turnover in tives are less adversely affected because the repricers is modeled as a function of a CEO has relatively more equity compared firm’s decision to include the CEO in the to options. Non-CEO repricing should, there- list of executives repriced. Controls include fore, be more common when equity-linked firm size, industry dummies for technology, compensation is option-oriented for execu- service, trade, and manufacturing, and tives but straight equity-oriented for CEOs, the number of shares held by the CEO, which is what we find. This would suggest since firms with high CEO shareholdings that repricing has a “re-incentivizing” role, are less likely to experience CEO turnover in which it corrects incentive imbalances (see, e.g., Denis et al., 1997). Column 1 within the top management team that arise reports estimates of a specification that from a negative return shock. includes a zero/one CEO repricing dummy In computing the variable “options,” we as an explanatory variable. The coefficient normalize the current stock price level to for CEO repricing is negative and signifi- $25. This accounts for differences in stock cant at 5%. price levels across firms, so options on a The dummy variable specification we stock trading at $30 are counted as being report in the first column of Table 9 ignores three times as valuable as options on a stock the endogeneity of the CEO repricing trading at $10. However, the normalization decision. Ignoring selection-bias potentially does not account for other differences such leads to inconsistent parameter estimates as differences in option maturities across (see, e.g., Greene, 1993, Chapter 22). We firms or across-firm variation in stock return correct for the selection-bias using the volatilities. These differences could be two-step procedure suggested in Heckman important and result in variation in option (1979). In the first step, we estimate a values across firms (see, e.g., Guay, probit model for the CEO repricing deci- 1999). Appendix A presents two alternate sion. In the second step, the inverse Mills approaches to calculating the dollar value ratio from the probit estimate replaces the of option holdings and recalculates CEO repricing dummy in the turnover regres- the variable “% Options (non-CEO) – % sion. Columns 2 and 3 report the results Options (CEO)” under these measures. The from the two-step procedure. Column 2 option-equity asymmetry between CEOs and reports the turnover probit estimates with non-CEO executives in CEO repricers the selection-bias correction. Column 3 and non-CEO repricers persists under the reports estimates of the first-step probit alternate measures of compensation asym- model for whether the CEO is repriced with metry as well. past returns, industry, firm age, and size, and the difference between the option percentage of the CEO and other executives 5.2. CEO repricing and turnover as explanatory variables. The negative sign in ESO repricers for repricing persists in this selection-bias As Table 16.8 shows, the rates of CEO corrected specification as well. Thus, ESO turnover for non-CEO repricers is higher repricers that reprice their CEOs are less than the rates of CEO turnover for CEO likely to witness CEO turnover. repricers. The difference in CEO attrition rates suggest that within firms that reprice their ESOs, the decision to reprice a CEO is 6. CONCLUSION related to the incidence of CEO turnover. Specifically, repricers that include the CEO The repricing of executive stock options has among repriced executives are less likely to received considerable attention recently, witness CEO turnover, while ESO repricers much of it unflattering. Institutional EXECUTIVE STOCK OPTION REPRICING 445

Table 16.9 Relation between CEO repricing and turnover

Table 16.9 reports estimates of probit models fitted to 213 firms on the 1998 EXECUCOMP database that reprice their executive stock options. The dependent variable is one if there is CEO turnover in the repricing year and zero otherwise. Independent variables include six-month returns for four non-overlapping periods beginning in month 24 where month zero is the repricing fiscal year-end, industry dummies for technology, services, trade, and manufacturing sectors, the natural logarithm of sales, firm age (years since listing), the natural logarithm of the shares held by the beginning-of-year CEO, a zero/one dummy variable for CEO repricing, and the inverse Mills ratio based on probit estimates in Column 3.The dependent variable in Column 3 is one for a CEO repricing and zero otherwise, and independent variables include, additionally, the option percentage (the number of executive options divided by the options plus the number of shares) for the non-CEO executives minus that for the CEO. Numbers in parentheses denote t-statistics.

Model 2

Dependent variable Model 1 CEO turnover CEO turnover CEO repricing

Independent variables Intercept 1.99 (2.51) 2.46 (3.19) 0.85 (1.83) Return [6, 0] 0.27 (0.92) 0.19 (0.63) 0.08 (0.32) Return [12, 6] 0.81 (1.78) 0.89 (1.97) 0.35 (1.17) Return [18, 12] 0.70 (1.71) 0.63 (1.53) 0.15 (0.61) Return [24, 18] 0.02 (0.10) 0.02 (0.07) 0.32 (1.39) Technology 0.86 (1.34) 0.89 (1.37) 0.55 (1.57) Services 0.66 (0.94) 0.66 (0.93) 0.06 (0.16) Trade 1.43 (2.12) 1.36 (2.01) 0.31 (0.74) Manufacturing 0.43 (0.63) 0.34 (0.49) 0.38 (1.00) Log (sales) 0.14 (1.42) 0.15 (1.56) 0.05 (0.65) Log (# CEO shares) 0.11 (3.77) 0.10 (3.03) Age firm 0.04 (1.39) % Options (non-CEO)– % Options (CEO) 0.64 (2.65) CEO repriced 0.66 (2.69) Inverse Mills ratio 0.40 (2.56) Model significance 2 χ11 (p-value) 38.41 (0.00) 35 (0.00) 22.88 (0.02)

investors, in particular, often frown upon the literature in three ways. First, we this practice, as it seems to reward present evidence on whether repricings are managers despite poor performance, pre- economically significant compensation cisely when they perhaps ought to be fired events. We find that while repricing grants instead. It is therefore natural to inquire are large, and in fact, comparable to total whether repricing represents an indictment annual compensation levels, compensation of poor shareholder governance systems, changes accompanying repricing tend to be reflecting entrenched and possibly over- small, and compensation levels in repricers compensated managers expropriating are not abnormally high. Thus, repricing shareholder wealth. is itself significant to the repriced execu- We contribute to the repricing debate tive, but there is little else unusual in the by developing new and extensive empirical compensation patterns in repricers. evidence on over 200 repricing events Our second contribution is methodolog- between 1992 and 1997. Besides a larger ical. In comparing repricers with control sample of repricings, our paper adds to firms that do not reprice despite negative 446 DIRECTORS’ REMUNERATION returns, a key issue is that control firms are We develop some first evidence on CEO not explicitly identified in the data except turnover in repricers, in perhaps the most through imperfect conditioning instruments direct examination of the entrenchment such as past returns or option grant prices. explanation for repricing. If repricing We implement statistical methods to account reflects the inability of shareholders to fire for unobserved control firms.The technique poorly performing managers because exploits information in the conditioning managers are entrenched, top management instruments but recognizes that the instru- turnover in repricers should be abnormally ments are imperfect pointers and produces low. In fact, repricers have abnormally a comparison between repricers and (never high rates of top management turnover. observed) control firms.We also implement Over 25% of firms making a repricing the technique by conditioning directly on the announcement experience a CEO change underwater nature of options rather than either in the year of repricing or the year through poor past returns. after. This attrition rate exceeds that in Finally, we expand the literature through both the universe of non-repricers and a richer cross-sectional analysis of the control firms of non-repricers with poor repricing decision. We present several new past returns. We thus find no evidence findings that shed light on alternative that repricers are especially reluctant to explanations for repricing. In particular,we initiate top management changes, as an offer an extensive evaluation of the agency entrenchment hypothesis suggests. and ineffective governance explanations that Finally, our paper presents some new underlie much of the public criticism of evidence on within-sample variation in repricing.We find, as in previous work, that repricers. We find that a significant 40% repricers are small firms. However, unlike proportion of repricings are “non-CEO” previous work, we find that other variables repricings, in which repricers do not include matter in explaining the repricing decision, their CEOs in the list of executives repriced. even after controlling for firm size. In these instances, repricing may redress We find that the typical repricer is a within-management incentive distortions young, rapidly growing firm that experiences created by negative return shocks. We a steep and somewhat abrupt shock to its also document that within repricers, CEO growth that is not reversed in the following turnover is lower at firms that include the two years. Repricers are more likely to belong CEO in their repricing. Thus, the inclusion to the technology sector,which accounts for of the CEO in a repricing lowers the close to 40% of all repricings in our sample. probability of turnover while exclusion of Repricers are also more likely to come the CEO is likely to be associated with from the services and trade sectors, and CEO departure. together the three sectors account for over Repricing is a key governance issue for two-thirds of all repricing announcements. institutional investors, who must decide There is no evidence that institutional own- whether to support this compensation prac- ership in repricers is low or that the tice. Much of the controversy surrounding ownership structure is especially fragmented repricing arises because repricing appears in repricers, or that CEOs in repricers are to confer rewards upon poorly performing entrenched by virtue of being long-tenured. executives, perhaps symptomatic of agency Finally, repricers tend to have smaller problems, entrenchment, or poor governance boards of directors. Collectively, our cross- of firms.The broader picture, however, tells sectional results reveal no evidence that a different story. Cross-sectional character- repricing is concentrated in firms whose istics of repricers, their compensation shareholder–manager agency problems are policies, the incidence of non-CEO repricing, especially prominent, nor do the results and the high rate of top management reveal any compelling evidence that repricers turnover in repricers do not support the are poorly governed. In fact, by at least one idea that repricing is primarily a manifes- metric, board size, repricers appear to be tation of poor governance or managerial better governed firms. entrenchment. EXECUTIVE STOCK OPTION REPRICING 447 APPENDIX A. VALUE OF significant (Wilcoxon z (p) 2.69 (0.00)). OPTION HOLDINGS The asymmetry in compensation between CEOs and non CEO executives across CEO In computing the variable “options” held for repricers and non-CEO repricers persists each executive, we normalize the current under this second measure of compensation stock price level to $25. This accounts for asymmetry, too. differences in stock price levels across firms, PAYOFF gives the intrinsic value of but does not account for other differences options but ignores the time value of options, such as differences in option maturities and in particular, assigns zero value to all across firms or across-firm variation in stock options that are out-of-the-money on the return volatilities. In this appendix, we con- reporting date. Our second measure sider two alternate measures that attempt to attempts to correct for this by exploiting capture the dollar value of outstanding information disclosed in repricing-year options for each executive. proxies in addition to that available through Estimating outstanding option values is EXECUCOMP. EXECUCOMP identifies not straightforward because of insufficient a list of repriced executives. For each exec- disclosure in proxy statements. This is a utive in the list, we obtain the total number well known problem that plagues empirical of outstanding options at the beginning of research on executive stock options. the repricing year as the sum of exercisable Proxies do not report the value of out- and unexercisable options. We next read standing options for executives, nor do they individual repricer proxies to ascertain report even basic inputs necessary for the number of options repriced for each option valuation such as strike prices and executive and obtain the number of non- maturities of outstanding options. Proxy repriced options as the difference between statements report two pieces of informa- the beginning of year outstanding options tion regarding outstanding options of and the repriced options. executives: the total number of exercisable We calculate the value of the repriced and unexercisable options outstanding and non-repriced options separately and (but not a breakup of these by strikes or sum to compute the value of total options moneyness) and the immediate payoff holdings. We compute the Black–Scholes from exercising in-the-money options but value of repriced options assuming they are little on out-of-the-money options. Some 40% out of the money (Brenner et al. and approximation of option values is required Chance et al. report similar moneyness using what is available in proxy statements. levels for repricing), using the maturity of Our first measure, PAYOFF, uses only the replacement options, risk-free interest information that is available in proxy rates of 8% per annum, and volatilities statements and coded by EXECUCOMP. based on 120-day daily returns prior to the PAYOFF is equal to the intrinsic value of fiscal year-end of repricing following Guay exercisable and unexercisable options that (1999).The value of non-repriced options is are in-the-money on the reporting date, or computed as the value of an at the-money i max(S Xi, 0), where S denotes the option with the same parameters plus the stock price on the reporting date and intrinsic value of exercisable plus unexercis- Xi denotes the strike price of the ith able options, i.e., PAYOFF. stock option outstanding. We recompute The additional data requirement for “% Options (non-CEO) – % Options (CEO)” developing this proxy shrinks our sample to by setting the value of the options portfolio 137 repricing announcements (84 CEO equal to PAYOFF and using the market repricers and 53 non-CEO repricers). The value of equity held by executives. Doing drop in sample size occurs because of so gives the median (mean) difference in missing repricing-year proxy statements option percentage between CEOs and non- and incomplete information in repricing or CEOs equal to 0.08% (8.65%) for CEO subsequent-year proxy statements. repricers and equal to 11.79% (23.47%) We use the second value measure to for non-CEO repricers, and the difference is recalculate the variable “% Options 448 DIRECTORS’ REMUNERATION (non-CEO) – % Options (CEO).” The approval (see, e.g., the case of General Data median (mean) difference in option percent- Comm Industries, Investor Relations Business, age between CEOs and non-CEOs is equal February 15, 1999). to 2.31% (5.04%) for CEO repricers, 2 Prior returns of repricers and non-repricers which is not significantly different from overlap in calendar time. Positive correlation z p across two samples tends to understate signifi- zero (Wilcoxon ( ) 0.71 (0.41)), cance, but correlation within samples can while the difference for non-CEO repricers overstate significance levels. To mitigate within- has a median (mean) value of 15.97% sample correlation, we also compute returns as an (14.02%) and is significant at 5% excess over median industry, time period, and size (Wilcoxon z (p) 2.09 (0.04)). The ratio decile matched portfolio returns. Patterns in these difference between CEOs and non-CEOs is excess returns are similar to those in raw returns. significant at 5% (Wilcoxon z (p) 2.17 Differences between repricer and matched returns (0.03)). The option-equity asymmetry for the six prior periods going back from [6, 0] between CEOs and non-CEO executives in have z statistics of 4.38, 10.05, 8.30, CEO repricers and non-CEO repricers 3.51, 0.25, and 0.99, respectively; the first four are significant at 1% but not the last two remains under this third measure of (p 0.80 and 0.32, respectively). compensation asymmetry as well. 3 The abnormal EBITDA and sales growth statistics for non-repricers show that industry adjustment on the lines of Barber and Lyon NOTES (1996) produces well-behaved statistics. The median abnormal operating performance (Panel † We are grateful to G. William Schwert (the B) and abnormal sales growth (Panel E) are editor) and two anonymous referees for very close to zero and economically and statistically helpful comments. We also thank Mary Ellen insignificant. The mean statistics are positive, Carter, Jonathan Karpoff, Scott Lee, Kai Li, suggesting that the metrics are positively Arvind Mahajan, Vojislav Maksimovic, Thomas skewed, as in Barber and Lyon (1996). Noe, Gordon Phillips, S. Abraham Ravid, 4 We are grateful to an anonymous referee for K. Geert Rouwenhorst, and seminar partici- motivating and developing this section pants at the Tuck/JFE Contemporaneous 5 While current-year grants in these control Corporate Governance Conference, the 2000 firms are lower than the repriced option grants, AFA and 2001 AFE meetings, Louisiana they exceed the new option grants in repricers; State University, Texas A&M University, median (mean) $123, 100 ($612, 200).This Tulane University, University of Massachusetts, can be attributed to the fact that repricers scale Amherst, University of Virginia, and Yale back grants of new options in the repricing University for their comments. Kristina Minnick fiscal year. and Subhankar Nayak provided excellent 6 For example, the board in Bell Sports Corp. research assistance. Special thanks to Kose sought ways to stay within the cap.The company John and David Yermack for extensive feedback writes in its November 1996 proxy that senior on an earlier draft. managers with long tenure were asked to cancel * Corresponding author. Rutgers Business 40% of existing stock options to increase stock School-Newark and New Brunswick, Rutgers options available for grants to other employees. University, 94 Rockafeller Road, Piscataway, 7 We are grateful to an anonymous referee for NJ 70118, USA. Tel.: 1-732-445-4446; suggestions on developing and implementing fax: 1-732-445-2333. E-mail address: this analysis. [email protected] (N.K. Chidambaran), 8 The requirement of lagged strike prices [email protected] (N.R. Prabhala). means that we lose data for fiscal 1992 and 1 In one unsuccessful instance, the State of 1993, the first two years of data in EXECU- Wisconsin Investment Board (SWIB) attempted COMP. We develop a second strike price ratio to force a computer networking company, Shiva proxy by augmenting the data with the two-year Corporation, to include such a proposal in the back market price of the stock from CRSP annual proxy statement, but failed to do so. whenever lagged strikes are missing. Repricers More recently, however, SWIB has successfully have similar median and mean strike price pushed through binding bylaw resolutions ratios (58.61% and 72.05%, respectively) that block repricing without shareholder under this alternative measure. EXECUTIVE STOCK OPTION REPRICING 449 REFERENCES Denis, D.J., Denis, D.K., Sarin, A., 1997. Ownership structure and top executive turnover. Abowd, J., Farber, H., 1982. Job queues and Journal of Financial Economics 45, 193–221. union status of workers. Industrial and Labor Gilson, S.C.,Vetsuypens, M.R., 1993. CEO com- Relations Review 35, 354–367. pensation in financially distressed firms: an Acharya, V., John, K., Sundaram, R., 2000. empirical analysis. Journal of Finance 48, Contract renegotiation and the optimality of 425–458. resetting executive stock options. Journal of Greene, W., 1993. Econometric Analysis. Financial Economics 57, 65–101. Macmillan, New York. Barber, B.M., Lyon, J.D., 1996. Detecting Guay,W., 1999.The sensitivity of CEO wealth to abnormal operating performance: the empiri- equity risk: an analysis of magnitudes and cal power and specification of test statistics. determinants. Journal of Financial Journal of Financial Economics 41, 359–399. Economics 53, 43–71. Brenner, M., Sundaram, R.K., Yermack, D., Hall, B.J., Murphy, K.J., 2000. Optimal exercise 2000. Altering the terms of executive stock prices for executive stock options. Unpublished options. Journal of Financial Economics 57, working paper, National Bureau of Economic 103–128. Research, Cambridge, MA. Carpenter, J.N., 1998. The exercise and valua- Heckman, J.J., 1979. Sample selection bias as tion of executive stock options. Journal of a specification error. Econometrica 47, Financial Economics 48, 127–158. 153–161. Carter, M.E., Lynch, L.J., 2000. Does account- Himmelberg, C.P., Hubbard, R.G., 2000. ing affect economic behavior? Evidence from Incentive pay and the market for CEOs: an stock option repricing. Unpublished working analysis of pay-for-performance sensitivity. paper, Columbia University, New York, NY. Unpublished working paper, Columbia Carter,M.E., Lynch, L.J., 2001. An examination University, New York, NY. of executive stock option repricing. Journal of Murphy, K.J., 1998. Executive compensation. Financial Economics 61, 207–225. Unpublished working paper, University of Chance, D.M., Kumar, R.,Todd, R.B., 2000.The Southern California, Los Angeles, CA. ‘repricing’ of executive stock options. Journal Saly, P.J., 1994. Repricing of executive stock of Financial Economics 57, 129–154. options in a down market. Journal of Core, J.E., Guay, W., 2001. Stock option plans Accounting and Economics 18, 325–356. for non-executive employees. Journal of Yermack, D., 1995. Do corporations award CEO Financial Economics 61, 253–287. stock options effectively? Journal of Cuny, C., Jorion, P., 1995. Valuing executive Financial Economics 39, 237–269. stock options with an endogenous departure Yermack, D., 1996. Higher market valuation of decision. Journal of Accounting and companies with a small board of directors. Economics 20, 193–205. Journal of Financial Economics 40, 185–211. Chapter 17

Julie Ann Elston1 and Lawrence G. Goldberg* EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY

Source: Journal of Banking & Finance, 27(7) (2003): 1391–1410.

ABSTRACT

With the growth of international mergers like DaimlerChrysler, interest in executive compensation practices abroad, particularly in Germany, has increased. Using unique data sources for Germany, we find that similar to US firms, German firms also have agency problems caused by the separation of ownership from control, with ownership dispersion leading to higher compensation. In addition, there is evidence that bank influence has a negative impact on compensation.

1. INTRODUCTION because it has a very different corporate governance structure, characterized by THE DIFFERING LEVELS OF EXECUTIVE concentrated firm ownership and a strong compensation across countries have recently bank presence. In this paper we examine the aroused significant public interest as cross- factors affecting the level of executive country mergers have increased. While the compensation in Germany, with particular academic literature has focused mostly on emphasis on the agency problem created compensation behavior in the US, recent by the separation of management and studies have begun to explore compensation ownership. practices in other countries, particularly in Agency problems caused by the separation Germany. Mergers between firms in different of ownership from control in large corpo- countries naturally raise issues of compara- rations were first popularized by Berle and bility of pay and incentives-for-performance Means (1932) and have since been exam- practices between countries. Numerous ined extensively in both the popular and studies have tried to identify the most academic literature. Jensen and Meckling important factors impacting executive com- (1976) formalized the agency problem and pensation. Recently several studies have stimulated interest in executive compensa- examined the agency problems of firms by tion, the most easily measured way that relating the level of executive compensation managers could take advantage of lack of to corporate control measures. Germany is a control by owners. country of particular interest not only Top executive compensation has grown because it has considerably lower levels dramatically in the United States, and conse- of compensation than the US, but also quently, public interest in compensation EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 451 levels has increased. Popular publications despite high performance. The article cites regularly reveal the compensation of the top 1997 German salaries of top executives executives and particularly the chief executive including: officers (CEO) of the large corporations. The Securities and Exchange Commission ● Jurgen Schremp of DaimlerChrysler (SEC) requires companies to reveal at 3 million DM; compensation components for top execu- ● Bernd Pischetsrieder of BMW at tives. Since US executives are the highest 3.5 million DM; paid in the world, the US has the largest ● Ulrich Hartmann of VEBA at economy, and data are readily available in 2.9 million DM; and the US, most popular and academic attention ● Heinrich Von Pierer of Siemens at has focused almost exclusively on the United 2 million DM. States. However, there has been increased interest in executive compensation in other These compensation levels seem low countries, and particularly in Germany.2 compared to equivalent executives in the Several bodies of literature have US such as: developed in order to determine whether management is taking advantage of the ● Bob Eaton of DaimlerChrysler at lack of ownership control.The first body of 20 million DM; literature examines the relationship between ● John Welch of GE at 69 million DM; executive compensation and corporate ● Andrew Grove of Intel at 89 million performance, while the second relates DM; and corporate performance to the degree of ● Richard Scrushy of Healthsouth at control that owners have over managers. Our 181 million DM. study contributes to the relatively smaller third body of literature relating corporate Although this difference between the two control directly to executive compensation countries has been narrowing recently as and thus directly tests the agency issue. German firms have increasingly turned to Section 2 reviews some of the most granting stock options to top executives relevant studies to date. (Kroll, 1999), corporate control systems Though we know less about executive remain quite different in the two countries.3 compensation in countries other than the One main difference is that many German US, researchers are currently investigating firms have a 2-tiered board system and very similar issues in some of these coun- also many firms have representation of tries, although data availability has banks on the supervisory board (SB) of the impeded this process. Germany has the firm. All the firms in our sample do in fact third largest economy in the world and have a 2-tiered board system. For many consequently is of considerable interest. reasons then, it is quite important to Internationalization has progressed identify the factors affecting executive recently at a rapid pace and German firms compensation in Germany and to assess are more commonly merging with firms the effect of corporate control mechanisms from other countries. Executive compensa- on executive compensation. tion becomes a particularly important issue In order to analyze this issue we have when a German firm merges with an compiled a unique and comprehensive data American firm, such as in the case of the set in order to analyze German executive Daimler–Benz merger with Chrysler. compensation. This data set is more exten- Corporate structure differs greatly between sive than data used in previous studies of the two countries. Do American CEOs get German compensation. This enables us to paid more than their German counterparts? contribute to the literature, based largely If so, can this be justified? A recent article on American data, by analyzing the factors in the German magazine Focus has claimed affecting executive compensation and the that German executives are underpaid effects of ownership control on executive 452 DIRECTORS’ REMUNERATION compensation in Germany. The results for measure or compare across companies. German executives are strikingly similar to Consequently, in this study we follow the those found previously for American practice of previous studies and compare executives.We find evidence that both sales monetary compensation. and ROE are positively correlated with The earliest empirical studies of executive compensation, and that greater ownership compensation in the United States used dispersion leads to higher levels of execu- simple correlation coefficients to identify tive compensation holding other factors the factors related to the level of com- constant. Thus, agency problems appear to pensation. Patton (1951) examines the exist in Germany as well as in the United distribution of pay among different types of States. Further there is evidence that bank executives and finds that executive compen- influence reduces compensation, and that sation is positively correlated with both large block ownership of stock by various profits and growth of the firm. Subsequent groups also has a negative effect on studies usually just used data relating to compensation. the CEO because of greater ease in obtaining Section 2 of the paper reviews the these data. Roberts (1956) and McGuire previous relevant literature. Section 3 et al. (1962) find a stronger relationship describes the German corporate gover- between sales and compensation than nance structure, including the two types of between profits and compensation. In a boards and the role of banks in corporate more recent study, Jensen and Murphy control. Section 4 presents the empirical (1990) find the opposite. Ciscel and Carroll model and the refutable hypotheses. (1980) correct econometric problems in Section 5 describes the data employed.The these models by using an instrumental empirical results are presented and variable for profits to reduce multi- discussed in Section 6. Section 7 concludes collinearity between sales and profits. This with implications for corporate control. literature plus more recent studies conclude that both profits and size are positively related to executive compensation. 2. LITERATURE REVIEW Executives are paid more when the company is more successful and are also paid Berle and Means (1932) raised a number more when the company is larger. This of issues about the operation of the modern provides incentives to executives to improve corporation and noted the problems that performance, a goal in line with stockholders’ arise with the separation of ownership and interest, but it also provides incentives to management. When owners do not control increase the size of the company, a goal the corporation, managers are able to pursue that might not be in the interest of their own economic interests. Dispersion of stockholders. ownership reduces the ability of shareholders Other studies have introduced additional to remove bad managers and also reduces explanatory factors. Among these variables the incentive for and the ability of share- are regulatory effects in transportation and holders to monitor managerial activity. utilities (Carroll and Ciscel, 1982), market Ownership dispersion also provides man- concentration and barriers to entry agers incentives to exploit their protected (Auerbach and Siegfried, 1974), the gender positions and extract benefits for them- of executives and capital investment in selves. The most direct benefit managers workers (Bartlett and Miller, 1988), sales could obtain is increased monetary growth (Murphy, 1985; Coughlin and compensation, and this benefit is also the Schmidt, 1985), and stock performance easiest one to measure. Non-monetary instead of accounting performance (Masson, benefits, such as the size and quality of the 1971; Murphy, 1985; Deckop, 1988). In support staff, the character of the work this study we follow the lead of previous environment, and access to corporate studies in choosing control variables and assets for personal use, are difficult to use the two types of measures, profitability EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 453 and size, that have been found to be Increased attention has been paid to important determinants of executive corporate governance issues recently in compensation. other countries. However, there are only a A second line of research in executive limited number of academic empirical compensation relates corporate perform- investigations of corporate governance in ance to ownership dispersion. Demsetz and other countries. Germany has probably Lehn (1985) find no relationship between been the country in which the greatest profitability and ownership concentration interest has been shown. Germany has more using a linear model. However,Morck et al. of a bank dominated financial system in (1988) find a significant non-monotonic contrast to the market dominated system relationship between both ownership con- in the United States. A number of studies, centration and profit rates and ownership such as Rubach and Sebora (1998), Kim concentration and Tobin’s Q—a measure of (1995), and Emmons and Schmid (1998), market valuation. As ownership concentra- have discussed corporate governance issues tion increases, Tobin’s Q rises, falls, and in Germany and compared them to the then rises again. United States. Chirinko and Elston (1996) The last series of studies is most closely systematically evaluate German bank influ- related to our study. These studies directly ence and find that bank influence does not test the agency problem with respect to lower finance costs nor have any discern- executive compensation. Stigler and able effect on profitability. However, they Friedland (1983) examine the relationship also find that control problems are between shareholder control and executive addressed by concentrated ownership and compensation for a sample of 92 firms for bank influence. Weigand and Lehmann 1937–1938. They find no significant (1998) examine the impact of ownership relationship between the average salary of structure on the performance of German the top three executives and the percent of firms, as measured by return on assets, stock held by the top 20 shareholders. from 1991 to 1996. They find that corpo- Santerre and Neun (1989) replicate this rate structure only affects performance for study, and after including profits as an firms not quoted on the stock exchange. In additional independent variable, find a addition, they show that blockholder own- negative relationship. Both Santerre and ership results in higher returns and that Neun (1986) and Dyl (1988) use more there are systematic interactive influences recent data to test the agency problem with of ownership concentration and the respect to executive compensation. Using identity of owners on performance for the only data for CEOs, both studies find a non-quoted companies. Franks and Mayer negative relationship between ownership (1998), however, do not find a significant concentration and CEO compensation. relationship between board turnover and More recently, Goldberg and Idson (1995) patterns of ownership. examined the agency question for Fortune Schwalbach (1991) was one of the first 500 companies. This study estimated the studies to examine the effects of manage- relationship between the dispersion of ment interests and profits on compensation corporate ownership and the compensation in Germany. Using 468 firms in 1988 from of top executives across the executive the Bonn Data, he shows that size and hierarchy and for different components of industry effects are important in determin- the compensation package in contrast to ing executive compensation. Conyon and the more limited previous studies.The study Schwalbach (1999) examine executive pay found a statistically significant agency effect, outcomes in the UK and Germany. Their though it was small relative to company study finds a positive and significant size. The effects were greatest for salaries, relationship between pay and company the most liquid form of remuneration, and performance in both countries. Finally, the executive with the strongest effect using 83 firms from 1968–1990 and 220 was the Chairman of the Board. firms from 1988–1992, Schwalbach and 454 DIRECTORS’ REMUNERATION Grasshoff (1997) find that firm size and responsibilities of American boards of industry are important determinants of directors.The SB appoints the MB and also compensation for the management board sets the compensation of the members of but that performance itself plays only a the MB. The MB is the top management minor role. of the corporation and has responsibility for The two previous studies most similar to the day-to-day operation of the corporation. our current study are FitzRoy and Commercial banks have a significant influ- Schwalbach (1990) and Schmid (1997). ence on corporate governance in contrast to The first study, using annual data for 95 the United States where the banks are firms from 1969–1985, finds a negative prohibited in general from directly owning effect of concentrated ownership on the equity in corporations. average annual salary of the management In Germany only certain legal identities board. Ownership is measured by a are required to have supervisory and Herfindahl index of equity capital distribu- managing boards. These include the com- tion in 1982. Since these data are only mon AG (Aktiengesellschaft) or stock held available for one year, the same ownership firm, as well as the more rare corporate distribution variable is employed for every form, the KgaA (Kommanditgesellschaft year for each company in the sample. Schmid auf Aktien).4 German law dictates that if (1997) provides an alternative approach to an AG has fewer than 2000 employees, the issue by examining the relationship for then the employees elect one third of the 1991 between compensation on the two members of the SB and two thirds are boards of 110 of the largest 120 traded elected by the shareholders. If the AG has German companies and shareholder struc- more than 2000 employees, the shareholders ture and firm performance. He tests various appoint half of the SB while the firm hypotheses and finds evidence that the employees select the other half. The chair- compensation of both boards is affected by person of the SB is elected by a two-thirds performance and by the firms’ shareholder vote of all board members for a maximum structures. His measure of ownership term of five years, with the possibility of concentration uses a Herfindahl index for reappointment.The law dictates the size of 110 firms in 1991 and has some advan- the SB for firms with more than 2000 tages over the measure that we use here. employees. The SB must have 12 members However, his study only covers one year for firms with 2000–10,000 employees, while our study contributes to the literature 16 members for firms with 10,000–20,000 by conducting a dynamic analysis of corpo- employees, and 20 members for larger rate governance effects on compensation firms with over 20,000 employees. The for a broad panel of German firms over a owner appointed members are frequently 17-year time period. bank representatives and the employee members are frequently labor representatives. Since SB members are paid an 3. GERMAN CORPORATE honorarium fee, and all members are paid GOVERNANCE STRUCTURE the same, we are more interested in the relationship of MB compensation and There are two main aspects of the German performance. Moreover,the members of the corporate governance structure that are dif- MB are comparable to the top executives ferent from the system in the United States. including CEOs of American corporations The first is that many large companies in that have been studied previously. Germany have a 2-tiered board structure A controversial attribute of the German comprised of the Aufsichtrat or SB and the corporate governance system is the extent Vorstand or managing board (MB). We are to which the German universal banks primarily interested in the MB rather than impact the governance of the firm. While the SB. The SB has oversight responsibil- little empirical evidence exists to measure ities for the corporation similar to the the precise extent of corporate control of EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 455 German banks, it is widely believed that these 4. MODEL AND HYPOTHESES universal bankers have a degree of control that extends beyond the traditional bound- The main objective of this paper is to find aries of the creditor–lender relationship. As the relationship between corporate control specified in Elston (1998), the primary factors in Germany and executive compen- spheres of influence of banks on firms is sation. We examine the compensation of through (1) bank share ownership and members of both the supervisory and the associated voting rights accrued from managing boards of German corporations. ownership and proxy votes, (2) through We are most concerned with the compen- bank representation on the SB (sometimes sation of the MB since its members com- as chair or deputy chair), and (3) through prise the top executives of the company. In bank lending and share underwriting. As fact, if there is an agency problem and shareholders, bank representatives regularly management is able to extract additional participate at annual shareholders meetings, compensation for itself, we would expect and are frequently represented on the this effect to be more pronounced for the firm’s SB with one or more representatives. MB than for the SB. This multi-faceted role provides a direct The methodology, which is well estab- line of influence between banks and firms lished in the US compensation literature, that goes beyond the traditional Anglo-Saxon requires least-squares analysis of the creditor–firm relationship. following regression model: Through the proxy voting system (Depotstimmrecht), banks can also obtain log(Compensation) 0 1 ROEjt voting rights from shares in trustee accounts 2 Concjt of bank customers.These votes count in gen- 3 log(Sizejt) eral board decisions, including appointments 4 Bankjt to the firm’s MB and their salary levels. In 5Djt t (1) fact almost half of the total shares issued are deposited in such bank trustee accounts. where Compensation is the total or per Adding the power of the proxy votes to the member salaries for the SB or MB, ROE is votes from direct ownership rights, the over- the return on equity, Conc is the ownership all proportion of votes controlled by banks concentration of the firm, Size is the in the largest 100 firms is 36%; and in the annual firm sales net of value added tax, top 10 firms this control jumps to over Bank is the bank influence dummy variable, 50%. Thus, though banks may own directly and D is the vector of industry and time only a small portion of the voting rights of dummy variables. the firm, through the collection of proxy We employ several variants of this model votes, the banks can have a significant influ- in order to test a series of related hypotheses ence on the decisions of the firm, including and insure robustness of results. Non- both the composition and compensation of reported regression variants, which produced the SB and MBs. From an agency problem consistent results, included use of return on perspective, since banks often hold signifi- assets as an alternative measure of returns cant shares of the firms over long periods of instead of ROE, number of employees and time, they also have both the incentive and total assets as alternative firm size measures the ability to engage in extensive and instead of sales, and lagged independent vari- ongoing monitoring of the firm.Therefore in ables. Correlation matrices were estimated analyzing the factors affecting supervisory to check for the impact of multicollinearity and managing board salaries in Germany, it among independent variables. is important to consider the unique Our main analysis contains four different structure of the German style system of equations, each with a different independent universal banking. Consequently, we include variable. For each of the two boards, SB in our analysis a composite variable and MB, we employ two different compen- measuring bank influence. sation measures, total compensation of the 456 DIRECTORS’ REMUNERATION board and compensation for the average definition in Section 5, and 0 otherwise.The board member.These measures are chosen refutable hypothesis is that banks will exert because the actual compensation is not influence and moderate executive compen- available for each individual board member. sation. Therefore, a negative coefficient is SB members are compensated equally, but expected. MB members are paid differentially. We also test the effects on executive com- However, these compensation differentials pensation of different types of ownership. do not approach the differentials among We employ dummy variables to indicate American executives. whether the firm is owned by another firm, We hypothesize a positive relationship owned by foreign interests, owned by a bank, between log Compensation and ROE and or family owned. Government control, mixed also a positive relationship between log ownership, and management control (no Compensation and log Sales.This is consis- large blocks) comprise the omitted cate- tent with the findings of previous studies gory. Because bank influence has a close that have found that executives are paid relationship with bank ownership, we run more when the company is more profitable regressions including these dummy variables and when the company is larger. both with and without the bank influence We cannot employ a Herfindahl index as variable. We expect executive compensation done by Schmid (1997) and FitzRoy and to be less when there is strong ownership Schwalbach (1990) because complete data control in these four categories. on ownership structure are not available Since agency problems may differ by for each firm for the 17-year span of our industry, there may be some heterogeneity data panel. This is because those who own in compensation practices across firms. A less than 25% of a firm are not required to final analysis examines bank influence by disclose ownership in Germany. We do, industry group. however, have complete interval data on ownership percentages greater than 25% for each firm for every year of the study. 5. DATA We therefore construct an alternative concentration measure based on the One of the greatest impediments to meas- ranked categorical source data on owner- uring the impact of corporate governance ship concentration. The highest numbered on executive compensation in Germany has concentration category represents the high- been the lack of reliable and comprehensive est degree of owner concentration. Initially data.This study uses two non-commercially we develop a variable that incorporates the available sources of unique data.The first is rank of the concentration class, and for this a firm-level database that tracks the variable we would expect a negative coeffi- financial performance of a comprehensive cient in the regression model because the set of German firms from 1961 to 1986. agency problem is expected to be most The second data source is derived from important when ownership is more dispersed. various sources, including Commerzbank’s That is, in the categories with greater Wer Gehort zu Wem?, Handbuch der dispersion, management may be able to Grossunternehemen, Leitende Maenner und extract greater compensation. Since this Frauen der Wirtschaft, Aktienfuehrer, formulation using the ownership categories various annual reports of the firm, and might assume a particular relationship, Boehm (1992). The combined information we alternatively employ a model with in these collected data exceeds the depth dummy variables for each category, omit- and breadth of commercially available ting the category with the most dispersed data for Germany, enabling us to directly ownership. examine the importance of corporate The bank influence variable is a dummy governance on compensation. The bank variable that takes a value of 1 when a firm influence variable for example is a composite is determined to be bank influenced, per our measure using information on percent of EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 457 bank ownership of the firm, percent of because of the substantial changes in the bankers on the SB and whether they are German accounting laws, which make it chair of the SB, and total number of votes impossible to compare the data without exercised by banks at annual shareholder considerable adjustment.12 The sample meetings.5 time period of 1970–1986 is important not We operationalized this information by only as point of comparison with current defining a dummy bank influence variable. stock of US studies available for this period, A firm is characterized as bank influenced but also to understand the pay–performance rather than independent if the bank owns relationship during this period for Germany. more than 25% of the shares of the firm, The number of firms in the sample is also if total votes of banks at shareholder meet- fairly representative because the German ings (including proxy votes) are greater exchange is considerably smaller than its than 50%, or if total votes are between American counterpart. For example in 25% and 50% and the chair of the SB is 1980 there were only about 459 listed a banker.6 This discrete measure is possibly firms incorporated as AG and KgaA.13 an imperfect measure of the influence of For these 91 firms we have annual firm- banks on firms, however it does succeed in level information detailing the identity of incorporating all of the available information firm owners. These discrete data are on bank–firm relations for each of these reported in mutually exclusive categories firms over each of the 17 years, thereby defining the identity and concentration of enabling a dynamic empirical analysis of the ownership structure for the firms. compensation behavior.7 Firms categorized as “Firm Controlled” As mentioned above, the data used also are those in which another domestic firm include a subset of the Bonn Data that is either owns more than 50% of the out- based on a collection of financial reports of standing shares or another firm owns at about 700 German industrial corporations least 25% and no one else owns more than quoted on the German stock exchange.8 25%.“Foreign Controlled” firms are those Because of mergers, bankruptcies, acquisi- with more than 50% ownership by foreign- tions, changes in legal status, and double ers.“Bank Controlled” firms are those with listing of consolidated and non-consolidated more than 50% ownership by financial information, about 295 unconsolidated institutions. “Family Controlled” firms are firms remained in 28 industry groups as of those with more than 50% ownership by 1986.9 We had information on the owner- family members, groups, or individuals. ship identity and concentration for exactly “Government Controlled” firms are those 100 firms over the full time period. This with more than 50% ownership by any was reduced to 91 firms in the analysis county or state. “Management Controlled” where we had both balance sheet and bank firms are those which have no block owner- relationship data.Thus we used all firms for ship percentage over 25%. Finally “Mixed which the appropriate data were available Control” denotes a situation where differ- rather than using a random sample of ent types of the above owners exist, each firms.10 owning anywhere from 25% to 50% of the We need to focus on these listed firms in shares.We combine the last three groups as order to study the phenomenon of executive the omitted category in the empirical compensation since these are the types of analysis. Most firms in this omitted group firms for which supervisory and managing have no block ownership over 25%. boards are mandatory. We used data from The concentration of firm ownership is 1970 to 1986 because prior to 1970 there measured from one to five, where five were too many missing key variables.11 defines the highest degree of concentration, Even for this time period, some variables with a single stockholder holding more were missing, thus requiring running than 75% of the firm’s shares. If two or regressions on smaller subsets of the full three stockholders hold more than 75% sample. Data after 1986 are not used of the shares and the firm does not qualify 458 DIRECTORS’ REMUNERATION for the higher concentration level, then The number of observations at the bottom concentration is set at four. Concentration of the table indicates the maximum num- is set at three if a single stockholder holds ber of observations for each variable; more than 50% of the shares, and two, if however, the actual regressions generally two or three stockholders own more than used fewer observations because of missing 50% of the shares. Concentration is set at data. Note that there appear to be substan- one for all cases in which the concentration tial differences across variable means for level is lower than for category two.14 Note the levels of ownership concentration. This that if the firm satisfies two or more con- justifies the separate analysis of each centration levels, the firm is placed in the category of ownership concentration. It higher concentration category.15 SB and should be noted that average compensation MB pay is defined as total salaries paid to per board member is lower for the higher supervisory and managing board members concentration ownership categories 3–5. by the firm for which they provided this Thus without controlling for other factors, service.16 Sales is measured as sales of the there is some evidence that higher levels of firm net of taxes.The data for variables are ownership concentration help to control the measured in terms of logs of millions of agency problem. Deutsche marks. The estimations were run Table 17.2 contains the results of OLS with both time and industry dummy fixed-effect regressions on the log of com- variables in order to control for macroeco- pensation measures of the two boards. We nomic shocks and industry-specific effects are most interested in the results dealing in the model. with the MB. Panel A of Table 17.2 reports results for regressions using the log of average salary of a management board 6. EMPIRICAL RESULTS member and the log of total salaries of the management board members of each firm The means and standard deviations of the as the dependent variables. The results for main variables are presented in Table 17.1. both measures of compensation are quite

Table 17.1 Decriptive statistics 1970–1986

Variable All firms Bank Independent Conc 5 Conc 4 Conc 3 Conc 2 Conc 1 influenced (high) (low)

ROA 0.0239 0.0254 0.0337 0.0197 0.0260 0.0224 0.0251 0.0239 (0.0230) (0.0166) (0.0238) (0.0155) (0.0184) (0.0172) (0.0341) (0.0639) ROE 0.0555 0.1013 0.0494 0.1109 0.0396 0.0355 0.0377 0.0363 (0.2168) (0.3474) (0.1921) (0.4269) (0.0369) (0.0315) (0.0695) (0.0639) Net sales 1453.04 808.869 1551.32 965.950 659.265 294.522 1111.32 915.88 (3869.95) (1890) (4080.08) (2245.39) (1129.85) (974.26) (1968.99) (1845.21) Total assets 1149.83 399.2058 1263.91 558.223 732.51 242.6478 672.0945 564.98 (3136.83) (862.966) (3335.37) (946.81) (1202.29) (654.896) (937.13) (941.51) Number of 9701.93 6747.21 10148.93 5731.71 3122.28 1891.28 9802.26 7299.25 employees (25957) (15974) (27126) (11665) (4405) (3914) (17112) (12950) SB salaries 0.2215 0.2306 0.2202 0.1363 0.2014 0.1555 0.3031 0.2492 (0.2511) (0.2315) (0.2539) (0.1349) (0.2333) (0.1628) (0.2478) (0.2453) MB salaries 1.4761 1.1722 1.5208 0.9477 1.2038 0.7049 1.6876 1.3918 (2.2285) (1.8093) (2.2808) (1.1923) (1.9360) (1.0046) (1.8959) (1.6029) Observations 1683 221 1462 517 192 288 257 429

Standard deviation in parenthesis below means. Net sales, total assets, and average board salaries are in logs of millions of DM. Conc stands for ownership concentration: Conc 5 highest concentration, Conc 1 lowest concentration. Bank influence is a composite measure of bank influence on the firm, incorporating information on: Ownership structure and concentration, proxy voting by banks, bank representation, and position on SB. Data sample covers 17 years for 100 firms; however, not all firms have data for each year for every variable.Therefore, actual observations used in calculations vary. EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 459 Table 17.2 OLS fixed-effects regressions of (Panel A) MB salaries and (Panel B) firm-level SB salaries

Ownership Variable ROE Sales concentration Bank influence Adj R2

Panel A Average MB 0.049* 0.210* 0.036* 0.108* 0.4184 member salary (2.326) (20.403) (3.419) (2.338) Firm total MB 0.027 0.502* 0.0841* 0.191* 0.8072 salary (1.100) (42.408) (7.020) (3.594) Panel B Average SB 0.094* 0.1577* 0.114* 0.002 0.3425 member salary (2.648) (9.096) (6.471) (0.019) Firm total SB 0.111* 0.410* 0.098* 0.1310** 0.5304 salary (2.937) (22.079) (5.185) (1.570)

All equations use time and firm dummies, and White’s error correction for obtaining heteroskedastic parameter estimates. *t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365. similar indicating that we are capturing the indicating that better performance and same compensation process with either larger size lead to greater compensation definition and can proceed to report results for the members of the SB board. However, only for the average salary per board mem- the bank influence variable is negative and ber in the following tables. The ownership not significant in the average SB member concentration variable has a negative salary equation and positive and significant coefficient and is statistically significant in at the 10% level in the firm total SB salary both equations as predicted by the agency equation. Since bankers sit on the SB they hypothesis. Firms with more concentrated may have conflicting motives regarding ownership provide less compensation for controlling compensation. Clearly bank MB members. ROE has a positive and influence is more important in controlling significant coefficient for the average MB compensation on the management board. member salary but is insignificant for the In results not reported here, we have firm total MB salary.The Sales variable has substituted ROA for ROE and Assets or a positive and statistically significant coeffi- Employees for Sales. The results including cient in both equations indicating that larger these variables were essentially the same as size leads to greater compensation for the those reported here. members of the MB.17 The bank influence In Tables 17.3 and 17.4 we add addi- dummy variable is negative in both cases at tional explanatory variables to the analysis. the 5% level. Bank influence appears to Instead of using a single summary variable reduce management compensation. to measure ownership concentration, we Panel B of Table 17.2 presents results employ four dummy variables in order to for the SB. In both the average salary examine the importance of the different equation and the total salary equation, the ownership concentration categories and ownership concentration variable again has omit the least concentrated category, a negative coefficient that is statistically Conc 1. We also include dummy variables significant. Greater ownership concentration for groups that have ownership control.18 leads to lower compensation for SB mem- Table 17.3 analyzes management board bers, confirming the hypothesis that lack of salaries and Table 17.4 analyzes SB ownership control permits greater personal salaries. In both tables three equations are rewards. Both ROE and Sales are positive estimated using OLS fixed-firm effects and significant in the two regressions, and generalized method of moments using 460 DIRECTORS’ REMUNERATION Table 17.3 Managing Board (MB) salaries

OLS MB salaries GMM MB salaries Dependent variables 1 23123

ROE 0.0091 0.0092 0.0136 0.0885 0.2061 0.0514 (0.23) (0.23) (0.27) (0.62) (1.59) (0.34) Sales 0.1372* 0.1555* 0.1495* 0.0907* 0.2596* 0.0754* (7.01) (8.16) (16.52) (7.99) (10.01) (7.87) Bank – 2.6262* 0.4831* – 2.4381* 0.1113** influence – (27.77) (7.96) – (18.19) (1.80) Concentration (1 low, 5 high) Conc 5 0.1045 – 0.2138* 0.3204* – 0.0552 (1.43) – (2.96) (3.50) – (0.74) Conc 4 0.1182 – 0.3070* 0.1392 – 0.0742 (1.21) – (2.79) (1.25) – (0.71) Conc 3 0.2085* – 0.7625* 0.5086* – 0.6557* (3.81) – (13.73) (8.88) – (12.47) Conc 2 0.1614** – 0.09256 0.1858 – 0.0699 (1.65) – (1.06) (1.61) – (0.62) Ownership Firm 2.3419* 2.5311* – 0.2597* 2.5152* – (18.41) (21.98) – (3.25) (16.70) – Foreign 2.3792* 2.6496* – 0.5550* 3.0891* – (11.32) (14.72) – (3.93) (12.52) – Bank 2.3848* – 0.2618* – – (20.73) – (3.50) – – Family 2.2640* 2.4781* – 0.2715* 2.4527* – (19.18) (24.07) – (3.83) (17.57) – Adj R2 0.7583 0.7530 0.5935 0.7994 0.8339 0.7705 Sargan – – – 146.693 121.2323 168.1391 (p-Value) – – – (0.2503) (0.2069) (0.2869)

All equations use time and firm dummies, and average MB salary. OLS fixed-effects model reports estimates corrected for heteroskedasticity. GMM heteroskedastic-consistent estimates use instruments of lagged ROE, ROA, and Sales variables from t 2,...,t 7. The Sargan statistic tests for over identifying restrictions in the model. Regressions 2 and 3 do not have the Bank ownership variable because of the high correlation with the composite bank-influence variable. *t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365. instrumental variables (GMM IV). All The results in Tables 17.3 and 17.4 reported coefficients are heteroscedastic- provide strong support for the agency consistent parameter estimates.Time dum- hypothesis, though the traditional variables, mies were also included in the regressions. ROE and Sales, do not perform as well as The first equation in each set of equations before. For the regressions with the MB includes the concentration dummies and average salary, ROE has a positive but four ownership dummy variables but non-significant coefficient while Sales is excludes the bank influence variable positive and significant in half of the equa- because it is highly correlated with bank tions but negative and significant in the ownership. The second equation excludes other half of the equations.The bank influ- the concentration dummies and the third ence variable however, has a negative and excludes the ownership dummies. significant coefficient in all four equations EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 461 Table 17.4 Supervisory Board (SB) salary

OLS SB salaries GMM SB salaries Dependent variables 1 2 3 1 2 3

ROE 0.0889 0.0900 0.1180 0.4408** 0.4130** 0.1113 (1.36) (1.36) (1.34) (1.88) (1.74) (0.37) Sales 0.0839* 0.1073* 0.5055* 0.1570* 0.1414* 0.4828* (2.62) (3.44) (35.09) (2.96) (3.01) (26.25) Bank – 4.4983* 0.3908* – 4.3470* 0.0638 influence – (29.54) (4.14) – (18.24) (0.58) Concentration (1 low, 5 high) Conc 5 0.1452 – 0.9720* 0.0365 – 0.8975* (1.31) – (8.32) (0.27) – (6.84) Conc 4 0.3126* – 1.1946* 0.0597 – 1.0521* (2.45) – (7.95) (0.43) – (6.81) Conc 3 0.2517* – 1.3873* 0.1708 – 1.1206* (2.81) – (15.07) (1.57) – (10.94) Conc 2 0.5086* – 0.0178 0.8345* – 0.4230** (3.40) – (0.13) (4.43) – (1.90) Ownership Firm 5.0138* 5.1999* – 5.3364* 5.1337* – (24.01) (27.59) – (15.71) (18.80) – Foreign 6.0995* 6.4531* – 6.6204* 6.5132* – (17.83) (21.87) – (11.63) (14.57) – Bank 4.2251* – – 4.5323* – – (22.52) – – (14.18) – – Family 4.3604* 4.3692* – 4.7789* 4.6326* – (22.80) (28.37) – (14.88) (18.79) – Adj R2 0.7285 0.7200 0.5065 0.7950 0.7865 0.6568 Sargan – – – 403.2899 422.5238 681.5953 (pValue) – – – (0.6655) (0.6972) (0.1247)

All equations use time and firm dummies, and average SB salary. OLS fixed-effects model reports estimates corrected for heteroskedasticity. GMM heteroskedastic-consistent estimates use instruments of lagged ROE, ROA, and sales variables from t 2,...,t 7. The Sargan statistic tests for over identifying restrictions in the model. Regressions 2 and 3 do not have the Bank ownership variable because of the high correlation with the composite bank-influence variable. *t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365. in which it is included. As expected, the and is negative in two, and it is statistically more highly concentrated ownership significant in each case.The bank influence categories 3–5 have negative coefficients in variable is negative in all four equations all but one case and many of the coefficients and is statistically significant in three of are significant. All of the ownership the equations.The ownership concentration dummies are negative as expected and all dummies are negative in all but one case are statistically significant. for the three most concentrated groups and The results are similar for the SB average are significant in many cases. Note that salary regressions reported in Table 17.4. Conc 2 is positive and significant in three of ROE has a positive coefficient in all cases the four cases where it is included. The and is statistically significant in two equa- ownership categories are again negative tions. Sales is positive in four equations and significant in all cases. 462 DIRECTORS’ REMUNERATION Table 17.5 Managing Board (MB) salaries by industry

Electronic and Dependent Metals and Chemicals precision Motor Food and Variables minerals and fiber instruments vehicles tobacco Textiles Other

ROE 1.765 0.024 0.035 0.253 0.063 0.099 1.442* (0.56) (0.56) (0.06) (0.32) (0.35) (0.30) (5.91) Sales 0.0426 0.148* 0.078* 0.181* 0.076** 0.002 0.213* (0.29) (3.32) (5.41) (8.64) (1.74) (0.01) (9.29) Bank 0.657** 0.204 0.948* 0.948* 0.780* 0.204 2.057* influence (1.66) (1.11) (5.36) (10.00) (2.89) (0.12) (19.62) Adj R2 0.5344 0.9472 0.8864 0.8025 0.0966 0.0167 0.5276 Observations 102 221 323 102 235 102 595

All equations use time and firm dummies, and average MB salary. OLS fixed-effects model reports estimates corrected for heteroskedasticity. * t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1683.

Table 17.5 reports regressions for MB the United States. There is evidence that salaries by industry using three independ- banks serve as monitors of executive com- ent variables. While ROE is insignificant in pensation for the MB and some evidence all five equations, Sales is positive and with respect to the SB. Block ownership by statistically significant in three of the five various groups also restrains compensation. industries. Bank influence is negative and The results of this study are consistent significant in five of the seven industries. with findings of Schmid (1997) and indi- Thus, even within most industries banks cate that in a different institutional setting appear to temper compensation when they the same types of economic problems relat- have significant influence. ing to executive compensation hold—that is, there is evidence of agency problems caused by the separation of ownership and 7. CONCLUSION control. When ownership is dispersed management can obtain greater monetary We have analyzed data on the compensa- compensation. Unfortunately, we do not tion of members of the supervisory and have information on non-monetary compen- managing boards of large German corpora- sation, but we might also expect that lack tions in order to satisfy the heightened of control by ownership may enable man- interest in the compensation of German agement to extract greater non-monetary executives. Most of the empirical work on compensation. executive compensation has dealt with the This study shows that the agency impact United States. This study extends one on executive compensation in Germany is important aspect of the research done on similar to that in the United States. In American executives to Germany. We find, addition we find that bank influence as do studies on the US, that the greater the reduces executive compensation and that ownership concentration the less the ability ownership structure can affect executive of executives to extract higher levels of compensation practices. compensation. The agency problem caused by the separation of ownership and control appears to exist in Germany as well as ACKNOWLEDGEMENTS elsewhere. We also find that performance and size are generally positively related to We would like to thank Joachim compensation, similar to previous studies in Schwalbach, Frank Seger, and three EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 463 anonymous referees for their insightful APPENDIX A. MEANS OF BANK suggestions on this research.We would also INFLUENCE VARIABLES like to thank conference participants at the May 2000 Financial Management Bequity: Bank percentage equity ownership Association meetings in Edinburgh of firm. Scotland, the July 2000 Western Economic Bsuper: Percent of bankers on the firm’s Association meetings in San Diego, CA, and SB. the November 2000 Southern Finance Big Votes: Big 3 banks total percentage of Association Meetings in Savannah, GA for votes at shareholder meetings. their useful comments. The first author Bchair: Dummy variable 1 when banker is acknowledges support from the American chair or deputy chair of SB 0 otherwise. Institute for Contemporary German All Votes: Total percentage votes of all Studies for this research. banks at shareholders meetings.

Bequity Bsuper Big Votes Bchair All Votes

All Firms 0.0550 0.0889 0.0952 0.2235 0.2072 (0.1233) (0.0755) (0.1743) (0.4172) (0.3111) Bank influence 0.1748 0.1724 0.3206 0.7342 0.6825 (0.1698) (0.0770) (0.2033) (0.4446) (0.2224) Independent 0.0188 0.0630 0.0271 0.0654 0.0627 (0.0742) (0.0550) (0.0842) (0.7477) (0.1476)

Own calculations from multiple sources. (See Section 5 for details.) Standard deviations in parenthesis. Above data are available for all firms for 1986, except Bequity which is available for all firms over each of the 17 years. A firm is considered bank influenced if: Bequity 25% or if (All Votes 50%) or (25% All Votes 50%) and (Chair or Deputy Chair of the SB is a banker). A firm is Independent if it is not bank influenced. Big 3 banks include: Commerzbank, Deutsche, and Dresdner.

NOTES SB does not appoint the MB. Further, until 1987 only a subset of the GmbHs was even * Corresponding author. Tel.: 1-305-284- required to adhere to public disclosure law, so 1869. that very little information was public before E-mail addresses: [email protected] 1987. For these reasons the GmbH firms are (J.A. Elston), [email protected] (L.G. not included in this study. Goldberg). 5 Appendix A reports descriptive statistics for 1 Tel.: 1-541-322-3165. bank influence variables. Note that there is a 2 Among the most important articles on substantial difference between the independent German compensation are FitzRoy and firms and the bank influenced firms. See Seger Schwalbach (1990), Schwalbach (1991), (1997) for a detailed study of governance Schwalbach and Grasshoff (1997), and Schmid effects on the performance of German firms. (1997). 6 Voting data are available for 1986 only. 3 Kroll finds no sensitivity of SB compensa- Ownership identity and concentration are tion to firm performance. available for each firm for every year of the study. 4 The Gesellschaft mit beschraenkter Haftung 7 As a secondary check for accuracy of the (GmbH) legal identity is also required to have a bank influence measure, we also checked the supervisory and managing board if the number results of this characterization on a firm by firm of employees regularly exceeds 500. However, basis over the sample period and found it the rights of the GmbH SB are not as extensive remarkably consistent with all available as that of the AG board. For example, the GmbH information on bank–firm relationships. 464 DIRECTORS’ REMUNERATION 8 The main sources for the data was the Bonn the dummy variables for groups that have Data, constructed at the Business and ownership control. The highest correlation Economics Institute of the University of Bonn. coefficient was 0.3416 between Conc and Firm. 9 We use unconsolidated data although there The correlation coefficients were respectively is no theoretical reason to believe that the 0.2755 for Foreign, 0.1225 for Bank, and compensation performance relationship is 0.0422 for Family. Thus we do not appear to impacted by the existence of a parent company have a multicollinearity problem. A full correla- nor are we aware of any previous research tion matrix for all relevant variables is available addressing this issue. In fact, Bond et al. (2002) from the authors. compare investment estimations for both consolidated and unconsolidated firms using these data and fail to find a significant differ- REFERENCES ence in estimation results between groups when examining investment sensitivity to liquidity Aktienfuehrer, 1970–1986. Hoppenstedt, constraints. Darmstadt. 10 Our data represent nearly 20% of what Auerbach, P.R.,Siegfried, J.J., 1974. Executive Edwards and Fischer (1994, p. 77), report as compensation and corporation control. the listed stock held by AG firms in Germany Nebraska Journal of Economics and Business in 1980. 11 We use data after 1965 because of the 13, 3–16. Corporation Act of 1965, under which the Bartlett, R.L., Miller, T.I., 1988. Executive accounting rules for the valuation of plant, earnings by gender: A case study. Social equipment, and inventories, as well as profits, Science Quarterly (December), 802–909. were tightened. For example, if BASFs 1981 Berle Jr., A.S., Means, G.C., 1932. The Modern equity was valued under US–SEC rules rather Corporation and Private Property. MacMillan, than under German law, the valuation would be New York. 40% higher than reported according to the new Boehm, J., 1992. Der Einfluss des German rules. Banken auf Grossunternehmen. Duisberger 12 An accounting law provision passed on Volkswirtschaftliche Schriften Band 13, December 12, 1985 that stated that unconsolidated firms must comply with the new standards S W Hamburg. by 1/1/1986. Bond, S., Elston, J.A., Mairesse, J., Mulkay, B., 13 See Edwards and Fischer (1994, p. 77) for 2002. A comparison of empirical investment a detailed discussion on legal incorporations. equations using company panel data for 14 Note that 25% is a key percentage France, Germany, Belgium, and the UK. because it represents a minority blocking vote at Review of Economics and Statistics, shareholders meetings and German law requires forthcoming. disclosure of ownership for any party owning Carroll, T.M., Ciscel, D.C., 1982. The effects of 25% or more of outstanding stock. regulation on executive compensation. The 15 We do not include data from consolidated Review of Economics and Statistics 64 annual reports because it represents a different level of firm activity from non-consolidated firms. (August), 505–509. 16 Note that salary data are reported annu- Chirinko, R., Elston, J.A., 1996. Finance, con- ally by firms only as a total. Individual salaries trol, and profitability: An evaluation of were calculated by dividing the corresponding German Bank influence. Caltech working salary by the number of board members. While paper 96–5. this was the only option available for these esti- Ciscel, D.H., Carroll, T.M., 1980. The determi- mations, it makes particular sense in the nants of executive salaries: An econometric German context where SB members are paid survey. The Review of Economics and the same amount (honorarium) and MB have Statistics 62 (February), 7–13. less variation in salaries as compared to Conyon, M.J., Schwalbach, J., 1999. Executive US firms. 17 Using sales that were lagged produced compensation: Evidence from the United essentially equivalent results and these results Kingdom and Germany. Manuscript, are available from the authors on request. November 1999. 18 We examine the correlation between the Coughlin, A.T., Schmidt, R.M., 1985. Executive summary concentration measure and each of compensation, management turnover, firm EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 465 performance: An empirical investigation. Germany. Law and Policy in International Journal of Accounting and Economics 7, Business 26 (2), 371–405. 43–66. Kroll, L., 1999. Warning: Capitalism is Deckop, J.R., 1988. Determinants of chief contagious. Forbes, May 18, 1999. executive officer compensation. Industrial Leitende Maenner und Frauen der Wirtschaft, and Labor Relations Review 41 (January), 1987. Hoppenstedt, Darmstadt. 215–226. Masson, R.T., 1971. Executive motivations, Demsetz, H., Lehn, K., 1985.The structure of cor- earnings, and consequent equity-performance. porate ownership: Causes and consequences. Journal of Political Economy (October), Journal of Political Economy 93, 1155–1177. 1278–1292. Dyl, E.A., 1988. Corporate control and McGuire, J.W., Chiu, J.S.Y., Elbing, A.O., 1962. management compensation: Evidence on the Executive incomes, sales and profits. The agency problem. Managerial and Decision American Economic Review 52 (September), Economics 9, 21–25. 753–761. Edwards, J.S., Fischer, K., 1994. The German Morck, R., Shleifer, A., Vishney, R.W., 1988. Financial System. Cambridge University Management ownership and market valuation: Press, Cambridge. An empirical analysis. Journal of Financial Elston, J.A., 1998. Investment, liquidity Economics 20, 293–315. constraints and bank relationships: Evidence Murphy, K.J., 1985. Corporate performance and from German manufacturing firms. In: managerial remuneration: An empirical Competition and Convergence in Financial analysis. Journal of Accounting and Markets: The German and Anglo-Saxon Economics 7, 11–42. Models. Advances in Finance, Investment, and Patton, A., 1951. Current practices in executive Banking and Finance Series, vol. 5. Elsevier compensation. Harvard Business Review 29 Science Publishers, Amsterdam. (January), 56–64. Emmons, W.R., Schmid, F.A., 1998. Universal Roberts, D.R., 1956. A general theory of banking, control rights, and corporate finance executive compensation based on statistically in Germany. In: Federal Reserve Bank of tested propositions. Quarterly Journal of St. Louis Review 80 (4), 19–42. Economics 70 (May), 270–294. FitzRoy, F.R.,Schwalbach, J., 1990. Managerial Rubach, M.J., Sebora, T.C., 1998. Comparative compensation and firm performance: Some corporate governance: Competitive implica- evidence from West Germany. WZB, tions of an emerging convergence. Journal of Discussion Paper FS IV 90–20. World Business 33 (2), 167–184. Focus, 1998. No. 49, pp. 290–297. Santerre, R.E., Neun, S.P.,1989. Managerial con- Franks, J., Mayer, C., 1998. Ownership, control trol and executive compensation in the 1930s: A and the performance of German Corporations. reexamination. Quarterly Journal of Business IFA working paper, London. and Economics 28 (Autumn), 100–118. Goldberg, L.G., Idson, T.L., 1995. Executive Santerre, R.E., Neun, S.P., 1986. Stock compensation and agency effects. The dispersion and executive compensation. The Financial Review 30 (2), 313–335. Review of Economics and Statistics 88 Handbuch der Grossunternehmen, 1987. (November), 685–687. Hoppenstedt, Darmstadt. Stigler, G.J., Friedland, C., 1983. The literature Jensen, M.C., Meckling, W.H., 1976. Theory of of economics: The case of berle and means. the firm: Managerial behavior, agency costs, Journal of Law and Economics 26 (June), ownership structure. Journal of Financial 237–268. Economics 3 (October), 305–360. Schmid, F.A., 1997. Vorstandbezuge, Jensen, M.C., Murphy, K.J., 1990. Performance Aufsichtsratsvergutung und Aktionarsstruktur. pay and top-management incentives. Journal Zeitschrift für Betriebswirtschaft 1, 67–83. of Political Economy 98 (April), 225–264. Schwalbach, J., 1991. Principal/agent-problem, Kim, H.-J., 1995. Markets, institutions, and Managervergutung und Unternehmenserfolg. corporate governance: Perspectives from WZB, Discussion Paper FS IV 91–5. 466 DIRECTORS’ REMUNERATION Schwalbach, J., Grasshoff, U., 1997. Weigand, J., Lehmann, E., 1998. Does the Managervergutung und Unternehmenserfolg. governed corporation perform better? Gov- Zeitschrift für Betriebswirtschaft 2, ernance structures and the market for 203–217. corporate control in Germany, Unpublished Seger, F., 1997. Einfluss der Banken auf manuscript. Unternehmenserfolg und Unternehmen- Wer Gehört zu Wem, 1970–1986. Commerzbank, sfinanzierung. Gabler Publishing, Wiesbaden. Germany. Chapter 18

John E. Core and David F.Larcker† PERFORMANCE CONSEQUENCES OF MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP*

Source: Journal of Financial Economics, 64(3) (2001): 317–340.

ABSTRACT

We examine a sample of firms that adopt “target ownership plans”, under which managers are required to own a minimum amount of stock. We find that prior to plan adoption, such firms exhibit low managerial equity ownership and low stock price performance. Managerial equity ownership increases significantly in the two years following plan adoption. We also observe that excess accounting returns and stock returns are higher after the plan is adopted.Thus, for our sample of firms, the required increases in the level of managerial equity ownership result in improvements in firm performance.

1. INTRODUCTION equilibrium behavior of optimizing firms, Demsetz and Lehn assume that firms can DESPITE THE CENTRAL IMPORTANCE OF THE continuously re-contract because there are ISSUE to corporate finance researchers, there no adjustment costs. The choice of one of is no theoretical or empirical consensus on these two extremes drives the design and whether managerial equity ownership interpretation of the results of any study affects firm performance. Studies of this that examines the relation between owner- issue generally take one of two very different ship and performance. It is perhaps not directions, as two seminal studies illustrate. surprising that there is no consensus on the On the one hand, Morck et al. (1988) find performance consequences of managerial that managers’ equity ownership and firm equity ownership. performance is too low for many firms. On We choose an alternative, middle the other hand, Demsetz and Lehn (1985) approach by relaxing some of the strong predict that when firms’ ownership levels assumptions of this prior research. are optimally determined, there will be no Specifically, we assume that firms choose relation between ownership and performance. optimal managerial equity incentives when These two schools of thought make very they contract (consistent with the literature different assumptions about the nature of that predicts no relation between ownership the adjustment costs of correcting subopti- and performance), but that transaction mal contracts. For example, Morck et al. costs prohibit continuous re-contracting implicitly assume that adjustment costs (consistent with the literature that docu- are so great that firms cannot contract ments a strong relation between ownership optimally. Therefore, some firms deliver and performance). Because ownership is poor performance to their shareholders. periodically re-optimized, we expect no Conversely, by concentrating on the association between ownership and firm 468 DIRECTORS’ REMUNERATION performance in a cross-sectional regression in Kole (1996)). Accordingly, firms that controls for the endogenous determi- adopting target ownership plans provide a nants of firms’ optimal ownership levels. unique and powerful sample for examining However,because contracting is not contin- the link between managerial ownership and uous, firms’ ownership levels gradually performance. deviate from the optimal level. We predict The remainder of the paper consists of that firms that are below optimum can five sections. Section 2 provides institutional improve their performance by increasing background on target ownership plans and ownership levels, and that a subset of these develops our three research hypotheses. firms can benefit sufficiently from the Section 3 describes the sample selection increased performance that it is worthwhile process and provides descriptive statistics for them to incur the recontracting costs of on target ownership plans. Section 4 exam- mandating the ownership increase. For this ines the ability of variables that measure sample of firms, required increases in the existence of governance problems to managers’ ownership should strengthen discriminate between adopting and non- firm performance. adopting firms. Section 5 describes the We implement our approach by accounting and stock market performance constructing a sample of firms that adopt consequences associated with target own- requirements specifying the minimum ership plan adoption. A summary of the amount of stock that must be held by paper and concluding remarks are provided executive officers. These contracts are in Section 6. generally termed “target ownership plans.” Before the plan is adopted, these firms deliver low stock returns and have low levels of equity ownership. However, in the 2. INSTITUTIONAL BACKGROUND two years after the board adopts the plan, AND RESEARCH HYPOTHESES managerial stock ownership increases significantly. Finally, excess accounting The section titled “Corporate Governance” returns are statistically higher in the two in Campbell Soup Company’s 1993 proxy years following plan adoption and excess statement illustrates common features of stock price returns are statistically higher target ownership plans: in the first six months of the fiscal year in which the plan is announced. Thus, increases in managerial equity ownership The Company is committed to shareholder- from suboptimal levels appear to result in sensitive corporate governance... By the improvements in firm performance. end of 1994, all officers (29 persons), as One advantage of our approach is that well as approximately 40 other execu- the board of directors is using the target tives, are required to own outright (i.e., ownership plan to mandate increased excluding options and restricted stock) equity ownership by executives. Thus, any Campbell stock valued at one-half to subsequent changes in firm performance three times base salary, depending on are likely to be related to the shifts in man- their positions. (Proxy statement dated agerial incentives brought about by October 8, 1993, p. 6) increases in equity ownership.This approach differs from using a sample in which the Proxy statement disclosures of target top executives voluntarily increase owner- ownership plans vary greatly in the level of ship. If we were to use this design, we could detail provided. The most common plan not be sure if firm performance improved disclosure makes an explicit formal state- because of increased ownership or if equity ment of the minimum level of managerial ownership increased in anticipation of stock ownership, where this ownership performance improvements (e.g., a form of requirement must be satisfied by outright insider trading on private information, as ownership of common stock (i.e., stock MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 469 option holdings do not satisfy the ownership options is delayed.Thus, by explicitly linking requirements). Seven percent of the plans in future equity compensation to a specific our sample express the target as a number minimum ownership goal, these target of shares. The remaining plans express the ownership plans are designed to motivate target as a percentage of salary, so that executives to increase their equity ownership the target becomes more difficult to attain and to maintain this increase. if the stock price decreases. Finally, the We hypothesize that target ownership typical disclosure specifies the maximum plans are adopted when the board of time allowed to achieve the ownership goal. directors recognizes that the firm has a Adopting firms indicate that their governance problem. The board adopts motivation for imposing minimum equity the plan in order to move the firm to a ownership levels is to ensure that their more appropriate governance structure.We managers have the appropriate incentives assume that stock returns that are lower to increase shareholder value. For example, than industry benchmark returns are the 1993 proxy statement of Morrison evidence of potential governance problems. Restaurants Inc. says: The United Shareholders Association (1992) uses this approach to detect Believing that equity ownership plays a governance problems, and the approach is key role in aligning the interests of verified empirically by Core et al. (1999). Company personnel with Company Another way to detect actual or potential stockholders, the Company encourages governance problems is to determine all employees to make a personal invest- whether a firm’s managerial stock ownership ment in Company stock. The Company’s is low relative to some comparison group. goal is that 10 percent of the Common This approach is similar to the method used Stock will be owned by employees by the by governance monitoring groups such as year 2000 and that 80 percent of Institutional Shareholder Services (1993). employees with more than two years of Thus, we predict that if a firm has low experience with the Company will own managerial stock ownership and poor stock Common Stock. (Proxy statement dated price performance, the board infers that the August 26, 1993, p. 12) level of managerial equity is insufficient to motivate good performance. One strategy Thus, target ownership plans are for mitigating this perceived governance designed to address the contention of some problem is to adopt a target ownership researchers and governance activists that plan that requires managers to own a min- stock ownership of senior-level executives is imum level of stock.Thus, our first research “too small” (e.g., Jensen and Murphy, hypothesis is as follows: 1990; Jensen, 1993; Norton, 1995). In addition to the implicit requirement H1. The likelihood of adopting a target created by the public announcement of the ownership plan is negatively related to ownership targets, 52 (27%) of our prior stock price performance and sample firms state an explicit penalty for managerial equity ownership levels. executives who do not meet the ownership target. The penalty falls into one of three We hypothesize that the board acts in categories, each of which involves changing the interests of shareholders to mitigate a some aspect of the executive’s equity perceived governance problem, and for that compensation. When executives do not reason it adopts a “substantive” target meet their targets, either (1) a fraction of ownership plan that mandates ownership their annual cash pay is paid as restricted increases by executives. stock, (2) their grants of options, restricted An alternative to this hypothesis is that stock and cash long-term incentives are management controls the board and reduced or eliminated, or (3) the vesting of convinces the board to adopt a plan that their outstanding restricted stock and does not actually force the managers to 470 DIRECTORS’ REMUNERATION increase their equity ownership. The most firms, and performance will improve purpose of this “symbolic” plan is to make if managerial equity ownership increases. it appear that managers have taken steps For example, McConnell and Servaes to improve governance, but without actu- (1990) find evidence of a positive relation ally having to bear the costs of increased between increases in ownership and firm ownership. This hypothesis assumes that performance so long as managerial owner- managers believe that outside investors are ship is less than 50%. Demsetz and Lehn unable to discriminate between symbolic (1985) predict, and Himmelberg et al. changes and substantive changes (e.g., (1999) find, that there is no relation Pfeffer, 1981). between equilibrium levels of ownership Another explanation for an association and firm performance. In other words, once between target ownership plan adoption the researcher recognizes the endogenous and equity ownership increases is that nature of equity ownership, there is little management has private information about evidence to indicate a relation between future firm performance and encourages equilibrium levels of ownership and firm the board to adopt a target ownership plan. performance (Himmelberg et al.). However, if this were true, we would not In contrast to this prior research, we expect changes in managers’ relative owner- examine firms that are below equilibrium ship positions after plan adoption, because levels of managerial equity ownership. We rational managers would take advantage of predict that as equity ownership rises, there their superior information and increase will be an increase in operating perform- their stockholdings prior to the announce- ance. In addition, if the stock market does ment of the plan. Because managers expect not completely anticipate this contractual to achieve the ownership targets, it will not change, the firm’s stock price will increase matter to them whether the plan includes a when the firm announces to the market penalty for noncompliance. that it has adopted a target ownership To distinguish between these competing plan. Thus, our third research hypothesis is explanations, we examine changes in man- as follows: agerial equity ownership following the adoption of the plan, and our second H3. Adopting a target ownership plan research hypothesis is as follows: will have a positive impact on subsequent operating and stock market per- H2. Managerial equity ownership formance. increases following the adoption of a target ownership plan.

If target ownership plans improve manage- 3. SAMPLE SELECTION AND rial incentives, adoption should have favor- DESCRIPTIVE STATISTICS ON able operating performance consequences TARGET OWNERSHIP PLANS for the firm. In saying this, we assume that when a firm with low ownership requires We identify our initial sample from that managers increase their ownership, selected articles that discuss target owner- this increase mitigates agency problems ship plans (e.g., Brill, 1993; Reese, 1993; and motivates managers to select actions McMillan and Sabow, 1994; Young, 1995) that are more consistent with shareholder and from a keyword search of all proxy objectives. statements on Lexis. We read the proxy The existing literature does not provide statement of each firm in the initial sample consistent evidence on the association to determine whether the firm has actually between managerial equity ownership and adopted a target ownership plan, and we firm performance. The evidence in Morck check each firm’s prior year’s proxy state- et al. (1988) and subsequent studies sug- ment to ensure that we have the proxy that gests that equity ownership is too low at first announces the plan.The earliest target MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 471 ownership plan adopters have December 1992 FYEs, 35.5% have 1993 FYEs, 31, 1991 fiscal year-ends. We include plan 36.9% have 1994 FYEs, and 9.2% have adopters with fiscal years ending up to 1995 FYEs. Table 18.1 describes the final and including December 31, 1995. sample, which comprises 195 firms across Approximately 1.5% of the sample firms 40 different industrial and service sectors have 1991 fiscal year-ends, 16.9% have of the economy.There is some concentration

Table 18.1 Industry composition of firms adopting target ownership plans

The sample consists of 195 firms that adopt target ownership plans.This table lists the number of sample firms in each two-digit standard industrial classification (SIC) code, and a description of that industry SIC code No. of firms Industry description

1 2 Agriculture–Crops 10 1 Mining 13 4 Petroleum and natural gas 16 1 Heavy construction 20 8 Foods 22 2 Textile 23 1 Apparel 24 1 Lumber 25 2 Furniture 26 2 Paper 27 5 Publishing and printing 28 20 Chemical 29 4 Petroleum refining 30 3 Tires and rubber 32 2 Stone, metal, and glass 33 3 Steel and nonferrous metals 34 1 Metal fabricating 35 15 Machinery 36 9 Electrical equipment 37 9 Automobile and aerospace equipment 38 8 Instruments and measuring equipment 39 2 Miscellaneous manufacturing 40 3 Railroad 41 1 Bus transit 44 2 Water transportation 48 9 Telephone and broadcast media 49 12 Electric and gas services 50 3 Durable wholesale 51 5 Nondurable wholesale 52 1 Building materials 53 3 Retail merchandise 57 1 Retail home furniture and equipment 58 3 Eating and drinking places 60 16 Commercial banking 61 6 Credit institutions 63 14 Insurance 64 1 Insurance agents 73 8 Business services 75 1 Automobile rental 80 1 Health services 472 DIRECTORS’ REMUNERATION

Timeline

Year -2 Year -1 Year 0 Year 1 Year 2

Plan adoption Proxy date announcing plan adoption

Figure 18.1 Timeline. in the chemical, machinery, utilities, banking, stock, which is expressed as a multiple of and insurance industries. base salary. Consistent with the prior sur- We measure variables relative to the vey results in Towers Perrin (1993) and fiscal year in which the proxy statement McMillan and Sabow (1994), Panel A announcing the plan adoption appears. shows that for the 138 firms disclosing a Figure 18.1 provides a timeline and illustra- target, the minimum level of ownership for tion. Proxy statements appear a few the median CEO is four times base salary, months after the fiscal year-end. For exam- and the minimum level of ownership for ple, a firm with a December 1993 fiscal other top executives is two and a half times year-end typically issues its proxy in April base salary. For the nine firms that express 1994. We denote the fiscal year in which the target as a number of shares, we con- the proxy statement appears as year 1, and vert the target into a salary multiple using the fiscal year that the proxy describes as the stock price at the end of year 0. Similar year 0. The SEC requires that the year 1 to the survey results in Hewitt Associates proxy statement describe compensation (1993) and Towers Perrin, the typical firm during year 0, ownership at the end of year allows the executives approximately five 0, and the actions of the compensation years to comply with the minimum level of committee for the period following the year stock ownership. 0 proxy statement through the year 1 proxy Panel B, Table 18.2, presents statistics statement. For example, the April 1994 for the actual ownership multiples of the proxy statement would describe compensa- sample firms. We compute the ownership tion payments during the fiscal year ended multiple (value of stock owned/salary) December 31, 1993 (year 0), executive using the salary and shares disclosed in the ownership as of the fiscal year-end proxy statement for year 0, and the stock (December 31, 1993), and the actions of price at the end of year 0.1 The size of our the compensation committee from May sample of target ownership firms is slightly 1993 (year 0) to April 1994 (year 1). It is less than 195 because we cannot compute during this period spanning the latter part the ownership multiple for one CEO (two of year 0 and the early part of year 1 that executive groups) due to insufficient proxy the board decides to adopt a plan. Eighty- disclosure. To compute the ownership vari- two firms (42%) do not specify when the able for the other (four) top executives, we board action to adopt the plan took place, use the data disclosed in the proxy state- 75 (39%) state that the plan was adopted ment for the (five) most highly compen- in year 0, and 38 (19%) indicate that the sated executives, and then exclude the CEO. plan was adopted early in year 1. We calculate a weighted average by sum- Table 18.2 provides descriptive statistics ming the stock values of each executive and for the target ownership plans and actual dividing this by their total salaries. The ownership levels. Recall that the plans median CEO owns 5.6 times his or her require executives to own a target value of salary in stock. The other executives own a MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 473 Table 18.2 Descriptive statistics

The sample consists of 195 firms adopting target ownership plans. Panel A contains descriptive statistics for the ownership multiples (stock value divided by salary) required for the CEOs and the other top executives, and the period allowed for the executives to reach this target. Number is the number of the 195 adopters whose proxy statements provide data. Panel B gives descriptive statistics for the actual multiple for the CEO and the other top executives. For the CEO, the multiple is (value of stock owned/salary). The other executives are the most highly compensated executives as a group, excluding the CEO.The multiple for the other top executives is a weighted average equal to the total value of stock owned by the other executives divided by their total salary. We compute the multiples by using the salary and shares owned as disclosed in the proxy statement for the adoption year 0, and the stock price at the end of year 0. Panel C gives descriptive statistics for the number of firms for which the actual multiple for the CEO and the other top executives is less than the minimum specified by the firm. When no minimum is specified, we impute the minimum using the median values shown in Panel A

Panel A Ownership minimums and compliance periods Number Mean Median Min Max

Ownership minimum multiple of base salary): Chief executive officer (CEO) 138 4.0 4.0 0.4 11.0 Other top executives 116 2.5 2.5 0.1 8.6 Compliance period (years) 103 5.0 5.0 0.0 10.0 Panel B Actual ownership multiples

Actual ownership (multiple of base salary): Chief executive officer (CEO) 194 32.2 5.6 0.0 1,111.7 Other top executives as a group 193 4.7 2.4 0.1 55.0 Panel C Actual ownership multiple is less than minimum multiple

Number Percent below below Number minimum minimum

Chief Executive Officer (CEO) 194 73 38% Other top executives as a group 193 94 49% At least one other top executive 193 156 81% At least one executive 195 163 84% (including CEO)

median of 2.4 times their salary in stock. to the actual plan minimum. To do these Because the distributions are skewed by comparisons, we use the plan targets if some very large observations, the mean disclosed (Panel A, Table 18.2), and if not values are much higher than the medians, disclosed we impute a target equal to the at 32.2 and 4.7 times salary, respectively. medians of 4.0 and 2.5 for the CEO and The median values suggest that it is a other executives, respectively. In Panel C, minority of CEOs who have not already we show that 38% of CEOs are below attained the target, and a minority of other minimum and that 49% of other executives executive teams that have not attained the (on a weighted average basis) are below minimum. To provide more direct evidence, minimum. Because the weighted average we compare the actual ownership multiple calculation can be distorted by one executive 474 DIRECTORS’ REMUNERATION with very large ownership, we also examine enables us to interpret the residual as the whether each other executive meets the percentage by which actual ownership devi- minimum. The results of this analysis indi- ates from expected ownership. We compute cate that for 81% of the firms, at least one this variable for the CEO and for the other other top executive does not meet his or her top executives. target. Accordingly, while only 38% of the Following Demsetz and Lehn (1985) firms have CEOs that do not exceed their and Baker and Hall (1998), we expect that minimums, 84% of the firms have a CEO managerial ownership will increase at a or at least one other top executive who decreasing rate as firm size increases. We does not meet his or her minimum. follow prior researchers in measuring size with the natural logarithm of the market value of equity at the end of year 0, denoted 4. ANALYSIS OF TARGET OWNER- as log(MV equity). Prior researchers such as Murphy (1999) find that executive SHIP PLAN ADOPTION salaries also increase at a decreasing rate as firm size increases. Because log(stock In this section, we describe how we value/salary) is equal to log(stock value)– construct a test of Hypothesis 1, which log(salary), we do not predict the direction predicts that target ownership adopters of the association between this variable and have lower pre-plan returns and ownership log(MV equity). than a control sample. We define our We expect that equity ownership will measures, describe how we construct the increase at a decreasing rate as monitoring control sample, and then describe our test. costs increase.We also follow Demsetz and Lehn in using stock return volatility as a 4.1. Measures for benchmark- proxy for noise that increases monitoring adjusted returns and ownership costs.We capture the hypothesized concave relation between increases in noise and We measure firm performance as the increases in equity ownership by including stock price return in the two years (years -2 the square of stock return volatility. and -1 in Figure 18.1) preceding the adop- We measure stock volatility by using the tion year 0, less the median stock price standard deviation of daily stock price performance during the same time period returns over the six months before the for the firms on the 1998 Compustat file fiscal year-end. Because several of our that have the same two-digit SIC code. We firms did not trade for all of year 0, we label the resulting variable “prior industry- maximize our sample size by measuring adjusted returns”, and we assume that daily stock volatility over the six months governance problems are an inverse function prior to the fiscal year-end. We predict that of prior industry-adjusted returns. stock volatility will be positively associated We compute our ownership benchmark with log(stock value/salary), and that stock by constructing a regression model compa- volatility squared will be negatively rable to that used by Demsetz and Lehn associated with log(stock value/salary). (1985). Because target ownership plans Similar to Smith and Watts (1992), we require executives to own a certain value of expect that CEO equity ownership will be common stock directly (exclusive of stock greater for firms with larger investment options) that is a multiple of that executive’s opportunity sets. Like Smith and Watts, we base salary, we examine the ownership use the book value of assets divided by the multiple computed at year 0 as described market value of assets as a proxy for above.2 To normalize the distribution of growth opportunities, and expect that firms this highly skewed variable, we transform it with greater growth opportunities will have by the natural logarithm and use log(stock lower book-to-market ratios.We measure the value/salary) as the dependent variable in book-to-market ratio at the end of year 0. our ownership regression.This transformation We expect to find a negative relation MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 475 between the book-to-market ratio and of 19.4% and 17.9%, respectively. All of log(stock value/salary). the reported statistical tests are two-tailed. To control for industry factors and other The log of the ownership multiple has a unspecified determinants that might affect statistically positive relation to firm size. the level of equity ownership, we also As we expected, we find a negative relation include 23 industry indicator variables. To to the book-to-market ratio (or a positive capture potential temporal differences, we relation to the investment opportunities include five indicator variables that corre- confronting the firm). We also observe that spond to the years of data collected from the log of the ownership multiple has a our sample of plan adopters and concave relation to the standard deviation ExecuComp. The resulting benchmark of stock returns. We find less managerial model for the level of stock ownership is as ownership for those firms with very low follows: or very high stock volatility. We label the residual from the benchmark model the Log(stock value/salary)it 0 “stock value residual”, and we assume that 1Log(MV equity)it governance problems are an inverse function 2 Stock volatilityit of the stock value residual. 3 Stock volatility squaredit 4.2. Results 4 Book-to-marketit 1...5 year Because Hypothesis 1 predicts differences indicatorsit between adopting and non-adopting firms, 1...23 industry we require a set of control firms that have indicatorsit it . (1) not adopted a target ownership plan. We create a control sample by deleting from We estimate this benchmark model using the sample described above all data for ordinary least squares. We create an 1996 and 1997 and all firms that are estimation sample by pooling annual own- included in our sample of target ownership ership data for the target ownership plan plan adopters. (Because we have the full adopters for years 0, 1, and 2 with sample of target ownership adopters prior ExecuComp data on managerial ownership to 1996, we know that none of our control for firm-years from 1992 to 1997. Because firms have adopted plans.) A total of 4,498 ExecuComp has no 1991 fiscal year data, firm-years on ExecuComp from 1992 to we include the 1991 adopters coded with a 1995 constitute our control sample. year indicator of 1992. The ExecuComp Because some firms are missing data on database meets our requirement for an prior industry-adjusted returns, the target accurate, convenient data source for mana- ownership sample reduces to 170 observa- gerial equity ownership. We include data tions and the control sample to 4,022 for the years 1996 and 1997 so that we observations. Again, to mitigate the influ- can use these residuals in later tests ence of outliers, we set the upper- and (described in Section 5.1) of whether the lower-most percentiles for each independent managerial equity ownership of the 1994 variable equal to the values at the first and and 1995 plan adopters increases in the 99th percentiles in each year, respectively. two years following adoption. To mitigate Univariate results are consistent with the influence of outliers, we set the upper- Hypothesis 1. We find that prior industry- and lower-most percentiles for each variable adjusted returns for the target ownership equal to the values at the first and 99th plan sample are significantly lower than percentiles in each year, respectively. those for the control sample by a mean of Table 18.3 presents the estimation results, 4.5 percentage points (p0.002) and which indicate that the models for the CEO median of 0.7 percentage points and other top executives are statistically (p0.05).3 In addition, the CEO stock significant (p0.001), with adjusted R2s value residual for the target ownership plan 476 DIRECTORS’ REMUNERATION

Table 18.3 OLS regression models of log(stock value/salary)t

This table summarizes regression results from estimating Eq. (1). The sample consists of 7,373 firm-year observations for years 1992–1997 for the target ownership plan adopters and for firms included on ExecuComp. We base t-statistics (in parentheses) on OLS standard errors. Log(CEO stock value/salary) is the natural logarithm of the value of stock owned by the CEO divided by the CEO’s salary. Log(other execs’ stock value/salary) is the natural logarithm of the total value of stock owned by the other executives divided by the total salary of the other executives. The other executives are the most highly compensated executives, excluding the CEO. We compute these variables by using the salary and shares owned as disclosed in the year t proxy statement, and the stock price at the end of year t. We compute all the explanatory variables at or for the period ending at year t. Log(MV equity) is the natural logarithm of the market value of equity. Stock volatility is the standard deviation of daily stock price returns over six months. Stock volatility squared is the square of stock volatility, and book-to-market is the book value of assets divided by the market value of assets. Coefficients on an intercept, five year indicators, and 23 industry indicators are not shown.

Dependent variable

Log (CEO stock Log (other execs’ stock Independent Predicted value/salary) value/salary) variable sign (1) (2)

*** *** Log(MV equity)t ? 0.16 0.23 (8.17) (13.20) *** * Stock volatilityt 5.92 0.96 (9.45) (1.80) Stock volatility 6.06*** 2.13*** squaredt (9.73) (4.02) *** *** Book-to-markett 3.17 2.23 (27.36) (22.55) N 7,373 7,291 Adjusted R2 19.4% 17.9%

***, **, * significant at a 0.01, 0.05, 0.10 level (two-tailed).

sample is statistically lower than that of where we hypothesize that 1 0 and the control sample by a mean of 36.0% 2 0. (p0.002) and median of 32.9% We estimate this model using logistic (p0.001).4 Finally, the other executives’ regression and present the results in stock value residual for the target owner- Table 18.4. In Columns 1 and 2, we see ship plan sample is statistically lower than that the individual coefficient estimates that of the control sample by a mean of (and changes in predicted probability) for 34.2% (p0.002) and median of 35.0% prior industry-adjusted returns and the (p0.002). stock value residual are statistically nega- Using “Plan” as an indicator variable tive at conventional levels. The changes in equal to one if a firm adopts a target own- predicted probability for prior industry- ership plan, and zero otherwise, we can adjusted returns indicate that if prior express the adoption decision as follows: industry-adjusted returns increase from the first quartile to third quartile and the other PLANit 0 1 Prior industry- independent variables remain at their mean adjusted returnsit values, the predicted probability of the firm 2 Stock value residualit having a target ownership plan decreases 1...3 year indicatorsit from 3.8% to 3.0%. Although this uit, (2) decrease seems small, it is arguably more MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 477 Table 18.4 Determinants of the decision to adopt a target ownership plan

This table summarizes estimation results of Eq. (2), which is a logistic model in which the dependent variable is equal to one if a firm adopts a target ownership plan and zero otherwise.The sample consists of 4,192 firm-year observations from 1992 to 1996. Pred. Prob. is the change in the predicted probability that occurs when the independent variable increases from its first to third quartile value, and is evaluated at the mean values of the remaining independent variables. We base t-statistics (in parentheses) on maximum likelihood standard errors. The prior industry-adjusted return is the stock price return in the two years preceding the fiscal year in which the plan is adopted, less the median stock price performance during the same time period for all firms contained in the 1998 Compustat file that have the same two-digit SIC code.The stock value residual is the residual obtained from estimating the model described by Eq. (1).We measure the stock value residual at the end of the year in which the plan is adopted. Coefficients on the intercept and three year indicator variables are not shown.

Coefficient estimate Coefficient estimate Coefficient estimate Independent [Pred. Prob.] (t-statistic) [Pred. Prob.] [Pred. prob.] variable (1) (2) (3)

Prior industry-adjusted 1.23*** 1.26*** 1.23** returns [0.82%] [0.84%] [0.76%] (2.92) (2.98) (2.88) CEO stock value 0.08** 0.05 residual [0.58%] [0.34%] (2.06) (1.22) Other execs’ 0.12** 0.10* stock value residual [0.69%] [0.51%] (2.44) (1.80) N 4,192 4,113 4,113 # adopting 170 168 168 # non-adopting 4,022 3,945 3,945 Pseudo R2 4.1% 4.5% 4.6% Model p-value 0.0001 0.0001 0.0001

***, **, * significant at a 0.01, 0.05, 0.10 level (two-tailed). relevant that it represents a relative decline important determinant of the decision to of 21% in the predicted probability adopt than is low ownership by the CEO. (DeAngelo et al., 2000, p. 341). The We note that there is a 0.39 correlation decreases in predicted probability for CEO between the two residuals, and that we also stock residual (Column 1) and for other obtain a significant coefficient (p0.01) if executives’ stock residual (Column 2) rep- we instead include as a single variable a resent relative declines of 16% and 18%, weighted average of the two residuals, in respectively, in the predicted probability of which the CEO (other executives) residual a target ownership plan. receives a 20% (80%) weight. If we When we include both the CEO and the impute a zero value for missing values of other executives’ stock value residuals in prior industry-adjusted returns (and the same model (Column 3), the coefficient include an indicator variable equal to one on the other executives’ stock residual is when prior industry-adjusted returns are significant (p0.10), but the coefficient on missing), the Table 18.4 results are robust the CEO stock residual is insignificant.This to this change in specification and increase evidence suggests that low ownership by in sample size. The only qualitative change executives other than the CEO is a more is that the coefficient on the other executives’ 478 DIRECTORS’ REMUNERATION stock value residual becomes more whether this difference indicates a significant significant in Column 3 (p0.01). increase. For the sample of firms for which Although not reported, we also obtain we can obtain proxy statement data on qualitatively similar results if we select the CEO ownership two years after plan adop- control sample by using a random match tion (n 174), we observe that the CEO that approximates the proportion of target stock value residual increases by a mean ownership adoptions per year.For example, (median) of 14.1% (27.5%). The mean is because 26.3% of the adoptions occur in marginally significant (p 0.13), and the the Compustat data year 1993, we randomly median is significant at a 0.001 level. For select 26.3% of all firms on the 1993 those firms with data on top executive ExecuComp file as a control group for the ownership changes for the two years after 1993 adoption group, and so forth. Finally, the plan adoption (n 173), we find that our results are robust to our inclusion of the other executives’ stock value residual firm size and industry indicators as increases by a mean (median) of 18.2% additional control variables. (11.7%), which is statistically significant Thus, consistent with Hypothesis 1, our at a 0.07 (0.01) level. results suggest that boards respond to There are three possible limitations to governance problems (as measured by low these results. First, the increases that we relative stock price returns and low relative document above might reflect mean rever- managerial equity ownership) by adopting sion in executive ownership, in which case target ownership plans for senior-level one would expect to observe increases for a executives. sample of firms known to have a low level of ownership relative to the population. In fact, when we regress the change in ownership for the full sample on the beginning 5. CONSEQUENCES OF TARGET residual, we find a negative and significant OWNERSHIP PLAN ADOPTION coefficient on the beginning residual, which means that firms with lower ownership The majority of this section discusses how experience greater subsequent increases. we conduct tests of our hypothesis that We address this limitation by matching performance improves for target ownership each sample firm to the control firm with plan adopters. However, before we examine the closest ownership residual in the the performance consequences of plan adoption year. We choose from among adoption, we must first determine if target those control firms for which we can obtain ownership plans actually cause executives proxy statement data on ownership two to increase their level of stock holdings, as years after the plan adoption. We then we predict in our second hypotheses. If we compute the two-year change in the stock do not find that managerial equity value residual for each sample firm and ownership increases after the adoption of compare it to the two-year change in the target ownership plans, we would not stock value residual for the matched expect to find improvements in performance control firm. associated with the plans. Benchmarking sample firm increases against control firm increases also addresses a second limitation, which is that 5.1. Changes in managerial stock managers’ equity ownership is likely to ownership increase with the time they have spent at the firm. This increase over time can occur To test whether equity ownership increases either because of mechanical reasons following adoption, we use the stock value related to the accumulation of stock residuals described in Section 4.1. We through stock compensation plans (i.e., the examine the difference between the year 2 exercise of stock options) or because of residual and the year 0 residual, and test economic reasons, such as the need to MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 479 prevent horizon problems with CEOs near Moreover, the ownership residuals for retirement. Palia (1998) and Core and Guay these CEOs are no longer statistically (1999) find evidence that CEO ownership different from zero, indicating that these increases with the CEO’s tenure. CEOs’ ownership levels have increased to We find that the sample firm increases equilibrium levels. We obtain qualitatively are significantly greater than those of the the same results if we examine CEO stock control firms. Paired t-tests (z-tests) indi- value residualtenure instead. The sample cate that the sample firm mean (median) firms’ mean (median) increase of 15.6% increase of 14.1% (27.5%) in the CEO (3.2%) is significantly greater than both stock value residual is significantly greater zero and the control firms’ mean (median) than the control firm mean (median) increase of 10.1% (10.1%), and the increase of 9.3% (1.7%) at a 0.04 year 2 ownership residuals for these CEOs (0.02) level. The sample firm mean are not statistically different from zero. (median) increase of 18.2% (11.7%) in This consistent evidence of ownership the other executives’ stock value residual is increases for the CEO and for the other also greater than the control firm mean top executives supports Hypothesis 2 that (median) increase of 12.0% (3.9%) target ownership plans are followed by at a 0.02 (0.02) level. significant increases in managers’ equity As a second means of controlling for ownership. CEO ownership increases over time, we add CEO tenure as an additional control variable in the regression model for log(CEO 5.2. Post-adoption operating stock value/salary) described by Eq. (1).We performance: methodology denote the residual from this model as CEO and results stock value residualtenure. Data on tenure for the other executives are not available from To examine whether plan adoption most firms’ proxy disclosures. Therefore, we improves firm performance, we first look at cannot control for tenure in the ownership whether the accounting return on assets model for the other executives. For the (ROA) is statistically positive over the two sample firms, we observe that the mean years (years 1 and 2) following the adop- (median) CEO stock value residualtenure tion of the target ownership plan. For increase of 12.4% (5.0%) is significantly accounting returns, we use the industry and greater than zero at a 0.12 (0.09) level, and performance match suggested by Barber also significantly greater than the control and Lyon (1996) to develop a comparison firm mean (median) change of 15.8% benchmark. Barber and Lyon find that this (12.1%) at a 0.004 (0.001) level. approach generates well-specified models Another limitation on our results is that that can test future abnormal performance the changes in CEO ownership from year 0 for firms that “...as a group, have histori- to year 2 might be contaminated if a cally experienced especially good or poor sample firm changes CEO during year 1 or performance” (p. 378), or whose perform- year 2. We delete the 42 firms with CEO ance “differs only slightly” from the popu- turnover, and examine separately the sam- lation (p. 396). Using this benchmark is ple of firms with the same CEO from the important, because in the two years prior year of adoption until at least two years to adoption, our sample firms underper- after the plan adoption (n 132). For form the control firms’ ROA by a mean of these firms, we observe that the CEO stock 0.9 percentage points (p0.01) and value residual increases by a mean median of 0.7 percentage points (p0.01). (median) of 39.9% (30.3%), which is sta- Also, by examining the future performance tistically different from zero at a p0.01 of historical poor performers that are less level. This increase is also greater than the likely to survive for two future years, this 12.2% (12.0%) increase of the control matching procedure mitigates any potential firms with the same CEO at a p0.01 level.5 sample selection bias. 480 DIRECTORS’ REMUNERATION To implement this procedure, we select a the ROA next closest to the sample firm comparison firm for each adopting firm. in the year of adoption.We still require that We match firms on their two-digit standard the control firm ROA be within 90% and industrial classification (SIC) codes, and 110% of the sample firm ROA. When we then select that firm with an ROA closest cannot find a matching firm in the same to the sample firm in the year prior to industry, we follow Barber and Lyon adoption.We require that the control firm’s (1996) and select the firm with the closest ROA be within 90% and 110% of the ROA regardless of its SIC code. As dis- sample firm’s ROA. We are unable to cussed below, Vijh (1997) uses a similar match five adopting firms because they splicing method when he computes excess have no Compustat data for year 0. stock returns. Sometimes, either the sample firm or a Although Barber and Lyon (1996) base control firm is missing data in year 1 or their ROA calculation on operating income year 2, either because it was acquired or before depreciation expense (Compustat for some other reason. All of the 190 data item 13), this data item is not avail- sample firms with year 0 data also have able on Compustat for certain financial data available for year 1, but nine of the institutions. To maximize our sample size, firms have no data available for year 2, so we also compute an ROA using operating we do not compute the excess ROA for income after depreciation expense these firms. If a matching firm has no data (Compustat data item 178). available for either year 1 or year 2, we use Table 18.5, Panel A, presents the the firm that is in the same SIC code with accounting performance comparisons. The

Table 18.5 Two-year post-adoption excess operating and stock performance for target ownership firms

The sample consists of 190 target ownership firms. We calculate excess ROA by using the matched-firm approach of Barber and Lyon (1996), where the matching firm is the firm in the same industry with the closest prior operating performance, and by using both operating income after depreciation and operating income before depreciation. We calculate excess stock returns using the matched-firm approach of Barber and Lyon (1997), where each sample firm is matched to the non-sample firm with the closest book-to-market ratio within that subset of firms whose market value lies between 70% and 130% of the sample firm market value.

n Mean p-value Median p-value

Panel A Operating performance

Excess ROA computed using operating income after depreciation: Year 0 190 0.0% 0.552 0.0% 0.321 Year 1 190 1.2% 0.028 0.5% 0.024 Years 1 and 2 181 1.8% 0.017 0.8% 0.002 Excess ROA computed using operating income before depreciation: Year 0 181 0.0% 0.843 0.0% 0.462 Year 1 181 1.2% 0.049 0.6% 0.017 Years 1 and 2 173 1.4% 0.068 0.7% 0.025 Panel B Stock price performance

Excess returns: First six months 190 3.8% 0.086 2.9% 0.041 of year 1 Year 1 190 5.7% 0.161 5.7% 0.160 Years 1 and 2 190 5.3% 0.442 7.9% 0.171 MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 481 first line of each half of the panel indicates zero (p 0.10). These short-window results that our matching procedure is successful do not support Hypothesis 3. in creating a control sample whose prior- Although target ownership plans lead to year performance is insignificantly different improvements in operating performance, we from that of our test sample. In the first might not see excess stock returns in the year after the contractual change and for short time period around the proxy date. the cumulated two years after adoption, the There are two reasons for this. First, target ownership sample has an ROA that although our searches of the Dow Jones is significantly greater than the control and PR newswires reveal no public discus- firms. The adopting firms statistically sion of the plans prior to the proxy date, outperform the ROA (after depreciation) of managers might have privately communi- the benchmark firms by a mean (median) cated the news of the adoption of a target of 1.2 (0.5) percentage points. In the two ownership plan to large shareholders.Thus, years after adoption, the adopting firms by the proxy date, the expected benefits of statistically outperform the compounded the plan might already be impounded into ROA of the benchmark firms by a mean the stock price.7 (median) of 1.8 (0.8) percentage points. Second, even if the market knows that The bottom half of Panel A shows that the plan exists, target ownership plans rep- our results are also robust for a reduced resent an innovation, and the positive or sample for which we use ROA before negative performance consequences of this depreciation. Again, the adopting firms innovation become clear only with the pas- statistically outperform the ROA of the sage of time. In this case, when there are benchmark firms by a mean (median) of improvements in operating performance, 1.2 (0.6) percentage points. In the two the market will be surprised and there will years after adoption, the adopting firms be positive abnormal returns for a period statistically outperform the compounded following the plan adoption. For example, ROA of the benchmark firms by a mean the stock market might react to changes in (median) of 1.4 (0.7) percentage points. managerial investment and financing deci- These results are consistent with sions, rather than to the contractual change Hypothesis 3 that target ownership plans disclosed in the proxy statement. We are followed by significant increases in explore these possibilities by assessing operating performance. whether stock price returns are statistically positive over the six-, 12-, and 24- 5.3. Post-adoption stock month periods starting in the year after performance: methodology and target ownership plan adoption (year 1). results Barber and Lyon (1997) and Lyon et al. (1999) show that the use of a control firm We assess stock price performance in the matched on size and book-to-market in announcement window around the release of computing buy-and-hold excess returns the proxy statement and for the six-, 12-, and (BHERs) for each sample produces test 24-month periods beginning in year 1, the statistics that are well specified in random year after target ownership plan adoption. samples of firms and for almost all non- We first examine stock market returns in the random samples of firms. We use their three-day window surrounding the date of methodology to develop our group of con- the proxy statement in which the target trol firms. We obtain the matching firms ownership plan is disclosed.6 Using standard from the list of all Center for Research in event study methodology (e.g., Brickley, Securities Prices firms with price data as 1986) and the statistical tests described in of the end of June of year 0 (where year 0 Patell (1976), we find that over the three- is the year of target ownership plan adop- day window, the cumulative average excess tion). We compute market value in June of return for the adopting firms is 0.15%, year 0, and the book-to-market ratio by which is statistically indistinguishable from using the last book value reported prior to 482 DIRECTORS’ REMUNERATION June of year 0. Using the subset of firms positive stock value residuals, 70 of the whose market value lies between 70% and firms have other top executives with 130% of the sample firm value, we pair positive stock value residuals, and 38 of the each sample firm with the firm that has the firms have both CEOs and other executives closest book-to-market ratio. (We note that with positive stock value residuals. This our control group is closely matched to our observation raises two issues. First, we test sample: on average the percentage wish to establish that our results are not difference in firm size is 4.0% and the driven by observations that already have percentage difference in the book-to-market appropriate equity incentives (for which we ratio is -0.0%.) We compute the BHERs expect the incentive effects of establishing starting the first month of year 1, the fiscal minimum ownership levels would be lower). year in which the firm releases the proxy A second issue is that if an executive has a statement announcement of the plan. positive stock value residual, the plan could Because the proxy is always released within create perverse incentives by forcing him or the first six months of the fiscal year-end, her to own “too much” stock. all of our BHER measures can capture any We address the first issue in two ways. announcement effects related to the plan First, we delete the 38 firms that have adoption. positive stock value residuals for both the If either a sample or a control firm stops CEO and the other executives as a group. trading because it was acquired (or for We then examine the performance of the some other reason) before the end of our remaining 157 firms and find that these accumulation period, we follow Vijh (1997) firms experience large, significant increases by ending the excess return calculation with in ownership for both the CEO and the the delisting month. If a control firm other executives as a group in the two years delists, we use the return for a firm that has after plan adoption. For this subsample, the book-to-market ratio next closest to the all means and medians of the excess sample firm in the year prior to adoption performance variables remain positive and and a firm size within 70% and 130% of significant, with two exceptions. First, while the sample firm size. excess ROA using operating income after Table 18.5, Panel B, presents the stock depreciation remains significant, excess price performance results.The sample firms ROA using operating income before statistically outperform the control group depreciation loses significance (this vari- for the first six months of the fiscal year of able is available for fewer observations). adoption with a mean (median) BHER of Second, the mean excess stock return for 3.8% (2.9%). Although the BHERs for the first six months becomes insignificant the 12- and 24-month periods are positive, (p 0.152), although the median remains they are insignificant at conventional levels. significant (p 0.056). Thus, our results Combined with our findings of significant are robust to deleting the subset of firms and positive operating performance, the with positive stock value residuals. BHER results suggest that the market Second, the plan may not require an reacts favorably to the adoption of the increase in ownership even if the ownership target ownership plan, and prices its residuals are negative (i.e., the plan is not expected benefits in the six-month period binding). If the plan is not binding on the around plan adoption. These stock price top executives, one might expect the results support Hypothesis 3. incentive effects of establishing minimum ownership levels to be lower.To address this 5.4. Sensitivity analysis issue, we delete from the sample of 157 firms with negative residuals the 15 (42) Our sample comprises firms with relatively firms for which the plan does not bind on at low levels of management stock ownership least the CEO or one of the other top exec- prior to the adoption of the new program. utives (the CEO or the other top executives However, 70 of the firms have CEOs with as a group), resulting in a subsample of MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 483 142 (115) firms. The inference for the ownership plans does not impose excess subsample of 142 firms is the same as that ownership on the executives. Moreover, we reported above for the 157-firm subsample: find no evidence that firms with CEOs or all of the excess performance variables other executives with positive stock value remain positive and the same variables are residuals have excess operating or stock significant. For the smaller 115-firm price performance that is significantly subsample for which the plan is binding on lower than the remainder of the sample. the CEO or the top executives as a group, Finally, we note the results of two addi- the inference is qualitatively similar to that tional sensitivity tests. The two-year CEO reported above for the 157-firm subsam- turnover rate of 24.1% for the target ple: all of the excess performance variables ownership adopters is greater than the remain positive, and the same variables are turnover rate of the control sample of significant except that the mean excess 19%, and the difference is marginally two-year ROA using operating income after significant (p 0.12 from a binomial test depreciation becomes insignificant of differences). Our accounting and stock (p 0.141), although the median remains price performance results are qualitatively significant (p 0.014). Thus, with the the same as in Table 18.5 if we examine exception that excess ROA using operating only those firms with the same CEO over income before depreciation loses signifi- the time period used for developing the cance in these subsamples, our results are performance tests. Finally, we note that we robust to examining the subsamples of have a concentration of financial firms in firms obtained by deleting firms with our sample (as shown in Table 18.1), and nonpositive stock value residuals and then these firms have different accounting by deleting firms without binding plans. conventions and operate in a very specific The finding that a subset of firms has industrial sector. We obtain qualitatively positive ownership residuals raises another the same performance results if we delete question. If a firm’s executives already own the 36 financial firms and examine a large amount of stock prior to the target separately the non-financial adopting firms; ownership plan adoption, the plan could the only difference is that the six-month force the executives to own “too much” excess return is only marginally significant stock, which could create perverse incentives (p 0.15). and lead to lower firm performance. Although this is a possibility, target ownership plans impose a floor on executive own- 6. SUMMARY AND CONCLUSION ership levels, and do not require increases in ownership levels for executives whose We construct a powerful test of the holdings are already above this floor.Of the hypothesis that re-contracting to require 70 sample firms with positive CEO stock managers to increase equity ownership from value residuals, none of the firms’ owner- suboptimal levels will improve incentives ship targets require increases in ownership and increase performance. We implement by the CEO. Similarly, of the 70 sample this test by examining a sample of firms firms with positive other executive stock that adopt minimum ownership levels for value residuals, none of the firms’ ownership executive officers. Because most managers targets require increases by all of the other in our sample have low ownership, these executives. However, eight of the targets plans generally require increases in mana- require increases in ownership by some of gerial ownership. We find that firms that the other executives. Finally, the actual adopt target ownership plans show lower increases in ownership for these firms stock price performance than do their either are significantly less than zero or are industry peers in the time period prior to not different from zero, depending on the plan adoption. These firms also have measure used for the analysis. Thus, it managers who own lower levels of equity appears that the design of these target relative to our benchmark model, which is 484 DIRECTORS’ REMUNERATION similar to that of Demsetz and Lehn E-mail address: [email protected] (1985). We also find that managers (J.E. Core). increase their stock ownership following 1 We use the value of stock disclosed in the the adoption of a target ownership plan. year 0 proxy (rather than in the year -1 proxy) These results indicate that target owner- to maximize the sample size. The disadvantage of the year -1 shareholdings is that they are ship plan adoption is an intervention by the available for a much smaller group of firms. If board of directors to improve incentives we used year -1 numbers, our sample size would and governance. be reduced substantially because some sample More important, we find that excess firms are not public at that time and it is not accounting returns are statistically higher always possible to determine ownership levels in the two years following plan adoption, from 1991 proxies for 1992 adopters. Because and that excess stock price returns are sta- ExecuComp does not have data prior to 1992 tistically higher in the first six months of (i.e., prior to the 1992 reforms in proxy disclo- the fiscal year in which the plan is sure), we also would have no control firms for announced. These results illustrate that our 1992 adopters. Although the board does not know the year 0 shareholdings at the time of when managers with below-equilibrium adoption, the interim ownership numbers that equity ownership are required to increase the board considers when it adopts the plan are their ownership levels, there are improve- probably close to the year 0 numbers. If some ments in firm performance. managers react to the plan adoption in year 0 We contribute to the literature on own- by increasing their stockholdings, our measure ership and performance by suggesting a of ownership at the time of adoption will be too way to reconcile the starkly contrasting large, and this will limit our ability to find predictions and findings of Morck et al. significant differences between the adopters and (1988) and Demsetz and Lehn (1985). non-adopters and to find significant increases Like Demsetz and Lehn, we assume that after adoption. 2 While our measure of ownership is consistent firms optimize ownership levels when they with the measure used in the target ownership contract, and that at the optimum there is plans, prior research has concentrated on either no association between ownership and firm the fraction of the firm owned or the value of performance. Like Morck et al. we expect ownership (not deflated by base salary). In that some firms are below optimum and addition, arguably a better proxy for managerial that their performance can be improved by equity incentives would also include the increasing ownership levels. Consistent incentives provided by the manager’s options with our approach, we find that mandatory (e.g., Core and Guay, 1999). If we use instead increases in suboptimal equity ownership the logarithm of the value of managerial stock are associated with increases in subsequent ownership or the logarithm of the total incentives provided by the manager’s stock and firm performance. option portfolio, our results are qualitatively the same as those discussed below. In particular, the adopting firms’ managers have significantly NOTES lower equity incentives than the comparison group, and these incentives increase significantly * A previous version was titled “Performance in the two years following adoption. Consequences of Mandatory Increases in CEO 3 A “percentage point” difference is the Stock Ownership”. We gratefully acknowledge difference in two returns, e.g., a 1% return is the helpful comments of Stanley Baiman, 1 percentage point lower than a 2% return. Wayne Guay, Paul Healy (the referee), Robert 4 As noted above, because our dependent Holthausen, Christopher Ittner, and Rebecca variable is log(stock value/salary), we can inter- Tsui. We thank Howard Yeh and Christine pret the residual as the percentage by which Phillips for research assistance. We gratefully actual ownership deviates from expected owner- acknowledge the help of John Moyer of Ernst & ship. Further,we can interpret the differences in Young LLP and the financial support of the two residuals as percentage differences, e.g., we Wharton School. term the difference in the -0.336 mean residual † Corresponding author. Tel.: 1–215–898– for the adopters and the 0.026 mean residual for 4821; fax: 1–215–573–2054. the control samples as a 36.2% difference. MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 485 5 The other executives’ stock value residual Core, J., Holthausen, R., Larcker, D., 1999. for the 130 of these observations with available Corporate governance, CEO compensation, data increased by a mean (median) of 16.5% and firm performance. Journal of Financial (14.2%), which is significantly greater than Economics 51, 371–406. both zero and the -19.9% ( 5.6%) change for DeAngelo, H., DeAngelo, L., Skinner, D., 2000. the control sample. Special dividends and the evolution of divi- 6 There is some debate on the desirability of using proxy statement release dates for dend signaling. Journal of Financial detecting the shareholder value consequences Economics 57, 309–354. associated with changes in compensation Demsetz, H., Lehn, K., 1985. The structure of contracts; Gaver et al. (1992) discuss this point. corporate ownership: causes and conse- 7 As noted above, most of the target owner- quences. Journal of Political Economy 93, ship plans are adopted in year 0, which raises 1155–1177. the possibility that the leakage could have Gaver,J., Gaver,K., Battistel, G., 1992.The stock occurred prior to the start of our event window market reaction to performance plan adop- at the beginning of year 1. To address this tion.The Accounting Review 67, 172–182. possibility, we examine excess returns for the six Hewitt Associates, 1993. Survey Findings: months ending with the start of year 1 for the set of firms (with available return data) that do Executive Stock Ownership Guidelines. not state that they adopted early in year 0. We Hewitt Associates, Lincolnshire, Ill. find no significant excess returns for either the Himmelberg, C., Hubbard, G., Palia, D., 1999. sample of 73 firms that specifically disclose Understanding the determinants of managerial that they adopted in year 0, or for the 152-firm ownership and the link between ownership subsample that does not state that they adopted and performance. Journal of Financial in year 1.This finding is consistent with the lack Economics 53, 353–384. of disclosure of the plans and with the Institutional Shareholder Services, 1993. The conjecture that the board action to adopt these ISS Proxy Voting Manual. Institutional plans occurs late in year 0. Shareholder Services, Washington, DC. Jensen, M., 1993. The modern industrial revolution, exit, and the failure of internal con- REFERENCES trol systems. Journal of Finance 48, 831–880. Jensen, M., Murphy, K., 1990. Performance pay Baker, G., Hall, B., 1998. CEO incentives and and top-management incentives. Journal of firm size. NBER Working paper 6868. Political Economy 98, 225–264. Barber, B., Lyon, J., 1996. Detecting abnormal Kole, S., 1996. Managerial incentives and firm operating performance: the empirical power performance: incentives or rewards? and specification of test statistics. Journal of Advances in Financial Economics 2, Financial Economics 41, 359–399. 119–149. Barber, B., Lyon, J., 1997. Detecting long-run Lyon, J., Barber, B., Tsai, C., 1999. Improved stock returns: the empirical power and methods for tests of long-run abnormal stock specification of test statistics. Journal of returns. Journal of Finance 54, 341–372. Financial Economics 43, 341–372. McConnell, J., Servaes, H., 1990. Additional Brickley, J., 1986. Interpreting common stock evidence on equity ownership and corporate returns around proxy statement disclosures value. Journal of Financial Economics 27, and annual shareholder meetings. Journal of 595–612. Financial and Quantitative Analysis 21, McMillan, J., Sabow, S., 1994. Encouraging 343–349. executive stock ownership. The Corporate Brill, J., 1993. Stock ownership guidelines for Board (January/February), 16–20. executives. The Corporate Executive Morck, R., Shleifer, A., Vishny, R., 1988. (March–April), 381–385. Management ownership and market valua- Core, J., Guay, W., 1999. The use of equity tion: an empirical analysis. Journal of grants to manage optimal equity incentive Financial Economics 20, 293–315. levels. Journal of Accounting and Economics Murphy, K., 1999. Executive compensation. In: 28, 151–184. Ashenfelter, O., Card, D. (Eds.), Handbook of 486 DIRECTORS’ REMUNERATION Labor Economics, Vol. 3. North-Holland, Smith, C., Watts, R., 1992. The investment Amsterdam. opportunity set and corporate financing, Norton, L., 1995. Velvet handcuffs. Barron’s dividends and compensation policies. Journal (July 24), 21–24. of Financial Economics 32, 263–292. Palia, D., 1998. The endogeneity of managerial Towers, P., 1993. Stocking up in the 90s: how compensation in firm valuation: a solution. companies are creating managers/owners. Working paper, Columbia University. CompScan. Patell, J., 1976. Corporate forecasts of earnings United Shareholders Association, 1992. 1992 per share and stock price behavior: empirical Shareholder 1,000: rating corporate tests. Journal of Accounting Research 14, America’s responsiveness to shareholders. 246–276. Vijh, A., 1997. Long-term returns for equity Pfeffer, J., 1981. Power in Organizations. carveouts. Journal of Financial Economics Pitman, Boston. 51, 273–308. Reese, J., 1993. Buy stock—Or die. Fortune Young, L., 1995. The owner mentality. The Wall (August 23), 14–15. Street Journal (April 12), R12. Part 5 Governance, performance and financial strategy

INTRODUCTION

HE PAPERS IN THIS PART DEAL WITH THE IMPACT of governance structure on Tcorporate performance, its investment and financing strategy.

Corporate governance and performance Erik Lehmann and Jürgen Weigand (Ch.19) focus on the relationship between governance structure in Germany and corporate performance. As noted above, Germany has a unique corporate governance structure and its performance effects are therefore a matter of considerable interest in evaluating the comparative merits of different corporate governance systems. Although there has been an intensive debate on the relative merits of different systems of corporate governance, empirical evidence on the link between corporate governance and firm performance almost exclusively refers to the market-oriented Anglo- American system. This paper, therefore, investigates the more network- or bank-oriented German system. In panel regressions for 361 German corporations over the time period 1991 to 1996, significantly the authors find ownership concentration to negatively affect profitability. However,this effect depends intricately on stock market exposure, the location of control rights, and the time horizon (short-run vs. long-run).They conclude that (1) the presence of large shareholders does not necessarily enhance profitability, (2) ownership concentration seems to be sub-optimal for many German corporations, and, finally, (3) having financial institutions as largest shareholders of traded corporations improves corporate performance. Corporate governance is manifested in a variety of ways and different metrics of corporate governance may offer different, even conflicting, perspectives on the effectiveness of governance. A composite metric that integrates these disparate metrics is therefore a useful step in establishing the relationship between corporate governance and performance. Shareholder rights vis-à-vis the firm are major pillars of corporate governance but vary widely across countries and even within a single country. Paul Gompers, Joy Ishii and Andrew Metrick (Ch.20) measure the level of corporate governance on the basis of shareholder rights that vary across firms in the US. Using the incidence of 24 governance rules, the authors construct a ‘Governance Index’ to proxy for the level of shareholder rights in about 1,500 large US firms during the 1990s. An investment strategy that bought firms in the lowest decile of the index (strongest rights) and sold firms in the highest decile of the index (weakest rights) would have earned abnormal returns of 8.5 percent per year during the sample period. They found that firms with stronger shareholder rights had higher firm value, higher profits, higher sales growth, lower capital expenditures, and made fewer corporate acquisitions. This study represents a landmark in the way corporate 488 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY governance is quantified and its methodology can be extended to other countries and to other aspects of corporate governance such as accounting disclosures and creditor rights.

Corporate governance and corporate investment strategy The governance structure of strategic investments and its impact on both the investment itself and the investing firms is a matter of interest to firms. Often strategic investments may fail due to poor governance structures as evidenced by numerous studies on post- merger integration (Sudarsanam, 2003, ch.22). Such governance problems also plague joint ventures and strategic alliances (Sudarsanam, 2003, ch.10). Gertjan Schut and Ruud van Frederikslust (Ch.21) examine the shareholder wealth effects of 233 joint venture announcements of Dutch public companies in the period 1987 till 1998.They find that on average, establishing joint ventures has a positive effect on the market value of Dutch companies. Using the strategic characteristics of joint ventures it is possible to explain and understand these wealth effects. The authors show that strategic intention, the context in which the strategy is unfolded and the extent to which the company has ownership and managerial control over the implementation of the joint venture, strongly explains the extent to which it can create value. Financial distress is another context in which the influence of corporate governance on the strategic and other decisions a firm makes can be studied. Firms in performance decline may choose a variety of restructuring strategies for recovery with conflicting welfare implications for different stakeholders such as shareholders, lenders and managers. Choice of recovery strategies is therefore determined by the complex interplay of ownership structure, corporate governance and lender monitoring of such firms. For a sample of 297 UK firms experiencing relative stock return decline during 1987–1993, Jim Lai and Sudi Sudarsanam (Ch.22) examine the impact of these factors as well as other control factors on their turnaround strategies. Strategy choices during the decline year and two post-decline years are modelled with logit regressions. The results show that turnaround strategy choices are significantly influenced by both agency and control variables. While there is agreement among stakeholders on certain strategies there is also evidence of conflict of interests among them. There is further evidence of shifting coalitions of stakeholders for or against certain strategies. This paper points to the rich and complex ways in which different corporate governance mechanisms interact, sometimes conflicting with and, at others, complementing one another.

Corporate governance and corporate financing strategy Retained profits may create the problem of free cash flow which entrenched managers may invest in value destroying projects such as ambitious acquisitions or reckless R&D. Whether they are allowed to do so or prevented by being forced to return free cash flow to shareholders depends at least partly on the effectiveness of corporate governance. Jorge Farinha (Ch.23) tests the agency explanation for the cross-sectional variation of corporate dividend policy in the UK by looking at the managerial entrenchment hypothesis. Consistent with its predictions, a significant U-shaped relationship between dividend payout ratios and insider ownership is observed for a large sample of over 600 UK companies and two distinct periods.These results strongly suggest the possibility of managerial entrenchment until insider ownership reaches a threshold of around 30 per cent. Beyond this level managers seek to offset the higher agency costs by increasing payout. Evidence is also presented that non-beneficial holdings by insiders can lead to entrenchment in conjunction with shares held beneficially. Farinha also provides evidence, as do Dahya et al. (see Part 3, Ch.13), that compliance with the Cadbury Code after 1992 results in GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 489 corporate decisions more aligned to shareholder interests, thus testifying to the greater effectiveness of the governance regime introduced by the Code. Entrenched managers may not only enjoy private benefits of control at the expense of shareholders and capture the pay setting process to award themselves excessive compensation but also prevent shareholders from finding out what they are up to. The earnings they report may have been ‘managed’ and therefore lack transparency. Whether such creative accounting is feasible depends on the corporate governance structure including shareholder rights (see Gompers et al., (Ch.20) on shareholder rights as a measure of corporate governance). Christian Leuz, Dhananjay Nanda and Peter Wysocki (Ch.24) propose an explanation for the systematic differences they observe in earnings management across 31 countries. Insiders, in an attempt to protect their private control benefits, use earnings management to conceal firm performance from outsiders. Thus, earnings management is expected to decrease in investor protection because strong protection limits insiders’ ability to acquire private control benefits, which reduces their incentives to mask firm performance.The findings are consistent with this prediction and suggest an endogenous link between corporate governance and the quality of reported earnings.

REFERENCE

Sudarsanam, P. Sudi (2003), Creating value from mergers and acquisitions. The challenges, an international and integrated perspective, New York: FT Prentice Hall.

Chapter 19

Erik Lehmann and Jürgen Weigand DOES THE GOVERNED CORPORATION PERFORM BETTER? GOVERNANCE STRUCTURES AND CORPORATE PERFORMANCE IN GERMANY

Source: European Finance Review, 4(2) (2000): 157–195.

ABSTRACT Although there has been an intensive debate on the relative merits of different systems of corporate governance, empirical evidence on the link between corporate governance and firm performance almost exclusively refers to the market-oriented Anglo-Saxon system.This paper therefore investigates the more network- or bank-oriented German system. In panel regressions for 361 German corporations over the time period 1991 to 1996, we find ownership concentration to affect profitability significantly negatively. However,this effect depends intricately on stock market exposure, the location of control rights, and the time horizon (short-run vs. long- run). We conclude from our results that (1) the presence of large shareholders does not necessarily enhance profitability, (2) ownership concentration seems to be sub-optimal for many German corporations, and, finally, (3) having financial institutions as largest shareholders of traded corporations improves corporate performance.

Shareholders are stupid and impertinent – stupid, because they give their funds to some- body else without adequate control, and impertinent, because they clamor for a dividend as a reward for their stupidity. Carl Fürstenberg (1850–1933), German financier

1. INTRODUCTION and performance among academics and business press. As firms face new challenges EVER SINCE BERLE AND MEANS (1932) from increased cross-border competition, STATED THAT in the modern corporation pressures to adapt to an internationally hired managers have enough discretion for integrated environment mount.The question corporate plundering, the issue of separating arises naturally whether established systems ownership from control and its resulting of corporate finance and corporate gover- impact on corporate performance has been nance are still appropriate to cope with the placed high on the agenda of economists. challenges ahead.1 Globalizing product and financial markets Ownership structures are a central have recently triggered renewed interest distinguishing feature of financial systems in the link between corporate governance (Mayer,1992; Moerland, 1995). Particular 492 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY attention has been paid in the corporate thwarted any serious attempts of hostile governance literature to ownership concen- takeovers in Germany.6 In view of high unem- tration as a key to more effective corporate ployment and sluggish growth, critics have governance and shareholder value maximiza- pointed to the German network-orientation tion. The presence of large shareholders as a root cause for entrepreneurial inertia, may curb managerial discretion, reduce risk aversion, and low investment in emerging agency costs and enhance performance new technologies and infant industries.7 (Stiglitz, 1985; Shleifer and Vishny, 1986). In the remainder of the paper we discuss The existing empirical evidence on the impact in Section 2 the link between corporate of ownership structures on corporate per- governance and firm performance in more formance refers almost exclusively to detail. Section 3 presents our empirical Anglo-Saxon firms and is rather mixed.2 analysis that employs a panel data set of This paper therefore focuses on German 361 German corporations over the time corporations. We investigate empirically period 1991 to 1996. Our main finding is a how ownership concentration, the location significantly negative effect of ownership of control rights, board representation of concentration on profitability. However,this owners, and stock market exposure affect effect depends intricately on stock market firm profitability (return on total assets). exposure, the location of control rights, and Ownership concentration is low in the the time horizon chosen (short-run vs. long- Anglo-Saxon countries which rely heavily run). Section 4 summarizes and concludes. on stock markets to channel the flow of capital, control its efficient use, and assure outside investors of maximizing the return 2. CORPORATE GOVERNANCE AND on their investments.3 The Anglo-Saxon FIRM PERFORMANCE financial system has been criticized for short-termism, neglect of interests other 2.1. The governed versus the than shareholders’, and inefficiency in managed corporation delivering effective corporate governance.4 By contrast, concentrated ownership is The separation of ownership a salient feature of the German system. and control Non-incorporated small and medium-sized firms tightly held by individuals or families A prime element of corporate governance is dominate the firm population. German the alignment of shareholders’ interests corporations are typically in the hands of with the interests of managers hired to run large blockholders who often command a the firm. The major concern is whether super-majority interest. Widely dispersed managers pursue their own interests (pet outside shareholdings as in the USA or UK projects, empire building, perks etc.) rather are rare. Ownership structures hardly change than maximize shareholder value. Public over time, implying that large shareholders corporations with widely dispersed outside stick to their blockholdings even in bad times. shareholdings may be particularly prone to An active market for corporate control does managerial discretion. Such corporations not exist yet despite the recent takeover- have a clear separation of control and battle between Mannesmann and Vodafone. ownership. Shareholders delegate decision- Although improving lately, the German making to managers and control to a stock market is still relatively small supervisory board. Managers are put in regarding listings and market capitaliza- place because it is expected that their tion.5 Close ties between industrial firms superior entrepreneurial skills allow them and financial institutions (banks) (e.g., via to better achieve success for the firm than cross-holdings, long-term lender-borrower the owners. The supervisory board is relations) apparently foster access to debt responsible for selecting, monitoring, and capital, thus reducing the need to attract replacing senior managers. equity capital on the stock market. This An extensive literature has discussed the network-like structure has effectively pros and cons of separating ownership from DOES THE GOVERNED CORPORATION PERFORM BETTER? 493 control and whether governance of such a governed corporation. For a corporation “managed” (manager-controlled) corpora- to be governed, investors must be different tion is inherently ineffective.8 Corporate from the investors of the managed supervisory boards are suspected of being corporation. The model of the governed inefficient or ‘entrenched’ monitors who corporation rests on committed owners only take action when true performance who actively participate in governing the disasters have already happened (Warner firm. Commitment means that investors et al., 1988).9 In principle, if the corporation do not sell out quickly in times of trouble. does not perform as expected, shareholders To guarantee that the company is soundly can oust the supervisory board in a joint managed, they participate in selecting the voting effort. However,coordinating a large top management and initiate replacements number of shareholders for joint voting is in case of inferior performance. difficult. The incentive to monitor and take For having one’s interests and concerns action against the board will be low for respected, the “active” (Jensen 1993, shareholders who command only a negligible p. 866) or “relationship” (Thompson, share in a firm. The costs of such efforts 1998, p. 27) investor needs to be a large usually outweigh the individual benefits. shareholder,that is, he must have sufficient Moreover, a shareholder gains from any control over the firm’s assets.10 Only other shareholder’s control effort without investors who control a substantial part having to contribute to the incurred costs. of the voting capital will be able to keep This free-rider problem makes it unattractive managers from diverting free cash flow into for a small shareholder to exercise and pet projects and force them to distribute enforce voting rights. Thus, managerial profits to shareholders.The incentive to take mistakes can “go uncorrected” in the action should increase with the investor’s managed corporation “until they become wealth commitment (Shleifer and Vishny, catastrophes” (Pound, 1995, p. 92). When 1986; Huddart, 1993; Admati et al., 1994). markets for equity capital are highly liquid, The model of the governed corporation dissatisfied shareholders can easily sell off thus suggests concentrating the stakes in their holdings. Therefore, shareholders of a firm in the hands of only a few share- the managed corporation rather favor a holders. The underlying hypothesis is that “cheap ‘exit’” over an “expensive ‘voice’” by re-integrating ownership and control (Bhide, 1994, p. 132) in times of crisis. corporate performance (profitability, produc- In active markets for corporate control, tivity, innovative thrust etc.) is going to be hostile takeovers may provide an effective enhanced. mechanism to reign in free-wheeling man- Demsetz (1983) has argued that agers and sanction underperforming firms concentrated ownership is no safeguard (Manne, 1965; Jensen and Ruback, 1983). against corporate plundering: “Where is it However, as shown by Grossman and Hart written that the owner-manager of a closely (1980), the free-rider problem also besets held firm prefers to consume only at the takeover mechanism. The threat of home?” (ibid., p. 381). Contesting the shareholders’ cheap exit by selling off to a model of the firm on which both the raider may thus be insufficient to discipline managerial discretion hypothesis and management. In addition, incumbent the agency approach rest, Demsetz states, managers can apply anti-takeover strate- “It is clearly an error to suppose that a gies to entrench their positions (Stulz, firm managed by its only owner comes 1988; Shleifer and Vishny, 1988; Jensen, closest to the profit-maximizing firm 1988; Fluck, 1999). postulated in the model firm of economic theory” (ibid., p. 383). The governed corporation Theoretically, the impact of ownership concentration on firm value is indeed not In view of these problems, Pound (1995) definite. Shleifer and Vishny (1986) show has suggested replacing the managed that concentrated shareholdings raise firm corporation with what he terms the value. In Huddart (1993) and Admati et al. 494 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY (1994) this outcome is reversed because depend on who they are. In other words, the of risk-averse large shareholders. Recent location of control rights can be a more contributions by Pagano and Röell (1998) important determinant of the degree of and Bolton and von Thadden (1998a,b) control exerted by owners than ownership stress the existence of a trade-off between concentration. Insider (internal) control control and liquidity. As the incentive and may represent a higher degree of control the ability to monitor increase with share at any given level of blockholdings than concentration, control will be more stringent outsider (external) control (Cubbin and and effective but this comes at the cost of Leech, 1983; Mayer, 1992). Family inter- liquidity. Constraining liquidity could affect ests, allied industrial firms, banks, and profitability negatively if the firm cannot holdings companies are frequently identi- fully exploit its investment opportunities. fied as insiders, while the shareholders of This reasoning may only apply to smaller diffusely held firms (e.g., institutional firms with the intention to go public. In investors such as pensions funds) are general, mature large corporations hardly viewed as outsiders. face financing constraints and finance It is not obvious though that insider investment overwhelmingly from retained control enhances performance. For instance, earnings.11 Cash-flush corporations may individuals or families are often large thus rather overinvest to keep funds in the blockholders because they are the founders firm instead of “wasting” them on the of the company or the heirs. Given an shareholders (Jensen, 1986; Hellwig, 2000). emotional involvement, these owners may Strong owners could prevent managers be more strongly interested in the success from stashing money away and thus reduce of the firm than investors for which the both free cash flow and inefficient invest- firm is just one piece in their portfolios. ments. Profitability should then be enhanced, A large shareholder “personally” attached but this is not for sure, since large share- to the firm possibly motivates managers holders may overdo monitoring. As shown and employees to more commitment and by Burkart, Gromb, and Panunzi (1997), determination so that firm performance too much monitoring can stifle managerial improves.The opposite may hold if running initiative. The governed corporation may the firm is impeded because the owners then miss out on profitable investment engage in infighting or take on a know-all opportunities. attitude and try to “mastermind” decision- Large shareholders can inflict substantial making. In this respect, investors without a costs on other shareholders as well as stake- personal interest may be better for corporate holders (managers, employees, creditors) in performance. Banks can be such neutral the form of an expropriation-like redistrib- investors. If they are more efficient monitors ution of wealth if their interests diverge (Diamond, 1984, 1991) and reduce agency from those of other investors or stakeholders costs, performance should be enhanced. (Shleifer and Vishny, 1997). Further, large Their actual governance role, either as own- share- or stakeholders and managers may ers, board representatives or proxy voters, collude to keep minority shareholders at bay and the impact on corporate performance (Hellwig, 2000). Finally, large shareholders has been disputed.12 Not much is known closely associated with the management about industrial firms and holdings as may refrain from attempts to dismiss blockholders of other industrial firms. underperforming managers if such an Empirical evidence on large diversified US intervention puts their own reputation at corporations (Lang and Stulz, 1994; Shin risk (Mayer, 1996). and Stulz, 1998) suggests that members of business groups may underperform because The identity of owners internal capital markets do not work efficiently. In short, it is open to empirical The commitment of owners and their testing whether the identity of owners willingness to intervene may crucially really matters. DOES THE GOVERNED CORPORATION PERFORM BETTER? 495 Product market competition and affect their performance”. Mayer (1996, endogenous ownership structures p. 17) interprets the empirical evidence as implying “benefits in the exercise of The mode of corporate governance may be corporate governance from modest levels less important if there is intensive competi- of concentrations of ownership”, but tion in product markets (e.g., Demsetz, “exploitation of private benefits” at high 1983; Hart, 1983; Schmidt, 1997; Allen levels of ownership concentration. and Gale, 1998). Competition forces firms Considering the high ownership concen- to adopt not only cost minimizing methods tration and the presence of presumably of production but also efficient governance strong insiders such as allied industrial and organization structures. More efficient firms, banks and families, many German firms will steal business from slack firms corporations seem to come close to the and drive the least efficient firms out of the model of the governed corporation. Only a market. At least in the longer run when few empirical studies have focused on entry is facilitated, this selection effect of German corporations so far to investigate competition eliminates inefficient structures the impact of governance structures on and generates effective governance. In the firm performance. The evidence from longer run, the mode of corporate governance regression analyses is similarly ambiguous may be simultaneously determined with as for Anglo-Saxon firms.13 firm performance by the forces of competi- In a pioneering study, Thonet and tive markets (Demsetz, 1983; Demsetz and Poensgen (1979) found significantly lower Lehn, 1985). returns on equity for owner-controlled than for manager-controlled German public 14 Conclusion companies. Other studies focused on banks as blockholders of industrial firms. The debate on the managed versus the Cable (1985),Weigand (1999) and Gorton governed corporation, or the insider versus and Schmid (2000) report a significantly the outsider model of the corporate gover- positive impact of bank involvement on nance, has generated conflicting hypotheses firm profitability, whereas profitability concerning the link between ownership, differences between bank- and non-bank control, and firm performance. The model firms are not significant in Chirinko and of the governed corporation suggests that Elston (1996).15 The positive influence of the tightly-held, insider-controlled firm banks is in line with the standard reasoning outperforms the managed, diffusely-held that banks are better monitors and thus firm. As there are costs of having large firm performance is enhanced. However, it shareholders, ownership concentration or may very well be that banks were not better increased monitoring through the owners in governing but in “picking winners”. may be beneficial only up to a certain Franks and Mayer (1997), analyzing extent. At a given level of wealth commit- quoted corporations only, and Goergen ment, the willingness of owners to control (1999), focusing on IPOs, could not find a may also be dependent on who they are. It significant impact of ownership indicators is thus an empirical question which of the on firm performance. Franks and Mayer advanced hypotheses are valid. interpret their results as inconsistent with the view that the German market of 2.2. Empirical evidence corporate control performs a disciplining function. Rather concentration of owner- The existing empirical evidence does not ship seems to further rent extraction. allow for clear-cut answers. Short’s survey Goergen views his results as support for the (1994, p. 227) finds no “conclusive evi- Demsetz (1983) hypothesis that ownership dence either in support of, or in opposition structures are chosen as to maximize firm to, the hypothesis that the ownership and value.16 Gedajlovic and Shapiro (1998) control structures of firms materially find a significantly negative and non-linear 496 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY impact of ownership concentration on the reporting firms with complete and consistent return on total assets for a sample of large data. According to their main field of quoted corporations. Thus profitability economic activity, the sample firms were first decreases in ownership concentration assigned to 30 different two-digit industries, and then, at higher levels of concentration, among them machinery (76 firms), chemi- rises again.17 Becht (1999) reports a cals and pharmaceuticals (60 firms), the negative effect of ownership concentration electronic products industry (56 firms), on liquidity (measured by the ratio of and iron and steel (37 firms). turnover to market capitalization) which Our preferred measure of corporate supports the hypothesis of a control-liquidity performance is the return on total assets trade-off.18 (ROA). We also report summary statistics In sum, “governed” (more concentrated for the return on equity (ROE) but ROE or owner-controlled) German firms seem to comparisons across firms may be distorted have enjoyed higher returns during the by the leverage effect and differences in 1970s and early 1980s (Cable, Gorton and the user cost of capital. For both ROA Schmid, Weigand). The positive impact and ROE the numerator is gross profits, vanishes or turns even negative when data calculated as turnover minus expenses for from the late 1980s and the 1990s are personnel and materials.19 Equity capital included (Becht, Chirinko and Elston, is defined as shareholders’ equity plus Franks and Mayer, Goergen, Gedajlovic reserves and pension liabilities.20 and Shapiro). The opening of markets and The firm-specific variables are constructed increased international competition may from balance sheet and profit-and-loss have altered the profitability-ownership account data as collected from either the concentration relation since the late Hoppenstedt Bilanzdatenbank (a commer- 1980s. In the subsequent section we cially sold data source), the Bundesanzeiger explore this conjecture using a large panel (a federal gazette, in which corporations data set of both quoted and non-quoted are obliged by law to publish their annual German corporations for the 1990s. financial statements), or annual reports directly obtained from the corporations.21 We used unconsolidated company data 3. OWNERSHIP CONCENTRATION, whenever available and excluded pure INSIDER CONTROL, AND holding companies.22 PERFORMANCE OF GERMAN FIRMS Secondary sources were consulted for identifying owners, share distributions, and 3.1. Data, variables, and sample the composition of managing and supervisory characteristics boards. The sources are Commerzbank’s Wer gehört zu wem? (Who owns whom?, The data set contains 361 firms from the 16th–18th edition), Bayerische Hypotheken- German mining and manufacturing indus- und Wechselbank’s Wegweiser durch tries. Although 300 of these firms have deutsche Aktiengesellschaften (Guide of the legal form of stock corporations German Stock Corporations, “Hypo- (Aktiengesellschaften), only 183 were offi- Guide”, annual issues 1988–1996), and cially listed and traded on German stock Hoppenstedt’s Börsenführer (Stock Market exchanges during the observation period. Guide, annual issues, 1988–1998). In As we are interested in the importance of combining these data sources it was stock market exposure, our sample also possible to obtain a fairly precise picture of includes non-traded stock corporations as voting stock ownership.23 For the purpose well as some limited liability corporations of this study, we have defined a ‘large’ (GmbH, 54 firms) and limited commercial shareholder as one who controls at least partnerships (GmbH & Co. KG, KGaA, 5 per cent of a firm’s voting capital. This 7 firms).The time period covered is 1991 to cut-off point of when a shareholder is large 1996, which yielded the largest number of rather than small is less important than in DOES THE GOVERNED CORPORATION PERFORM BETTER? 497 Anglo-Saxon studies, since almost all the extent of “governance” exercised by sample firms have large shareholders who firm owners. In this study as in previous control at least 25% of the voting capital. ones, we measure ownership concentration As Table 19.1 shows, 65% of the companies by the Herfindahl index of outstanding in the sample have one large shareholder voting stock and, alternatively, by the who, on average, controls 89% of the percentage stake of the largest shareholder. voting stocks and faces a group of small The share of the largest shareholder indi- shareholders with an aggregate share in the cates her fundamental voting power,that is, voting capital of about 11%. the ability to outvote other shareholders In the empirical literature ownership or initiate major changes by herself (e.g., concentration is the standard indicator for ousting the supervisors, introduce a new

Table 19.1 Ownership structures: sample of 361 German manufacturing firms, 1991–1996

Largest shareholder’s Block of widely Group of firms stake dispersed shares

Full sample Percentage Number of large Quoted firms relative to the Mean (std. dev.) Mean (std. dev.) shareholders Non-quoted firms group of firms in per cent in per cent

0 2 0.55 0.00 (0.00) 100.00 (0.00) 2 1.09 0.00 (0.00) 100.00 (0.00) ––– – 1 235 65.10 89.07 (21.54) 10.93 (21.54) 100 54.64 76.55 (27.14) 16.50 (27.14) 135 75.84 98.35 (7.88) 1.65 (7.88) 2 77 21.33 56.55 (18.75) 18.62 (19.45) 50 27.32 55.08 (20.87) 26.80 (18.53) 27 15.17 59.26 (13.96) 3.48 (9.60) 3 28 7.76 44.07 (18.41) 24.43 (23.47) 19 10.38 40.26 (20.01) 36.00 (19.65) 9 5.06 52.11 (11.65) 0.00 (0.00) 4 15 4.16 52.87 (29.71) 25.67 (25.74) 11 6.01 43.82 (19.30) 35.00 (23.84) 4 2.25 77.75 (41.82) 0.00 (0.00) 5 1 0.28 52.00 (NA) 15.00 (NA) ––– – 6 1 0.28 55.00 (NA) 0.00 (NA) ––– – 1 0.56 55.00 (NA) 0.00 (NA) 7 2 0.55 53.00 (4.24) 0.00 (0.00) ––– – 2 1.12 53.00 (4.24) 0.00 (0.00) Total 361 100.00 75.27 (28.72) 14.53 (22.73) 183 100.00 63.98 (28.56) 27.67 (25.53) 178 100.00 86.49 (24.39) 1.54 (7.61)

The sample consists of 361 German firms which operated in 30 industries of the mining and manufacturing sector during the period 1991 to 1996. Of the sample firms 300 have the legal form of stock corporations (Aktiengesellschaften) with 183 listed and traded on German stock exchanges. The data set further includes 54 limited liability corporations (GmbH) and 7 limited commercial partnerships (GmbH & Co. KG, KGaA). For identification of owners and share distributions we used Commerzbank, Wer gehört zu wem? (Who owns whom?, editions 16th to 18th, publication years 1988, 1992, 1994), Bayerische Hypotheken- und Wechselbank, Wegweiser durch deutsche Aktiengesellschaften (“Hypo-Guide”, Guide of German Stock Corporations, annual issues 1988–1996), and Hoppenstedt’s Börsenführer (annual issues, 1988–1998).We define a large shareholder as controlling at least 5 per cent of a corporation’s voting capital. 498 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY corporate charter). The Herfindahl index, them to the group of firms controlled by defined as the sum of squared individual families or (voting pools of) individuals. stakes, has the advantage of accounting for For some firms such as Bayer, an asymmetric dispersion of shares among Mannesmann, Schering, Siemens, or VEBA different shareholders.24 more than 90% of the voting capital is Ownership concentration may not suffice generally reported as “widely dispersed”. as an indicator of the degree of “gover- The study of Baums and Fraune (1995) nance”. The identity of owners may play shows that in the 1992 annual shareholders’ a more crucial role. Officially reported meetings banks controlled an aggregate of shareholdings often do not reflect the more than 90% of those firms’ voting true extent of voting control exercised by capital via associated investment compa- owners of German corporations (Becht nies and proxy votes. These sample firms and Boehmer, 1997; Boehmer, 1999). were therefore added to the group of Complex cross-holding arrangements and firms with financial institutions as largest unreported fiduciary and dormant voting shareholders. rights complicate the identification of To relate back to the discussion on the actual controllers. Nevertheless, given the managed vs. the governed firm, or the information available, we attempt to assess insider vs. outsider model, corporations the relevance of the owner’s identity for the having individuals or families (FAMILY), link between profitability and ownership banks (FININST), or different independent concentration. large shareholders (MIX) may be inter- For the purpose of grouping firms by the preted as potentially governed by committed location of control rights five broad cate- insiders. By contrast, for firms owned by gories of direct owners can be identified: another industrial firm (INDFIRM) it another industrial firm or holding company might be managers rather than the ultimate (INDFIRM), families or (voting pools of) owners who control the managers of the individuals (FAMILY), financial institutions owned company. Firms of this group, (banks or bank-owned investment compa- mostly subsidiaries of large quoted stock nies, insurance companies, FININST), corporations (e.g., BASF, Thyssen), may different large shareholders (MIX, e.g., then be considered as managed. However, industrial firms and investment companies), we make no attempt here to argue that and foreign owners (FOREIGN). A sixth a certain location of control rights indeed group, CHANGE, is defined to account implies “more” or “less” governance. for 26 firms for which a turnover of Although we took great efforts to trace blockholdings from one of these owner ownership back to the ultimate owners for categories to another was observed during all firms, it is not entirely clear whether the observation period. these ultimate owners really “govern”, or The groups are mutually exclusive. A whether hired managers nonetheless exercise firm that, for instance, had another industrial control even if ownership concentration firm as largest shareholder in some years is high and firms are owned by families and a bank in other years only appears in for example. the CHANGE group but not in the groups To investigate whether stock market INDFIRM or FININST. Due to the lack of exposure makes a difference for the potential detailed ownership information, the foreign- link between governance indicators and owned firms were also treated as a separate corporate performance we distinguish group. A few firms are owned by foundations between corporations traded on the stock which have no owner in a strict sense. It is exchange (QUOTED) and non-traded or thus unclear who really controls these firms. non-stock corporations (NON-QUOTED). As the foundations are often close to the Table 19.2A presents the mean and founding families of the firms, as in the median values of selected variables for the case of Bosch GmbH (a leading electronic full sample and the groups of quoted and products firm), it seems tenable to assign non-quoted firms. The last column reports DOES THE GOVERNED CORPORATION PERFORM BETTER? 499

Table 19.2A Summary statistics: sample of 361 German manufacturing firms, 1991–1996. Sample split by stock market exposure

Group of firms Test statistics

Variable All firms (361) Quoted (183) Non-quoted (178) t/Mann-Whitney

Return on total assets (ROA) Mean/median 0.2958/0.2875 0.3159/0.3028 0.2755/0.2705 5.39***/5.75*** Return on equity Mean/median 0.5515/0.4993 0.5593/0.5475 0.5436/0.5116 0.90/3.09*** Ownership concentration (OC) Herfindahl index (bounded) Mean/median 7,465/10,000 5,130/4,543 8,628/10,000 27.48***/482.2*** Largest shareholder’s stake Mean/median 0.8584/1.0000 0.7340/0.7800 0.9787/1.0000 28.91***/29.60*** Board representation (B) Mean/median 0.2802/0.0000 0.2811/0.0000 0.2793/0.0000 0.09/0.07 Turnover Mean/median 3,283/552 4,327/477 2,221/631 4.95***/0.89 Employment Mean/median 11,063/2,094 14,805/2,291 7,259/1,843 5.16***/3.29*** Firm size (S) Mean/median 19.897/19.779 19.982/19.621 19.810/19.887 2.25**/0.88 Firm growth (G) Mean/median 0.0219/0.0220 0.0359/0.0296 0.0078/0.0159 2.64***/3.06*** Capital intensity (K) Mean/median 12.177/12.085 12.088/12.036 12.267/12.144 6.46***/4.82*** Capital structure (C) Mean/median 0.5841/0.5974 0.5979/0.6112 0.5699/0.5753 3.49***/2.99*** Market concentration (H) Mean/median 579.52/327.70 610.19/316.70 548.33/350.60 1.95*/0.27

Calculations are based on data from annual reports of 361 German corporations (see Table 19.1). The sample split is based on stock market quotation for at least three out of the six sample years. Return on total assets (ROA) is gross profits (calculated from the firms’ profit-and-loss statement as turnover minus expenses for personnel and materials) divided by the book value of total assets (as reported in the firms’ balance sheet). Return on equity is gross profits divided by equity (calculated as shareholders’ equity plus reserves and pension liabilities). Ownership concentration is measured as the sum of squared individual percentage stakes, thus bounded to lie in the interval [0; 10,000], and alternatively, as the percentage stake of the largest shareholder. Board representation is an indicator variable taking on unit value if an identified owner served on the board of executive directors in a certain year, otherwise it has zero value. Turnover is in millions DM. Employment is the reporting year average number of full-time employees as provided in the firms’ annual report or in secondary sources. In some cases end-of-the-year figures had to be used because information on average employment was not available. We measure absolute firm size as the natural logarithm (log) of total assets to reduce the skewness of the firm size distribution. Firm growth is the annual logarithmic change in real turnover (turnover deflated by the 1995 GDP deflator, source: German Statistical Office). Capital intensity is defined as log (total assets/employees). Capital structure is equity (as defined above) over total capital (book values). Market concentration is measured by a standard Herfindahl index at the two-digit industry level of the German industry classification for the manufacturing sector (“SYPRO”, source: German Statistical Office). Means and medians of all variables are calculated for the period 1991 to 1996 with the exception of firm growth which, due to its definition, is available for the period 1992 to 1996. We compare the variables’ means and medians between the two groups of firms, quoted versus non-quoted firms. The last column of the table provides the t-statistic for the test of equal group means and the Mann– Whitney statistic for the test of equal group medians. ***/**/* Significant at the 0.01/0.05/0.10 error level respectively. 500 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY statistics for testing the hypotheses of firms had significantly higher growth of equal group means (t-statistic) and medians turnover than non-quoted firms. (Mann–Whitney statistic).The table shows Table 19.2B contains summary statistics that quoted firms have significantly higher of selected variables for the groups of firms average returns on total assets and also as classified by the identity of the largest higher median returns on equity.The quoted owner. For any pair of owner categories we firms are less levered. Further,as measured conducted pairwise t- and median tests for by turnover, employment, and the natural the variables listed in Table 19.2B but do log of total assets, the quoted firms are not report these results separately.25 The substantially larger on average but the tests reveal that the INDFIRM firms have median firms do not differ significantly in significantly lower rates of return than any turnover or total assets. Finally, quoted other owner group except for the equity

Table 19.2B Summary statistics: sample of 361 German manufacturing firms, 1991–1996. Sample split by the location of control rights

Location of control rights

INDFIRM FAMILY FININST MIX FOREIGN CHANGE Number of firms 81 122 20 54 58 26

Return on total assets Mean 0.2408 0.3162 0.2725 0.2918 0.3439 0.2913 Median 0.2253 0.3120 0.2536 0.2999 0.3129 0.2844 Return on equity Mean 0.4288 0.6359 0.4386 0.5173 0.6084 0.5690 Median 0.3500 0.5626 0.4091 0.5106 0.5487 0.4982 Ownership concentration (bounded Herfindahl index) Mean 8,818 7,485 1,897 3,318 9,045 4,078 Median 10,000 9,900 342 3,322 10,000 3,563 Turnover Mean 2,880 1,918 16,241 3,427 2,065 3,389 Median 853 333 3,378 889 613 503 Firm growth Mean 0.0016 0.0428 0.0363 0.0221 0.0018 0.0204 Median 0.0074 0.0378 0.0256 0.0191 0.0076 0.0082 Capital structure Mean 0.5930 0.5541 0.6367 0.5999 0.6115 0.5626 Median 0.6019 0.5666 0.6673 0.6068 0.6264 0.5464

Calculations are based on data from annual reports and secondary sources of 361 German manufacturing firms (see Table 19.1). The sample split is based on the location of control rights (voting stock). We define a large shareholder as controlling at least 5 per cent of a corporation’s voting capital and distinguish six identities of large shareholders: 1. INDFIRM is defined as firms having another independent industrial firm or a holding company as largest shareholder (e.g., THYSSEN GUSS AG, subsidiary of THYSSEN concern). 2. FAMILY is defined as firms having (pools of) individuals or families as largest shareholders (e.g., BAUSCH AG, BOSCH GmbH). 3. FININST is defined as firms having banks, insurance companies, or associated investment companies as largest shareholders (e.g., LINDE, larger stakes owned by Deutsche Bank and Commerzbank), or, having widely dispersed shareholdings, but banks control at least 75 per cent of the voting capital through proxy voting rights (e.g., BAYER). 4. MIX is defined as firms having different independent large shareholders (e.g., BOSCH- SIEMENS Hausgeräte GmbH, owned equally by BOSCH GmbH and SIEMENS AG). 5. FOREIGN is defined as firms having foreign companies as largest shareholders (e.g., OPEL AG, owned by GM). 6. CHANGE is defined as firms which experienced a change in the identity of blockholders through turnovers of blocks from one of the owner categories 1–5 to another (e.g., AQUA SIGNAL AG). See Table 19.2A for definitions of the other variables. DOES THE GOVERNED CORPORATION PERFORM BETTER? 501 return of firms controlled by financial supplier concentration at the two-digit institutions. Ownership concentration is industry level (source: German Statistical the highest for the managed firms. Family- Office).These variables will be discussed in controlled firms have the highest ROE and more detail below.27 the second highest ROA but their share of Due to the combination of time series equity capital is the lowest of all groups. and cross-section data we can decompose Bank-controlled firms have the highest the regression disturbance uit into a classical share of equity capital and thus the lowest white noise error ‡it and effects specific to degree of leverage. Foreign-owned firms individual firms, ai, time, t, and industries, aj. have the highest ROA and the second We use standard panel regression techniques highest ROE. (see Hsiao, 1986; Baltagi, 1995) to estimate (1). As ownership concentration and profitability could be determined 3.2. Regression model and simultaneously in competitive and well- hypothesis functioning markets, we test for simultaneity bias in (1) by applying instrumental variable estimation techniques as well. Details Regression model on estimation techniques and specification To investigate the impact of the mode tests will be provided in the tables. of corporate governance on corporate performance we use the following panel Hypotheses regression 1. Ownership concentration, board represen- ROAit b1OCit b2Bit b3Sit b4Git tation, and stock market exposure j b5Kit b6CI,t1 b7H it uit, Our main interest lies in estimating the (1) coefficients b1 and b2. If ownership concentration indicates tighter and perform- in which the subscript i 1,...361 identi- ance-enhancing governance exerted by fies individual firms, j 1,...,30 indicates owners, b1 0 is to be expected. If manage- the respective two-digit industry a firm rial discretion is facilitated in traded operates in, and t 1992,...1996 denotes corporations because ownership concentra- time periods. tion tends to be lower for traded than for In model (1) we regress the return on non-traded firms, we expect to observe a total assets ROA on ownership concentration stronger positive impact of ownership OC and a set of other variables. For lucidity concentration on ROA for the traded firms. and space restrictions, we will only present However, if there is less asymmetric infor- regression results using for OC the unbounded mation in traded corporations because they Herfindahl index as in Demsetz and Lehn are more transparent to outside investors (1985).26 B indicates the presence of the than non-traded firms, the differential largest shareholder on the executive board. impact of ownership concentration on ROA The following variables which have frequently between traded and non-traded corpora- been employed in the Industrial Organization tions may go the other way. In the extreme, (IO) literature to explain profitability if the market mechanism indeed provides differences across firms or industries, serve effective corporate control, as suggested, as right-hand side control variables: absolute e.g., by Demsetz and Lehn (1985), b1 0 firm size S (natural logarithm of total should hold for the traded firms. By con- assets), firm growth G (logarithmic annual trast, b1 0 may indicate inefficiency or change in turnover), capital intensity K (log rent extraction due to the presence of large of total assets divided by the number shareholders. The same reasoning as for of employees), capital structure C (share- ownership concentration can be applied to holders’ equity plus reserves divided by the representation of the largest share- total capital), and the Herfindahl index of holder on the board of executive directors. 502 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 2. Firm size and firm growth concentration as a summary measure of Economic theory offers no clear-cut industry characteristics, reflecting current predictions for the impact of firm size and production technologies (potential scale firm growth on profitability. A positive economies), demand (price elasticity) as influence of absolute firm size is implied by well as the intensity of competition. the economies of scale and scope argument 6. Industry characteristics (e.g., Baumol, 1959), while a negative effect Industry characteristics such as knowledge may result from organizational inefficiency conditions and technological opportunities (X-inefficiency,Leibenstein, 1966). As larger may simultaneously determine governance firms tend to be more diversified, lower risk structures, investment, and profitability premiums could render b negative. Firm 3 (Audretsch and Weigand, 1999). In growth may, on the one hand, reflect better industries such as optical instruments or investment opportunities and thus higher machinery, production technology or the profitability, b 0. On the other hand, if 4 knowledge on which firm know-how is managers have sufficient discretion to divert based does not require large firm size free cash flow and overinvest, profitability per se. Cost advantages (scale economies) should be lower for growing firms, b 0. 4 from large-scale operations (production, 3. Capital intensity R&D) are not ubiquitous. Therefore, these The coefficient on capital intensity may help industries offer a favorable environment to identify overinvestment. If decreasing for smaller firms, and indeed that is where scale economies prevail, profitability should we find small family-owned firms to be decrease in capital intensity, b5 0. The very common. By contrast, in the chemical opposite may hold if high capital intensity and pharmaceutical industry large stock is a barrier to entry and exit, which allows corporations dominate. Clearly, production incumbents to earn rents, as frequently technologies require larger firm sizes. argued in the IO literature. Different ownership structures are also necessary to satisfy the increased capital 4. Capital structure needs and to better spread the higher risk With imperfect capital markets capital involved in large-scale operations. Based structure matters for investment decisions on this reasoning we expect industry and firm profitability. Capital structure characteristics, if not fully reflected by choices vary across firms and industries market concentration, to affect profitability depending on informational asymmetries, and ownership structures. transaction costs, and growth prospects. If a higher equity share implies lower risk 7. Firm-specific effects of bankruptcy, an inverse relation between The firm-specific effects control for system- profitability and the share of equity capital, atic variation in profitability not captured C, can be expected, b6 0, since the return by the explanatory variables (e.g., differences on investment required to compensate for in risk-taking, user cost of capital). If our risk-taking decreases in risk.28 To reduce explanatory variables do miss systematic potential simultaneity bias between ROA ROA variation, the firm-specific effects and capital structure, we use beginning-of- should turn out to be statistically significant. the-period values for C. 8.Time-specific effects 5. Market concentration The time-specific effects subsume the macro- Supplier concentration is implied by oligopoly economic shocks common to all firms. theory to be positively correlated with Their inclusion is suggested by the fact that profitability. This prediction is supported during the sample period the German econ- by a vast empirical literature. We therefore omy first experienced the reunification expect b7 0. As it is not the concern of boom of 1990/91 and then slid into the this paper to discuss the appropriate inter- recession of 1992/93, the worst downturn in pretation of the relationship, we take market Germany’s post-war history. DOES THE GOVERNED CORPORATION PERFORM BETTER? 503 3.3. Results The specification tests show that there is systematic variation in the explained vari- Specifying the relationship ables picked up by either individual firm or between profitability and corporate industry effects. Firm-specific effects are governance indicators fixed rather than random. There is no gain by endogenizing OC. Put differently, OC can Table 19.3A summarizes the panel regression be taken as an exogenous variable in the estimates for Equation (1), the extended ROA regression. For the OC regression, the model, which includes the set of variables specification test indicates simultaneity highlighted in the IO literature, and also bias at least in the extended model, that is, for the parsimonious variant which excludes ROA should be treated as an endogenous the IO variables.Table 19.3B takes owner- variable.30 In short, model E is supported. ship concentration as the dependent variable. As they lack time series variation, the indus- For now, we selectively focus on the impact try indicator variables cannot be included of the governance indicators OC and B, the in Model E because they would be perfectly importance of individual firm and industry correlated with the fixed firm effects (see effects as well as the possible simultaneity Baltagi, 1995, p. 11). We will return to the of OC and ROA.29 problem of estimating the impact of time- All regressions in Tables 19.3A and 19.3B constant variables in a fixed effects model imply a negative relation between the return below. The time-specific effects are highly on total assets and ownership concentration. significant in the parsimonious variant of The estimated coefficients on OC in the Model E but insignificant in the extended extended and parsimonious ROA regressions variant. Further investigation reveals that (Table 19.3A) are statistically significant at including capital intensity in the regression the 0.05 level in Models C to G which take renders the time effects insignificant. We advantage of the pooled data. With one will offer an explanation shortly. exception the same holds for ROA in the OC regression (Table 19.3B). Board represen- Does stock market exposure affect tation has a positive coefficient in the the profitability-corporate ROA regression and a negative coefficient governance relationship? in the OC regression in almost all models. However, the statistical significance of the The negative impact of ownership concen- estimated coefficient is clearly affected by tration on ROA supports the view of the way unobserved systematic influences inefficient ownership concentration and are modelled.The coefficient of determina- rent extraction. So far it does not refute tion (adjusted R squared) reveals how the Demsetz–Lehn hypothesis which much the respective variables add to the implicitly assumes firms to be exposed to explanatory power of the regression. the market for corporate control. This Beginning with the parsimonious variant in assumption holds only for about half of our Model C, the governance indicators explain sample firms. Therefore, we consider stock about 1% of the ROA variation, while market exposure next. ROA and board representation are of little As the Durbin–Watson statistic in help in explaining the variation of ownership Table 19.3A indicates problems of first- concentration across firms and over time. order serial correlation in our favored In terms of explanatory power, the Model E, we used an estimation technique combination of random firm-specific and which corrects for both heteroskedasticity fixed industry-specific effects with other and first-order serial correlation.Table 19.4A firm variables is not very useful either contains these robust estimates for the (Models F and G). By far the highest extended model.To allow for a comparison explanatory power is achieved by includ- with the non-robust coefficient estimates ing fixed individual firm effects (Model E). for OC and B in Table 19.3A, column 1 of 504 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 19.4A ignores stock market exposure. concentration are highly significant. Larger Column 2 then presents the robust estimates firms and firms with a higher share of equity for the quoted firms, while column 3 gives capital (lower leverage) have significantly the coefficient differences with respect to the lower returns, which is consistent with diver- (not separately reported) regression coeffi- sification and risk-return considerations. cients of the non-quoted firms.We also report Firm growth affects ROA positively which the group means of the firm-specific effects. supports the argument that growing firms The fixed effects measure the average have better investment opportunities. systematic difference in ROA across groups The negative coefficient on capital intensity of firms not accounted for by the model’s seems to imply decreasing returns to scale. explanatory variables. There is no statisti- Given the above-mentioned fact that the cally significant difference in the means of time effects become insignificant when these firm-specific intercepts between the capital intensity enters the regression, this traded and non-traded firms.31 Stock variable is apparently a proxy for capital market exposure does also not change the utilization and thus picks up business cycle profitability-ownership concentration rela- effects. Market concentration has a tionship. The negative impact of ownership positive but statistically insignificant concentration is a bit weaker for the quoted effect. Stock market exposure does not firms (0.0027) than for the non-quoted change the impact of the IO variables on firms (0.0049) but still significant at the ROA in any significant way. 0.05 error level. The coefficient difference ( 0.0022) is not statistically significant. Does the identity of owners matter? These results contradict both the hypothesis of the beneficial large shareholder as As in the case of the industry indicators, we well as the Demsetz–Lehn hypothesis of no cannot consider the identity of owners in a ownership effect. However, the estimated fixed-effects regression directly, since the coefficient on OC for the full (non-split) indicators for the location of control rights sample implies a quantitatively small effect. lack time series variation and thus will be Evaluated at the sample means, doubling perfectly correlated with the firm-specific the unbounded Herfindahl index would lower effects.We therefore follow two approaches. ROA by about 8%. In the first approach, we investigate the There is a notable difference between the indirect impact of the location of control two groups with respect to board represen- rights on ROA by interacting the identity tation.The coefficient on B is insignificantly indicators with all time-varying right-hand positive (and remains so when ownership side variables of (1).32 The second approach concentration is excluded from the regres- homes in on the estimated fixed firm sions) for the non-split sample and for the effects.The fixed effects can be interpreted as quoted firms but significantly positive for the autonomous component of ROA neither the non-traded firms. For the non-traded explained by the explanatory variables nor firms the presence of the largest owner on by purely stochastic disturbances. By the board of executive directors thus regressing the fixed effects on the identity improves profitability. Assuming that infor- indicators we obtain a direct multivariate mational problems are more pronounced test on how the location of control rights for non-traded firms this result implies that affects ROA, after having controlled for tighter control through the owner reduces the time-varying explanatory variables’ agency costs and improves firm perform- influence. ance.The estimated coefficient on B for the The regression results from the indirect non-quoted firms suggests an increase of approach are contained in Table 19.4B.The average ROA by about 3% if average group of firms with another firm as largest board representation raises by 1%. shareholder, INDFIRM, serves as the base For the IO variables we find that all group. Column 1 presents the coefficient coefficients except for the one on market estimates for this base group, while the DOES THE GOVERNED CORPORATION PERFORM BETTER? 505 ) *** *** *** *** *** *** *** *** *** *** Table 19.3A continued ( 0.10160.00960.10170.0117 – 0.2829 – 0.2021 0.3636 0.3646 0.5005 0.4509 0.4589 0.2480 * ** *** *** ** *** * 0.0009 (0.05) 0.7302 1.6699 0.0091 (0.33) – – 0.0313 (3.59) 0.0365 (4.19) 0.0187 (2.20) 0.0247 (2.83) 0.0098 (0.51)0.0137 (0.91)0.0272 (1.76) 0.0073 (0.52)0.0198 (1.38) 0.7763 0.0351 0.0330 0.0071 1.5598 0.2203 0.2777 0.3022 ** ** *** ** ** ** * ** ** ** 0.0007 (0.87) 0.0309 (1.68) 0.0009 (1.11) 0.0254 (1.49) 0.2058 – 0.0009 (1.04)0.0007 (0.93)0.0008 (2.14) 0.0375 (2.04) 0.0142 (0.40) 0.2939 – 0.0009 (2.41) 0.0030 (1.98) 0.0014 (2.03) 0.0013 (2.05) 0.0078 (0.71) 0.0010 (2.91) 0.0013 (1.95) 0.0009 (2.27) 0.0029 (2.28) 0.0014 (2.24) 0.0079 (0.79) 0.0018 (0.06) – – Regression coefficient (absolute t-statistic) Regression diagnostics Ownership concentrationExtended model Board representation Extended model Panel regression estimates of the return on total assets equation, of the return estimates regression Panel 1992–1996. Full sample with fixed industry effects with fixed and slopes) C.Total (common intercept A. Between (cross-section) B. Between (cross-section) Table 19.3A Dependent variable: ROARegression model of selected explanatory variables industry effects D.Total with fixed effects) E. firm (fixed Within Parsimonious model effects)F.RE firm (random G. industry effects RE with fixed H. Within-2SLS Parsimonious model Adj. R sq. DW 506 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject *** *** *** *** *** *** *** *** ** *** *** *** *** *** *** 16.01 14.29 232.48 240.79 228.71 193.31 56.24 3.23 58.79 4.72 50.28 0.8216 Accept 66.07 68.91 0.4728 Accept 49.81 60.58 2.24 Accept Chisq(4) F(29, 330) F(29, 1767) F(360, 1436) Chisq(1) Chisq(1) Chisq(3) F(29, 1767) F(4, 1767) Test statistic Extended model Continued Model F vs.Model G Model industry effects No fixed F(29, 1762) Model C vs.Model D Model C vs.Model E Model C vs.Model F Model D vs.Model G Model industry effects No fixed E vs.Model F Model effects firm No fixed effects firm No random effects firm No random DModel F(29, 1762) effects Random firm F(360, 1431) Chisq(1) Chisq(1) Chisq(7) effects No time F(4, 1762) Model E vs.Model H Model bias No simultaneity Chisq(6) Table 19.3A Specification test A vs.Model B Model Null hypothesis industry effects No fixed F(29, 325) Parsimonious model Accept/Reject null DOES THE GOVERNED CORPORATION PERFORM BETTER? 507 (1) . it rge shareholders has rge shareholders Reject Reject Reject firm-specific intercepts). firm-specific time- Other cept from which individual firms may firms individual which cept from , is – variable a specific picking since rm-specific effects to be constant and rm-specific s OC as an endogenous right-hand side s OC as an endogenous right-hand ions). B adds 29 industry Model dum- the are (which estimates Within-2SLS l exogenous right-hand side variables in variables side right-hand l exogenous ustry an unre- effects) against or no time y the industrial organization literature. The organization y the industrial n this case under the name “Wu-Hausman” the name case under n this ics as provided in parentheses are based on in ics as provided ls (Bhargava et al.,ls (Bhargava 1982). See Hsiao (1986) ultiplier test by Breusch and Pagan suggested ultiplier *** *** *** 7.13 5.98 1.6914.66 Accept respectively, disturbance regression and a classical white noise t F(4, 1767) F(4, 1792) is given by is given , j a , i at time t.at time of variables. model See definitions In the parsimonious 19.2A for Table a j extended model , it u it j operating in industry in operating H i 7 b 1 I, t C 6 b it K 5 b it are the return on total assets,are the return index), (unbounded Herfindahl concentration ownership board representation, size, firm firm growth, G 4 H ). A, Models B, by OLS. C and D are estimated estimator). (LSDV,Within-OLS Variables Least E is Squares Dummy Model Models b it it , and S 3 b , can be decomposed firm-, into effects, industry- and time-specific it it it B u 2 plus the number of large shareholders as an additional instrument included to identify the regression equation.The of la the regression number to identify included instrument as an additional of large shareholders plus the number b t B, S, G, K, C it j OC a 1 b i a it Significant at the 0.01/0.05/0.10 error level respectively. level at the 0.01/0.05/0.10 error Significant * / B, S, G, K, C, H it ** / We present specification test statistics these models. for present specification We A standard to test F-test a restricted is employed (e.g., model no ind ROA Model A is a standard cross-section regression with the regression variables averaged over the observation period (361 observat period the observation over averaged variables with the regression regression A is a standardModel cross-section u or Baltagi (1995) for a detailed discussion of panel regression models and specification tests. and specification models estimates’ regression of panel The discussion a detailed (1995) for or Baltagi absolute t-statist estimator. variance-covariance (1980) heteroskedasticity-robust White’s efficient if OC is indeed endogenous).“Durbin–Watson” tests for first-order serial correlation (AR1) of panel regression residua endogenous).“Durbin–Watson” if OC is indeed regression (AR1) of panel efficient correlation serial first-order tests for (1980). Hausman’s (1978) specification test is used to decide on fixed versus random firm-specific effects.The same test, firm-specific (1980). random versus test i on fixed is used known to decide Hausman’s (1978) specification (1) is endogenous) with Equation if OC in are inefficient (which (Wu, estimates Within-OLS to compare the 1972) is also applied stricted model (allowing for such effects).To test for random firm-specific effects no such effects versus we use the Lagrange firm-specific random test such effects).To for for stricted (allowing m model variable (i.e.,variable with correlated being constant effects (e.g., cannot be included. dummies) industries effects (i.e., firm-specific F assumes random Model a common inter effects E but randomly). as Model deviate treat firm-specific F.Model to Model G adds 29 industry fixed Model dummies H includes admittedly was excluded. pair of the IO variables when any other different not significantly arbitrary.The – results were identical for all firms. for identical C. to Model D adds 29 industry Model dummies effects (361 (constant) firm-specific E assumes fixed Model (1), been excluded from the second-stage OC regression (to be presented in the second-stage from OC regression been excluded 19.3B). of variables exclusions Table tested alternative We mies to Model A. to Model mies data (1805 pooled observations). time-series C to H use the pooled cross-section Models C restricts Model the fi F and G use GLS. Model H is Within-2SLS using the fitted values of OC from the reduced-form (first-stage) regression of OC on al (first-stage) regression the reduced-form of OC from F and G use values GLS. the fitted using Within-2SLS H is Model Model GModel effects No time F(4, 1762) Model EModel in as described firms I.The 361 German sample contains estimating Table The effects No time F(4, 1762) *** in which ROA,in OC, capital intensity, capital structure, of firm concentration and market we only include ownership concentration and board representation as right-hand side variables and ignore all variables implied all variables and ignore variables side as right-hand and board representation b concentration ownership include we only regression disturbance regression 508 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY following columns contain the estimates of Finally,Table 19.4B shows differences in coefficient differences with respect to the the means of the firm-specific intercepts other owner categories. The coefficient on across the groups. The firms controlled by ownership concentration is positive but financial institutions have the highest mean insignificant for this base group of presum- intercept, while the FAMILY and CHANGE ably managed rather than governed firms. firms have the lowest. Compared to the The negative impact of ownership concen- INDFIRM group, FAMILY and CHANGE tration as reported for the non-split sample firms have a signifcantly lower autonomous (Table 19.4A, column 1) can be traced ROA, while having financial institutions as back to the firms owned, and perhaps largest shareholders significantly improves governed by FAMILY,MIX, FOREIGN, and profitability. To investigate these differ- CHANGE. The coefficient difference is ences further is our second, direct approach statistically significant for the FAMILY to assessing the impact of the identity and MIX firms. Taking the number of indicators on ROA. shareholders as given, the Herfindahl index Ignoring time effects for simplicity, the is lower when the variance of stakes is estimates of the firm-specific fixed effects lower.33 Therefore, the higher the stake of are given by (see Hsiao, 1986, p. 30) the largest shareholder,and the smaller the ’ individual stakes of other shareholders, the aˆi yi Xi bˆ, (2) larger will be the variance of stakes cet. par. (i.e., the more asymmetric is the stake where ^ indicates an estimated parameter, distribution). In other words, the presence the bar denotes the firm-specific time mean of another large shareholder improves of a variable, y is the dependent variable, profitability. This result is in line with X is the matrix of time-varying right-hand Boehmer (2000) who finds the presence of side variables, a and b are the vectors of a second or third large shareholder to raise firm-specific effects and slope coefficients. firm value. We interpret (2) as depicting the long-term Interestingly, ownership concentration is or autonomous component of the ROA beneficial to firms potentially under the model (1). The fixed effects can be control of financial institutions. The coeffi- regressed on a set of variables, z, which cient difference is significantly positive. This vary across firms but not over time, such as group of firms has by far the lowest degree the identity or industry indicators. of ownership concentration (cf.Table 19.2B) ’ and can, by German standards, be viewed aˆi zi i. (3) as dispersedly held.The group also includes firms in which banks accumulated proxy The coefficient vectors and in votes. The positive impact of ownership Equation (3) can be consistently estimated concentration suggests that firm perform- by OLS ifˆb is a consistent estimate and if ance is enhanced when banks increase their the number of cross-section units is large cash flow stakes. and the time-series component is not With respect to the IO variables, larger tending to infinity (Hsiao 1986, p. 50).The firm size and a higher equity share are first requirement is satisfied, since we can significantly more important for FAMILY insert for ˆb the estimates described in firms, which are the smallest on average Table 19.4A. With 361 firms and variables and at the median, than for any other group averaged over five years we feel confident of firms. The choice of absolute firm size that the second condition is not completely and capital structure can be interpreted as violated.Therefore, we present in Table 19.5 reflecting the risk-return trade-off. As the the results from estimating regression (3) risk of bankruptcy is lower for larger firms by OLS. with more equity, the rate of return When we regress the firm-specific required to compensate for risk-taking can effects on the identity indicators only, be lower as well. taking INDFIRM as the base group and DOES THE GOVERNED CORPORATION PERFORM BETTER? 509 ) *** *** *** *** *** *** *** *** *** *** Table 19.3B continued ( 0.0014 – 0.0209 0.0234 0.13640.00880.14410.2082 – – 0.0398 0.3230.4645 0.0414 0.0146 0.0229 0.0219 0.0247 0.0003 0.0263 * *** *** ** ** ** *** 0.5887 (0.51) 1.8985 (1.63) 0.1489 – 0.0471 (0.04) 1.4862 (3.18) 0.0091 (0.33) – – 0.6368 (1.28) 0.1163 (0.22)1.4332 (1.08)1.6843 (1.19)1.2139 (2.40) 1.4840 (2.89) 1.1100 (2.17) 1.3782 (2.65) 0.2021 0.9680 0.9674 0.4589 0.9930 0.9793 1.3606 (0.97) – – 1.6834 (3.43) ** ** ** ** * ** ** ** ** 3.2924 (0.86) 2.0470 (1.69) 3.5145 (1.05) 1.8990 (0.58) 2.8930 (2.14) 4.3864 (1.12) 1.9620 (1.55) 3.4874 (2.32) 1.1429 (1.90) 1.1929 (2.33) 1.1986 (2.31) 0.3924 (0.23) 2.9901 (2.19) 1.4324 (2.32) 1.3055 (2.41) 1.3877 (2.52) 6.9690 (1.58) Regression coefficient (absolute t-statistic) Return on total assetsExtended model Board representation Extended model Panel regression estimates of the ownership concentration equation, concentration of the ownership estimates regression 1992–1996.Panel Full sample with fixed industry effects with fixed (common intercept and slopes) (common intercept B.Between (cross-section) C.Total A. Between (cross-section) Table 19.3B Dependent variable: OCRegression model of selected explanatory variables industry effects D.Total with fixed effects) E. firm (fixed Within Parsimonious model effects)F.RE firm (random G. industry effects RE with fixed H. Within-2SLS Parsimonious model Regression diagnostics Adj. R sq. DW 510 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject Reject *** *** *** *** *** *** *** *** *** *** *** 128.97 149.05 *** ** 66.03 64.49 68.51 61.01 0.25 Reject 11.94 0.13 Reject 13.19 19.97 0.86 Accept 3.51 Accept 19.99 20.27 7.14 6.41 0.63 Accept Chisq(3) F(4, 1767) F(29, 330) F(29, 1767) F(360, 1436) Chisq(1) Chisq(1) Chisq(2) F(29, 1767) Test statistic Extended model Continued Model E vs.Model H Model bias No simultaneity Chisq(2) Model F vs.Model G Model industry effects No fixed F(29, 1762) Model C vs.Model D Model C vs.Model E Model C vs.Model F Model D vs.Model G Model industry effects No fixed E vs.Model F Model effects firm No fixed effects firm No random effects firm No random DModel F(29, 1762) effects Random firm F(360, 1431) Chisq(1) Chisq(1) Chisq(4) effects No time F(4, 1762) Model A vs.Model B Model industry effects No fixed F(29, 325) Table 19.3B Specification test Null hypothesis Parsimonious model Accept/Reject null DOES THE GOVERNED CORPORATION PERFORM BETTER? 511 , , LS (2) se GLS. rminants B, S, G, N Reject Reject can be decomposed into model we only include the include we only model it u nown in this case under the name case under this in nown ustry an effects) against or no time ous right-hand side variable (i.e., variable side ous right-hand being enous) with the Within-2SLSenous) with the estimates ndividual firms may deviate randomly). may deviate firms ndividual parsimonious ecific effects to be constant and identical ecific ions). B adds 29 industry Model dummies ute t-statistics as provided in parentheses in ute t-statistics as provided e ht-hand side variables in (2), in variables ht-hand side ific intercepts).ific effects time-constant Other ion residuals (Bhargava et al., (Bhargava residuals ion 1982). See cluded from the second-stagecluded from ROA regres- on disturbance nge multiplier test by Breusch and suggested multiplier nge ir of variables was excluded. ir of variables As theory is rather silent on potential dete silent As theory is rather * *** t. . it 0.73 Accept 2.04 5.63 0.51 Accept at time j F(4, 1792) F(4, 1767) , it u it j H 7 e -1 I, t C 6 e it LS N 5 e respectively, disturbance regression noise and a classical white t , it j a G 4 , i e a , index), (unbounded Herfindahl on total assets, the return concentration are ownership board representation, size, firm growth firm H it S 3 , and e C , it LS it B 2 e t it B, S, G, N j ). A, Models B, by OLS. C and D are estimated (LSDV, estimator).Within-OLS Variables E is Model Least Squares Dummy F and G u Models a it ROA 1 e I a it Significant at the 0.01/0.05/0.10 error level respectively. level at the 0.01/0.05/0.10 error Significant it * OC u / plus capital intensity as an additional instrument (included to identify the regression equation). the regression to identify (included instrument as an additional plus capital intensity has been ex Capital intensity ** / We present specification test statistics these models. for present specification We A standard to test F-test a restricted is employed (e.g., model no ind Model A is a standard cross-section regression with the regression variables averaged over the observation period (361 observat period the observation over averaged variables with the regression A is a standard regression Model cross-section are based on White’s (1980) heteroskedasticity-robust variance-covariance estimator. variance-covariance (1980) heteroskedasticity-robust White’s are based on Hsiao (1986) or Baltagi (1995) for a detailed discussion of panel regression models and specification tests.The and specification models estimates’ regression of panel discussion a detailed (1995) for Hsiao (1986) or Baltagi absol “Wu-Hausman” (Wu, 1972) is also applied to compare the Within-OLS estimates (which are inefficient if OC in Equation (1) is endog Equation if OC in “Wu-Hausman” (which are inefficient estimates (Wu,Within-OLS to compare the 1972) is also applied unrestricted model (allowing for such effects).To test for random firm-specific effects no such effects versus we use the Lagra firm-specific such effects).To random test for for unrestricted (allowing model (1980). effects.Pagan firm-specific random versus test on fixed same test, is used to decide The Hausman’s (1978) specification k endogenous). if OC is indeed are efficient (which regress (AR1) of panel “Durbin–Watson” correlation serial first-order tests for return on total assets and board representation as right-hand side variables and ignore all other ad-hoc variables. all other regressi and ignore The variables side as right-hand on total assets and board representation return sion (Table 19.3A). We tested alternative exclusions of variables.The results were not significantly different when pa any other different not significantly of variables.The results were 19.3A). exclusions tested alternative (Table sion We of ownership concentration,of ownership (2) is an ad-hoc specification. regression of variables. See definitions 19.1 and 19.2A for In Tables th the number of large shareholders,the number industry in capital structure, i operating of firm concentration and market Model GModel effects No time F(4, 1762) Model EModel by in as described is given model firms 1.The 361 German regression sample contains extended estimating Table The effects No time F(4, 1762) *** Model H is Within-2SLS using the fitted values of ROA from the reduced-form (first-stage) regression of ROA on all exogenous rig of ROA on all exogenous (first-stage) regression Model the reduced-form of ROA from values the fitted using Within-2SLS H is firm-, effects, industry- and time-specific to Model A.to Model data (1805 pooled observations). time-series C to H use Models the pooled cross-section C restricts Model the firm-sp all firms.for C. D adds 29 industry to Model effects Model dummies (361 firm-spec (constant) firm-specific E assumes fixed Model C, H in which OC, which in ROA, (e.g., cannot be included. dummies) industries effects (i.e., firm-specific F assumes random Model which i from a common intercept Model G adds 29 industry dummies to Model F.Model H includes fixed firm-specific effects E but as Model treats OC as an endogen firm-specific F.Model G adds 29 industry to Model Model dummies fixed H includes correlated with correlated 512 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 19.4A Robust estimates of the return on total assets equation, 1992–1996. Full sample and sample split by stock market exposure Regression coefficients (absolute t-statistic)

Dependent variable: ROA Difference to non- Explanatory variable All firms (1) Quoted firms (2) quoted firms (3)

Ownership 0.0031 (2.54)** 0.0027 (2.00)** 0.0022 (0.69) concentration Board representation 0.0113 (0.43) 0.0157 (0.59) 3.0411 (7.35)*** Firm size 0.0954 (7.69)*** 0.0975 (6.14)*** 0.0069 (0.26) Firm growth 0.1232 (12.49)*** 0.1226 (8.33)*** 0.0014 (0.07) Capital intensity 0.0695 (5.66)*** 0.0793 (4.79)*** 0.0196 (0.76) Capital structure (t 1) 0.0938 (3.23)*** 0.1375 (3.08)*** 0.0830 (1.40) Market concentration 0.2361 (1.31) 0.3484 (1.45) 0.2468 (0.68) Mean firm intercept 3.0996 (183.42)*** 3.1038 (188.26)*** 0.0082 (0.36) Adj. R squared 0.78 (0.78) Durbin–Watson 1.56**, 1.86

The estimating sample contains 361 German firms as described in Table 19.1. The split criterion is stock market quotation which divides the full sample into 183 traded and 178 non-traded firms. The estimating regression is j ROAit =b1OCit b2Bit b3Sit b4Git b5Kit b6CI,t 1 b7H it d1Dit OCit d2DitBit d3DitSit d4DitGit j d5Dit Kit d6DitCi,t1 d7DitH it ai t it in which ROA, OC, B, S, G, K, C, and H are the return on total assets, ownership concentration (unbounded Herfindahl index), board representation, firm size, firm growth, capital intensity, capital structure, and market concentration of firm i operating in industry j at time t. See Table 19.2A for definitions of variables. ai is a firm- specific effect, t a time dummy, and it the stochastic disturbance. The split sample estimates are based on the full sample regression with the set of explanatory variables (except for the fixed firm-specific effects) interacted with a dummy variable D which takes on unit value for non- quoted firms and zero value for quoted firms.The regression coefficients b estimate the impact of the respective explanatory variables on ROA for the quoted firms, while the coefficients d yield the difference between the coefficients b and the respective coefficients for the non-quoted firms, to be obtained by adding d’s to matching b’s, or directly, by estimating the regression a second time but with D defined the other way around. The “All firms” estimates (column 1) come from the regression excluding the set of interacted variables. For the governance variables, these latter estimates can be compared to the Within-OLS estimates presented in Table 19.3A (Model E) which are not robust to first-order serial correlation. The estimates shown above are robust to heteroskedasticity (White, 1980) and first-order serial correlation (AR1) (see Hsiao, 1986, pp. 55).The Within-OLS estimator was employed to generate consistent first-step estimates. We used the Within residuals to estimate a firm-specific serial correlation coefficient rho. Alternatively, we restricted the model to have a serial correlation coefficient common to all firms.The rhos were used to transform the original data. Regressions based on the transformed data were then estimated by OLS (firm-specific rho) or Maximum Likelihood (common rho). The drawback of the “firm-specific rho” approach is that due to the larger number of firm-specific coefficients to be estimated the coefficient on board representation was not estimable for the group of non-quoted firms.Therefore, we only report the estimates based on the “common rho” approach. However, the coefficients on all other explanatory variables can be estimated in the “firm-specific rho” approach. As these latter coefficient estimates are qualitatively identical and quantitatively only marginally different from those shown below, we do not report them separately.The estimates are available from the authors on request. The first Durbin–Watson statistic refers to the first-step Within-OLS residuals, the second statistic is based on the residuals from the AR1 regression.The adjusted R squared is given for both the original and, in parentheses, the transformed data. ***/**/*Significant at the 0.01/0.05/0.10 error level respectively. DOES THE GOVERNED CORPORATION PERFORM BETTER? 513 Table 19.4B Robust estimates of the return on total assets equation, 1992–1996. Sample split by location of control rights

Regression Coefficient difference Dependent coefficients (absolute t-statistic) variable: ROA (abs. T-statistic)

Explanatory variable INDFIRM FAMILY FININST MIX FOREIGN CHANGE

Ownership concentration 0.0028 0.0060 0.0129 0.0146 0.0061 0.0098 (1.00) (1.72)* (2.09)** (2.56)** (1.14) (2.73)*** Board representation 0.0852 0.1024 0.1242 0.1176 (1.63) (1.57) (1.30) (1.23) Firm size 0.0529 0.0633 0.0442 0.0498 0.0360 0.0464 (1.74)* (1.74)* (0.43) (0.96) (0.78) (0.84) Firm growth 0.1104 0.0131 0.0822 0.0726 0.0176 0.0256 (6.50)*** (0.48) (1.19) (1.67)* (0.69) (0.68) Capital intensity 0.0520 0.0023 0.0634 0.0014 0.0021 0.1104 (1.74)* (0.70) (1.55) (0.03) (0.54) (1.69)* Capital structure (t1) 0.0180 0.2539 0.1021 0.0208 0.0091 0.0655 (0.35) (3.55)*** (1.15) (0.16) (0.75) (0.56) Market concentration 0.3048 0.2177 1.3709 0.7218 0.2214 0.0010 (1.05) (0.39) (0.78) (0.77) (0.54) (0.00) Mean intercept 3.1126 0.0590 0.1174 0.0080 0.0455 0.0782 (136.7)*** (1.94)* (1.97)** (0.23) (1.15) (1.64)* Number of firms (obs.) 81 (405) 122 (610) 20 (100) 54 (270) 58 (290) 26 (130) adj. R squared 0.78 (0.78) Durbin–Watson 1.56**, 1.86

The estimating sample contains 361 German firms as described in Table I. The split criterion is the location of control rights (identity of voting block owners).The estimating regression is

j ROAit b1OCit b2Bit b3Sit b4Git b5Kit b6Ci,t 1 b7H it 5 7 z1k1 dkzDzi xkzitait it in which ROA, OC, B, S, G, K, C, and H are the return on total assets, ownership concentration (unbounded Herfindahl index), board representation, firm size, firm growth, capital intensity, capital structure, and market concentration of firm i operating in industry j at time t. ai is a firm-specific effect, t a time dummy, and it the stochastic disturbance. The split sample estimates are based on the full sample regression with the set of explanatory variables x {OC, B, S, G, K, C, H} interacted with a set of dummy variables Dz which take on unit value for firms from z {FAMILY, FININST, MIX, FOREIGN, CHANGE} respectively and zero value for other firms. See Tables 19.2A and 19.2B for definitions of variables.The regression coefficients b estimate the impact of the respective explanatory variables on ROA for firms from the INDFIRM group, arbitrarily chosen as the base group, while the coefficients d yield the difference between the coefficients b and the coefficients for other groups of firms as represented by z. Adding the respective “coefficient difference” to the base group’s “regression coefficient”, yields the regression coefficients for the other groups of firms. The estimates shown above are robust to heteroskedasticity (White, 1980) and first-order serial correlation (AR1) (see Hsiao, 1986, pp. 55). The Within-OLS estimator was employed to generate consistent first-step estimates. We used the Within residuals to estimate a firm-specific serial correlation coefficient rho. Alternatively, we restricted the model to have a serial correlation coefficient common to all firms.The rhos were used to transform the original data. Regressions based on the transformed data were then estimated by OLS (firm-specific rho) or Maximum Likelihood (common rho). The coefficient on board representation was not estimable for the firms in MIX and CHANGE due to an insufficient number of observations. In line with Table 19.4A, we report the estimates based on the “common rho” approach. Again, the coefficients of all other explanatory variables can be estimated in the “firm-specific rho” approach.These latter coefficient estimates are qualitatively identical and quantitatively only marginally different from those shown below so we do not report them separately.The estimates are available from the authors on request.The first Durbin–Watson statistic refers to the first-step Within-OLS residuals, the second statistic is based on the residuals from the AR1 regression.The adjusted R squared is given for both the original and, in parentheses, the transformed data. – Not estimable due to an insufficient number of observations in the respective group. ***/**/* Significant at the 0.01/0.05/0.10 error level respectively. 514 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 19.5 OLS estimates of the firm-specific fixed effects equation, 1992–1996 Dependent variable: fixed firm-specific effect Regression coefficient (absolute t-statistic)

Explanatory variable (1) (2) (3)

INDFIRM (base group) 3.1346 (53.87)*** FAMILY 0.0541 (1.70)* FININST 0.0996 (1.96)** MIX 0.0279 (0.84) FOREIGN 0.0290 (0.74) CHANGE 0.0857 (1.85)* QUOTED 0.0140 (0.56) QUOTED INDFIRM (base group) 3.1406 (37.48)*** 0.9066 (4.19)*** FAMILY 0.0233 (0.48) 0.0368 (1.18) FININST 0.1432 (2.26)** 0.0593 (1.54) MIX 0.0373 (0.69) 0.0508 (1.48) FOREIGN 0.0517 (0.73) 0.1156 (2.24)** CHANGE 0.0415 (0.70) 0.0509 (1.44) Ownership concentration 0.0048 (2.62)*** Board representation 0.0131 (0.55) NON-QUOTED INDFIRM 0.0320 (0.62) 0.1112 (2.61)*** FAMILY 0.0030 (0.07) 0.0805 (2.00)** FININST 0.1009 (1.20) 0.1021 (1.65)* MIX 0.0938 (1.98)** 0.0227 (0.63) FOREIGN 0.0093 (0.13) 0.0377 (0.64) CHANGE 0.0748 (0.59) 0.0312 (0.78) Ownership concentration 0.0066 (2.62)*** Board representation 0.0079 (0.21) Firm size 0.0872 (15.42)*** Firm growth 0.1404 (2.01)** Capital intensity 0.0145 (0.92) Capital structure 0.2529 (5.83)*** Market concentration 0.1009 (0.52) Industry effects F(29,326) 42.91*** F(29,312) 52.49*** Adj. R2 0.0377 0.1854 0.6804

The estimating sample contains 361 German firms as described in Table 19.1.The estimating regression is

âi 1 2FAMILYi 3FININSTi 4MIXi 5FOREIGNi 6CHANGEi d1 Di INDFIRMi d2Di FAMILYi d3Di FININSTi d4Di MIXi d5Di FOREIGNi d6Di CHANGEi aj i with j aˆi ROAi bˆ1OCi bˆ2Bi bˆ3Si bˆ4Gi bˆ5Ki bˆ6Ci bˆ7Hi. j The firm-specific effects are denoted by ROAi, OCi, Bi, Si, Gi, Ki, Ci, Hi are the time means (1992–1996) of the return on total assets, ownership concentration (unbounded Herfindahl index), board representation, firm size, firm growth, capital intensity, capital structure, and market concentration of firm i operating in industry j. See Table 19.2A for variable definitions. INDFIRM, FAMILY, FININST, MIX, FOREIGN, and CHANGE are the mutually excluding groups referring to the location of control rights as defined in Table 19.2B. aj is an industry- specific effect and ‡i the stochastic disturbance. The coefficients ˆbk (k 1,..., 7) are the robust estimates as shown in Table 19.2A, column 1. The coefficients 2 to 6 estimate the difference between the average fixed firm-specific effect 1 of the chosen base group INDFIRM and the average fixed firm-specific effect of the other groups. Adding the respective coefficient difference 2 to 6 to the base group’s regression coefficient 1, yields the regression coefficients for the other groups of firms.The dummy variable D takes on unit value for non-quoted firms and zero value for quoted firms. The coefficients d estimate the difference in the coefficients between quoted and non-quoted firms. The OLS cross-section estimates below are robust to heteroskedasticity (White, 1980). ***/**/*Significant at the 0.01/0.05/0.10 error level. DOES THE GOVERNED CORPORATION PERFORM BETTER? 515 the regression constant in (3) as its Controlling for differences in ownership regression coefficient, we obtain the mean concentration, firm size and capital structure firm-specific effects and the corresponding changes the magnitude, sign and significance t-statistics on the differences as reported in of the coefficients on the identity indica- Table 19.4B. Column 1 of Table 19.5 adds tors compared with column 2. Our in stock market exposure and industry multivariate test now reveals that the indicators as further determinants of non-quoted FININST significantly under- autonomous ROA. The coefficient on stock performed with respect to the base group market exposure is positive but not signifi- of quoted INDFIRM firms, whereas the cantly so. This result is consistent with quoted FOREIGN firms and the non-quoted Table 19.4A where we ignored the location INDFIRM and FAMILY firms have of control rights and industry effects when significant larger fixed effects than the comparing quoted and non-quoted firms. base group.35 All other differences are The included industry effects are highly statistically insignificant at conventional significant. error levels. In column 2 of Table 19.5, we interact Summing up, the changes in the estimates the identity and stock market indicators to on the identity indicators underline that, explore whether there is a combined impact given firm size, investment opportunities of the location of control rights and stock (growth) and capital structure, it is the market exposure on the fixed firm-specific combination of the size of the equity stake effects. Significant differences emerge for held, the identity of owners and stock the MIX firms for which the lack of stock market exposure that matters for firm market exposure lowers profitability.Further, performance in the longer run. as previously found, firms with financial institutions as controlling stakeholders perform better than the “managed” base 4. SUMMARY AND CONCLUSION group. However,the regressions of columns 1 and 2 ignore that the firm-specific effects Almost a decade ago Michael C. Jensen may be correlated with the explanatory 34 (1993, p. 873) laid out the research variables of the original ROA regression. agenda for the new millenium in the field of Therefore, column 3 adds in the regres- corporate governance, corporate finance, sion the time means of these variables, and corporate performance: distinguishing also ownership concentration and board representation by stock For those with a normative bent, making market exposure. the internal control systems of corpora- Ownership concentration significantly tions work is the major challenge facing improves the autonomous ROA for the economists and management scholars in quoted firms but lowers it for the non-quoted the 1990s. For those who choose to take firms. As the quoted firms are significantly a purely positive approach, the major less concentrated (see Table 19.2A), this challenge is understanding how these result supports the view that the presence systems work, and how they interact of large shareholders enhances the long-term with other control forces (in particular performance of traded corporations. For the product and factor markets, legal, the non-quoted firms higher concentration political, and regulatory systems, and implies weaker performance both in the the capital markets) impinging on the short (Table 19.4A) and in the long run corporation. (Table 19.5). Board representation, market concentration and capital intensity do This paper contributes to the positive not make a difference. Autonomous ROA approach. We focused on German firms is significantly larger for larger and because ownership concentration is an growing firms as well as for firms with important feature setting the German more equity capital. system of corporate governance apart 516 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY from the Anglo-Saxon. In the corporate profitability (by exploiting investment governance literature ownership concentra- opportunities better). The location of tion is often understood as reflecting a control rights seems to be more important stronger governing effort of owners. By for non-quoted firms.While the representa- reducing informational asymmetries between tion of the largest shareholder on the board owners and managers as well as between of executive directors does not affect the the firm and external investors ownership ROA of quoted firms, it makes a significant concentration is expected to affect firm positive difference for the non-quoted firms profitability positively. Contrary to this in the short run but not in the long run. argument, we find a significantly negative Finally, can we answer the question posed impact of ownership concentration on in this paper: Do governed corporations profitability as measured by the return perform better? If one is willing to view on total assets. This result holds for both firms owned by families, financial institu- quoted and non-quoted firms and supports tions, or a mix of large shareholders as the view that large shareholders inflict governed rather than managed, the answer costs on the firm (e.g., rent extraction, too to the question seems to be yes. These much monitoring, or infighting). groups have significantly higher mean and Ownership concentration can be an median profitabilities than the group of insufficient or misleading indicator of the firms owned by another industrial firm. control owners actually exert. We therefore After controlling for other potential ROA checked the location of control rights and determinants such as firm size, stock also the time horizon. When focussing on market exposure, industry effects, or the short-run ROA, the negative effect of owner- time horizon, these differences between ship concentration originates primarily with owner categories are less pronounced but the firms owned by families (individuals) or still present. Most of the variation in ROA different large shareholders as well as across firms is explained by firm- or firms which experienced a change in owners industry-specific effects. To get a more during the observation period. A positive comprehensive picture of how governance impact of ownership concentration on structures affect firm performance, future profitability is found for firms with financial research faces the challenge to identify institutions as largest shareholders. This what is behind these effects. latter result is consistent with the view that In conclusion, our study finds systematic banks are better (more efficient) monitors influences of ownership concentration, so that agency costs are lower. stock market exposure, and the location of When we look at the long-term perform- control rights on the profitability of German ance we find that ownership concentration corporations. Our results imply that (1) the significantly improves the ROA of firms presence of large shareholders does not exposed to the stock market. This result is necessarily improve performance as meas- consistent with the standard view advanced ured by ROA, (2) concentrated ownership in the literature that in the bank- or network- appears to be a sub-optimal choice for oriented system of corporate governance, many of the tightly held and non-traded committed owners take a long-term horizon German corporations, and, finally, (3), on their investments at the expense of a having financial institutions as largest higher short-term return. However, and the shareholders of traded corporations qualification warrants attention, this con- enhanced profitability. clusion does not hold for the non-quoted firms. For the non-quoted firms ownership concentration influences profitability ACKNOWLEDGMENTS negatively, not only in the short but particularly so in the long run. An interpretation The authors are indebted to Bob Chirinko is that the owners favor control over both for extensive and insightful comments on higher liquidity (by going public) and higher the very first draft of the paper and, most DOES THE GOVERNED CORPORATION PERFORM BETTER? 517 of all, for his encouragement. We are also 2 See Short (1994) and Shleifer and Vishny grateful to an anonymous referee for (1997) for surveys of the field. substantial advice which greatly improved 3 Allen and Gale (2000, ch. 2 and 3) provide the paper. Many helpful suggestions on an excellent overview of the financial systems in various drafts were received from Rashjree the USA, UK, Germany, Japan, and France. 4 See, e.g., Porter (1992), Jensen (1993), Agarwal, David Audretsch, Matthew Roe (1994), Bhide (1994), Pound (1995). As Bennett, Hans Degryse, Paul De Grauwe, Bhide (1994, p. 129, p. 131) argues,“Unwittingly, Julie Ann Elston, Oliver Fabel, Rainer the system [of U.S. securities regulations Feuerstack, Günther Franke, Carolin Frohlin, and disclosure rules] nurtures market liquidity George Gelauff, Sabine Langner, Manfred at the expense of good governance....U.S. Neumann, Marco Pagano, Hans-Jürgen rules that protect investors don’t just sustain Ramser, and seminar participants at the market liquidity, they also drive a wedge 1999 Meetings of the European Finance between shareholders and managers. Instead of Association (EFA, Helsinki), European yielding long-term shareholders who concen- Economic Association (EEA, Santiago de trate their holdings in a few companies, where they provide informed oversight and counsel, Compostela), European Association for the laws promote diffused, arm’s length stock- Research in Industrial Economics (EARIE, holding”.The benefits typically attributed to this Torino), and Western Economic Association so-called market-based system are seen in a (WEA, San Diego) as well as at the better provision of finance to innovative start-up University of Central Florida Orlando, firms and higher returns to investors. Humboldt University Berlin, University of 5 In December 1998, 323 domestic corpora- Konstanz, and Katholieke Universiteit tions with an aggregate market capitalization of Leuven. The second author would like to 1,822,103 million DM (equivalent to 48% of thank Ulrich Hege, Piet Moerland, and the 1998 German GDP) were quoted and offi- Luc Renneboog for stimulating discussions cially traded on the Frankfurt stock exchange. At the same time, 2,399 domestic corporations on corporate governance. As a matter of (3,829,375 million DM, 182% of GDP) were fact, any remaining errors are the authors’ listed on the London stock exchange and 2,722 sole responsibility. Generous financial (17,373,835 million DM, 105% of GDP) support was provided by the Bundesland on the New York stock exchange. (Source: Mecklenburg-Vorpommern (first author) Deutsche Börse, Factbook, Frankfurt, http:// and the German Science Foundation (DFG, www.exchange.de; own calculations.) Boehmer second author). Main parts of the paper (1999) provides a thorough analysis of owner- were written while the second author was ship structures and location of control rights for visiting Georgia State University (School the 430 stock corporations which were officially of Policy Studies, Atlanta) and Indiana traded on the Frankfurt stock exchange in 1996. According to this study, banks, industrial University (School of Public and firms, holdings, and insurance companies Environmental Affairs, Bloomington). The controlled as large blockholders almost 80% hospitality and support of these institutions of the median firm’s voting rights, or roughly is gratefully acknowledged. This research 50% of the overall market value of firms is not related to any CPB Netherlands officially listed. Bureau for Economic Policy Analysis 6 The high tax on the sale of company cross- projects. The opinions expressed here do holdings (to be abolished soon) and the rigid not necessarily reflect those of the CPB. German takeover code (under revision) which requires shareholders to be bought out with cash have contributed to making takeovers unattractive. NOTES 7 See, e.g., Edwards and Fischer (1994), Perlitz and Seeger (1994), Audretsch and 1 See for a discussion and references, e.g., Elston (1997), or Wenger and Kaserer (1998). Porter (1992), Jensen (1993), Blair (1995), The implementation of the Neuer Markt (new Berglöf (1997), Shleifer and Vishny (1997), market) on which shares in innovative young Mayer (1998), Hopt et al. (1998), Barca and firms are traded can be seen as a step in the Becht (1999), OECD (1999), Allen and Gale direction of a more market-based system of (2000), and Vives (2000). corporate finance. 518 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 8 See, e.g., Fama and Jensen (1983), Short corporations over the observation period 1965 (1994) and Shleifer and Vishney (1997) for dis- to 1990 but due to incomplete ownership infor- cussion and references to seminal contributions. mation cannot take advantage of the panel data 9 In Germany, the case of Metallgesellschaft properties. Weigand uses a panel data set which may serve as an example of supervisory board includes detailed ownership data (identity of failure. Deutsche Bank, as a major debtholder owners, outstanding shares) for 240 stock represented on the supervisory board, was corporations during 1965 to 1986. regarded responsible for the disaster and 16 Franks and Mayer (1997) apply a panel scolded in the business press for insufficient data set of 171 quoted corporations over the and bad monitoring. See Frankel and Palmer period 1989 to 1994 and use a fixed effects (1996) for further discussion. See also Franks panel regression approach. There is only weak and Mayer (1998) on the dubious role of banks evidence of a relationship between either in the few German corporate takeovers. management or supervisory board turnover and 10 What insufficient control over a firm’s performance in firms with concentrated owner- assets means, is nicely depicted by the General ship. This result is consistent with Kaplan’s Motors example cited in Roe (1994, XII). In (1994) findings. Goergen’s (1999) sample 1990, two ‘large’ (relative to all other share- consists of 86 IPOs during the period 1981 to holders) institutional investors, being dissatisfied 1988. In both static and dynamic panel data with the company’s bad performance, wanted to regressions, ownership concentration did not negotiate the implementation of a new CEO have an impact on profitability (measured by with GM’s leaders. The management could cash flow over total assets). calmly decline the request, since each ‘large’ 17 Gedajlovic and Shapiro (1998) studied the shareholder only accounted for less than one blockholdings-performance link for the largest per cent of the voting capital. In 1992 GM’s top firms from five major industrialized countries. management had to take action after all. As Their sample includes 99 publicly traded German losses piled up to over $6 billion, CEO Robert stock corporations over the period 1986 to Stempel was eventually fired. 1991. The U-formed relationship between 11 See Hubbard (1998) for a recent survey ownership concentration and ROA was also on investment and capital market imperfections. found for the US firms. 12 See, e.g., Edwards and Fischer (1994) or 18 Becht (1999) uses the corporations Hellwig (2000) for discussion and references. represented in the DAX100 stock market index 13 Boehmer (1999, 2000) provides in-depth during 1996 to 1998. descriptive analyses of governance structures 19 This definition is equivalent to earnings and stock market performance of traded German before interest, taxes, and depreciation corporations. As his approach is different from (EBITDA). the regression studies surveyed here, we will 20 Pension liabilities have been added for two refer to some of his results below. reasons. First, it is peculiar to the German 14 A firm was defined as owner-controlled if system of accounting that pension assets individuals or families held at least the blocking and pension liabilities are not netted out in minority (25% of voting capital plus one vote). companies’ balance sheets. Further, pension A drawback of the study is that ownership liabilities are not paid into a trust (pension structures were identifiable only for about 90 fund) but remain within the firm. They are out of 300 sample firms. Thus in the regression available to the firm as a source of internal analysis owner-controlled firms had to be long-term finance. Pension liabilities thus can be compared to a mixture of presumably manager- seen as ‘quasi’ equity. Second, the shareholders’ controlled firms and firms with unknown equity of limited liability companies (GmbH) governance structures, rendering the results is, by legal construction, extremely low. Adding questionable. reserves and pension liabilities to shareholders’ 15 Cable (1985), Chirinko and Elston (1996) equity helps avoid generating unrealistically and Gorton and Schmid (2000) apply a cross- high returns on equity for these firms compared section regression approach. Cable’s analysis to stock corporations. for the time period 1968 to 1972 is based on 48 21 We referred to the Bundesanzeiger or out of Germany’s 100 largest corporations in requested annual reports from firms to double- 1970. The Gorton and Schmid study is also check and correct some entries in the based on a very small sample and only two years Hoppenstedt Bilanzdatenbank (inconsistencies (82 firms in 1975) and (56 firms in 1986). in reporting, missing values, obviously wrong Chirinko and Elston investigate 300 stock entries etc.). DOES THE GOVERNED CORPORATION PERFORM BETTER? 519 22 This exclusion refers to the subsequent 30 With respect to the IO variables in the empirical analysis of profitability only. Of extended model we find ownership concentra- course, a holding company can be the owner of tion to be significantly lower for firms which a sample firm. are larger, use more capital per employee, have 23 In many cases the Hypo-Guide also lists more equity capital, and, hardly surprising, blocks smaller than 25% and indicates indirect have more large shareholders. By contrast, ownership (voting rights granted to a large growing firms and firms operating under more shareholder from other shareholders). Sometimes concentrated market structures have more con- even very small blockholdings (5% and less) centrated ownership. In an alternative try, are reported. we endogenized board representation and the 24 As disclosure of smaller blockholdings was number of large shareholders. The results were not mandatory during the observation period, qualitatively identical. The Wu–Hausman test the calculated Herfindahl index can only be an did not indicate the endogeneity of board repre- approximation to the true degree of ownership sentation or the number of large shareholders. concentration. Two alternative Herfindahl 31 As some of the variables in the regression indices were constructed for corporations with are expressed in the natural logarithm, the missing information on this “dispersed” portion means of the firm-specific intercepts are of of shareholdings. One measure treats the “dis- different magnitude than mean ROA in the persed” portion as one block. Using the means summary statistics tables. in Table 19.1 for corporations with only one 32 This model specification gives estimates large shareholder as an example, the Herfindahl identical to those from separate regressions for index is (89.07)2 (10.93)2 8,053. The each group of owners but considers the across- other measure assumes that the dispersed por- groups variation of the regression residuals in tion is uniformly distributed among an unknown the variance-covariance matrix. number of shareholders, each holding at most 33 Recall that the Herfindahl index can be 1% of the shares outstanding, which yields a decomposed into n1 nV where, in our case, n Herfindahl index of (89.07)2 10 (1.00)2 and V denote the number of (large) shareholders (0.93)2 7,944.The results to be presented and the variance of individual stakes respectively. below are not affected by the choice of the 34 This is suggested by the Hausman specifi- calculation method. cation test in Table 19.3A (model E vs. model F) 25 The results are available from the authors from which we concluded that the firm- on request. specific effects are fixed rather than random. 26 The Herfindahl-Index in its standard The test does indeed check whether these definition is restricted to take on values between effects are correlated with the right-hand 0 and 10,000. The logit transformation log variables or not. [H/(10,000 H)] yields an “unbounded” 35 There are only two non-quoted firms with variable. In the case of Mannesmann, for which bank involvement. These two firms operated ownership of voting shares is reported to be with mean ROAs of 0.16 and 0.09 respectively “100% dispersed”, H was set at 1 in the in the notoriously declining and restructuring transformation. For firms with only one owner mining industry. The mean ROA for the quoted holding 100% of voting stock, H was set at firms with bank involvement is 0.28. 9,999. In alternative regression runs we used the bounded Herfindahl-Index as well as the share of the largest shareholder plus its squared value as in Gedajlovic and Shapiro (1998). REFERENCES Further, we replaced ROA by ROE. As the estimates with respect to governance indicators Admati, A., Pfleiderer,P., and Zechner,J. (1994), do not differ significantly, we do not report Large shareholder activism, risk sharing, the results from these alternative specifications and financial market equilibrium, Journal here. They are available from the authors on of Political Economy 102, 1097–1130. request. Allen, F. and Gale, F. (1998), Corporate 27 See for a recent discussion, e.g., Church and Ware (2000, ch. 12). Governance and Competition, Working paper, 28 See the discussion in Hall and Weiss New York University. (1967). Allen, F. and Gale, D. (2000), Comparing 29 We discuss the impact of the IO variables Financial Systems. MIT Press, Cambridge below. and London. 520 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Audretsch, D. and Elston, J. (1997), ‘Financing Bolton, P. and von Thadden, E.-L. (1998b), the German Mittelstand’, Small Business ‘Liquidity and control’, Journal of Institutional Economics Journal 9, 97–110. and Theoretical Economics 154, 177–215. Audretsch, D. and Weigand, J. (1999), Does Breusch,T. and Pagan, A. (1980), ‘A simple test Science Make a Difference? Investment, for heteroscedasticity and random coefficient Finance and Corporate Governance in variation’, Econometrica 47, 1287–1294. German Industries, CEPR working paper Burkhart, M., Gromb, D., and Panunzi, F. No. 2056, London. (1997), ‘Large shareholders, monitoring, and Baltagi, B. H. (1995), Econometric Analysis of the value of the firm’, Quarterly Journal of Panel Data. Wiley, New York. Economics 112, 693–728. Baumol, W. (1959), Business Behavior, Value, Cable, J. (1985), ‘Capital market information and Growth. Macmillan, New York. and industrial performance: The role of Baums, T. and Fraune, C. (1995), ‘Institution- West German banks’, Economic Journal 95, elle Anleger und Publikumsgesellschaft: 118–132. Eine empirische Untersuchung’, Die Chirinko, R. and Elston, J. (1996), Finance, Aktiengesellschaft 40, 97–112. Control, and Profitability: An Evaluation of Barca, F. and Becht, M. (eds.) (1999), German Bank Influence, American Institute Ownership and Control in Europe. Oxford for Contemporary German Studies:The Johns University Press, Oxford. Hopkins University, Working Paper No. 28, Becht, M. (1999), ‘European corporate gover- Washington, D.C. nance: Trading off liquidity against control’, Church, J. and Ware, R. (2000), Industrial European Economic Review 43, 1049–1056. Organization: A Strategic Approach. Becht, M. and Boehmer,E. (1997), Transparency McGraw-Hill, New York. of Ownership and Control in Germany, Cubbin, J. and Leech, D. (1983), ‘The effect Working paper, Humboldt University, Berlin. of shareholding dispersion on the degree of Bergloef, E. (1997), ‘Reforming corporate control in British companies: Theory and governance: Redirecting the European evidence’, Economic Journal 93, 351–369. agenda’, Economic Policy, 93–123. Demsetz, H. (1983),‘The structure of ownership Berle, A. and Means, G. (1932), The Modern and the theory of the firm’, Journal of Law Corporation and Private Property. MacMillan, and Economics 26, 375–390. New York. Demsetz, H. and Lehn, K. (1985),‘The structure Bhargava, A., Franzini, L., and Narendanathan, of corporate ownership: Causes and conse- W. (1982), ‘Serial correlation and the fixed quences’, Journal of Political Economy 93, effects model’, Review of Economic Studies 1155–1177. 49, 533–549. Diamond, D. (1984), ‘Financial intermediation Bhide, A. (1994), ‘Efficient Markets, Deficient and delegated monitoring’, Review of Governance’, Harvard Business Review, Economic Studies 51, 393–414. November–December, 129–139. Diamond, D. (1991),‘Monitoring and reputation: Blair, M. (1995), Ownership and Control: The choice between bank loans and directly Rethinking Corporate Governance for the placed debt’, Journal of Political Economy 99, Twenty-First Century, The Brookings 689–721. Institution, Washington, D.C. Edwards, J. and Fischer, K. (1994), Banks, Boehmer, E. (1999), Who Controls Germany? Finance, and Investment in Germany. An Exploratory Analysis, Working paper, Cambridge University Press, Cambridge. Humboldt University, Berlin. Fama, E. and Jensen, M. (1983), ‘Separation of Boehmer, E. (2000), ‘Business groups, bank ownership and control’, Journal of Law and control, and large shareholders: An analysis Economics 26, 301–325. of German takeovers’, Journal of Financial Fluck, Z. (1999), ‘The dynamics of the Intermediation 9, 117–148. management-shareholder conflict’, Review Bolton, P. and von Thadden, E.-L. (1998a), of Financial Studies 12, 379–404. ‘Blocks, liquidity, and control’, Journal of Frankel, A. and Palmer, D. (1998), Finance 53, 1–25. ‘The management of financial risks at DOES THE GOVERNED CORPORATION PERFORM BETTER? 521 German nonfinancial firms: The case of Huddart, S. (1993), ‘The effects of a large Metallgesellschaft, in S. Black and M. Moersch, shareholder on corporate value’, Management eds., Competition and Convergence in Financial Science 39, 1407–1421. Markets: The German and Anglo-American Jensen, M. C. (1986),‘Agency costs of free cash Models, Advances in Finance, Investment flow, corporate finance and takeovers’, and Banking, vol. 5, Elsevier Science, North- American Economic Review 76, 323–329. Holland, Amsterdam, pp. 41–78. Jensen, M. C. (1988), ‘Takeovers: Their causes Franks, J. and Mayer, C. (1997), Ownership, and consequences’, Journal of Economic Control and the Performance of German Perspectives 2, 21–48. Corporations, Paper presented at the Jensen, M. C. (1993), ‘The modern industrial Center for Financial Studies Conference, revolution, exit, and the failure of internal November 19–20, Frankfurt a.M. control systems’, Journal of Finance 48, Franks, J. and Mayer, C. (1998), ‘Bank control, 831–880. takeovers, and corporate governance in Jensen, M. C. and Ruback, R. (1983).‘The market Germany’, Journal of Banking and Finance for corporate control: The scientific evidence’, 22, 1385–1403. Journal of Financial Economics 11, 5–50. Gedajlovic, E. and Shapiro, D. (1998), Kaplan, S. (1994), ‘Top executives, turnover, ‘Management and ownership effects: Evidence and firm performance in Germany’, Journal from five countries’, Strategic Management of Law, Economics, and Organization 10, Journal 19, 533–553. 142–159. Goergen, M. (1999), Corporate Governance Lang, L. and Stulz, R. (1994),‘Tobin’s q, corpo- and Financial Performance. Edward Elgar, rate diversification, and firm performance’, Cheltenham. Journal of Political Economy 102, Gorton, G. and Schmid, F. (2000), ‘Universal 1248–1280. banking and the performance of German Leibenstein, H. (1966),‘Allocative efficiency vs. firms’, Journal of Financial Economics 58, “X-Efficiency”’, American Economic Review forthcoming. 56, 392–415. Grossman, S. and Hart, O. (1980), ‘Takeover Manne, H. G. (1965), ‘Mergers and the market bids, the free-rider problem and the theory of for corporate control’, Journal of Political the corporation’, Bell Journal of Economics Economy 73, 110–120. 11, 42–64. Mayer, C. (1992), Corporate finance, in Hausman, J. A. (1978), ‘Specification Tests in P. Newman, M. Milgate and J. Eatwell, eds., Econometrics’, Econometrica 46, 1251–1272. The New Palgrave Dictionary of Money & Hall, M. and Weiss, L. (1967), ‘Firm size and Finance, New York. profitability’, Review of Economics and Mayer, C. (1996), ‘Corporate governance, com- Statistics 47, 319–331. petition, and performance’, OECD Economic Hart, O. (1983), ‘The market mechanism as an Studies 27, 7–34. incentive scheme’, Bell Journal of Economics Mayer, C. (1998), ‘Financial systems and 14, 366–382. corporate governance: A review of the inter- Hellwig, M. (2000),‘On the economics and politics national evidence’, Journal of Institutional of corporate finance and corporate control’, and Theoretical Economics 154, 144–165. in Xavier Vives, ed., Corporate Governance, Moerland, P. W. (1995), ‘Corporate ownership Cambridge University Press, Cambridge. and control structures: An international Hopt, K., Kanda, H., Roe, M. J., Wymersch, E., comparison’, Review of Industrial Organization and Prigge, S. (1998), Comparative Corporate 10, 443–464. Governance: The State of the Art and OECD (1999), Principles of Corporate Emerging Research, Clarendon Press, Oxford. Governance, OECD, Paris. Hsiao, C. (1986), Analysis of Panel Data, Pagano, M. and Röell, A. (1998), ‘The choice of Cambridge University Press, Cambridge. stock ownership structure: Agency costs, Hubbard, R. G. (1998), ‘Capital-market monitoring, and the decision to go public’, imperfections and investment’, Journal of Quarterly Journal of Economics 113, Economic Literature 36, 193–225. 187–225. 522 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Perlitz, M. and Seger, F. (1994), ‘The role of Stulz, R. (1988), ‘Managerial control of voting universal banks in German corporate rights: Financing policies and the market for governance’, Business and the Contemporary corporate control’, Journal of Financial World 4. Economics 20, 25–54. Porter, M. E. (1992), ‘Capital disadvantage: Thompson, J. K. (1998),‘Shareholder value and America’s failing capital investment system’, the market for corporate control in OECD Harvard Business Review, September– countries’, OECD Financial Market Trends October, 65–82. 69, February, 15–37. Pound, J. (1995), ‘The promise of the governed Thonet, P. J. and Poengsen, O. (1979), corporation’, Harvard Business Review, ‘Managerial control and economic perform- March–April, pp. 89–98. ance in West Germany’, Journal of Industrial Roe, M. (1994), Strong Managers, Weak Economics 28, 23–37. Owners: The Political Roots of American Vives, X. (ed.) (2000), Corporate Governance, Corporate Finance, Princeton: Princeton Cambridge University Press, Cambridge. University Press. Warner, J., Watts, R., and Wruck, K. (1988), Schmidt, K. (1997), ‘Managerial incentives ‘Stock prices and top management changes’, and product market competition’, Review of Journal of Financial Economics 20, Economic Studies 64, 191–213. 461–492. Shin, H.-H. and Stulz, R. (1998), ‘Are internal Weigand, J. (1999): Corporate Governance, capital markets efficient?’, Quarterly Journal Profitability, and Capital Structure, unpublished of Economics 113, 531–552. post-doctoral thesis (Habilitationsschrift), Shleifer, A. and Vishny, R. (1986), ‘Large Nuernberg. shareholders and corporate control’, Journal Wenger, E. and Kaserer, C. (1998), ‘The German of Political Economy 95, 461–488. system of corporate governance – A model Shleifer,A. and Vishny, R. (1988),‘Management that should not be imitated’, in S. Black and entrenchment: The case of manager-specific M.Moersch, eds., Competition and Convergence investment’, Journal of Financial Economics in Financial Markets: The German and 25, 123–139. Anglo-American Models, Advances in Finance, Shleifer, A. and Vishny, R. (1997), ‘A survey of Investment and Banking, vol. 5. Elsevier corporate governance’, Journal of Finance Science, North-Holland, Amsterdam. 52, 737–783. White, H. (1980),‘A heteroscedasticity-consistent Short, H. (1994), ‘Ownership, control, finan- covariance matrix estimator and a direct cial structure, and the performance of test for heteroscedasticity’, Econometrica 48, firms’, Journal of Economic Surveys 8, 817–838. 203–249. Wu, D. (1972), ‘Alternative tests of inde- Stiglitz, J. (1985), ‘Credit markets and the pendence between stochastic regressors control of capital’, Journal Money, Credit and disturbances’, Econometrica 41, and Banking 17, 133–152. 733–750. Chapter 20

Paul Gompers, Joy Ishii and Andrew Metrick CORPORATE GOVERNANCE AND EQUITY PRICES*

Source:The Quarterly Journal of Economics, 118(1) (2003): 107–155.

ABSTRACT Shareholder rights vary across firms. Using the incidence of 24 governance rules, we construct a “Governance Index” to proxy for the level of shareholder rights at about 1500 large firms during the 1990s. An investment strategy that bought firms in the lowest decile of the index (strongest rights) and sold firms in the highest decile of the index (weakest rights) would have earned abnormal returns of 8.5 percent per year during the sample period. We find that firms with stronger shareholder rights had higher firm value, higher profits, higher sales growth, lower capital expenditures, and made fewer corporate acquisitions.

I. INTRODUCTION Twenty years ago, large corporations had little reason to restrict shareholder rights. CORPORATIONS ARE REPUBLICS. THE ULTIMATE Proxy fights and hostile takeovers were rare, authority rests with voters (shareholders). and investor activism was in its infancy. By These voters elect representatives (directors) rule, most firms were shareholder democra- who delegate most decisions to bureaucrats cies, but in practice management had much (managers). As in any republic, the actual more of a free hand than they do today.The power-sharing relationship depends upon the rise of the junk bond market in the 1980s specific rules of governance. One extreme, disturbed this equilibrium by enabling which tilts toward a democracy, reserves hostile-takeover offers for even the largest little power for management and allows public firms. In response, many firms added shareholders to quickly and easily replace takeover defenses and other restrictions directors. The other extreme, which tilts of shareholder rights. Among the most toward a dictatorship, reserves extensive popular were those that stagger the terms power for management and places strong of directors, provide severance packages restrictions on shareholders’ ability to for managers, and limit shareholders’ replace directors. Presumably, shareholders ability to meet or act. During the same accept restrictions of their rights in time period, many states passed anti- hopes of maximizing their wealth, but takeover laws giving firms further defenses little is known about the ideal balance of against hostile bids. By 1990 there was power. From a theoretical perspective, considerable variation across firms in the there is no obvious answer. In this paper strength of shareholder rights.The takeover we ask an empirical question—is there market subsided in the early 1990s, but a relationship between shareholder rights this variation remained in place throughout and corporate performance? the decade. 524 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Most research on the wealth impact of a “Governance Index” as a proxy for the takeover defenses uses event-study method- balance of power between shareholders ology, where firms’ stock returns are and managers. Our index construction is analyzed following the announcement of straightforward: for every firm we add one a new defense.1 Such studies face the point for every provision that reduces difficulty that new defenses may be driven shareholder rights. This reduction of rights by contemporaneous conditions at the firm; is obvious in most cases; the few ambiguous i.e., adoption of a defense may both change cases are discussed. Firms in the highest the governance structure and provide a decile of the index are placed in the signal of managers’ private information “Dictatorship Portfolio” and are referred to about impending takeover bids. Event as having the “highest management power” studies of changes in state takeover laws or the “weakest shareholder rights”; firms are mostly immune from this problem, but in the lowest decile of the index are it is difficult to identify a single date for an placed in the “Democracy Portfolio” and event that is preceded by legislative negoti- are described as having the “lowest ation and followed by judicial uncertainty. management power” or the “strongest For these and other reasons, some authors shareholder rights.” argue that event-study methodology cannot In Section III we document the main identify the impact of governance provisions.2 empirical relationships between gover- We avoid these difficulties by taking nance and corporate performance. Using a long-horizon approach. We combine a performance-attribution time-series regres- large set of governance provisions into an sions from September 1990 to December index which proxies for the strength of 1999, we find that the Democracy Portfolio shareholder rights, and then study the outperformed the Dictatorship Portfolio by empirical relationship between this index a statistically significant 8.5 percent per and corporate performance. Our analysis year.These return differences induced large should be thought of as a “longrun event changes in firm value over the sample study”: we have democracies and dictator- period. By 1999 a one-point difference in ships, the rules stayed mostly the same the index was negatively associated with for a decade—how did each type do? Our an 11.4 percentage-point difference in main results are to demonstrate that, in the Tobin’s Q. After partially controlling for 1990s, democracies earned significantly differences in market expectations by higher returns, were valued more highly, using the book-to-market ratio, we also and had better operating performance. Our find evidence that firms with weak share- analysis is not a test of market efficiency. holder rights were less profitable and Because theory provides no clear prediction, had lower sales growth than other firms in there is no reason that investors in 1990 their industry. should have foreseen the outcome of this The correlation of the Governance Index novel experiment. Also, because this “exper- with returns, firm value, and operating iment” did not use random assignment, we performance could be explained in several cannot make strong claims about causality, ways. Section IV sets out three hypotheses but we do explore the implications and to explain the results. Hypothesis I is that assess the supportive evidence for several weak shareholder rights caused additional causal hypotheses.3 agency costs. If the market underestimated Our data are derived from publications of these additional costs, then a firm’s stock the Investor Responsibility Research Center. returns and operating performance would These publications provide 24 distinct have been worse than expected, and the corporate-governance provisions for approx- firm’s value at the beginning of the period imately 1500 firms since 1990.4 In Section would have been too high. Hypothesis II is II we describe these provisions and data that managers in the 1980s predicted poor sources in more detail. We divide the rules performance in the 1990s, but investors did into five thematic groups and then construct not. In this case, the managers could have CORPORATE GOVERNANCE AND EQUITY PRICES 525 put governance provisions in place to common stock (less than 10 percent of the protect their jobs. While the provisions total).5 The IRRC universe covers most of might have real protective power,they would the value-weighted market: even in 1990 not have caused the poor performance. the IRRC tracked more than 93 percent Hypothesis III is that governance provi- of the total capitalization of the combined sions did not cause poor performance (and New York Stock Exchange (NYSE), need not have any protective power) but American Stock Exchange (AMEX), and rather were correlated with other charac- NASDAQ markets. teristics that were associated with abnormal The IRRC tracks 22 charter provisions, returns in the 1990s. While we cannot bylaw provisions, and other firm-level rules identify any instrument or natural experi- plus coverage under six state takeover laws; ment to cleanly distinguish among these duplication between firm-level provisions hypotheses, we do assess some supportive and state laws yields 24 unique provisions. evidence for each one in Section V. For Table 20.1 lists all of these provisions, and Hypothesis I we find some evidence of Appendix 1 discusses each one in detail.We higher agency costs in a positive relation- divide them into five groups: tactics for ship between the index and both capital delaying hostile bidders (Delay); voting expenditures and acquisition activity. In rights (Voting); director/officer protection support of Hypothesis III we find several (Protection); other takeover defenses observable characteristics that can explain (Other); and state laws (State). up to one-third of the performance The Delay group includes four provisions differences.We find no evidence in support designed to slow down a hostile bidder. For of Hypothesis II. Section VI concludes takeover battles that require a proxy fight the paper. to either replace a board or dismantle a takeover defense, these provisions are the most crucial. Indeed, some legal scholars II. DATA argue that the dynamics of modern takeover battles have rendered all other II.A. Corporate-governance defenses superfluous [Daines and Klausner 2001; Coates 2000]. The Voting group provisions contains six provisions, all related to share- Our main data source is the Investor holders’ rights in elections or charter/bylaw Responsibility Research Center (IRRC), amendments.The Protection group contains which publishes detailed listings of six provisions designed to insure officers corporate-governance provisions for indi- and directors against job-related liability vidual firms in Corporate Takeover Defenses or to compensate them following a termi- [Rosenbaum 1990, 1993, 1995, 1998]. nation. The Other group includes the six These data are derived from a variety of remaining firm-level provisions. public sources including corporate bylaws These provisions tend to cluster within and charters, proxy statements, annual firms. Out of (22 * 21)/2 231 total reports, as well as 10-K and 10-Q pairwise correlations for the 22 firm-level documents filed with the SEC. The IRRC’s provisions, 169 are positive, and 111 of universe is drawn from the Standard & these positive correlations are significant.6 Poor’s (S&P) 500 as well as the annual In contrast, only 9 of the 62 negative lists of the largest corporations in the correlations are significant. This clustering publications of Fortune, Forbes, and suggests that firms may differ significantly Businessweek. The IRRC’s sample in the balance of power between investors expanded by several hundred firms in 1998 and management. through additions of some smaller firms The IRRC firm-level data do not include and firms with high institutional-ownership provisions that apply automatically under levels. Our analysis uses all firms in the state law. Thus, we supplement these data IRRC universe except those with dual-class with state-level data on takeover laws as 526 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 20.1 Governance provisions

Percentage of firms with governance provisions in

1990 1993 1995 1998

Delay Blank check 76.4 80.0 85.7 87.9 Classified board 59.0 60.4 61.7 59.4 Special meeting 24.5 29.9 31.9 34.5 Written consent 24.4 29.2 32.0 33.1 Protection Compensation plans 44.7 65.8 72.5 62.4 Contracts 16.4 15.2 12.7 11.7 Golden parachutes 53.1 55.5 55.1 56.6 Indemnification 40.9 39.6 38.7 24.4 Liability 72.3 69.1 65.6 46.8 Severance 13.4 5.5 10.3 11.7 Voting Bylaws 14.4 16.1 16.0 18.1 Charter 3.2 3.4 3.1 3.0 Cumulative voting 18.5 16.5 14.9 12.2 Secret ballot 2.9 9.5 12.2 9.4 Supermajority 38.8 39.6 38.5 34.1 Unequal voting 2.4 2.0 1.9 1.9 Other Antigreenmail 6.1 6.9 6.4 5.6 Directors’ duties 6.5 7.4 7.2 6.7 Fair price 33.5 35.2 33.6 27.8 Pension parachutes 3.9 5.2 3.9 2.2 Poison pill 53.9 57.4 56.6 55.3 Silver parachutes 4.1 4.8 3.5 2.3 State Antigreenmail law 17.2 17.6 17.0 14.1 Business combination law 84.3 88.5 88.9 89.9 Cash-out law 4.2 3.9 3.9 3.5 Directors’ duties law 5.2 5.0 5.0 4.4 Fair price law 35.7 36.9 35.9 31.6 Control share acquisition law 29.6 29.9 29.4 26.4 Number of firms 1357 1343 1373 1708

This table presents the percentage of firms with each provision between 1990 and 1998. The data are drawn from the IRRC Corporate Takeover Defenses publications [Rosenbaum 1990, 1993, 1995, 1998] and are supplemented by data on state takeover legislation coded from Pinnell [2000]. See Appendix 1 for detailed information on each of these provisions. The sample consists of all firms in the IRRC research universe except those with dual class stock. given by Pinnell [2000], another IRRC provisions given in other groups.We discuss publication. From this publication we code these analogues in subsection II.B. the presence of six types of so-called The IRRC data set is not an exhaustive “second-generation” state takeover laws listing of all provisions. Although firms can and place them in the State group.7 Few review their listing and point out mistakes states have more than three of these laws, before publication, the IRRC does not and only Pennsylvania has all six.8 Some update every company in each new edition of these laws are analogues of firm-level of the book, so some changes may be CORPORATE GOVERNANCE AND EQUITY PRICES 527 missed. Also the charter and bylaws are not require any judgments about the efficacy available for all companies and thus the or wealth effects of any of these provisions; IRRC must infer some provisions from we only consider the impact on the balance proxy statements and other filings. Overall, of power. the IRRC intends its listings as a starting For example, consider Classified Boards, point for institutional investors to review a provision that staggers the terms and governance provisions. Thus, these listings elections of directors and hence can be are a noisy measure of a firm’s governance used to slow down a hostile takeover. If provisions, but there is no reason to suspect management uses this power judiciously, it any systematic bias. Also, all of our analysis could possibly lead to an increase in overall uses data available at time t to forecast shareholder wealth; if management uses performance at time t 1 and beyond, so this power to maintain private benefits of there is no possibility of look-ahead bias control, then this provision would decrease induced by our statistical procedures. shareholder wealth. In either case, it is To build the data set, we coded the data clear that Classified Boards increase the from the individual firm profiles in the power of managers and weaken the control IRRC books. For each firm we recorded the rights of large shareholders, which is all identifying information (ticker symbol, that matters for constructing the index. state of incorporation) and the presence of Most of the provisions can be viewed in each provision. Although many of the a similar way. Almost every provision gives provisions can be made stronger or weaker management a tool to resist different types (e.g., supermajority thresholds can vary of shareholder activism, such as calling between 51 and 100 percent), we made no special meetings, changing the firm’s strength distinctions and coded all provi- charter or bylaws, suing the directors, or sions as simply “present” or “not present.” just replacing them all at once. There are This methodology sacrifices precision for two exceptions: Secret Ballots and the simplicity necessary to build an index. Cumulative Voting. A Secret Ballot, also For most of the analysis of this paper,we called “confidential voting” by some firms, match the IRRC data to the Center for designates a third party to count proxy Research in Security Prices (CRSP) and, votes and prevents management from where necessary, to Standard and Poor’s observing how specific shareholders vote. Compustat database. CSRP matching was Cumulative Voting allows shareholders to done by ticker symbol and was supple- concentrate their directors’ votes so that a mented by handchecking names, exchanges, large minority holder can ensure some and states of incorporation. These proce- board representation. (See Appendix 1 for dures enable us to match 100 percent of fuller descriptions.) These two provisions the IRRC sample to CRSP, with about are usually proposed by shareholders and 90 percent of these matches having opposed by management.9 In contrast, none complete annual data in Compustat. of the other provisions enjoy consistent shareholder support or management oppo- II.B. The Governance index sition; in fact, many of these provisions receive significant numbers of shareholder The index construction is straightforward: proposals for their repeal [Ishii 2000]. for every firm we add one point for every Also, both Cumulative Voting and Secret provision that restricts shareholder rights Ballots tend to be negatively correlated (increases managerial power). This power with the presence of other firm-level provi- distinction is straightforward in most sions (19 negative out of 21 for Cumulative cases, as is discussed below. While this Voting; 11 out of 21 for Secret Ballot). simple index does not accurately reflect Thus, we consider the presence of Secret the relative impacts of different provisions, Ballots and Cumulative Voting to be it has the advantage of being transparent increases in shareholder rights. For each and easily reproducible. The index does not one we add one point to the Governance 528 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Index when firms do not have it. For all subindex and its Other subindex, but only other provisions we add one point when one point (not two) would be added to its firms do have it.10 overall G index. Thus, G has a possible Thus, the Governance Index (“G”) is just range from 1 to 24 and is not just the sum the sum of one point for the existence of the five subindices. (or absence) of each provision. We also Table 20.2 gives summary statistics for construct subindices for each of the five G and the subindices in 1990, 1993, 1995, categories: Delay, Protection, Voting, Other, and 1998. Table 20.2 also shows the fre- and State. Recall that there are 28 total quency of G by year, broken up into groups provisions listed in the five categories, beginning with G 5, then each value of G of which 24 are unique. For the state from G 6 through G 13, and finishing laws with a firm-level analogue, we add one with G 14. These ten “deciles” are simi- point to the index if the firm is covered lar but not identical in size, with relative under the firm-level provision, the state sizes that are fairly stable from 1990 to law, or both.11 For example, a firm that 1995. In the remainder of the paper we has an Antigreenmail provision and is also pay special attention to the two extreme covered by the Antigreenmail state law portfolios: the “Dictatorship Portfolio” of would get one point added to both its State the firms with the weakest shareholder rights

Table 20.2 The Governance Index

1990 1993 1995 1998

Governance index Minimum 2 2 2 2 Mean 9.0 9.3 9.4 8.9 Median 9 9 9 9 Mode 10 9 9 10 Maximum 17 17 17 18 Standard deviation 2.9 2.8 2.8 2.8 Number of firms G 5 (Democracy Portfolio) 158 139 120 215 G 6 119 88 108 169 G 7 158 140 127 186 G 8 165 139 152 201 G 9 160 183 183 197 G 10 175 170 178 221 G 11 149 168 166 194 G 12 104 123 142 136 G 13 84 100 110 106 G 14 (Dictatorship portfolio) 85 93 87 83 Total 1357 1343 1373 1708 Subindex means Delay 1.8 2.0 2.1 2.1 Protection 2.4 2.5 2.5 2.1 Voting 2.2 2.1 2.1 2.2 Other 1.1 1.2 1.1 1.0 State 1.8 1.8 1.8 1.7

This table provides summary statistics on the distribution of G, the Governance Index, and the subindices (Delay, Protection, Voting, Other, and State) over time. G and the subindices are calculated from the provisions listed in Table 20.1 as described in Section II. Appendix 1 gives detailed information on each provision. We divide the sample into ten portfolios based on the level of G and list the number of firms in each portfolio.The Democracy Portfolio is composed of all firms where G 5, and the Dictatorship Portfolio contains all firms where G 14. CORPORATE GOVERNANCE AND EQUITY PRICES 529 (G 14), and the “Democracy Portfolio” and 1998. Of the ten largest firms in the of the firms with the strongest shareholder Democracy Portfolio in 1990, six of them rights (G 5).These portfolios are updated are still in the Democracy Portfolio in at the same frequency as G. 1998, three have dropped out of the port- Most of the changes in the distribution folio and have G 6, and one (Berkshire of G come from changes in the sample due Hathaway) disappeared from the sample.13 to mergers, bankruptcies, and additions of The Dictatorship Portfolio has a bit more new firms by the IRRC. In 1998 the sample activity, with only two of the top ten size increased by about 25 percent, and firms remaining in the portfolio, four firms these new firms tilted toward lower values dropping out with G 13, and three firms of G. At the firm level, G is relatively stable. leaving the sample though mergers or the For individual firms the mean (absolute) addition of another class of stock.14 Thus, change in G between publication dates 40 percent (eight out of 20) of the largest (1990, 1993, 1995, 1998) is 0.60, and firms in the extreme portfolios in 1990 the median (absolute) change between were also in these portfolios in 1998. This publication dates is zero.12 is roughly comparable to the full set of Table 20.3 shows the correlations firms: among all firms in the Democracy between pairs of subindices. The Delay, and Dictatorship Portfolios in 1990, Protection, Voting, and Other subindices 31 percent were still in the same portfolios all have positive and significant pairwise in 1998. correlations with each other. State,how- There is no obvious industry concentration ever, has negative correlations with Delay, among these top firms; the whole portfolios Protection, and Voting. It could be that are similarly dispersed. Classifying firms firms view some of the state laws as sub- into 48 industries as in Fama and French stitutes for the firm-level provisions, but [1997], the portfolios appear to be broadly then it would be surprising that Other, similar to each other in all years, with a which contains three provisions that are mix of old-economy and new-economy direct substitutes for state laws, is the only industries.15 Each portfolio has an impor- subindex that is positively correlated with tant technology component. “Computers” State. Overall, it appears that coverage is the largest industry by market value in under state laws is not highly correlated the Democracy Portfolio in 1990, with with the adoption of firm-level provisions. 22.4 percent of the portfolio, falling to This fact has implications for the analysis third place with 12.3 percent of the value of causality, as is discussed in Section IV. in 1998. “Communications” does not Table 20.4 lists the ten largest firms (by make the top five in market value for the market capitalization) in the Democracy Dictatorship Portfolio in 1990, but rises and Dictatorship Portfolios in 1990 and to first place with 25.3 percent of the gives the value of G for these firms in 1990 portfolio in 1998.

Table 20.3 Correlations Between the Subindices

Delay Protection Voting Other

Protection 0.22** Voting 0.33** 0.10** Other 0.43** 0.27** 0.19** State 0.08** 0.04 0.07* 0.05

This table presents pairwise correlations between the subindices, Delay, Protection, Voting, Other, and State in 1990.The calculation of the subindices is described in Section II.The elements of each subindex are given in Table 20.1 and are described in detail in Appendix 1. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. 530 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 20.4 The Largest Firms in the Extreme Portfolios 1990 Democracy portfolio

1990 Governance 1998 Governance State of incorporation index index

IBM New York 5 6 Wal-Mart Delaware 5 5 Du Pont de Nemours Delaware 5 5 Pepsico North Carolina 4 3 American International Group Delaware 5 5 Southern Company Delaware 5 5 Hewlett Packard California 5 6 Berkshire Hathaway Delaware 3 — Commonwealth Edison Illinois 4 6 Texas Utilities Texas 2 4

1990 Dictatorship Portfolio

1990 Governance 1998 Governance State of incorporation index index

GTE New York 14 13 Waste Management Delaware 15 13 General Re Delaware 14 16 Limited Inc Delaware 14 14 NCR Maryland 14 — K Mart Michigan 14 10 United Telecommunications Kansas 14 — Time Warner Delaware 14 13 Rorer Pennsylvania 16 — Woolworth New York 14 13

This table presents the firms with the largest market capitalizations at the end of 1990 of all companies within the Democracy Portfolio (G 5) and the Dictatorship Portfolio (G 14).The calculation of G is described in Section II.The companies are listed in descending order of market capitalization.

III. GOVERNANCE: EMPIRICAL characteristics with G, the next two RELATIONSHIPS columns give the mean value in the Democracy and Dictatorship Portfolios, III.A. Summary statistics and the final column gives the difference between these means. These results are Table 20.5 gives summary statistics and descriptive and are intended to provide correlations for G (and subindices) with a some background for the analyses in set of firm characteristics as of September the following sections. 1990: book-to-market ratio, firm size, share The strongest relation is between G and price, monthly trading volume, Tobin’s Q, S&P 500 inclusion.The correlation between dividend yield, S&P 500 inclusion, past these variables is positive and significant— five-year stock return, past five-year sales about half of the Dictatorship Portfolio is growth, and percentage of institutional drawn from S&P 500 firms compared with ownership. The first four of these char- 15 percent of the Democracy Portfolio. acteristics are in logs. The construction Given this finding, it is not surprising of each characteristic is described in that G is also positively correlated with Appendix 2.The first column of Table 20.5 size, share price, trading volume, and gives the correlation of each of these institutional ownership. S&P firms tend to CORPORATE GOVERNANCE AND EQUITY PRICES 531 Table 20.5 Summary Statistics

Correlation Mean, democracy Mean, dictatorship Difference with G portfolio portfolio

BM 0.02 0.66 0.54 0.12 (0.10) SIZE 0.15** 12.86 13.46 0.60** (0.21) PRICE 0.16** 2.74 3.14 0.40** (0.12) VOLUME 0.19** 16.34 17.29 0.95** (0.24) Q 0.04 1.77 1.47 0.30* (0.14) YLD 0.03 4.20% 7.20% 3.00% (4.34) SP500 0.23** 0.15 0.49 0.34** (0.06) 5-year return 0.01 90.53% 85.41% 5.12% (20.74) SGROWTH 0.08** 62.74% 44.78% 17.96% (9.83) IO 0.14** 25.89% 34.44% 8.55%* (3.36)

This table gives descriptive statistics for the relationship of G with several financial and accounting measures in September 1990.The first column gives the correlations for each of these variables with the Governance Index, G. The second and third columns give means for these same variables within the Democracy Portfolio (G 5) and the Dictatorship Portfolio (G 14) in 1990. The final column gives the difference of the two means with its standard error in parentheses. The calculation of G is described in Section II, and definitions of each variable are given in Appendix 2. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. have relatively high levels of all of these multivariate analysis suggesting no robust characteristics. In addition, the correla- relationship between G and changes in tion of G with five-year sales growth is institutional ownership. negative and significant, suggesting that high-G firms had relatively lower sales III.B. Governance and returns growth over the second half of the 1980s, the period when many of the provisions If corporate governance matters for firm were first adopted. performance and this relationship is fully Correlations at other times in the sample incorporated by the market, then a stock period (not shown in the table) are similar. price should quickly adjust to any relevant Overall, it appears that firms with weaker change in the firm’s governance.This is the shareholder rights tend to be large S&P logic behind the use of event studies to firms with relatively high share prices, analyze the impact of takeover defenses. If institutional ownership and trading volume, such a reaction occurs, then expected relatively poor sales growth, and poor returns on the stock would be unaffected stock-market performance.The 1990s were beyond the event window. However, if a time of rising activism by institutional governance matters but is not incorporated investors and more attention to governance immediately into stock prices, then real- provisions; thus, we might expect to see ized returns on the stock would differ some reduction in the institutional owner- systematically from equivalent securities. ship of high-G firms. In untabulated tests, In this section we examine the relationship we find no evidence of such a reduction, between G and subsequent returns. An with both pairwise correlations and investment of $1 in the (value-weighted) 532 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY

Dictatorship Portfolio on September 1, performance attribution.Thus, we interpret 1990, when our data begin, would have the estimated intercept coefficient,“alpha,” grown to $3.39 by December 31, 1999. In as the abnormal return in excess of what contrast, a $1 investment in the Democracy could have been achieved by passive Portfolio would have grown to $7.07 over investments in the factors. the same period. This is equivalent to The first row of Table 20.6 shows the annualized returns of 14.0 percent for the results of estimating (1) where the Dictatorship Portfolio and 23.3 percent for dependent variable Rt is the monthly return the Democracy Portfolio, a difference of difference between the Democracy and more than 9 percent per year. Dictatorship Portfolios. Thus, the alpha in What can explain this disparity? One this estimation is the abnormal return on possible explanation is that the performance a zero-investment strategy that buys the differences are driven by differences in the Democracy Portfolio and sells short the riskiness or “style” of the two portfolios. Dictatorship Portfolio. For this specifica- Researchers have identified several equity tion the alpha is 71 basis points (bp) per characteristics that explain differences in month, or about 8.5 percent per year.This realized returns. In addition to differences point estimate is statistically significant at in exposure to the market factor (“beta”), the 1 percent level. Thus, very little of the a firm’s market capitalization (or “size”), difference in raw returns can be attributed book-to-market ratio (or other “value” to style differences in the two portfolios. characteristics), and immediate past returns The remaining rows of Table 20.6 (“momentum”) have all been shown to summarize the results of estimating (1) for significantly forecast future returns.16 If all ten “deciles” of G, including the extreme the Dictatorship Portfolio differs signifi- deciles comprising the Democracy (G 5) cantly from the Democracy Portfolio in and Dictatorship (G 14) Portfolios. As these characteristics, then style differences the table shows, the significant perform- may explain at least part of the difference ance difference between the Democracy in annualized raw returns. and Dictatorship Portfolios is driven both Several methods have been developed by overperformance (for the Democracy to account for these style differences in a Portfolio) and underperformance (by the system of performance attribution. We Dictatorship Portfolio). The Democracy employ one method here and use another in Portfolio earns a positive and significant Section V.The four-factor model of Carhart alpha of 29 bp per month, while the [1997] is estimated by Dictatorship Portfolio earns a negative and significant alpha of 42 bp per month. Rt 1 * RMRFt 2 * SMBt The results also show that alpha 3 * HMLt 4 *Momentumt decreases as G increases. The Democracy t, (1) Portfolio earns the highest alpha of all the deciles, and the next two highest alphas, 24 where Rt is the excess return to some asset and 22 bp, are earned by the third (G 7) in month t, RMRFt is the month t value- and second (G 6) deciles, respectively. weighted market return minus the risk-free The Dictatorship Portfolio earns the lowest rate, and the terms SMBt (small minus big), alpha, and the second lowest alpha is HMLt (high minus low), and Momentumt earned by the eighth (G 12) decile. are the month t returns on zero-investment Furthermore, the four lowest G deciles factor-mimicking portfolios designed to earn positive alphas, while the three highest capture size, book-to-market, and momen- G deciles earn negative alphas. More for- tum effects, respectively.17 Although there mally, a Spearman rank-correlation test of is ongoing debate about whether these the null hypothesis of no correlation factors are proxies for risk, we take no between G-decile rankings and alpha rank- position on this issue and simply view ings yields a test statistic of 0.842, and is the four-factor model as a method of rejected at the 1 percent level. CORPORATE GOVERNANCE AND EQUITY PRICES 533 Table 20.6 Performance-Attribution Regressions for Decile Portfolios

RMRF SMB HML Momentum Democracy- 0.71** 0.04 0.22* 0.55* 0.01 Dictatorship (0.26) (0.07) (0.09) (0.10) (0.07)

G 5 (Democracy) 0.29* 0.98** 0.24** 0.21** 0.05 (0.13) (0.04) (0.05) (0.05) (0.03) G 6 0.22 0.99** 0.18** 0.05 0.08 (0.18) (0.05) (0.06) (0.07) (0.04) G 7 0.24 1.05** 0.10 0.14 0.15** (0.19) (0.05) (0.07) (0.08) (0.05) G 8 0.08 1.02** 0.04 0.08 0.01 (0.14) (0.04) (0.05) (0.06) (0.04) G 9 0.02 0.97** 0.20** 0.14** 0.01 (0.12) (0.03) (0.04) (0.05) (0.03) G 10 0.03 0.95** 0.17** 0.00 0.08** (0.11) (0.03) (0.04) (0.04) (0.03) G 11 0.18 0.99** 0.14* 0.06 0.01 (0.16) (0.05) (0.05) (0.06) (0.04) G 12 0.25 1.00** 0.11* 0.16** 0.02 (0.14) (0.04) (0.05) (0.06) (0.04) G 13 0.01 1.03** 0.21** 0.14* 0.08* (0.14) (0.04) (0.05) (0.06) (0.04) G 14 (Dictatorship) 0.42* 1.03** 0.02 0.34** 0.05 (0.19) (0.05) (0.06) (0.07) (0.05)

We estimate four-factor regressions (equation (1) from the text) of value-weighted monthly returns for portfolios of firms sorted by G.The calculation of G is described in Section II.The first row contains the results when we use the portfolio that buys the Democracy Portfolio (G 5) and sells short the Dictatorship Portfolio (G 14).The portfolios are reset in September 1990, July 1993, July 1995, and February 1998, which are the months after new data on G became available.The explanatory variables are RMRF, SMB, HML, and Momentum.These variables are the returns to zero-investment portfolios designed to capture market, size, book-to-market, and momentum effects, respectively. (Consult Fama and French [1993] and Carhart [1997] on the construction of these factors.) The sample period is from September 1990 through December 1999. Standard errors are reported in parentheses and significance at the 5 percent and 1 percent levels is indicated by * and **, respectively.

Table 20.7 reports several variations The first row of Table 20.7 replicates the of the abnormal-return results. In each baseline portfolio construction used above. variation we estimate the performance- The remaining rows of the table summarize attribution regression in equation (1) on tests using industry-adjusted returns (row 2), the return difference between the Democracy two alternative constructions of the extreme and Dictatorship Portfolios, while changing portfolios (rows 3 and 4), fixed portfolios some aspect of the portfolio construction built with 1990 levels of G (row 5), a sub- or return calculation. We perform all of sample that includes only Delaware firms these tests using both value-weighted (VW) (row 6), and subsamples split between the and equal-weighted (EW) portfolios. first half and the second half of the sample These tests allow us to estimate the fraction period (rows 7 and 8). Details of each of of the benchmark abnormal returns that these constructions are given in the table can be attributed to industry composition, note.The main themes of these results are, choice of cutoffs for the extreme portfolios, first, that the VW returns (Democracy new provisions during the decade, legal minus Dictatorship) are economically large variation across states, and different time in all cases and, second, the EW abnormal periods. returns are usually about two-thirds of the 534 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 20.7 Performance-Attribution Regressions under Alternative Portfolio Constructions , Value-weighted , Equal-weighted

(1) Democracy-Dictatorship 0.71** 0.45* (0.26) (0.22) (2) Industry-adjusted 0.47* 0.30 (0.22) (0.19) (3) Big portfolios 0.47* 0.39* (0.21) (0.19) (4) Small portfolios 0.78* 0.45 (0.33) (0.25) (5) 1990 portfolio 0.53* 0.33 (0.24) (0.22) (6) Delaware portfolio 0.63 0.42 (0.34) (0.26) (7) Early half 0.45 0.58* (0.23) (0.28) (8) Late half 0.75 0.04 (0.40) (0.27)

This table presents the alphas from four-factor regressions for variations on the Democracy (G 5) minus Dictatorship (G 14) Portfolio. The calculation of G is described in Section II. The portfolios are reset in September 1990, July 1993, July 1995, and February 1998, which are the months after new data on G became available. The sample period is September 1990 to December 1999. The first row uses the unadjusted difference between the monthly returns to the Democracy and Dictatorship Portfolios. The second row contains the results using industry-adjusted returns, with industry adjustments done relative to the 48 industries of Fama and French [1997]. The third and fourth rows use alternative definitions of the Democracy and Dictatorship Portfolios. In the third row, firms are sorted on G and the two portfolios contain the smallest set of firms with extreme values of G such that each has at least 10 percent of the sample. This implies cutoff values of G for the Democracy Portfolio of 5, 5, 6, and 5 for September 1990, July 1993, July 1995, and February 1998, respectively. The cutoffs for the Dictatorship Portfolio are always 13. In the fourth row, the two portfolios contain the largest set of firms such that each has no more than 10 percent of the sample. The cutoff values of G for the Democracy Portfolio are 4, 4, 5, and 4 for September 1990, July 1993, July 1995, and February 1998, respectively, and they are always 14 for the Dictatorship Portfolio. In the fifth row, portfolio returns are calculated maintaining the 1990 portfolios for the entire sample period. As long as they are listed in CRSP, we neither delete nor add firms to these portfolios regardless of subsequent changes in G or changes in the IRRC sample in later editions. The sixth row shows the results of restricting the sample to firms incorporated in Delaware. In the seventh and eighth rows, the sample period is divided in half at April 30, 1995, and separate regressions are estimated for the first half and second half of the period (56 months each).The explanatory variables are RMRF, SMB, HML, Momentum, and a constant. These variables are the returns to zero-investment portfolios designed to capture market, size, book-to-market, and momentum effects, respectively. (Consult Fama and French [1993] and Carhart [1997] on the construction of these factors.) All coefficients except for the alpha are omitted in this table. Standard errors are reported in parentheses, and significance at the 5 percent and 1 percent levels is indicated by * and **, respectively.

VW abnormal returns. Most of the return of causality and omitted-variable bias in differential can be attributed to within- Section V, so we defer a detailed analysis state variation already in place in 1990, until then. and this return differential is apparent in both halves of the sample period. III.C. Governance and the value Overall, we find significant evidence that of the firm the Democracy Portfolio outperformed the Dictatorship Portfolio in the 1990s. We It is well established that state and also find some evidence of a monotonic national laws of corporate governance relationship between G and returns. It would affect firm value. La Porta et al. [2001] be useful to know which subindices and pro- show that firm value is positively associated visions drive these results. We address this with the rights of minority shareholders. issue in depth within the broader analysis Daines [2001] finds that firms incorporated CORPORATE GOVERNANCE AND EQUITY PRICES 535 in Delaware have higher valuations than We rejected this alternative mainly because other U. S. firms. In this section we study there are few changes over time in the whether variation in firm-specific gover- Governance Index, and the inclusion of nance is associated with differences in fixed effects would force identification of firm value. More importantly, we analyze the G coefficient from only these changes. whether there was a change in the gover- In effect, our chosen method imposes a nance/value relationship during the 1990s. structure on the fixed effects: they must be Since there is evidence of differential stock a linear function of G or its components. returns as a function of G, we would expect Table 20.8 summarizes the results. The to find relative “mispricing” between 1990 first column gives the results with G as the and 1999 as a function of G. key regressor. Each row gives the coeffi- Our valuation measure is Tobin’s Q, cients and standard errors for a different which has been used for this purpose in year of the sample; the last row gives the corporate-governance studies since the average coefficient and time-series standard work of Demsetz and Lehn [1985] and error of these coefficients. The coefficients Morck, Shleifer, and Vishny [1988]. We on G are negative in every year and signifi- follow Kaplan and Zingales’ [1997] cantly negative in nine of the ten years.The method for the computation of Q (details largest absolute value point estimate are listed in Appendix 2) and also compute occurs in 1999, and the second largest is the median Q in each year in each of the in 1998.The point estimate in 1999 is eco- 48 industries classified by Fama and nomically large; a one-point increase in G, French [1997]. We then regress equivalent to adding a single governance provision, is associated with an 11.4 percent- Q’it atbtXitctWiteit, (2) age point lower value for Q. If we assume that the point estimates in 1990 and 1999 where Q’it is industry-adjusted Q (firm Q are independent, then the difference minus industry-median Q), Xit is a vector of between these two estimates (11.4 governance variables (G, its components, or 2.2 9.2) is statistically significant. inclusion in one of the extreme portfolios) In the second column of Table 20.8, we and Wit is a vector of firm characteristics. restrict the sample to include only firms in As elements of W, we follow Shin and Stulz the Democracy and Dictatorship [2000] and include the log of the book Portfolios. We then estimate (2) using a value of assets and the log of firm age as dummy variable for the Democracy of December of year t.18 Daines [2001] Portfolio. The results are consistent with found that Q is different for Delaware and the previous regressions on G. The point non-Delaware firms, so we also include a estimate for 1999 is the largest in the Delaware dummy in W. Morck and Yang decade, implying that firms in the Democracy [2001] show that S&P 500 inclusion has Portfolio have a Q that is 56 percentage a positive impact on Q, and that this points higher,other things being equal, than impact increased during the 1990s; thus, do firms in the Dictatorship Portfolio. This we also include a dummy variable for S&P compares with an estimated difference of 500 inclusion in W. 19 percentage points in 1990. While the Using a variant of the methods of Fama difference in coefficients between 1990 and MacBeth [1973], we estimate annual and 1999 is not statistically significant, it cross sections of (2) with statistical is similar to the total EW difference in significance assessed within each year (by abnormal returns estimated in Table 20.7.19 cross-sectional standard errors) and across There is no real pattern for the rest of the all years (with the time-series standard decade, however, and large standard errors error of the mean coefficient).This method toward the end of the sample period of assessing statistical significance deserves prevent any strong inference across years. some explanation. In particular,one logical The final columns of Table 20.8 give alternative would be a pooled setup with firm results for a single regression using the fixed effects and time-varying coefficients. five governance subindices: Delay, Voting, 536 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 20.8 Q Regressions

(2) (1) Democracy (3) (4) (5) (6) (7) G Portfolio Delay Protection Voting Other State

1990 0.022** 0.186 0.015 0.035 0.015 0.031 0.004 (0.008) (0.127) (0.022) (0.018) (0.030) (0.026) (0.020) 1991 0.040** 0.302* 0.033 0.048 0.012 0.059 0.003 (0.012) (0.143) (0.034) (0.028) (0.047) (0.040) (0.031) 1992 0.036** 0.340* 0.041 0.039 0.021 0.054 0.011 (0.010) (0.151) (0.027) (0.023) (0.038) (0.032) (0.025) 1993 0.042** 0.485* 0.023 0.055* 0.009 0.060 0.062* (0.011) (0.204) (0.029) (0.026) (0.038) (0.035) (0.027) 1994 0.031** 0.335* 0.032 0.012 0.032 0.029 0.047* (0.009) (0.161) (0.023) (0.020) (0.031) (0.028) (0.022) 1995 0.039** 0.435* 0.046 0.062* 0.086* 0.023 0.022 (0.011) (0.217) (0.030) (0.027) (0.041) (0.036) (0.028) 1996 0.025* 0.299 0.029 0.030 0.078 0.018 0.024 (0.011) (0.195) (0.031) (0.028) (0.041) (0.037) (0.028) 1997 0.016 0.210 0.017 0.007 0.055 0.001 0.017 (0.013) (0.196) (0.035) (0.032) (0.047) (0.042) (0.032) 1998 0.065** 0.203 0.023 0.096* 0.132 0.058 0.012 (0.020) (0.404) (0.052) (0.049) (0.070) (0.066) (0.052) 1999 0.114** 0.564 0.067 0.171* 0.294** 0.006 0.033 (0.027) (0.602) (0.071) (0.067) (0.098) (0.090) (0.073) Mean 0.043** 0.336** 0.033** 0.056** 0.065 0.025* 0.020* (0.009) (0.040) (0.005) (0.015) (0.030) (0.010) (0.007)

The first column of this table presents the coefficients on G, the Governance Index, from regressions of industryadjusted Tobin’s Q on G and control variables.The second column restricts the sample to firms in the Democracy (G 5) and Dictatorship (G 14) Portfolios and includes as regressors a dummy variable for the Democracy Portfolio and the controls.The third through seventh columns show the coefficients on each subindex from regressions where the explanatory variables are the subindices Delay, Protection, Voting, Other, and State, and the controls. We include as controls a dummy variable for incorporation in Delaware, the log of assets in the current fiscal year, the log of firm age measured in months as of December of each year, and a dummy variable for inclusion in the S&P 500 as of the end of the previous year.The coefficients on the controls and the constant are omitted from the table. The calculation of G and the subindices is described in Section II. Q is the ratio of the market value of assets to the book value of assets: the market value is calculated as the sum of the book value of assets and the market value of common stock less the book value of common stock and deferred taxes. The market value of equity is measured at the end of the current calendar year, and the accounting variables are measured in the current fiscal year. Industry adjustments are made by subtracting the industry median, where medians are calculated by matching the four-digit SIC codes from December of each year to the 48 industries designated by Fama and French [1997]. The coefficients and standard errors from each annual cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors are given in the last row. * and ** indicate significance at the 5 percent and 1 percent levels, respectively.

Protection, Other, and State. The table it difficult to draw strong conclusions. shows that all subindices except Voting For example, even in 1999 we cannot have average coefficients that are negative reject the null hypothesis that the coeffi- and significant (assuming independence cient on Voting is equal to the coefficient across years). Over the full sample period, on Delay. Delay and Protection have the most Overall, the results for returns and prices consistent impact, while the largest tell a consistent story. Firms with the absolute coefficients are for Voting at the weakest shareholder rights (high values of G) end of the sample period.The subindices are significantly underperformed firms with the highly collinear, however, and the resulting strongest shareholder rights (low values of large standard errors and covariances make G) during the 1990s. Over the course of the CORPORATE GOVERNANCE AND EQUITY PRICES 537 1990s, these differences have been at least operational measures are the net profit partially reflected in prices. While high-G margin (income divided by sales), the firms already sold at a significant discount return on equity (income divided by book in 1990, this discount became much larger equity), and one-year sales growth. All of by 1999. these measures are industry-adjusted by subtracting the median for this measure in III.D. Governance and operating the corresponding Fama-French [1997] industry.This adjustment uses all available performance Compustat firms.To reduce the influence of Table 20.9 shows the results of annual large outliers—a common occurrence for regressions for three operational measures all of these measures—we estimate median on G (or a Democracy dummy). The three (least-absolute-deviation) regressions in

Table 20.9 Operating Performance

(1) (2) (3) (4) (5) (6) Net profit margin Return on equity Sales growth

Democracy Democracy Democracy G Portfolio G Portfolio G Portfolio

1991 0.70 10.61 1.19* 13.54 2.30 3.52 (0.39) (7.12) (0.60) (11.30) (1.38) (17.83) 1992 0.52 9.45 0.42 2.54 1.43 0.10 (0.58) (10.43) (0.61) (9.21) (1.06) (11.52) 1993 0.76 7.77 0.34 2.51 3.35** 18.55 (0.48) (9.98) (0.79) (10.98) (1.17) (17.71) 1994 0.83 10.94 1.07 2.69 2.71* 12.58 (0.48) (6.59) (0.61) (10.36) (1.10) (22.81) 1995 0.72 7.56 1.39 14.77 0.89 7.91 (0.67) (8.30) (0.75) (9.88) (1.70) (19.67) 1996 0.43 2.17 0.90 2.30 2.44 14.84 (0.40) (7.22) (0.65) (12.09) (1.39) (19.36) 1997 0.21 9.61 0.66 17.54 0.01 4.28 (0.55) (9.99) (0.81) (9.83) (1.64) (26.61) 1998 0.73 3.99 1.28 13.62 1.45 15.65 (0.63) (7.15) (1.01) (15.10) (1.50) (23.36) 1999 1.27* 4.59 0.93 15.53 0.52 15.38 (0.58) (11.58) (0.85) (10.38) (1.92) (26.10) Mean 0.64** 3.91 0.26 1.59 1.68** 5.10 (0.13) (2.46) (0.33) (3.98) (0.37) (3.84)

The first, third, and fifth columns of this table give the results of annual median (least absolute deviation) regressions for net profit margin, return on equity, and sales growth on the Governance Index, G, measured in the previous year, and the book-to-market ratio, BM. The second, fourth, and sixth columns restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) portfolios and include as regressors a dummy variable for the Democracy Portfolio and BM. The coefficients on BM and the constant are omitted from the table.The calculation of G is described in Section II. Net profit margin is the ratio of income before extraordinary items available for common equity to sales; return on equity is the ratio of income before extraordinary items available for common equity to the sum of the book value of common equity and deferred taxes; BM is the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous fiscal year to size at the close of the previous calendar year. Each dependent variable is net of the industry median, which is calculated by matching the four-digit SIC codes of all firms in the CRSP-Compustat merged database in December of each year to the 48 industries designated by Fama and French [1997]. The coefficients and standard errors from each annual cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors are given in the last row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and standard errors are multiplied by 1000. 538 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY each case. While our sample does not HYPOTHESIS III. Governance provisions do include a natural experiment to identify G not cause higher agency costs, but their as the cause of operational differences, presence is correlated with other we attempt to control for “expected” cross- characteristics that earned abnormal sectional differences by using the log returns in the 1990s. book-to-market ratio (BM) as an additional explanatory variable. Most explanations of the Section III The odd-numbered columns give the results can be fit within these three results when G is the key regressor.We find hypotheses. Under Hypothesis I, a reduc- that the average coefficient on G is negative tion in shareholder rights causes an unex- and significant for both the net-profit-margin pectedly large increase in agency costs and sales-growth regressions, and is negative through some combination of inefficient but not significant for the return-on-equity investment, reduced operational efficiency, regressions. The even-numbered columns or self-dealing. If shareholders find it diffi- give the results for the subsample of firms cult or costly to replace managers, then from the extreme deciles, with a dummy managers may be more willing and able to variable for the Democracy Portfolio as the extract private benefits.This is the standard key regressor.For all three operating meas- justification for takeover threats as the ures, the average coefficient on this dummy strongest form of managerial discipline variable was positive but insignificant. [Jensen 1986]. For Hypothesis I to be cor- Thus, these results are consistent with the rect, these additional agency costs must evidence for the full sample but not signifi- have been underestimated in 1990. cant on their own. In untabulated results, Under Hypothesis II, governance does we also regressed these same measures not affect performance, but there must be a on the five subindices. The results show no perception that governance provisions are clear pattern of differential influence for any protective for management. In this case, the particular subindex, with most coefficients stock in these companies would have been having the same sign as G. Overall, we relatively overvalued in 1990, even though find some significant evidence that more objective measures (e.g., Q regressions) democratic firms have better operating per- would suggest that it was undervalued formance and no evidence that they do not. relative to observable characteristics. When the poor operating performance occurs, the market is surprised, but the managers are IV. GOVERNANCE: THREE not.The protective provisions then supply a HYPOTHESES shield, real or imagined, for managerial jobs and compensation. Section III established an empirical Under Hypothesis III, all of the results in relationship of G with returns, firm value, the previous section would be driven by and operating performance. Since firms did omitted-variable bias. Since governance not adopt governance provisions randomly, provisions were certainly not adopted this evidence does not itself imply a causal randomly, it is plausible that differences in role by governance provisions. Indeed, industry, S&P 500 inclusion, institutional there are several plausible explanations for ownership, or other firm characteristics our results: could be correlated both with G and with abnormal returns. Under this hypothesis, HYPOTHESIS I. Governance provisions cause governance provisions could be completely higher agency costs. These higher costs innocuous, with no influence either on were underestimated by investors in 1990. managerial power or on agency costs. HYPOTHESIS II. Governance provisions do Ideally, we would distinguish among these not cause higher agency costs, but rather three hypotheses by using random variation were put in place by 1980s managers in some characteristic that was causal for who forecasted poor performance for G. Unfortunately, we have not been able to their firms in the 1990s. identify such an instrument. One candidate CORPORATE GOVERNANCE AND EQUITY PRICES 539 would be the subset of state laws, with the Section IV. Subsection V.A covers State subindex as a proxy. Though in some Hypothesis I, subsection V.B covers states these laws were passed at the urging Hypothesis II, and subsection V.C cov- of large corporations, it seems reasonable ers Hypothesis III. Subsection V.D to assume that their passage was exoge- summarizes and discusses the evidence. nous to most firms. But the State subindex has three flaws as an instrument. First, firms can choose to reincorporate into V.A. Evidence on Hypothesis I different states; enough firms have done Increased agency costs at high-G firms can so that exposure to state laws is not truly directly affect firm performance in several exogenous [Subramanian 2001]. Second, ways. In the specific case of state takeover many firms have opted out of the protec- laws, where causality is easier to establish, tions of some of the most stringent of these researchers have found evidence of increased laws, so that a firm’s state of incorporation agency costs through a variety of mecha- is only a noisy measure for its actual legal nisms. Borokhovich, Brunarski, and Parrino exposure. Third, as shown in Table 20.3, [1997] show that compensation rises for the State subindex is not positively or CEOs of firms adopting takeover defenses. consistently correlated with the other Bertrand and Mullainathan [1999a, components of G. Other potential instru- 1999b, 2000] find a similar result for ments have different problems. For example, CEOs and other employees in firms newly if takeover protections were adopted during covered by state takeover laws. They also industry-specific takeover waves, then we find that these laws cause a decrease in might be able to use industry as an instru- plant-level efficiency, measured either ment for G. Unfortunately, this would by total factor productivity or return on render it impossible to distinguish between capital. Garvey and Hanka [1999] show G or industry as the cause of poor returns that state takeover laws led to changes in in the 1990s. leverage consistent with increased corporate In Section V our tests consist of a search slack. These studies provide the cleanest for evidence supportive of each hypothesis, evidence in support of Hypothesis I, but, while acknowledging the impossibility of a of course, do not make use of the full perfect test to distinguish among them. variation embodied in the G index. We First, if Hypothesis I is correct, then we supplement these findings by examining the should observe some “unexpected” differ- empirical relationship of G with two other ences in agency costs across firms. We possible sources of agency costs: capital discuss several previous studies on this expenditure and acquisition behavior. topic and look for such differences in our A substantial literature, dating back at sample by analyzing capital expenditure least to Baumol [1959], Marris [1964], and acquisition behavior. Second, for and Williamson [1964], holds that managers Hypothesis II we analyze insider-trading may undertake inefficient projects in order activity as a function of G. If governance to extract private benefits. This problem is provisions were put in place by prescient particularly severe when managers are managers, these same managers might be entrenched and can resist hostile takeovers net sellers of the stock in their firms. [Jensen and Ruback 1983; Shleifer and Finally, for Hypothesis III we test whether a Vishny 1989]. Under this view, if capital large set of observable firm characteristics expenditure increases following the adoption can explain the empirical relationship of new takeover defenses, this increase between returns and G. would be a net negative for firm value.20 To examine the empirical relationship between capital expenditure and governance, V. GOVERNANCE: TESTS we estimate annual median regressions for capital expenditure (CAPEX), scaled In this section we examine the evidence by either sales or assets, and net of the for each of the hypotheses described in industry median.To control for the different 540 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY investment opportunities available at value significant in both sets of regressions. and growth firms, we include the log of the Consistent with these results, we find that book-to-market ratio (BM) as a control the average coefficient on the Democracy variable in all specifications. Table 20.10 dummy is negative and significant in both summarizes the results, with BM coeffi- sets of regressions. We conclude that, other cients omitted. Columns (1) and (3) give things equal, high-G firms have higher results for the full sample, with G as the key CAPEX than do low-G firms. regressor; columns (2) and (4) give results Another outlet for capital expenditure is for the sample restricted to firms in the for firms to acquire other firms. Some of Democracy and Dictatorship Portfolios, with the strongest evidence for the importance a Democracy dummy as the key regressor. of agency costs comes from the negative The average coefficient on G is positive and returns to acquirer stocks after a bid is

Table 20.10 Capital expenditure

(1) (2) (3) (4) CAPEX/Assets CAPEX/Sales

Democracy Democracy G Portfolio G Portfolio

1991 1.32** 13.02** 0.70* 9.28 (0.27) (4.28) (0.32) (4.96) 1992 0.42 7.03 0.54 7.23 (0.35) (4.86) (0.35) (6.01) 1993 0.81* 6.06 0.09 1.68 (0.37) (4.48) (0.34) (4.98) 1994 0.51 7.84 0.07 4.82 (0.32) (5.21) (0.37) (4.76) 1995 0.35 3.40 0.32 9.80 (0.39) (6.83) (0.39) (5.90) 1996 0.75 6.90 0.31 3.26 (0.39) (5.55) (0.33) (6.36) 1997 0.74* 4.23 0.70 8.05 (0.34) (3.50) (0.40) (5.71) 1998 0.80* 10.57 0.37 6.43 (0.37) (6.75) (0.35) (5.63) 1999 0.15 3.12 0.32 3.49 (0.39) (4.20) (0.38) (5.52) Mean 0.62** 6.21** 0.30* 5.23** (0.13) (1.53) (0.11) (1.41)

The first and third columns of this table present the results of annual median (least absolute deviation) regressions of CAPEX/Assets and CAPEX/Sales on the Governance Index, G, measured in the previous year,and BM. The second and fourth columns restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) portfolios and include as regressors a dummy variable for the Democracy portfolio and BM.The coefficients of BM and the constant are omitted from the table. The calculation of G is described in Section II. CAPEX is capital expenditures, and BM is the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous fiscal year to size at the close of the previous calendar year.Both dependent variables are net of the industry median, which is calculated by matching the four-digit SIC codes of all firms in the CRSP-Compustat merged database in December of each year to the 48 industries designated by Fama and French [1997]. The coefficients and standard errors from each annual cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors are given in the last row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and standard errors are multiplied by 1000. CORPORATE GOVERNANCE AND EQUITY PRICES 541 announced. Considerable evidence shows year, and the average coefficient on G is that these negative returns are correlated positive and significant. Consistent with with other agency problems, including low this result, the average coefficient on the managerial ownership [Lewellen, Loderer, Democracy dummy is negative for both sets and Rosenfeld 1985], high free-cash flow of regressions and is significant for [Lang, Stulz, and Walkling 1991], and Acquisition Count. diversifying transactions [Morck, Shleifer, One interpretation of these results is that and Vishny 1990]. In addition to negative high-G firms engaged in an unexpectedly announcement returns, there is also long- large amount of inefficient investment dur- run evidence of negative abnormal per- ing the 1990s.This interpretation is consis- formance by acquirer firms [Loughran and tent with contemporaneous unexpected Vijh 1997; Rau and Vermaelen 1998].21 differences in profitability, stock returns, Taken together,these studies suggest acqui- and firm value. This inefficient investment sitions as another pathway through which does not necessarily mean that firms are governance affects performance. attempting to maximize their size in a form To analyze the relation between acquisi- of empire building. Indeed, empire building tion activity and G, we use the SDC database would be inconsistent with the negative to identify all transactions in which a relationship between sales growth and G sample firm acted as either the acquirer or found in Table 20.9. Instead, managers the seller during the sample period. From may be attempting to stave off “empire January 1991 through December 1999, collapse” with high expenditure and acqui- there are 12,694 acquisitions made by sition activity. In that case, the results sample firms; SDC gives the acquisition of this section are consistent with the price for just under half of these. For each evidence of Table 20.9. firm, we count the number of acquisitions (“Acquisition Count”). We also calculate the sum of the price of all acquisitions in V.B. Evidence on Hypothesis II each calendar year and divide this sum by the firm’s average market capitalization It is well established that insider trading for the first day and last day of the year can forecast returns. Firms whose shares (“Acquisition Ratio”). have been intensively sold (bought) by Table 20.11 summarizes the results of insiders tend to underperform (overper- annual regressions for both Acquisition form) benchmarks in subsequent periods.22 Count and the Acquisition Ratio in year t If some 1980s insiders forecasted poor on G (or a Democracy dummy), the log of performance for their firms, we might size, the log of the book-to-market ratio, expect them to have looked for ways to and 48 industry dummies, all measured at keep the shareholders from firing them, year-end t 1. Coefficients on all control either through voting or takeovers. In this variables are omitted from the table. Since case, weak shareholder rights would be a many firms make no acquisitions in a year, symptom of insiders’ superior information, the dependent variables are effectively left- but would not necessarily be the cause of censored at zero. To account for this cen- the poor performance in the subsequent soring, we estimate Poisson regressions for decade. Acquisition Count and Tobit regressions for To study this possibility, we use data the Acquisition Ratio. Columns (1) and (3) collected by Thomson Financial from the give results for the full sample, with G as required SEC insider-trading filings. For the key regressor; columns (2) and (4) give each firm in our sample, we sum all (split- results for the sample restricted to firms in adjusted) open-market transactions for all the Democracy and Dictatorship insiders in each year, with purchases enter- Portfolios, with a Democracy dummy as the ing positively and sales entering negatively. key regressor. For both sets of regressions, We then normalize this sum by shares the coefficients on G are positive in every outstanding at the beginning of the year to 542 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 20.11 Acquisitions

(1) (2) (3) (4) Acquisition count (Poisson Acquisition ratio (Tobit regressions) regressions)

Democracy Democracy G Portfolio G Portfolio

1991 1.58 50.81 0.51 0.14 (1.46) (26.12) (0.47) (5.03) 1992 1.64 31.39 0.10 7.91 (1.44) (24.61) (0.50) (6.42) 1993 1.75 47.67 0.70 6.31 (1.42) 24.51 (0.56) (6.85) 1994 4.09** 13.10 0.75 1.82 (1.27) (21.02) (0.48) (4.14) 1995 2.57* 60.92** 0.41 2.95 (1.15) (17.85) (0.44) (4.42) 1996 2.69* 66.06** 1.33* 24.22** (1.14) (20.48) (0.60) (9.41) 1997 2.34* 63.81** 0.99* 9.24 (1.12) (19.03) (0.51) (6.78) 1998 2.42* 52.03** 1.47 11.11 (1.09) (17.67) (0.76) (8.51) 1999 0.52 47.64** 0.84 20.87* (1.01) (17.27) (0.74) (9.68) Mean 2.18** 48.16** 0.79** 7.21 (0.33) (5.60) (0.14) (3.49)

The first column of this table presents annual Tobit regressions of the Acquisition ratio on the Governance Index, G, measured in the previous year, SIZE, BM, and industry dummy variables. The third column presents annual Poisson regressions of Acquisition count on the same explanatory variables. In the second and fourth columns, we restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) Portfolios, and we include as a regressor a dummy variable that equals 1 when the firm is in the Democracy Portfolio and 0 otherwise.The coefficients on SIZE, BM, and the industry dummy variables are omitted from the table.The calculation of G is described in Section II. Acquisition ratio is defined as the sum of the value of all corporate acquisitions during a calendar year scaled by the average of market value at the beginning and end of the year. Acquisition count is defined as the number of acquisitions during a calendar year.The data on acquisitions are from the SDC database. SIZE is the log of market capitalization at the end of the previous calendar year in millions of dollars, and BM is the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous fiscal year to size at the close of the previous calendar year. Industry dummy variables are created by matching the four-digit SIC codes of all firms in the CRSP-Compustat merged database in December of each year to the 48 industries designated by Fama and French [1997]. The coefficients and standard errors from each annual cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors are given in the last row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and standard errors are multiplied by 100.

arrive at a “Net Purchases” measure for Purchases on G (or a Democracy dummy), each firm in each year. If insiders put BM, and log of size. For some firm-years, new provisions in place when they forecast the Net Purchase measure is dominated poor performance, then we would expect by one large transaction. While large Net Purchases to be negatively correlated transactions might have information with G. content, they might also reflect liquidity We employ two regression specifications. or rebalancing needs. In an OLS regression, First, we estimate OLS regressions of Net firms with large outliers will dominate. CORPORATE GOVERNANCE AND EQUITY PRICES 543 Thus, we also estimate ordered logit a Democracy dummy as the key regressor. regressions on the same OLS regressors, Coefficients on all control variables are in which the dependent variable is equal omitted from the table. We find no signifi- to one if Net Purchases is positive, zero cant relationships between governance and if Net Purchases is zero, and negative one if insider trading. Two of four sets of regres- Net Purchases is negative. sions have positive average coefficients, Table 20.12 summarizes the results of two have negative average coefficients, and these regressions. Columns (1) and (3) give none of these average coefficients are results for the full sample, with G as the key significant. In untabulated results we also regressor; columns (2) and (4) give results estimated median regressions, replicated all for the sample restricted to firms in the of the above results using all transactions Democracy and Dictatorship Portfolios with (the main difference is the inclusion of

Table 20.12 Insider Trading

(1) (2) (3) (4) OLS Ordered logit

Democracy Democracy G Portfolio G Portfolio

1991 0.07* 0.14 8.85 345.18 (0.04) (0.53) (21.34) (295.15) 1992 0.10 1.47 66.92** 499.93 (0.07) (1.50) (21.70) (310.53) 1993 0.10 0.23 32.40 797.17* (0.07) 0.51 (21.41) (326.87) 1994 0.07 0.61 28.09 323.07 (0.04) (1.23) (20.58) (290.11) 1995 0.04 0.17 4.66 153.33 (0.02) (0.20) (22.00) (308.90) 1996 0.15 0.62 12.01 93.95 (0.14) (1.05) (21.67) (321.18) 1997 0.01 0.89 46.08 781.42* (0.10) (0.66) (24.33) (369.78) 1998 0.12 2.41 1.88 146.49 (0.20) (3.17) (24.31) (342.22) 1999 0.36 1.36 4.41 117.36 (0.48) (2.91) (21.09) (323.85) Mean 0.09 0.15 19.16 204.25 (0.04) (0.40) (8.66) (140.02)

The first and third columns of this table present annual OLS and ordered logit regressions of Net insider purchases on G measured in the previous year, SIZE, BM, and a constant. In the second and fourth columns, we restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) Portfolios and we include as a regressor a dummy variable that equals 1 when the firm is in the Democracy Portfolio and 0 otherwise. The coefficients on SIZE, BM, and the constant are omitted from the table. The calculation of G is described in Section II. Net insider purchases is the sum of split-adjusted open market purchases less split-adjusted open market sales during a year scaled by shares outstanding at the end of the previous calendar year.The ordered logit regressions use a dependent variable that equals 1 if Net insider purchases is positive, 0 if it is zero, and 1 if it is negative. The data on insider sales are from the Thomson database. SIZE is the log of market capitalization in millions of dollars measured at the end of the previous calendar year, and BM is the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous fiscal year to size at the close of the previous calendar year.The coefficients and standard errors from each annual cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors are given in the last row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and standard errors are multiplied by 1000. 544 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY option-exercise transactions), and estimated and industry-adjusted, for the full sample long-horizon regressions using all years of of firms in each month with G as the key data for each firm. In none of these cases independent variable. In both regressions did we find a robust relationship between the average coefficient on G is negative but governance and insider trading. Overall, we not significant. The point estimates are not find no support for Hypothesis II in the small. For example, the point estimate for insider-trading data. the coefficient on G in column (3) implies a lower return of approximately four bp per month ( 48 bp per year) for each V.C. Evidence on Hypothesis III additional point of G, but it would require estimates nearly twice as large before What other factors might be driving the statistical significance would be reached. return difference between the Democracy The next two columns give the results and Dictatorship Portfolios? We saw in when the sample is restricted to stocks Table II that G is correlated with several in either the Democracy (G 5) or firm characteristics, including S&P 500 Dictatorship (G 14) Portfolios. In the membership, institutional ownership, trading first column the dependent variable is volume, and past sales growth. If returns the raw monthly return for each stock. In to stocks with these characteristics differed the second column the dependent variable in the 1990s in a way not captured by the is the industry-adjusted return for each model in equation (1), then a type of omitted stock, where industry adjustments are variable bias may drive the abnormal-return relative to the Fama and French [1997] results. In this section we explore this pos- 48 industries. The key independent variable sibility using a cross-sectional regression in these regressions is the Democracy approach. In addition to providing evidence dummy, set equal to one if the stock is in on Hypothesis III, this method also supple- the Democracy Portfolio and zero if the ments the analysis of subsection III.B by stock is in the Dictatorship Portfolio. For allowing a separate regressor for each both the raw and industry-adjusted returns, component of G. the coefficient on this dummy variable is For each month in the sample period, positive and significant at the 1 percent September 1990 to December 1999, we level. The average point estimate can be estimate interpreted as a monthly abnormal return. These point estimates, 76 bp per month rit at btXit ctZit eit (3) raw and 63 bp per month industry- adjusted, are similar to those found in the where, for firm i in month t, rit are the factor models, and provide a further returns (either raw or industry-adjusted), robustness check to the benchmark result. Xit is a vector of governance variables Here, industry adjustments explain about (either G, its components, or inclusion in one-sixth of the raw result. In the factor- one of the extreme portfolios), and Zit is model results of Table 20.7, the industry a vector of firm characteristics. As adjustment explained about one-third of elements of Z, we include the full set the raw result. of regressors used by Brennan, Chordia, Columns (5) and (6) of Table 20.13 give and Subrahmanyam [1998], plus five-year the results for the full sample of firms sales growth, S&P 500 inclusion, and insti- when the five subindices are used as the tutional ownership.23 Variable definitions components of X. In principle, these are given in Appendix 2. regressions could help us distinguish We estimate (3) separately for each between Hypotheses I and III. If gover- month and then calculate the mean and nance provisions cause poor performance, time-series standard deviation of the then we might expect certain provisions to 112 monthly estimates of the coefficients. play a stronger role. In the absence of such a Table 20.13 summarizes the results. finding, we should wonder whether the results The first two columns give the results, raw are driven by some other characteristic. CORPORATE GOVERNANCE AND EQUITY PRICES 545 Table 20.13 Fama-MacBeth Return Regressions

(2) (4) (6) (1) Industry- (3) Industry- (5) Industry- Raw adjusted Raw adjusted Raw adjusted

G 0.04 0.02 (0.04) (0.03) Democracy 0.76* 0.63* Portfolio (0.32) (0.26) Delay 0.03 0.02 (0.10) (0.07) Protection 0.07 0.01 (0.08) (0.06) Voting 0.08 0.08 (0.13) (0.10) Other 0.01 0.04 (0.08) (0.07) State 0.02 0.04 (0.08) (0.06) NASDUM 0.83 0.42 8.23 10.36 2.60 0.29 (6.94) (5.26) (6.45) (5.94) (6.39) (4.98) SP500 0.19 0.20 0.42 0.21 0.19 0.24 (0.49) (0.42) (0.49) (0.41) (0.45) (0.40) BM 0.04 0.14 0.06 0.11 0.06 0.15 (0.19) (0.12) (0.38) (0.29) (0.20) (0.11) SIZE 0.17 0.22 0.47 0.02 0.19 0.24 (0.27) (0.16) (0.38) (0.32) (0.27) (0.17) PRICE 0.26 0.20 0.28 0.44 0.20 0.16 (0.26) (0.20) (0.31) (0.31) (0.28) (0.22) IO 0.61 0.10 0.78 0.16 0.59 0.14 (0.47) (0.33) (0.67) (0.60) (0.44) (0.33) NYDVOL 0.11 0.21 0.49 0.03 0.13 0.21 (0.29) (0.18) (0.36) (0.31) (0.28) (0.18) NADVOL 0.01 0.13 0.09 0.48 0.06 0.15 (0.43) (0.29) (0.41) (0.39) (0.43) (0.29) YLD 10.85 10.94 15.74 9.23 6.21 8.76 (10.54) (7.25) (14.62) (11.56) (11.63) (7.70) RET2–3 0.48 0.93 2.04 1.82 0.57 1.03 (1.40) (1.04) (2.33) (1.73) (1.43) (1.07) RET4–6 0.68 0.48 2.21 1.12 0.58 0.55 (1.33) (0.92) (1.89) (1.36) (1.33) (0.93) RET7–12 2.42* 0.89 0.12 1.67 2.69** 1.06 (1.00) (0.65) (1.35) (1.03) (0.99) (0.65) SGROWTH 0.00 0.03 0.75 0.27 0.01 0.02 (0.26) (0.18) (0.47) (0.40) (0.25) (0.18) Constant 0.53 0.18 1.17 1.86 0.03 0.16 (2.55) (1.71) (3.43) (2.99) (2.39) (1.69)

This table presents the average coefficients and time-series standard errors for 112 cross-sectional regressions for each month from September 1990 to December 1999.The dependent variable is the stock return for month t.The results are presented using both raw and industry-adjusted returns, with industry adjustments done using the 48 industries of Fama and French [1997].The first and second columns include all firms with data for all right-hand side variables and use G, the Governance index, as an independent variable. In the third and fourth columns, the sample is restricted to firms in either the Democracy (G 5) or Dictatorship (G 14) Portfolios, and we use the independent variable, Democracy Portfolio, a dummy variable that equals 1 when the firm is in the Democracy Portfolio and 0 otherwise. In the fifth and sixth columns, we again include all firms with data for each explanatory variable and use the subindices, Delay, Protection, Voting, Other, and State as regressors.The calculation of G and the subindices is described in Section II. Definitions for all other explanatory variables are provided in Appendix 2. All regressions are estimated with weighted least squares where all variables are weighted by market value at the end of month t 1. Significance at the 5 percent and 1 percent levels is indicated by* and **, respectively. 546 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY For example, some legal scholars argue the possibility that some unobserved that the Delay provisions are the only characteristic is correlated with G and is defenses with deterrent value [Coates also the main cause of abnormal returns. 2000; Daines and Klausner 2001]. If This type of omitted-variable bias could be managers also believe this, then the Delay something prosaic, such as imperfect indus- subindex should also be the most important try adjustments or model misspecification, driver of the results. or something more difficult to quantify, such Unfortunately, large standard errors, due as a partially unobservable or immeasurable in part to the substantial multicollinearity “corporate culture.” Under the latter between the regressors, makes it difficult explanation, management behavior would to construct a powerful test. None of the be constrained by cultural norms within the subindex coefficients are statistically signifi- firm, and democracy and dictatorship cant in either specification, but many of would be a persistent feature of a corpo- the point estimates are economically large. rate culture; G would be a symptom, but In the end, we cannot precisely measure not a cause, of this culture. In this case, all the relative importance of Delay or any the results of the paper could be explained other subindex. This is similar to the if investors mispriced culture in 1990, just problem that occurred in the Q regressions as they appear to have mispriced its of Table 20.8. For example, in both proxy, G. The policy impact of reducing G Tables 20.8 and 20.13 the coefficients on would be nonexistent unless it affected the Voting suggest potentially enormous culture of managerial power that was the economic significance, but large standard true driver of poor performance. errors prevent any meaningful statistical In addition to the three hypotheses inference. considered above, other explanations fall In untabulated tests, we also included all into the general class of “Type I” error. For 28 provisions from Table 20.1 as separate example, one could argue that investors in regressors in (3). Regressing raw returns 1990 had rational expectations about the on these 28 provisions plus the same expected costs and benefits of takeover controls as in Table 20.13, we find that 16 defenses, where the expected costs are of the coefficients are negative, and only more severe agency problems and the one (Unequal Voting) is significant. (With expected benefits are higher takeover pre- this many regressors, we would expect one miums. Then, when the hostile takeover to appear “significant” just by chance.) market largely evaporated in the early Results for industry-adjusted returns are 1990s—perhaps because of macro- similar.These results highlight and magnify economic conditions unrelated to takeover the lack of power in the subindex regres- defenses—Dictatorship firms were left sions. Indeed, many of the point estimates with the costs but none of the benefits of imply return effects above 20 basis their defenses. Over the subsequent decade, points per month (2.4 percent per year), the expected takeover premiums eroded as but are still far from being statistically investors gradually learned about the weak significant. This result also suggests that takeover market. Simple calculations the Democracy-minus-Dictatorship return suggest that this explanation cannot be differences are not driven by the presence that important. Suppose that in 1990 or absence of any one provision. the expected takeover probability for Dictatorship firms was 30 percent, and the V.D. Discussion expected takeover premium conditional on takeover was also 30 percent. Further The evidence in subsections V.A, V.B, and suppose that both of these numbers were V.C must be interpreted with caution. zero for Democracy firms.Then, the uncon- Since this is an experiment without random ditional expected takeover premium for assignment, no analysis of causality Dictatorship firms would have been only can be conclusive. The main problem is 9 percent, which is approximately the CORPORATE GOVERNANCE AND EQUITY PRICES 547 relative underperformance of these firms rules of corporate governance. Beginning for only a single year. in the late 1980s, there is significant and In sum, we find some evidence in support stable variation in these rules across of Hypothesis I and no evidence in different firms. Using 24 distinct corpo- support of Hypothesis II. For Hypothesis III rate-governance provisions for a sample of we find that industry classification can about 1500 firms per year during the explain somewhere between one-sixth and 1990s, we build a Governance Index, one-third of the benchmark abnormal denoted as G, as a proxy for the balance of returns, but we do not find any other power between managers and shareholders observable characteristic that explains the in each firm.We then analyze the empirical remaining abnormal return. The subindex relationship of this index with corporate regressions, which might be helpful in dis- performance. tinguishing between Hypotheses I and III, We find that corporate governance is are not powerful enough for strong strongly correlated with stock returns inference. We conclude that the remaining during the 1990s. An investment strategy performance differences, which are eco- that purchased shares in the lowest-G nomically large, were either directly caused firms (“Democracy” firms with strong by governance provisions (Hypothesis I), or shareholder rights), and sold shares in the were related to unobservable or difficult- highest-G firms (“Dictatorship” firms to-measure characteristics correlated with with weak shareholder rights), earned governance provisions (Hypothesis III). abnormal returns of 8.5 percent per year. What do these hypotheses imply about At the beginning of the sample, there is abnormal returns in the future? None already a significant relationship between suggests any obvious pattern for the valuation and governance: each one-point relationship between G and returns. Under increase in G is associated with a decrease Hypothesis I, if we interpret our test as a in Tobin’s Q of 2.2 percentage points. By long-run event study, then there is no the end of the decade, this difference reason to expect any relationship once the has increased significantly, with a one- market has fully priced the underlying point increase in G associated with a “event” of corporate governance. The fact decrease in Tobin’s Q of 11.4 percentage that this price adjustment is taking such a points. The results for both stock returns long time does not seem so surprising in and firm value are economically large and light of the lengthy intervals necessary for are robust to many controls and other much more tangible information to be firm characteristics. incorporated into prices.24 Thus, to the We consider several explanations for the extent that end-of-sample price adjustment results, but the data do not allow strong is incomplete, complete, or has overre- conclusions about causality. There is some acted, the future relationship between G evidence, both in our sample and from and returns could be negative, zero, or pos- other authors, that weak shareholder rights itive. Under Hypothesis II there is a similar caused poor performance in the 1990s. It is dependence on whether past insider infor- also possible that the results are driven by mation has been fully incorporated into some unobservable firm characteristic. prices. Under Hypothesis III future return These multiple causal explanations have differences would be driven by the relevant starkly different policy implications and omitted characteristic; clearly, this hypoth- stand as a challenge for future research. esis yields no clear prediction. The empirical evidence of this paper establishes the high stakes of this challenge. If an 11.4 percentage point difference in firm VI. CONCLUSION value were even partially “caused” by each additional governance provision, then the The power-sharing relationship between long-run benefits of eliminating multiple investors and managers is defined by the provisions would be enormous. 548 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY APPENDIX 1: CORPORATE- authority to determine voting, dividend, GOVERNANCE PROVISIONS conversion, and other rights.While it can be used to enable a company to meet changing This appendix describes the provisions listed financial needs, its most important use is to in Table 20.1 and used as components of implement poison pills or to prevent the Governance Index. The shorthand title takeover by placing this stock with friendly of each provision, as used in the text of the investors. Because of this role, blank check paper, is given in boldface. These descrip- preferred stock is a crucial part of a “delay” tions are given in alphabetical order and strategy. Companies that have this type of are similar to Rosenbaum [1998]. For a preferred stock but require shareholder few provisions we discuss their impact on approval before it can be used as a takeover shareholder rights or the logic behind their defense are not coded as having this categorization in Table 20.1. provision in our data. Antigreenmail. Greenmail refers to a Business Combination laws impose a transaction between a large shareholder moratorium on certain kinds of transac- and a company in which the shareholder tions (e.g., asset sales, mergers) between a agrees to sell his stock back to the com- large shareholder and the firm, unless the pany, usually at a premium, in exchange for transaction is approved by the Board of the promise not to seek control of the Directors. Depending on the State, this company for a specified period of time. moratorium ranges between two and five Antigreenmail provisions prevent such years after the shareholder’s stake passes a arrangements unless the same repurchase prespecified (minority) threshold. These offer is made to all shareholders or laws were in place in 25 states in 1990 and approved by a shareholder vote. Such two more by 1998. It is the only state provisions are thought to discourage takeover law in Delaware, the state of accumulation of large blocks of stock incorporation for about half of our sample. because one source of exit for the stake is Bylaw and Charter amendment limita- closed, but the net effect on shareholder tions limit shareholders’ ability to amend the wealth is unclear [Shleifer and Vishny governing documents of the corporation. 1986; Eckbo 1990]. Five states have This might take the form of a supermajority specific Antigreenmail laws, and two other vote requirement for charter or bylaw states have “recapture of profits” laws, amendments, total elimination of the ability which enable firms to recapture raiders’ of shareholders to amend the bylaws, or the profits earned in the secondary market. We ability of directors (beyond the provisions consider recapture of profits laws to be a of state law) to amend the bylaws without version of Antigreenmail laws (albeit a shareholder approval. stronger one). The presence of firm-level Control-share Cash-out laws enable Antigreenmail provisions is positively shareholders to sell their stakes to a “con- correlated with 18 out of the other 21 firm- trolling” shareholder at a price based on level provisions, is significantly positive in 8 the highest price of recently acquired of these cases, and is not significantly shares.This works something like fair-price negative for any of them. Furthermore, provisions (see below) extended to non- states with Antigreenmail laws tend to pass takeover situations. These laws were in them in conjunction with laws more clearly place in three states by 1990 with no designed to prevent take-overs [Pinnell additions during the decade. 2000]. Since it seems likely that most firms A Classified Board (or “staggered” and states perceive Antigreenmail as a board) is one in which the directors are takeover “defense,” we treat Antigreenmail placed into different classes and serve over- like the other defenses and code it as a lapping terms. Since only part of the board decrease in shareholder rights. can be replaced each year, an outsider who Blank Check preferred stock is stock gains control of a corporation may have to over which the board of directors has broad wait a few years before being able to gain CORPORATE GOVERNANCE AND EQUITY PRICES 549 control of the board.This slow replacement above those of other stakeholders [Pinnell makes a classified board a crucial compo- 2000]. We treat firms in these two states nent of the Delay group of provisions, and as though they had an expanded directors’ one of the few provisions that clearly retains duty provision unless the firm has explicitly some deterrent value in modern take-over opted out of coverage under the law. battles [Daines and Klausner 2001]. Fair-Price provisions limit the range of Compensation Plans with changes- prices a bidder can pay in two-tier offers. in-control provisions allow participants in They typically require a bidder to pay to all incentive bonus plans to cash out options or shareholders the highest price paid to any accelerate the payout of bonuses if there during a specified period of time before the should be a change in control. The details commencement of a tender offer, and do may be a written part of the compensation not apply if the deal is approved by the agreement, or discretion may be given to board of directors or a supermajority of the compensation committee. the target’s shareholders. The goal of this Director indemnification Contracts are provision is to prevent pressure on the contracts between the company and par- target’s shareholders to tender their shares ticular officers and directors indemnifying in the front end of a two-tiered tender offer, them from certain legal expenses and and they have the result of making such an judgments resulting from lawsuits pertain- acquisition more expensive. Also, 25 states ing to their conduct. Some firms have had Fair-Price laws in place in 1990, and both “Indemnification” in their bylaws or two more states passed such laws in 1991. charter and these additional indemnification The laws work similarly to the firm-level “Contracts.” provisions. Control-share Acquisition laws (see Golden Parachutes are severance Supermajority, below). agreements that provide cash and noncash Cumulative Voting allows a shareholder compensation to senior executives upon to allocate his total votes in any manner an event such as termination, demotion, or desired, where the total number of votes is resignation following a change in control. the product of the number of shares owned They do not require shareholder approval. and the number of directors to be elected. While such payments would appear to deter By allowing them to concentrate their votes, takeovers by increasing their costs, one this practice helps minority shareholders could argue that these parachutes also ease to elect directors. Cumulative Voting and the passage of mergers through contractual Secret Ballot (see below) are the only two compensation to the managers of the target provisions whose presence is coded as an company [Lambert and Larcker 1985]. increase in shareholder rights, with an While the net impact on managerial additional point to the Governance Index if entrenchment and shareholder wealth is the provision is absent. ambiguous, the more important effect is Directors’ Duties provisions allow direc- the clear decrease in shareholder rights. tors to consider constituencies other than In this case, the “right” is the ability of a shareholders when considering a merger. controlling shareholder to fire management These constituencies may include, for without incurring an additional cost. Golden example, employees, host communities, or Parachutes are highly correlated with all suppliers. This provision provides boards the other takeover defenses. Out of 21 pair- of directors with a legal basis for rejecting wise correlations with the other firm-level a takeover that would have been beneficial provisions, 15 are positive, 10 of these to shareholders. Thirty-one states have positive correlations are significant, and only Directors’ Duties laws allowing similar one of the negative correlations is signifi- expansions of constituencies, but in cant.Thus, we treat Golden Parachutes as a only two of these states (Indiana and restriction of shareholder rights. Pennsylvania) are the laws explicit that the Director Indemnification uses the bylaws, claims of shareholders should not be held charter, or both to indemnify officers and 550 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY directors from certain legal expenses and are used to count proxy votes, and the judgments resulting from lawsuits pertain- management usually agrees not to look at ing to their conduct. Some firms have both individual proxy cards. This can help this “Indemnification” in their bylaws or eliminate potential conflicts of interest for charter and additional indemnification fiduciaries voting shares on behalf of “Contracts.” The cost of such protection others, and can reduce pressure by can be used as a market measure of the management on shareholder-employees or quality of corporate governance [Core shareholder-partners. Cumulative Voting 1997, 2000]. (see above) and Secret Ballots are the only Limitations on director Liability are two provisions whose presence is coded as charter amendments that limit directors’ an increase in shareholder rights, with an personal liability to the extent allowed by additional point to the Governance Index if state law. They often eliminate personal the provision is absent. liability for breaches of the duty of care, Executive Severance agreements assure but not for breaches of the duty of loyalty high-level executives of their positions or or for acts of intentional misconduct or some compensation and are not contingent knowing violation of the law. upon a change in control (unlike Golden or Pension Parachutes prevent an acquirer Silver Parachutes). from using surplus cash in the pension fund Silver Parachutes are similar to Golden of the target to finance an acquisition. Parachutes in that they provide severance Surplus funds are required to remain the payments upon a change in corporate property of the pension fund and to be used control, but differ in that a large number of for plan participants’ benefits. a firm’s employees are eligible for these Poison Pills provide their holders with benefits. Since Silver Parachutes do not special rights in the case of a triggering protect the key decision makers in a event such as a hostile takeover bid. If a merger, we classified them in the Other deal is approved by the board of directors, group rather than in the Protection group. the poison pill can be revoked, but if the Special Meeting limitations either deal is not approved and the bidder pro- increase the level of shareholder support ceeds, the pill is triggered. Typical poison required to call a special meeting beyond pills give the holders of the target’s stock that specified by state law or eliminate the other than the bidder the right to purchase ability to call one entirely. Such provisions stock in the target or the bidder’s company add extra time to proxy fights, since bidders at a steep discount, making the target must wait until the regularly scheduled unattractive or diluting the acquirer’s annual meeting to replace board members voting power. Poison pills are a crucial or dismantle takeover defenses. This delay component of the “delay” strategy at is especially potent when combined with the core of modern defensive tactics. limitations on actions by written consent Nevertheless, we do not include poison pills (see below). in the Delay group of provisions, but Supermajority requirements for approval include it in the Other group because the of mergers are charter provisions that pill itself can be passed on less than one- establish voting requirements for mergers or day’s notice, so it need not be in place for other business combinations that are higher the other Delay provisions to be effective. than the threshold requirements of state law. The other provisions in this group require a They are typically 66.7, 75, or 85 percent, shareholder vote, so they cannot be passed and often exceed attendance at the annual on short notice. See Coates [2000] and meeting. In practice, these provisions are Daines and Klausner [2001] for a discussion similar to Control-Share Acquisition of this point. laws. These laws require a majority of Under a Secret Ballot (also called disinterested shareholders to vote on confidential voting), either an independent whether a newly qualifying large share- third party or employees sworn to secrecy holder has voting rights.They were in place CORPORATE GOVERNANCE AND EQUITY PRICES 551 in 25 states by September 1990 and one 13F quarterly filings, as provided by additional state in 1991. Thomson Financial.We use the most recent Unequal Voting rights limit the voting quarter as of the end of month t – 1, with rights of some shareholders and expand shares outstanding (from CRSP) measured those of others. Under time-phased voting, on the same date. shareholders who have held the stock for a NADVOL—The dollar volume of trading given period of time are given more votes in month t – 2 for stocks that trade on the per share than recent purchasers. Another NASDAQ. Approximated as stock price at variety is the substantial-shareholder the end of month t – 2 multiplied by share provision, which limits the voting power of volume in month t – 2. For New York Stock shareholders who have exceeded a certain Exchange (NYSE) and American Stock threshold of ownership. Exchange (AMEX) stocks, NADVOL Limitations on action by Written equals zero. Consent can take the form of the establish- NASDUM—A dummy variable equal to ment of majority thresholds beyond the one if the firm traded on the NASDAQ level of state law, the requirement of Stock Market at the beginning of month t unanimous consent, or the elimination of and zero otherwise. the right to take action by written consent. NYDVOL—The dollar volume of trading Such requirements add extra time to many in month t – 2 for stocks that trade on the proxy fights, since bidders must wait until NYSE or AMEX. Approximated as stock the regularly scheduled annual meeting price at the end of month t – 2 multiplied to replace board members or dismantle by share volume in month t – 2. For takeover defenses. This delay is especially NASDAQ stocks, NYDVOL equals zero. potent when combined with limitations for PRICE—Price at the end of month t – 2. calling special meetings (see above). Q—The market value of assets divided by the book value of assets (Compustat item 6), where the market value of assets is APPENDIX 2: DEFINITIONS FOR computed as book value of assets plus the THE REGRESSION VARIABLES market value of common stock less the sum of the book value of common stock This list includes all variables used as (Compustat item 60) and balance sheet regressors or for summary statistics in deferred taxes (Compustat item 74). All Tables 20.5 and 20.13. All components are book values for fiscal year t (from drawn from the CRSP monthly files and all Compustat) are combined with the market variables are in natural logs unless explic- value of common equity at the calendar itly noted otherwise. Variables are listed in end of year t. alphabetical order in boldface. RET2–3—Compounded gross returns BM—The ratio of book value of common for months t – 3 and t – 2. equity (previous fiscal year) to market RET4–6—Compounded gross returns value of common equity (end of previous for months t – 6 through t – 4. calendar year). Book value of common RET7–12—Compounded gross returns equity is the sum of book common equity for months t – 12 through t – 7. (Compustat item 60) and deferred taxes SGROWTH—The growth in sales (Compustat item 74).This variable, and all (Compustat item 12) over the previous five other variables that use Compustat data, fiscal years (not in logs). are recalculated each July and held SIZE—Market capitalization in millions constant through the following June. of dollars at the end of month t – 2. 5-Year Return—The compounded return SP500—membership in the S&P 500 as from month t – 61 to month t – 2. of the end of month t – 1. Value is equal to IO—Shares held by institutions divided one if the firm is in the index, and zero by total shares outstanding (not in logs). otherwise. Data are from CRSP S&P 500 Institutional holdings are from SEC Form constituent file. 552 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY VOLUME—The dollar volume of trading 5 We omit firms with dual-class common in month t – 2 NADVOL NYDVOL. stock because the wide variety of voting and YLD—The ratio of dividends in the ownership differences across these firms makes previous fiscal year (Compustat item 21) it difficult to compare their governance struc- to market capitalization measured at tures with those of single-class firms. 6 Unless otherwise noted, all statements calendar year-end (not in logs). about statistical significance refer to significance at the 5 percent level. 7 These laws are classified as “second- NOTES generation” in the literature to distinguish them from the “first-generation” laws passed by * We thank Franklin Allen, Judith Chevalier, many states in the sixties and seventies and held John Core, Robert Daines, Darrell Duffie, to be unconstitutional in 1982. See Comment Kenneth French, Gary Gorton, Edward Glaeser, and Schwert [1995] and Bittlingmayer [2000] Joseph Gyourko, Robert Holthausen, Steven for a discussion of the evolution and legal status Kaplan, Sendhil Mullainathan, Krishna of state takeover laws and firm-specific Ramaswamy, Roberta Romano, Virginia takeover defenses. The constitutionality of Rosenbaum, Andrei Shleifer, Peter Siegelman, almost all of the second-generation laws and the Robert Stambaugh, Jeremy Stein, René Stulz, firm-specific takeover defenses was clearly Joel Waldfogel, Michael Weisbach, Julie Wulf, established by 1990. All of the state takeover three anonymous referees, and seminar partici- laws cover firms incorporated in their home pants at the University of Chicago, Columbia, state. A few states have laws that also cover Cornell and Duke Universities, the Federal firms incorporated outside of the state that have Reserve Board of Governors, Georgetown significant business within the state.The rules for University, Harvard University, INSEAD, “significant” vary from case to case, but usually Stanford University, the Wharton School, Yale cover only a few very large firms. We do not University, the 2001 NBER Summer Institute, attempt to code for this out-of-state coverage. and the New York University Five-Star 8 The statistics of Table 20.1 reflect exactly Conference for helpful comments. Yi Qian and the frequency of coverage under the default law Gabriella Skirnick provided excellent research in each state. A small minority of firms elect to assistance. Gompers acknowledges the support “opt out” of some laws and “opt in” to others. of the Division of Research at Harvard Business We code these options separately and use them School. Ishii acknowledges support from an in the creation of our index. NSF Graduate Research Fellowship. 9 In the case of Secret Ballots, shareholder 1 Surveys of this literature can be found in fiduciaries argue that it enables voting without Bhagat and Romano [2001], Bittlingmayer threat of retribution, such as the loss of [2000], Comment and Schwert [1995], and investment-banking business by brokerage- Karpoff and Malatesta [1989]. house fiduciaries. See Gillan and Bethel [2001] 2 See Coates [2000] for a detailed review of and McGurn [1989]. these arguments. 10 Only two other provisions—Antigreenmail 3 Other papers that analyze relationships and Golden Parachutes—seem at all ambiguous. between governance and either firm value or Since both are positively correlated with the performance have generally focused on board vast majority of other firm-level provisions and composition, executive compensation, or insider can logically be viewed as takeover defenses, we ownership [Baysinger and Butler 1985; Bhagat code them like other defenses and add one point and Black 1998; Core, Holthausen, and Larcker to the index for each. See their respective 1999; Hermalin and Weisbach 1991; Morck, entries in Appendix 1 for a discussion. Shleifer, and Vishny 1988; Yermack 1996]. See 11 Firms usually have the option to opt out of Shleifer and Vishny [1997] for a survey. state law coverage. Also, a few state laws 4 These 24 provisions include 22 firm-level require firms to opt in to be covered. The firms provisions and six state laws (four of the laws that exercise these options are listed in the are analogous to four of the firm-level provi- IRRC data. When we constructed the State sions). For the remainder of the paper we refer subindex, we ignored these options and used the interchangeably to corporate governance default state coverage. When we constructed “laws,” “rules,” and “provisions.” We also the G index, we included the options and used refer interchangeably to “shareholders” and actual coverage. “investors” and refer to “management” as 12 The IRRC gives dates for some of the comprising both managers and directors. provision changes—where available, these data CORPORATE GOVERNANCE AND EQUITY PRICES 553 suggest that the majority of the provisions were but still allow for some underperformance for adopted in the 1980s. Danielson and Karpoff acquisitions financed by stock. A related debate [1998] perform a detailed study on a similar set on whether diversifying acquisitions destroy of provisions and demonstrate a rapid pace of value has grown too large to survey here. The change between 1984 and 1989. seminal works are Lang and Stulz [1994] and 13 Berkshire Hathaway disappeared because Berger and Ofek [1995]. Recent work is it added a second class of stock before 1998. summarized in Holmstrom and Kaplan [2001] Firms with multiple classes of common stock and Stein [2001]. are not included in our analysis. 22 See Seyhun [1998] for a comprehensive 14 NCR disappeared after a merger. It review of this literature and a discussion of SEC reappeared in the sample in 1998 as a spinout, rules, filing requirements, and available data. but since it received a new permanent number 23 All of these additional variables are from CRSP,we treat the new NCR as a different correlated with G (see Table 20.3) and, in prior company. studies, with either firm value or abnormal 15 The industry names are from Fama and returns. See Lakonishok, Shleifer, and Vishny French [1997], but use a slightly updated [1994] (sales growth), Gompers and Metrick version of the SIC classification of these indus- [2001] (institutional ownership), and Morck tries that is given on Ken French’s website (June and Yang [2001] (Q). 2001). In Sections III and V we use both this 24 For example, there is evidence that updated classification and the corresponding earnings surprises [Bernard and Thomas 1989], industry returns (also from the French website). dividend omissions [Michaely, Thaler, and 16 See Basu [1977] (price-to-earnings ratio), Womack 1995], and stock repurchases Banz [1981] (size), Fama and French [1993] [Ikenberry, Lakonishok, and Vermaelen 1995] (size and book-to-market), Lakonishok, Shleifer, have long-term drift following the event, and all and Vishny [1994] (several value measures), seem to be relatively simple events compared and Jegadeesh and Titman [1993] (momentum). with changes in governance structure. 17 This model extends the Fama-French [1993] three-factor model with the addition of a momentum factor. For details on the REFERENCES construction of the factors, see Fama and French [1993] and Carhart [1997]. We are grateful to Ken French for providing the factor Banz, Rolf, “The Relation between Return and returns for SMB and HML. Momentum returns Market Value of Stocks,” Journal of were calculated by the authors using the Financial Economics, XXXVIII (1981), procedures of Carhart [1997]. 269–296. 18 Unlike Shin and Stulz [2000], we do not Basu, Sanjoy, “The Investment Performance of trim the sample of observations that have Common Stocks in Relation to Their Price- extreme independent variables. Results with a to-Earnings: A Test of the Efficient Markets trimmed sample are nearly identical and are Hypothesis,” Journal of Finance, XXXII available from the authors. (1977), 663–682. 19 Table 20.7, first row, second column, shows an alpha of 45 bp per month for the Baumol, William, Business Behavior, Value, and EW difference between the Democracy and Growth (New York, NY: Macmillan, 1959). Dictatorship Portfolios. Over 112 months Baysinger, Barry D., and Henry N. Butler, this produces a difference of approximately “Corporate Governance and the Board of 50 percent, as compared with the 56 19 Directors: Performance Effects of Changes 37 percent difference estimated for the Q in Board Composition,” Journal of Law, regressions. We use the EW alpha as a Economics, and Organization, I (1985), comparison because the Q regressions are also 101–124. equal-weighted. Berger, Philip G., and Eli Ofek, “Diversification’s 20 For an alternative view, see Stein [1988, Effect on Firm Value,” Journal of Financial 1989]. Empirical evidence on this issue is given by Daines and Klausner [2001], Johnson and Economics, XXXVII (1995), 39–65. Rao [1997], Meulbroek et al. [1990], Pugh, Bernard, Victor, and J. K. Thomas, “Post- Page, and Jahera [1992], and Titman, Wei, and Earnings Announcement Drift: Delayed Price Xie [2001]. Response or Risk Premium?” Journal of 21 Mitchell and Stafford [2000] have chal- Accounting Research, Supplement, XXVII lenged the magnitude of this longrun evidence, (1989), 1–36. 554 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Bertrand, Marianne, and Sendhil Mullainathan, Antitakeover Measures,” Journal of “Is There Discretion in Wage Setting?” Financial Economics, XXXIX (1995), 3–43. Rand Journal of Economics, XXX (1999a), Core, John, “On the Corporate Demand for 535–554. Directors’ and Officers’ Insurance,” Journal Bertrand, Marianne, and Sendhil Mullainathan, of Risk and Insurance, LXIV (1997), 63–87. “Corporate Governance and Executive Pay: ——, “The Directors’ and Officers’ Insurance Evidence from Takeover Legislation,”Working Premium: An Outside Assessment of the Paper,MIT Department of Economics, 1999b. Quality of Corporate Governance,” Journal of Bertrand, Marianne, and Sendhil Mullainathan, Law, Economics, and Organization, XVI “Enjoying the Quiet Life? Managerial (2000), 449–477. Behavior Following Anti-Takeover Core, John E., Robert W. Holthausen, and David Legislation,” Working Paper, Department of F. Larcker, “Corporate Governance, Chief Economics, Massachusetts Institute of Executive Officer Compensation, and Firm Technology, 2000. Performance,” Journal of Financial Bhagat, Sanjai, and B. S. Black, “The Economics, LI (1999), 371–406. Relationship between Board Composition and Daines, Robert, “Does Delaware Law Improve Firm Performance,” in Comparative Corporate Firm Value?” Journal of Financial Governance: The State of the Art and Economics, LXII (2001), 525–558. Emerging Research, K. Hopt, M. Roe, and Daines, Robert, and Michael Klausner,“Do IPO E. Wymeersch, eds. (Oxford, UK: Clarendon Charters Maximize Firm Value? Antitakeover Press; New York, NY: Oxford University Protection in IPOs,” Journal of Law, Press, 1998). Economics, and Organization, XVII (2001), Bhagat, Sanjai, and Roberta Romano, “Event 83–120. Studies and the Law: Part II—Empirical Danielson, Morris G., and Jonathan M. Karpoff, Studies of Corporate Law,” Working Paper, “On the Uses of Corporate Governance Yale University, 2001. Provisions,” Journal of Corporate Finance, Bittlingmayer, George, “The Market for IV (1998), 347–371. Corporate Control (Including Take-overs),” in Demsetz, Harold, and Kenneth Lehn, “The Encyclopedia of Law and Economics, Structure of Corporate Ownership: Causes Vol. III—The Regulation of Contracts, and Consequences,” Journal of Political Boudewijn Bouckaert and Gerrit de Geest, eds. Economy, XCIII (1985), 1155–1177. (Cheltenham, UK, and Northampton, MA: Eckbo, B. Espen,“Valuation Effects of Greenmail Edward Elgar, 2000). Prohibitions,” Journal of Financial and Borokhovich, Kenneth A., Kelly R. Brunarski, Quantitative Analysis, XXV (1990), 491–505. and Robert Parrino, “CEO Contracting and Fama, Eugene F., and Kenneth R. French, Antitakeover Amendments,” Journal of “Common Risk Factors in the Returns on Finance, LII (1997), 1495–1518. Bonds and Stocks,” Journal of Financial Brennan, Michael J., Tarun Chordia, and Economics, XXXIII (1993), 3–53. Avanidhar Subrahmanyam,“Alternative Factor Fama, Eugene F., and Kenneth R. French, Specifications, Security Characteristics, and “Industry Costs of Equity,” Journal of the Cross Section of Expected Stock Financial Economics, XLIII (1997), 153–194. Returns,” Journal of Financial Economics, Fama, Eugene F., and James D. MacBeth,“Risk, XLIX (1998), 345–375. Return, and Equilibrium: Empirical Tests,” Carhart, Mark,“On Persistence in Mutual Fund Journal of Political Economy, LXXXI Performance,” Journal of Finance, LII (1973), 607–636. (1997), 57–82. Garvey, Gerald T., and Gordon Hanka, “Capital Coates, John,“Takeover Defenses in the Shadow Structure and Corporate Control: The Effect of the Pill: A Critique of the Scientific of Antitakeover Statutes on Firm Leverage,” Evidence,” Texas Law Review, LXXIX Journal of Finance, LIV (1999), 519–546. (2000), 271–382. Gillan, Stuart, and Jennifer Bethel,“Managerial Comment, Robert, and G. William Schwert, Control of the Proxy Process: The Impact on “Poison or Placebo? Evidence on the Shareholder Voting,” Working Paper, Babson Deterrence and Wealth Effects of Modern College, 2001. CORPORATE GOVERNANCE AND EQUITY PRICES 555 Gompers, Paul A., and Andrew Metrick, Lambert, Richard A., and David F. Larcker, “Institutional Investors and Equity Prices,” “Golden Parachutes, Executive Decision- Quarterly Journal of Economics, CXIV Making and Shareholder Wealth,” Journal of (2001), 229–260. Accounting and Economics, VII (1985), Hermalin, Benjamin, and Michael Weisbach, 179–203. “The Effects of Board Composition and Lang, Larry H. P., and René M. Stulz,“Tobin’s Q, Direct Incentives on Firm Performance,” Corporate Diversification, and Firm Financial Management, XX (1991), Performance,” Journal of Political Economy, 101–112. CII (1994), 1248–1280. Holmstrom, Bengt, and Steven N. Kaplan, Lang, Larry H. P., René M. Stulz, and Ralph A. “Corporate Governance and Merger Activity Walkling, “A Test of the Free Cash Flow in the United States: Making Sense of the Hypothesis: The Case of Bidder Returns,” 1980s and 1990s,” Journal of Economic Journal of Financial Economics, XXIX Perspectives, XV (2001), 121–144. (1991), 315–335. Ikenberry, David, Josef Lakonishok, and Theo La Porta, Rafael, Florencio Lopez-de-Silanes, Vermaelen, “Market Underreaction to Open Andrei Shleifer, and Robert Vishny, “Investor Market Share Repurchases,” Journal of Protection and Corporate Valuation,” Financial Economics, XXXIX (1995), Working Paper, Harvard University, 2001. 181–208. Lewellen, Wilbur, Claudio Loderer, and Ahron Ishii, Joy L., “Corporate Governance in the Rosenfeld, “Merger Decisions and Executive 1990s: New Evidence on the Role of Stock Ownership in Acquiring Firms,” Shareholder Proposals,” Senior Honors Journal of Accounting and Economics, VII Thesis, Harvard University, 2000. (1985), 209–231. Jegadeesh, Narasimhan, and Sheridan Titman, Loughran, Tim, and Anand M. Vijh, “Do Long- “Returns to Buying Winners and Selling Term Shareholders Benefit from Corporate Losers: Implications for Stock Market Acquisitions?” Journal of Finance, LII Efficiency,” Journal of Finance, XLVIII (1997), 1765–1790. (1993), 65–91. Marris, Robin, The Economic Theory of Jensen, Michael, “Agency Costs of Free Cash Managerial Capitalism (New York, NY: Free Flow, Corporate Finance, and Takeovers,” Press of Glencoe, 1964). American Economic Review, LXXVI (1986), McGurn, Patrick S., Confidential Proxy Voting 323–329. (Washington, DC: Investor Responsibility Jensen, Michael, and Richard Ruback, “The Research Center Inc., 1989). Market for Corporate Control: The Scientific Meulbroek, Lisa K., Mark L. Mitchell, J. Harold Evidence,” Journal of Financial Economics, Mulherin, Jeffrey M. Netter, and Annette B. XI (1983), 5–50. Poulsen, “Shark Repellents and Managerial Johnson, Mark S., and Ramesh P. Rao, “The Myopia: An Empirical Test,” Journal of Impact of Antitakeover Amendments on Political Economy, XCVIII (1990), Corporate Financial Performance,” Financial 1108–1117. Review, XXXII (1997), 659–690. Michaely, Roni, Richard Thaler, and Kent Kaplan, Steven N., and Luigi Zingales, “Do Womack, “Price Reactions to Dividend Investment-Cash Flow Sensitivities Provide Initiations and Omissions,” Journal of Useful Measures of Financing Constraints?” Finance, XXXVIII (1995), 1597–1606. Quarterly Journal of Economics, CXII Mitchell, Mark L., and Erik Stafford, (1997), 169–216. “Managerial Decisions and Long-Term Stock Karpoff, Jonathan M., and Paul H. Malatesta, Price Performance,” Journal of Business, “The Wealth Effects of Second-Generation LXXIII (2000), 287–330. State Takeover Legislation,” Journal of Morck, Randall, Andrei Shleifer, and Robert Financial Economics, XXV (1989), 291–322. Vishny, “Management Ownership and Market Lakonishok, Josef, Andrei Shleifer, and Robert Valuation: An Empirical Analysis,” Journal Vishny,“Contrarian Investment, Extrapolation, of Financial Economics, XX (1988), 293–315. and Risk,” Journal of Finance, XLIX (1994), Morck, Randall, Andrei Shleifer, and Robert 1541–1578. Vishny,“Do Managerial Objectives Drive Bad 556 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Acquisitions?” Journal of Finance,XLV Manager-Specific Investments,” Journal (1990), 31–48. of Financial Economics, XXV (1989), Morck, Randall, and Fan Yang,“The Mysterious 123–140. Growing Value of the S&P 500 Shleifer, Andrei, and Robert Vishny, “A Survey Membership,” Working Paper, University of of Corporate Governance,” Journal of Alberta, 2001. Finance, LII (1997), 737–783. Pinnell, Maria Carmen S., State Takeover Laws Stein, Jeremy C., “Takeover Threats and (Washington, DC: Investor Responsibility Managerial Myopia,” Journal of Political Research Center Inc., 2000). Economy, XCVI (1988), 61–80. Pugh, William M., Daniel E. Page, and ——, “Efficient Markets, Inefficient Firms: A John S. Jahera Jr., “Antitakeover Charter Model of Myopic Firm Behavior,” Quarterly Amendments: Effects on Corporate Journal of Economics, CIV (1989), Decisions,” Journal of Financial Research, 655–669. XV (1992), 57–67. ——, “Agency, Information, and Corporate Rau, P. Raghavendra, and Theo Vermaelen, Investment,” Handbook of the Economics of “Glamour, Value and the Post-Acquisition Finance, George Constantinides, Milton Performance of Acquiring Firms,” Journal of Harris, and René Stulz, eds. (Amsterdam: Financial Economics, XLIX (1998), Elsevier Science, 2001). 223–253. Subramanian, Guhan, “The Influence of Rosenbaum, Virginia, Corporate Takeover Antitakeover Statutes on Reincorporation Defenses (Washington, DC: Investor Choice: Evidence on the ‘Race’ Debate and Responsibility Research Center Inc., 1990, Antitakeover Overreaching,” Working Paper, 1993, 1995, 1998). Harvard Business School, 2001. Seyhun, H. Nejat, Investment Intelligence from Titman, Sheridan, K. C. John Wei, and Feixue Insider Trading (Cambridge, MA: MIT Press, Xie, “Capital Investments and Stock 1998). Returns,” Working Paper, University of Texas Shin, Hyun-Han, and René M. Stulz, “Firm at Austin, 2001. Value, Risk, and Growth Opportunities,” Williamson, Oliver, The Economics of NBER Working Paper No. 7808, 2000. Discretionary Behavior: Managerial Shleifer,Andrei, and Robert Vishny,“Greenmail, Objectives in a Theory of the Firm White Knights, and Shareholders’ Interest,” (Englewood Cliffs, NJ: Prentice Hall, 1964). Rand Journal of Economics, XVII (1986), Yermack, David, “Higher Market Valuation for 293–309. Firms with a Small Board of Directors,” Shleifer, Andrei, and Robert Vishny, Journal of Financial Economics, XL (1996), “Management Entrenchment: The Case of 185–211. Chapter 21

Gertjan Schut and Ruud van Frederikslust SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES*

Source: Multinational Finance Journal, 8(3–4) (2004): 211–225.

ABSTRACT We investigate the shareholder wealth effects of 233 joint venture announcements of Dutch public companies in the period 1987 till 1998.The research shows that, on average, establishing joint ventures has a positive effect on the market value of Dutch companies. Using the strategic characteristics of joint ventures it is possible to explain and understand these wealth effects. Our research shows that the factors of strategic intention, the context in which the strategy is unfolded and the extent to which the company has control over the implementation strongly explains the extent to which a joint venture can create value.

I. INTRODUCTION joint venture strategy on the market value of companies. Section IV presents the WE INVESTIGATE SHAREHOLDER WEALTH summary and conclusions. EFFECTS of 233 joint venture announcements by Dutch public companies in the period 1987 till 1998.The study focuses on II. DATA AND METHODOLOGY joint ventures in which the consequences for the shareholders of the parent compa- A. Data nies are central. Setting up a joint venture involves establishing a separate legal entity, This research is based on the event study with its own identity, liability and share methodology developed by Fama et al. capital. Most companies have experienced (1969).The initial announcement of a joint stagnation in their market value growth venture is defined as the ‘event’, while the and cash flow margins up to three years market value is studied by examining before the establishment of a joint venture; the development of the share price. The e.g., Mohanram and Nanda (1998) and announcements were found in the Dutch Bergman and Friedman (1977). This arti- financial daily Het Financieele Dagblad. cle addresses the joint venture strategy The study analyzed a sample of 233 factors and how they have an impact on non-financial joint ventures whose announce- the market value of parent companies. ments met the following criteria: Section II outlines the sample survey and the applied research methodology. Section III The shares of at least one of the joint presents the findings of the impact of the venture partners were being traded on 558 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY

Construction Technology Food Marketing Efficiency Publishers

Chemicals

Industry Steel/Rubber

Electronics

Transportation

Services

0501020 30 40 60 70 80 Number of joint ventures

Figure 21.1 Motive for the Joint Ventures by Sector.

Table 21.1 Location of the Joint Venture and Nationality of the Partners Nationality Japan & of Partner Netherlands Europe Nafta S. Korea China Other Total

Dutch 25 1 0 0 1 3 30 European 10 27 0 0 2 5 44 Nafta 11 9 11 0 0 2 33 Japanese & S. Korean 3 1 1 13 0 2 20 Chinese 0 0 0 0 21 0 21 Other 0 1 1 0 1 32 35 Total 49 39 13 13 25 44 183

Note: When location could not be clearly identified, the joint venture was not included.

the Amsterdam Stock Exchange at the public companies in the Netherlands time of the announcement. accounted for the 233 joint ventures. During the ( 20, 20) days in which the Figure 21.1 shows the motives underlying market reaction was measured, no other the joint ventures by sector. Chemical and relevant announcements regarding the petrochemical companies are strongly companies participating in the alliance represented in the sample; companies in appeared in Het Financieele Dagblad. this sector of industry formed nearly one- third of the total number of joint ventures. The companies involved were not financial The most common motive was market institutions1.Forty-eight Dutch compa- development, followed by technology and nies, that is, approximately 21% of the efficiency. SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 559 Table 21.1 presents the distribution by process of firms; e.g.; Copeland (1995).To location of joint ventures and nationality explain wealth effects of joint ventures, the of the partners. Approximately a quarter of dimensions that characterize the strategy the joint ventures are based in the should be able to explain the observed Netherlands. Most of the joint ventures variance in CAR. are established in the country of one of the In the decision-making process involved partners. This is due to the fact that the in formulating a joint venture, three strat- Netherlands is a small country with a long egy dimensions are distinguished: Strategy international trading history. Content, Strategy Context and Strategy Control. The strategy content of a joint B. Methodology venture concerns the characteristics of the actual strategy itself. Two characteristics The analysis of the value effect of joint of the strategy content are taken into ventures includes only the abnormal part of account: the functional motive underlying a the market reaction in respect to the parent joint venture and the degree of diversifica- company share price.2 The expected returns tion realized by the joint venture. Research on the relevant days were estimated by into joint venture motivations distinguishes means of the market model approach. The motives stemming from market and tech- abnormal return (ARi,t) of a share i at time nology developments; e.g., Koh and t is calculated as follows:3 Venkatraman (1991) and Das, Sen and Sengupta (1998). These studies show that ARi,t Ri,t (ai biRm,t), (1) alliances primarily motivated by technological issues have a more positive impact where Ri,t is the effective return of share i on the market value of companies than at time t, ai a constant term of share i, bi is marketing-oriented joint ventures. These the systematic risk of share i, Rm,t is the studies did not examine efficiency improve- market return at time t and ai biRm,t is ment as a motive. The influence of the the expected return on share i according to degree of diversification of the joint venture the estimated market model. is not clear. Depending on the theoretical To measure the full effect of the concepts used, diversification influences the announcement, the abnormal return (ARi,t) value effect of the alliance either positively of the announcement day and the following (Balakrishnan and Koza [1993]) or day were computed and averaged for all negatively (Koh and Venkatraman [1991]). sample companies to obtain the cumulative To explain the value effects, the authors abnormal return (CAR). The standard respectively applied the transaction cost deviation of the abnormal return of each approach and the strategic behavior share in the sample was estimated through perspective. observations made during the estimation The strategy context defines the decision- period (from 200 to 51 days before the making parameters of the environment. announcement). A t-test was used to deter- Important elements are partner selection, mine whether the market reaction on and the nationalities involved, and the associated around the day of the announcement signif- cultural differences. Mohanram and Nanda icantly deviates from zero (Brown and (1998) and Koh and Venkatraman (1991) Warner [1985]). In addition to CAR, the found that companies gain more excess standardized cumulative abnormal return, returns if they enter into an alliance with a or SCAR, was included in the analysis.4 larger partner. The relative size of the To gain further insight into why some partner appears to be a relevant factor in reactions are negative while others are establishing the value potential of a joint positive, we accomplish a regression analy- venture. Closely examined in the relation- sis of the CARs using several different ship between partners is the effect of the variables. Joint ventures and similar capital relatedness5 of the partners, i.e., the degree investments come forth out of the strategy of similarity between their activities. 560 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Research conducted by Koh and company can exert on the joint venture.The Venkatraman (1991) and Balakrishnan distribution of control within joint ventures and Koza (1993) produced conflicting is discussed in depth in qualitative studies; conclusions and, as a consequence, have e.g., Bleeke and Ernst (1993). Koh and been a source of debate. Studies which Venkatraman (1991) found that one party investigated the nationality of chosen having more control within an alliance had partners and its impact on joint venture a positive effect. In contrast, Bleeke and wealth effects yielded few results so far.The Ernst (1993) argue that an equal balance studies focused on American companies of control makes it easier to manage a joint (Lee and Wyatt [1990] and Borde, Whyte, venture and consequently increases the Wiant and Hoffman [1998]) and produced probability of success. None of the studies contradictory lists of countries and economic in Table 21.2 developed an integral model regions where joint ventures were either to determine the conditions under which a successful or unsuccessful in generating joint venture creates value. shareholder value. Following Datta and Puia (1995) and Kogut and Singh (1988), for the purposes III. RESULTS of this study cultural difference, measured by the variable individualism, was selected A. Results of the event study as the distinguishing indicator of nationality, since it was also used in qualitative Average CAR is positive but there is studies by Bleeke and Ernst (1993) and substantial variance, with 57% of joint others. As a rule, these studies show that ventures associated with positive share price cultural differences can lead to manage- impacts, and 43% negative. Table 21.3 ment problems. We therefore expected this provides the results obtained by testing the dimension to have a negative effect. CAR and SCAR of the 233 joint venture Finally, Strategic control is the extent to announcements. Over the two-day testing which an enterprise can exert influence period positive market reactions were on the development of the joint venture. found with a significance level of 0.01 for The ownership structure is a clear manifes- both performance criteria (CAR and tation of the degree of influence that a SCAR). The CAR is equal to 0.40%.

Table 21.2 Overview of the Results Found in the Literature Dimension Variable Expectation Authors

Content Technology Das, Sen and Sengupta (1998) Chan, Marketing Kensinger, Keown and Martin (1997) Efficiency Diversification Koh and Venkatraman (1991) Mohanram and Nanda (1998) Context Partner relatedness Koh and Venkatraman (1991) Balakrishnan and Koza (1993) Individualism Datta and Puia (1995) Bleeke and Ernst (1993) Relative size Mohanram and Nanda (1998) Koh and Venkatraman (1991) McConnel and Nantell (1985) Control Majority Bleeke and Ernst (1993) Copeland et al. (1995) Equality Minority SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 561 Table 21.3 Test Results of CAR and SCAR Period t(i,j) CAR (%) SCAR

(1,0) 0.40*** 3.50*** (1,1) 0.39*** 3.15*** (2,2) 0.39** 2.15**

Note: The table shows the CAR and SCAR for various event windows for a sample of 233 joint venture announcements by Dutch public companies in the period 1987 to 1998. ***p 0.01 and **0.01 p 0.05.

According to CAR, the reactions to joint analysis, the CAR is the dependent variable ventures were both strongly positive for the following variables: (1.77% on average) and negative Technology ( 1.34% on average). Carj b0 b1 To gain further insight into why some Dummy j reactions are negative while others are Efficiency positive, we accomplish a regression b2 analysis of the CARs using a strategic Dummy j explanation model. b3(Diversification)j

b4(Individualism)j

B. Impact of the joint venture b5(RelSize)j strategy on CAR Majority b6 None of the American studies attempted to Dummy j develop an integrated model that reveals the dynamics among the strategic factors Minority b7 and the value effects of the joint ventures. Dummy j The distribution of the factors according to strategic dimensions is addressed above in Risk-free b8 section II. Investor’s interpretation of Rate j the strategy content, the context in which this strategy must function, and the extent Industry of the company’s control over the imple- b9 ej (2) Dummy j mentation of the strategic option determine how much shareholder value is generated. where e N(0,2). An explanation of The variables investment climate (risk-free j 6 the independent variables of the model rate ) and industry were included in the follows below: analysis as control variables to ensure that they did not have any impact on the research results. Strategy content Three dichotomous variables are used to test the impact on the parent company Joint venture strategy of the underlying motive of the joint venture. This section describes the variables used The value of the dichotomous variable is 1 to identify factors that influence the share- if there is a distinguishable motive, holder wealth effects. Each strategic dimen- otherwise it is 0. The dummy variables are sion is discussed in terms of what factors technology, in the case of a technology were examined and how the variables were development joint venture; marketing, for examined and calculated. In the regression a market development joint venture; and 562 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY efficiency, for an efficiency-driven joint and thus economic growth. In their long venture. term study they find empirical support for The diversification variable represents their thesis. From the proxies used to char- the relatedness between the activities of the acterize cultural distance, it became clear parent company and those of the joint that the cultural dimension Individualism had venture. The relatedness is the difference the most explanatory power in explaining between the primary three-digit US-SIC economic growth.The dimension of individ- number for the parent company7 and the ualism is a criterion for the way in which an joint venture.8 The relatedness is divided by individual views his or her relationship with 899, the maximum possible distance used the rest of the collective10. Frank, Hofstede by Balakrishnan and Koza (1993) in their and Bond (1991) constructed scores on research. this dimension based on most major countries. Datta and Puia (1995), Kogut and Singh (1988), Erramilli (1991) and Strategy context Shane (1992) used these bipolar scores The relative size of the partner is calculated to investigate cultural distance. This study by dividing the number of staff the partner used these scores in calculating cultural employs by the number of staff working for distance measured by the individualism the company whose shares are analyzed.9 dimension. The variable Individualism is In discussing the success and failure of calculated by: cooperative agreements, cultural differences 2 are often used as a reason for failure (Ij INL) Individualism , (3) (Bleeke and Ernst; 1993). Franke, Hofstede V and Bond (1991) argue that differences in cultural values, rather than in material and where Ij is an index-score on the individu- structural conditions, are ultimate determi- alism index of the country of origin of nants of human organization and behavior, the partner, INL is the index-score of the

Relatedness Diversification Motivation Nationality Ownership Partner Structure

Non-Horizontal Joint Venture Efficiency Netherlands Equal Alliance activities are Improvement Ownership unrelated

Europe

Technology driven Minority Nafta-countries interest

China

Horizontal Joint Venture Marketing Other Majority Alliance activities are driven countries interest related

Figure 21.2 Autonomous relationships. SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 563 Table 21.4 Estimated Functions and Test Results

SCAR CAR

Strategy Variable Expected Result Result Result

Constant 1.357*** 2.492*** Content Technology 0.832*** 1.584*** Efficiency 0.843** 1.570*** Diversification 2.877*** 5.180*** Context Inividualism 0.287** 0.433** Rel. Size 0.057*** 0.101*** Control Majority 0.975** 2.134*** ** ** Control variable Rf 0.244 0.478 Publishers 2.685* 5.245*

R2 0.492 0.492 Adj.R2 0.399 0.399 F-value 5.530*** 5.303*** N 110 110

Note: This table shows the results of the estimated functions of the cumulative abnormal returns of a sample of 110 joint venture announcements of Dutch public companies in the period 1987 till 1998. ***p 0.01, **0.01 p 0.05 and *0.05 p 0.10. individualism index of the Netherlands and return is equal to the yield of ten-year V is the variance of individualism scores of government bonds. all countries. The study also takes industry effects into account. Based on SIC number,11 the companies in the sample are classified into the Strategy control following sectors of industry: construction, food, publishing, (petro) chemical, steel and The ownership structure within the joint rubber, electronics, logistics, and services. venture is represented by three dummy Industry effects are measured using variables.Three structures are possible: the dummy variables which take the value of company can have a majority, a minority or one if the firm is in that particular industry an equal distribution of shares in the joint otherwise zero. venture. It is assumed that the ownership structure correlates with the degree of con- Autonomous relationships trol in the joint venture. The dummy variables take the value of one if the firm has Figure 21.2 shows the autonomous that particular ownership structure and the relationships uncovered by the study. value of zero, otherwise. Technology joint ventures are mostly initiated between non-horizontal partners in NAFTA territory. Efficiency driven joint Control variables ventures are often between horizontal partners within the E.U. region. Marketing In order to prevent investment climate driven joint ventures are frequently setup in and industry specific factors to influence emerging markets where for China the the overall results of this study we control Dutch firms mostly have a majority stake. for them by using control variables in In China it is still obligatory to form joint the regression. To measure the effect of ventures; the Chinese government prevents investment climate the variable risk-free foreign firms to have full control of Chinese rate (Rf) is added to the model. This subsidiaries. 564 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Estimated functions and ventures can provide rays of hope. In this testing results study, the intensity of these rays is expressed through characteristics of the A regression analysis was conducted to strategy dimensions: Strategy Content, measure the explanatory power of the syn- Strategy Context and Strategy Control. thesis model (2). There appeared to be little difference between the results of the various performance criteria. The results A. Strategy content presented here are based on the CAR and SCAR performance criteria. Only the sig- The substantive function that a joint ven- nificant effects are shown in table 4. The ture fulfils for a company is an important results reveal that all the given strategic component in the evaluation of a joint ven- factors help to explain the performance of ture. As shown by the results of Koh and joint ventures.The model that incorporates Venkatraman (1991) for joint ventures and the strategic dimensions provides a Das, Sen and Sengupta (1998) for stronger, more substantial explanation for alliances without share participation, tech- the variation in the cumulative abnormal nology development joint ventures have a returns than the control variables alone. higher impact on a company’s market value The contribution of the control variables than market development joint ventures. on R2 of the estimated functions is 5% on The average value created by efficiency- average. The variables from the synthesis driven joint ventures lies somewhere in model appear to be relevant to the expla- between. nation of the shareholder wealth effects of The less the activities of the joint venture joint ventures. The estimated regressions are related with those of the parent com- are checked for multicollinearity using pany, the less enthusiastic the investor VIF-statistics. The few correlations reacts. In order to create shareholder value between the explanatory variables are not it is essential for a company to utilize its significant, as the value of VIF-statistics of core competencies. With an increasing dis- approximately 1 is low. tance between the new activities comes increasing difficulty in effectively deploying the company’s competencies. Moreover, IV. SUMMARY AND CONCLUSIONS there is less potential for the joint venture to affect the core activities and, since most of the corporate value stems from the core On average the joint ventures that Dutch activities, a distant joint venture is less companies entered in the period 1987–98 likely to be able to generate significant generated value. Notably, companies tended impact. It seems that there is an optimal to enter joint ventures when their perform- degree of diversification for joint venture ance was less than adequate. The statistics contracts. Investors are usually not in favor show that on average the company per- of companies using joint ventures for rea- formance was below the average stock sons of strong diversification. market performance in the years before the joint venture announcement. This means that we now have an explanation for the B. Strategy context negative relationship between return and joint venture activities observed by Berg The American studies show the various and Friedman (1977). Our results demon- impacts of partner nationality. In our study strate that companies were already per- we found that joint venture partners that forming poorly prior to deciding to enter a scored high on individualism have a positive joint venture and that joint venture deci- impact on value creation. As no other sions were a reaction to the poor results economic variables appear to make any rather than their cause. In bad times, when contribution, it is plausible that investor shareholder value is deteriorating, joint reactions are partly based on the perception SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 565 of the economic power of the partner’s venture is especially positive for companies country of origin and business location. that have a majority interest. A majority Research shows that the relatedness of interest gives a company a dominant posi- the partner’s activities is of less impor- tion in the collaboration, ensuring that the tance in determining the value generated by company has more control over the a joint venture. Investors focus on the achievement of the objectives of the joint activities of the joint venture. However, the venture. The results contradict the prevail- analysis reveals that horizontal partners are ing paradigms that assert the importance usually selected for horizontal joint of equality in cooperative alliances. Bleeke ventures and vice versa. Neither conven- and Ernst (1993), Harrigan (1988) and tional testing of differences in means nor Copeland et al. (1995) underlined the regression analysis uncovered any impact importance of equality in the ownership on value creation by joint ventures. The structure. If shareholder value is the tar- overlap in the partners’ activities is the get, then equality within an alliance is not result, rather than the cause, of the joint essential. It is more important that the venture and its objective. The debate con- ratios should be fair and bear a connection ducted in the literature on the impact of to the value of the resources (knowledge, overlapping strategic activities appears to capital and other assets) in which each be irrelevant in our research. party invests in the joint venture. It appears Partner size has a strong effect on the that the equal distribution of share capital performance of a joint venture.The variable at any cost is not appreciated by investors is highly significant and positive for all per- and has a negative impact on market value. formance criteria. In their qualitative Furthermore, investors have more confi- study, Bleeke and Ernst (1993) concluded dence in joint ventures with one captain that a stronger partner is a prerequisite for than those with two. a successful joint venture. The regression The synthesis of the aggregate strategic results and the univariate tests support this dimensions explains a great deal of the theory. The results of this study show that share price reaction. Clearly, investors venturing with relative large partners respond consistently enough to joint ven- result in higher wealth effects. For smaller ture announcements to justify the develop- firms the joint venture can underline the ment of an analytical model based on these value of its business proposition through factors. the acceptance of larger partners. The initial impact of venturing with smaller partners doesn’t seem to be very high. When NOTES there is much excitement surrounding a joint venture with a smaller partner, this * We acknowledge comments on prior drafts can be attributed to the promise of tech- by the anonymous referee and the editor Peter nology/innovation. Theodossiou. 1 Financial institutions were not included in the sample because they are not comparable to C. Strategy control other sectors in that they are regulated by the Dutch Central Bank. Moreover their annual Joint venture announcements with an accounts differ strongly from those of other unequal partner shares gain higher wealth companies. effects then with an equal (50/50) partner 2 In order to verify the stability of the esti- share. From the data observed, it seems mated market model, Cannella and Hambrick (1993), three alternative performance criteria that joint ventures with an equal share dis- have been calculated. They are based on the tribution receive a negative premium. following: (i) Returns corrected for the market. Minority partner shares receive on average This method comprises the effective return positive wealth effects, but report less minus the market return, i.e. without estimating strong significance in the multivariate the parameters alpha and beta as in performance regression tests. The valuation of a joint criteria (1). (ii) Returns corrected for the 566 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY average. The return is calculated by subtracting Borde, S.F.; Whyte, A.M.; Wiant, K.J; and the average return of the share in the estimation Hoffman L.L. 1998. New evidence on factors period from the effective return. (iii) Effective that influence the wealth of international joint return. The uncorrected return during the ventures. Journal of Multinational Financial announcement. In addition a verification Management 8: 63–77. analysis was also performed based on the Brown, S.J. and Warner, J.B. 1985. Using daily ‘buy and hold’ investment strategy. The extraordinary buy and hold return is also based on stock returns: The case of event studies. the market model; however a composite return Journal of Financial Economics 14: 3–31. is the starting point.The results of the alternative Cannella Jr, A.A. and Hambrick, D.C. 1993. methodologies are similar to those reported in Effects of executive departures on the per- the paper. formance of acquired firms. Strategic 3 The parameters a and b were estimated Management Journal 14: 137–152. using the least square method, based on obser- Chan, S.H.; Kensinger, J.W.; Keown, A.J; and vations made during the estimation period. The Martin J.D. 1997. Do strategic alliances cre- estimation period runs from 200 to 51 days ate value? Journal of Financial Economics before the announcement day. 46: 199–221. 4 Unlike CAR, SCAR is normally distributed for each hypothesis (Strong [1992]). Copeland, T.; Koller, T.; and Murrin J. 1995. 5 To avoid confusion the term relatedness is Valuation: Measuring and managing the used to characterize similarities between value of companies. John Wiley & Sons, Inc: parents and the term diversification is reserved New York. to characterize similarities between the activi- Das, S.; Sen, P. K.; and Sengupta, S. 1998. ties of the joint venture and the parent that is Impact of strategic alliances on firm valua- under investigation. tion. Academy of Management Journal 41: 6 In accordance to economic theory, when 27–41. the risk free rate lowers, so does the cost of Datta, D.K. and Puia G. 1995. Cross-border capital. With the cost of capital lower it is more acquisitions: an examination of the influence attractive for companies to invest. So the risk free rate is a proxy of the investment climate. of relatedness and cultural fit on shareholder This control variable helps in identifying over-or value creation in U.S. acquiring firms. undervaluation of the venture due to the Management International Review 35: economic situation at that particular time. 337–359. 7 Source: Worldscope. Erramilli, M.K. 1991. The experience factor in 8 Source: KPMG Dealwatch. foreign market entry behavior of service 9 Das el al. (1998) also used this proxy for firms. Journal of International Business the partners in the joint venture. Studies 22: 479–501. 10 Also used in Hofstede (1980). Fama, E.F; Fisher,L.; Jensen, M.C.; and Roll, R. 11 Standard industry classifications. 1969. The adjustment of stock prices to new information. International Economic Review 10: 1–21. REFERENCES Franke, R.H; Hofstede, G.; and Bond, H. 1991. Cultural roots of economic performance: A Balakrishnan, S. and Koza, M.P. 1993. research note. Strategic Management Information asymmetry, adverse selection Journal 12: 165–173. and joint-ventures. Journal of Economic Harrigan, K.R. 1988. Joint ventures and com- Behavior and Organization 20: 99–117. petitive strategy. Strategic Management Berg, S. and Friedman P. 1977. Joint ventures, Journal 9: 141–158. competition and technological complemen- Kogut, B. and Singh, H. 1988. The effect of taries. Southern Economic Journal 43: national culture on the choice of entry mode. 1330–1337. Journal of International Business Studies 19: Bleeke, J. and Ernst D. 1993. Collaborating 411–432. to Compete. John Wiley & Sons, Inc: Koh, J. and Venkatraman, N. 1991. Joint New York. venture formations and stock market SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 567 reactions: An assessment in the information Mohanram, P. and Nanda, A. 1998. When do technology sector. Academy of Management joint ventures create value? Working paper. Journal 34: 869–892. Harvard Business Studies. Lee, I. and Wyatt S.B. 1990. The effects Shane, S.A. 1992.The effects of cultural differ- of international joint ventures on share- ences in perceptions of transaction costs on holder wealth. The Financial Review 25: national differences in the presence for 641–649. licencing. Management International Review McConnell, J.J and Nantell, T.J. 1985. 32: 295–311. Corporate combinations and common stock Strong, N. 1992. Modeling abnormal returns: A returns: The case of joint ventures. The review article. Journal of Business Finance Journal of Finance 40: 519–536. and Accounting 19: 533–553. Chapter 22

Jim Lai and Sudi Sudarsanam CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE: IMPACT OF OWNERSHIP,GOVERNANCE AND LENDERS*

Source: European Finance Review, 1(2) (1997): 197–233.

ABSTRACT

Firms in performance decline may choose a variety of restructuring strategies for recovery with conflicting welfare implications for different stakeholders such as shareholders, lenders and managers. Choice of recovery strategies is therefore determined by the complex interplay of ownership structure, corporate governance and lender monitoring of such firms. For a sample of 297 U.K. firms experiencing relative stock return decline during 1987–93, we examine the impact of these factors as well as other control factors on their turnaround strategies. Strategy choices during the decline year and two post-decline years are modelled with logit regressions. Our results show that turnaround strategy choices are significantly influenced by both agency and control variables. While there is agreement among stakeholders on certain strategies there is also evidence of conflict of interests among them.There is further evidence of shifting coalitions of stakeholders for or against certain strategies.

1. INTRODUCTION improve its efficiency and profitability, asset sales to raise cash to meet its financial FIRMS WHICH HAVE EXPERIENCED SUBSTANTIAL commitments to, say, lenders, renegotiate decline in their financial performance adopt its debt to relieve the immediate burden of a variety of strategies to reverse that decline. financial commitments, issue new equity These strategies range from operational to finance its operations or reconfigure its actions to rationalise production and reduce business strategy by making strategic costs, changes to financial and management disposals of businesses or investing in new structure to asset sales. Some of these businesses. A precondition to firm revival strategies are of a fire fighting nature with may often be the removal of existing man- a short term cash flow increasing focus agement. These strategies may be grouped whereas others are of a longer term broadly into an operational, asset, financial strategic nature with no immediate cash and managerial restructuring. flow implications. Operational, asset, financial and mana- A firm faced with performance decline gerial restructuring may be motivated by may choose operational restructuring to the need to get the firm back on its feet. CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 569 However, any restructuring strategy has choices is examined for a variety of different implications for different stake- strategies. holders – shareholders, lenders, managers Our results show that turnaround and employees often leading to conflict of strategy choices are significantly influenced interests among them. While one stake- by both agency variables and control holder group may find its control diluted or variables. While there is agreement among investment eroded, another group may have stakeholders on certain strategies there its stake strengthened. is also evidence of conflict of interests An understanding of the nature of the between lenders and managers and between restructuring process – the range of managers and some block shareholders. restructuring choices available, the Lenders’ preference for cash generative determinants of these choices and the action is in direct conflict with shareholders’ relative effectiveness of different types incentive to avoid such action. Entrenched of restructuring – enables firms suffering managers are less likely to restructure performance decline to design feasible and the firm and replace themselves. Non- effective restructuring programmes to institutional rather than institutional achieve turnaround.The extant literature in shareholders appear to be active monitors. finance and strategy provides a fragmented Independent non-executive directors, perspective on the issues raised by firm however,seem to be effective in their overperformance decline and turnaround. In sight of managers as they instigate more this study we attempt a more comprehen- turnaround strategies. sive and integrated analysis of corporate The paper is organised as follows. restructuring in response to performance Section 2 discusses the theoretical decline. framework for the choice of turnaround We examine the restructuring strategy strategies by declining firms. The method- choices made by firms which have experi- ology and data for our empirical analysis enced relative performance decline in terms are described in Section 3. The results are of stock market returns to equity holders. presented and interpreted in Section 4. These choices are made by managers Section 5 provides a summary and the who are subject to a variety of pressures conclusions. from different stakeholders and also to conflicts of interest. The choices made by declining firms are analysed within the 2. THEORETICAL FRAMEWORK agency model of the relationship among shareholders, lenders and managers. The The choice of turnaround strategies is central thesis of this paper is that these contingent upon a number of factors. Since different stakeholders have preferences different strategies may have different, and for different restructuring strategies. often conflicting, welfare implications for We hypothesise that choice of recovery managers, shareholders and lenders, the strategies is determined by the complex choice of any strategy can only be made interplay of the ownership structure, cor- as a trade off among these contending porate governance and lender monitoring stakeholders. The restraints on any single of the firms in decline. stakeholder group such as managers For a sample of 297 U.K. firms whose maximising their own self-interest to the stock returns declined from being in the top detriment of other stakeholders is a func- 50% of all U.K. listed firms to the bottom tion of the governance structure and the 20% we assemble empirical evidence on mechanics of agency monitoring in a firm the extent and variety of restructuring (Gilson, 1990). Thus, an understanding of undertaken by declining firms. The impact the nature and sources of these restraints of ownership, governance and lender is necessary to make the appropriate monitoring on the restructuring strategic turnaround strategy choices. 570 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 2.1. Impact of agency conflicts such debt restructuring. On the other hand, on turnaround strategy choice turnaround based on fresh infusion of equity is likely to be preferred by lenders The restraints on managerial choice of but frowned upon by shareholders. Dividend turnaround strategies may be examined cuts may be loathed by shareholders but within the context of the agency conflicts supported by lenders. among shareholders, managers and Strategic or asset restructuring in the lenders. The motivations of these players form of divestments may be favoured by also provide the impetus to the pursuit of shareholders provided the divestment turnaround strategies so that firm value proceeds are not used to pay down debt. is enhanced and its ability to meet its Lenders may support divestment provided financial commitments is restored. their debt is paid off. New investments of While both lenders and shareholders a high risk nature may be preferred by have a common interest in restoring firm shareholders but not necessarily by lenders viability and its ability to generate (Myers, 1977). adequate returns to their investment in the firm, in the turnaround process either group 2.2. Impact of lender monitoring may gain at the expense of the other. Shareholders may benefit from a transfer on managerial choice of wealth from creditors when managers In the agency model of the firm posited by undertake risky investments (Myers, 1977). Jensen (1989a) a highly leveraged firm Likewise, lenders may benefit from a will react faster to poor performance than wealth transfer from shareholders when less leveraged ones due to a desire to avoid sale of assets to pay off debts eliminates the breaching debt covenants. This early option value of those assets (Lang et al., response preserves the going-concern value 1995). Similarly, managers may pursue of highly leveraged firms as compared to turnaround strategies which least harm less-leveraged firms. them while the burden of turnaround is Ofek (1993) examines the role of lender borne by shareholders or lenders or both. monitoring within the agency paradigm in Managers’ pursuit of self-serving objectives influencing the choice of restucturing may manifest itself in their choice of strategies of poorly performing companies strategies. However, managerial discretion in the U.S.A. He finds that high gearing in choice of strategy may be tempered significantly increases the probability of by the agency control mechanism in place financial and operational restructuring. in the firm. Gilson, John and Lang (1990) find no Broadly, the turnaround options available relation between gearing and financial to declining firms include: operational, restructuring.1 business strategic, managerial and financial High leverage is also found by Storey et restructuring. Not all of these actions will al. (1987) to be more positively associated appeal equally to shareholders, managers with equity rights issues (i.e. financial and lenders since they demand different restructuring) in failing firms than in degrees of sacrifice from these stakeholders non-failing firms. They attribute this to the during the turnaround process. monitoring pressure from bank creditors Managerial restructuring, e.g. replacement who are only willing to continue financial of the top managers, is obviously unlikely to support conditional upon shareholders be favoured by managers and where the sharing a part of the burden of turnaround. governance structure is weak and the man- Impact of debt on managerial choice of agement is entrenched such replacement turnaround strategies may depend on the may not happen. Similarly, where financial characteristics of debt such as ownership, restructuring involves additional borrowing maturity structure and security available. or dilution of the covenants protecting Debt ownership by informed bank creditors existing lenders they are likely to resist promotes more effective monitoring than CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 571 by other types of debt holders.This increased The efficiency of lender monitoring is effectiveness arises from the banks’ close almost beyond question. Leverage may relations with firms and their access to have a positive and significant relation with private information, a right established by the incidence of all four generic turnaround loan covenants or through ongoing bank strategies. The primary motivation is debt relationship. repayment. Lenders are expected to favour Banks’ reputational capital provides asset sales proceeds to be applied to debt them with the economic incentives to repayment rather than retained by the firm monitor firm actions. Hirschey et al. (Slatter,1984; Lang et al., 1995).They are (1990) find that the higher the proportion likely to favour cut or omission of dividends of bank debt in total debt the higher is to conserve cash or equity issues to the positive return on announcement of a increase liquidity (Storey et al., 1987). sell-off to the divestor shareholders. This Lenders may expect extensive asset sales, superior valuation is attributed to the more operational cost cutting and management effective and credible monitoring by banks changes as a prerequisite for debt restruc- with a large stake. turing. Debt restructuring may be the last James (1987) argues that banks provide resort after exhausting other forms of some special service not available from restructuring. other lenders. He finds evidence of a larger positive stock price reaction to new bank 2.3. Impact of ownership structure credit agreements than to announcement on managerial choice of private placements or public straight debt offerings. In Gilson’s (1989) study The share ownership in the declining firm bank lenders frequently initiate senior may provide an agency mechanism for management changes in financially controlling managerial discretion in the distressed firms. choice of turnaround strategy. Block share- Maturity structure of debt is likely to holders may provide effective oversight influence the borrower firm’s restructuring leading to value maximising behaviour on decisions since the greater the proportion the part of managers (Shleifer and Vishny, of short term debt the greater should be 1986). Where managers hold significant the level of monitoring. The credit renewal shares, their interests may be aligned to process associated with short-term debt those of shareholders in general. subjects firm managers to more frequent The role of block shareholders as agency monitoring than long-term debt and monitors has been studied by many increases the bargaining power of banks researchers. Large shareholders provide an over managerial decisions such as liqui- efficient mechanism for resolving the dation (Diamond, 1993; Rajan, 1992; agency conflict which arises in a firm Gertner and Scharftein, 1991) and the use owned by atomistic shareholders (Jensen of proceeds from asset sales (Brown et al., and Meckling, 1976). Demsetz and Lehn 1994). Empirically, Ofek (1993) finds that (1985) argue that as the size of large share- short-term leverage increases the probability holding increases, monitoring effectiveness of all the restructuring strategies examined also increases. Block shareholders may by him except financial restructuring. be institutional or non-institutional, and In addition to debt ownership and associated to incumbent management or maturity structure, the security for the debt independent of it. Agency monitoring may also impact on the restructuring effectiveness varies across these different decision. A high proportion of unsecured blockholder categories. debt is likely to be associated with more For the U.S., Hill and Snell (1989) find effective monitoring because of the unpro- a positive relation between large share- tected nature of this debt. Lack of security holding and firm productivity. McConnell may induce more intense monitoring by and Servaes (1990) find a significant unsecured lenders. relation between Tobin’s q and the level of 572 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY institutional block shareholding. Agrawal issue (e.g. Schipper and Smith, 1986) and and Mandelker (1990) provide evidence of asset sales where proceeds are utilised to a positive relation between institutional pay down debts (Lang et al., 1995). ownership and stockholder wealth effects of various types of anti-takeover amendments 2.4. Impact of corporate in target companies. Jarrel and Poulsen (1987) show that firms that adopt the governance structure most value-reducing forms of anti-takeover Corporate governance structure as a charter amendments also have lower insti- monitoring mechanism to reduce the agency tutional shareholding than do other firms. problem between shareholders and man- Brickley et al. (1988) find evidence that agers has recently received much attention institutional investors who do not have (Cadbury Report, 1992). Composition of business dealings with corporate manage- the board of directors is an important part ment are more likely to vote against of this structure, influencing its policing anti-takeover amendments. All these effectiveness. Board composition is there- U.S.-based results are consistent with fore likely to impact significantly on the the reduction of agency costs due to large choice of turnaround strategies. shareholders monitoring.2 Boards of directors differ in a number of The positive valuation impact of large ways: the relative importance of executive share acquisitions has been evidenced in a versus non-executive directors; the leader- number of studies. Barclay and Holderness ship of the board by an executive or (1991), Mikkelson and Ruback (1985), non-executive chairman and the separation Holderness and Sheehan (1985) and Choi of the roles of the chairman of the board (1991) report, for the U.S., that block and the CEO. In the context of declining acquisitions in excess of 5% generate firms, their performance decline may have significant wealth gains for target share- been caused by managerial entrenchment holders. For the U.K., Sudarsanam (1996) and weak governance structure may have reports similar results. contributed to this entrenchment. Turn- Shivdasani (1993) emphasises the need to around may, therefore, demand managerial differentiate associated from non-associated restructuring with the top management non-institutional block holders. Shareholders being replaced. Whether managerial associated with incumbent management, restructuring can be carried out depends e.g. family trusts or company pension funds upon the independence and strength of the are less likely to provide effective monitoring board as well as the power of block share- of managers. Shivdasani (1993) finds holders and lenders. evidence that unassociated shareholders According to Fama and Jensen (1983), increase the probability of hostile takeovers, the separation of decision management and invoking their disciplinary motives. decision control in the decision-making Bethel and Liebeskind (1993) report process can alleviate the agency problem. that block share ownership is associated Whilst inside directors are responsible for with corporate restructuring. Sudarsanam decision management, decision control (1995) finds substantial asset, financial should be left with outside directors. and managerial restructuring following Outside directors has an incentive to large block acquisitions and value monitor management actions since they have increases attendant upon such acquisitions staked their reputation as professional are maintained or enhanced over the corporate referees. Consequently, the higher following three years. the proportion of non-executive to executive In contrast, shareholders are also known directors, the more effective would be the to frown upon certain strategies which are board monitoring of management. painful to themselves such as dividend Empirically, Weisbach (1988) finds cut/omission (Asquith and Mullins, 1986; that CEO turnover is highly correlated with DeAngelo and DeAngelo, 1990), rights proportion of outside directors to inside CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 573 directors. The monitoring function of turnaround strategies of poorly performing outside directors is also supported by firms. We examine the individual as well Rosenstein and Wyatt (1990), who find as the combined effects of the three positive share price reactions to the mechanisms. appointment of outside directors. Further, Boeker and Goodstein (1993) report that strong insider presence significantly influ- 2.5. Impact of multiple agency ences CEO replacement decisions. However, control mechanisms Mallette and Fowler (1992) observe empirically that the proportion of outside In the choice of restructuring strategies, directors has no bearing on the adoption of the influences representing ownership, poison pills. Hermalin and Weisbach (1992) board composition and lenders may often and Shivdasani (1993) are unable to be reinforcing but at other times working document any systematic relation between at cross purposes. In other words, the outside directors, firm performance and monitoring role of owners, governance and the probability of hostile takeovers.3 lenders may be complementary, substitu- Where one person combines the roles of tory or contradictory. Lenders and outside board chairman and CEO his or her powers directors may complement each other, say are considerable. This duality of roles can in forcing management changes in declining promote focused objectives and a clear line firms. However, high leverage and high of command. On the other hand, duality lender influence on management change may entrench the top manager to the may also substitute for the lack of pressure detriment of shareholders and possibly from outside directors for the same action, lenders, reduce the oversight function where the proportion of outside directors of the board and weaken the governance in the board is low. An example of contra- structure. dictory influence arises when lenders press The role of board Chairman is to oversee for asset sales and rights issue to generate the performance of board members from cash for the purpose of paying down debt. an informed perspective. Therefore, if Lenders’ preference, in the latter case, the board is chaired by a non-executive clearly contradicts owners’ desire to avoid Chairman, the external and part-time injecting fresh equity funds and their nature of his position coupled with infor- preference for lenders to increase or at mation asymmetry problems can present least maintain financial support. him with difficulty in effectively controlling the board. Consequently, a non-executive Chairman strengthens CEO control and 2.6. Impact of dominant contributes to potential managerial stakeholders entrenchment. An executive Chairman, on the other hand, is likely to strengthen the Our discussion so far has ignored the board’s monitoring role. relative bargaining powers of the different Mallette and Fowler (1992) find support stakeholder groups in declining firms for the entrenchment hypothesis in their when strategy choices are made.The choice empirical study with duality increasing the of a strategy is likely to be decided by the probability that poison pills are adopted relative strength and dominance of these whereas separation diminishes the proba- stakeholders. We develop the concept of bility. However,Rechner and Dalton (1989) stakeholder dominance to take into account find no significant difference in firm per- the complex interactions between the formance between dual and non-dual firms. various stakeholders or agency monitors The primary focus of this study is the indicated above. Five types of stakeholder impact of three broad categories of agency dominance are examined – lender,manager- monitoring mechanisms – ownership, owner, blockholder, CEO and collective leverage and board composition – on the board dominance. 574 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 2.6.1. Lender dominance through operational restructuring or asset divestment and prefer size increasing Where a firm is highly leveraged and has 4 acquisitions or capital expenditure. suffered a severe decline, lenders are Likewise, dominant managers are likely deemed dominant in influencing the firm’s to disfavour the ‘final resort’ strategy – decision-making machinery. Recent work debt restructuring – which is adopted by Jensen (1989a,b) suggests that leverage only when all efforts to pay off (or buy is an important determinant of how decision out) creditors fail. In a debt restructuring rights are allocated among claimholders. exercise, lenders frequently insist on dealing Therefore, when a firm is severely distressed, with a credible management team leading to with equity shareholders occupying a very installation of a new management team. low position in the repayment queue, Gilson (1989, 1990), and Murphy and lenders have the ultimate say and influence Zimmerman (1993) find lenders to insti- on the firm’s restructuring choice. gate significant top management changes Lenders would generally prefer cash in distressed firms. generative strategies to facilitate debt repayment. They may frequently insist on removal of top managers and freeze in 2.6.3. Blockholder dominance investments as a condition for continuing Where neither lenders nor manager-owners financial support. Removal of top managers are dominant, and unassociated block- poses a serious conflict with managers’ holding is high, blockholders may dominate. interest, but if the firm’s finance is dire, Operational restructuring, which is the managers have little power to avoid least controversial of all strategies, is displacement even when they hold a high expected to be favoured by dominant block- equity share holding. Asset sales may holders. Extant literature (e.g. Schipper and pose a weak conflict of interest with block Smith, 1986; Asquith and Mullins, 1986) shareholders as they deem the sale of assets indicating shareholders’ dislike for equity- extinguishes the option value attached to based strategies such as dividends cut and assets sold. In the final analysis, lender omission and equity issue, would mean that dominance prevails over other stakeholders’ they are shunned by dominant blockholders. preferences as lenders’ continued support is As discussed above, shareholders also shun key to the survival of the firm. Management asset sales as they extinguish the option may therefore be forced to implement value attached to the assets sold. Following cash generative actions and refrain from from dominant shareholders’ dislike for cash consuming asset investment and cash generative actions (equity issues operational strategies. and asset sales), we can expect them to disfavour investments which necessitate 2.6.2. Manager-owner dominance such cash generative actions. Dominant blockholders are expected to If the firm’s decision-making process is not favour debt restructuring as lenders fre- dominated by lenders, and managerial and quently provide additional working capital, associated shareholdings are high, manager- forgive loans or interests or make other owners are deemed to be entrenched and form of concessions, though reluctantly, in possess dominant influence. In the circum- the hope of realising higher debt repayments stance, entrenched managers are expected, when the distressed firm is eventually in the least, to refrain from adopting turned around. Similarly, dominant block- managerial restructuring strategies. The holders who possess significant influence literature (e.g. Maris, 1964) suggests that over management are expected to initiate entrenched managers favour large size as top management replacement. power and compensation are related to size. In summary, blockholder dominance Consequently, dominant managers may is expected to be positively associated refrain from downsizing their operations with operational, managerial and debt CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 575 restructuring but negatively associated change in either simple or industry adjusted with all other strategies. accounting ratios such as return on assets and some others based on stock returns. Given the central importance of value 2.6.4. CEO and collective maximisation as an objective function board dominance within the agency model, we prefer to define performance decline in terms of stock When the firm is not lender, manager-owner market value changes i.e. stock returns. or blockholder dominated, corporate control Ofek (1993) defines performance decline is expected to lie with the board of directors. in terms of the change in the annual stock However, where the board is chaired by return ranking of a firm among all the a dual CEO, the dual CEO is expected firms in the market from being in the top to dominate the board and hence the 67% in one year (the base year) to the firm’s decision-making process. CEO board bottom 10% in the following year (the dominance is expected to favour strategies distress year).This decline may range from akin to manager-owner dominance firms, a maximum of 100% to a minimum of i.e. shun managerial restructuring, prefer 23%.This steep fall in value is regarded as investments, avoid operational restructuring sufficient to trigger various restructuring and cash generative actions. actions by the distressed firms. When the firm is not lender, manager- We employ a definition broadly similar owner, blockholder or CEO dominated, to Ofek’s but arguably more stringent. corporate control is expected to lie ‘collec- A firm is defined as having experienced tively’ with the board of directors. Since performance decline when it falls in stock the collective interests of all stakeholders return ranking of all firms on the London are in the avoidance of a crisis and recovery, Stock Exchange to the bottom 20% in a collective board dominance is expected to be year (the decline year) after having been in positively associated with all restructuring the top 50% in each of the two preceding strategies. years. In the decline year the maximum No study to date has examined the decline is 100% and the minimum is 30%. relationships between agency monitors’ With this definition, in contrast to Ofek’s, motivations and restructuring strategy the fall in rank has to be much steeper for choice in a comprehensive manner.Although inclusion in our sample. Further, the fall is Ofek (1993) finds that different agency from a stable high performance.This condi- variables are associated with different tion avoids sampling companies whose restructuring strategies, his approach lacks performance decline is due to short term a robust theoretical underpinning. More volatility of their share prices.5 importantly, Ofek does not examine the Firms satisfying the above criterion of relative dominance of stakeholders in performance decline are sampled and the shaping restructuring strategy choices.This restructuring actions they take subsequent to research attempts to fill the empirical decline are tracked. The agency monitoring gap by exploring the impact of agency determinants of choice of these actions by monitoring on specific strategy choice and the sample firms are identified using the the role of dominant stakeholders. logit regression methodology.

3. METHODOLOGY AND DATA 3.1. Definitions of dependent variables The first step in our empirical analysis is to The various restructuring actions declining devise an operational definition of perform- firms choose are the dependent variables for ance decline. In the turnaround literature in the logit regressions. These actions fall into corporate strategy and finance a range of the four generic strategies – operational, definitions has been used, some based on asset, managerial and financial. Ofek (1993) 576 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY distinguishes between actions resulting in decomposed into bank debt, short-term short term cash inflow and those with no debt and unsecured debt (as a proportion such cash inflow since cash generation to of debt and equity). Debt comprises all meet the firm’s financial commitments may interest-bearing liabilities. Equity refers to be necessary to alleviate financial distress shareholders funds i.e. equity and reserves. and avoid default on them. Accordingly, we Short-term debt is any debt that matures in define combinations of restructuring less than 12 months. strategies which generate cash and those Share ownership categories are proxied which do not. by directors’ shareholding and block hold- Operational restructuring covers cost ing in excess of 5% prior to 1990 and 3% reduction, improved financial control, and after 1990. Block holding is divided into closures and integration of production and shareholding by financial institutions and other facilities. Asset restructuring includes non-institutional holding. The latter is fur- both asset reduction and new investment. ther split into associated and unassociated Asset reductions comprise divestment, blocks. Associated blocks are held by fami- management buy-out, spin off, sale and lies or trusts associated with the directors leaseback, and other asset sales. Investment and company pension schemes. includes acquisitions and internal capital Board composition is proxied by three expenditures. Internal capital expenditure variables: proportion of outside or non- is measured by significant expenditure in executive directors on the board, whether plant and machinery, exceeding routine the board is chaired by a non-executive asset replacements. Significant expenditure director,CEO cum Chairman (CEO duality) is taken as that in excess of 10% of the where the two posts are held by the same pre-decline year total assets.6 person. Managerial restructuring covers replace- The empirical literature suggests that ment of Chairman or CEO or managing turnaround strategy choices are also dic- director. Financial restructuring refers to tated by non-agency monitoring factors. both equity- and debt-based strategies. These additional variables are included in Equity issues (excluding those issued in an our regressions as control variables. acquisition-related share exchange) and Severity of decline dictates both the pace dividend cuts and omission are part of an of restructuring and choice of particular equity-based restructuring. Debt restruc- actions. For example, asset investment or turing includes debt refinancing and acquisitions may be less likely in more seri- renegotiation of the terms of existing debt. ously distressed firms as they consume Cash generating strategies include asset scarce cash resources. sales and equity issues. Managerial and Economic and industry condition also operational changes are deemed non-cash may influence choice of strategy. For exam- generative in the short term. The above ple, where the industry as a whole is restructuring actions are summarised in depressed, asset sales and divestments may Table 22.1. In the logit regressions each not raise as much cash as otherwise restructuring action is coded as a dummy (Shleifer and Vishny, 1992). Economic variable. downturns may emphasise the need for operational cost cutting actions and pre- 3.2. Explanatory variables clude equity issues with depressed stock markets. Economic condition is measured The main explanatory variables are the by the Gross Domestic Product (GDP) agency monitoring variables representing growth rate whilst industry condition is different aspects of share ownership, lever- represented by the firm’s Financial Times- age and board composition. Leverage is Actuaries (FTA) industry return. Size of book value of total debt over book values the firm is a proxy for both the flexibility of debt and equity. The total debt is also and internal slack available to the declining CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 577 Table 22.1 Definition of restructuring strategies

Restructuring strategies selected by firms experiencing stock return performance decline are identified and defined. Information on strategies is from press releases to the London Stock Exchange which are documented by Extel Financial News Summary from 1987, with the exception of capital expenditure. Capital expenditure is based on annual reports and accounts. Supplementary information is also collected from Andersen Corporate Register and Hambro Company Guide, Datastream International, and Company Reports and Accounts. These alternative sources are also used for cross-checking information reported in the Extel Financial News Summary.

Strategy Definition

Operational restructuring Operational restructuring Cost rationalisation, layoffs, closures and integration of business units. Asset restructuring Asset sales Divestment of subsidiaries, management buy-outs, spin-offs, sale-and-leaseback, and other asset sales. Acquisitions Full and partial acquisitions of businesses. Capital expenditure Internal capital expenditure on fixed assets such as plant and machinery. Managerial restructuring Managerial restructuring Removal of Chairman or Chief Executive Officer Officer/Managing Director (retirement under the age of 65 is included). Financial restructuring Dividend cut or omission Omission or reduction of dividends from previous year. Equity issue Issue of equity for cash. Debt restructuring Debt refinancing involving extending, converting or forgiving of debt and interest. Combination strategies Cash generative actions Asset sales and cash equity issue.

firm and the opportunities for certain 3.3. DATA strategies such as divestment. A large firm may be able to negotiate debt restructuring As stated earlier, sample firms are those and may be able to avoid dividend cuts. which experience a sharp decline in their Where the firm’s performance decline relative stock return performance. On a has been caused by internal, firm-specific ranking of annual stock returns of all factors such as mistaken business strategy, London Stock Exchange listed firms, firms bad implementation of strategy or lack of which fall into the bottom 20% in the financial control, any restructuring has to decline year after having been in the top reverse the firm specific causes. Again, the 50% in the previous two years (base years) choice of restructuring will be dictated by are sampled. This sampling criterion is the existence of significant internal causes called the 50 : 50 : 20 rule. The sample of decline. covers the period 1985–1993, with 1985– The above explanatory variables are 1991 as the base years and 1987–1993 as summarised in Table 22.2. Chairman cum the decline years. CEO, non-executive Chairman and internal Datastream International is the data cause of decline are each represented by a source for annual stock returns. An initial dummy variable. sample of 415 declining firms satisfying 578 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 22.2 Definition of explanatory variables

The table defines three groups of variables representing firms’ agency monitoring mechanism which are expected to influence the choice of restructuring strategies by poor performing firms. Debt structure is based on accounting information provided by Datastream International and Extel Company Research. Ownership and governance data are extracted from Andersen Corporate Register,Hambro Company Guide and Annual Reports and Accounts. Block shareholding is holding of 5% or more (3% or more since 31st May 1990) as disclosed in the company annual reports. Industry is the Financial Times-Actuaries Index industry group to which the sample firm belongs.

Variable Definition

Debt structure Leverage Total book debt over total book debt and equity Short term leverage Short term debt over total book debt and equity Bank leverage Bank debt over total book debt and equity Unsecured leverage Unsecured debt over total book debt and equity Ownership structure Managerial shareholding Shareholding by members of the board of directors Associated block shareholding Shareholding by family members or trusts of members of the board and company pension plans Institutional block shareholding Shareholding by institutional investors Non-institutional unassociated Shareholding by non-institutional blockholders block shareholding unassociated with management Governance structure Chairman cum CEO Combined role of Chief Executive and Chairman (CEO duality) Non-executive Chairman Chairman in a non-executive capacity Proportion of outside directors Non-executive directors as a percentage of total number of directors Control variables Economic condition GDP growth in the restructuring year Industry condition FTA industry return in the restructuring year Internal cause of decline Reported internal problems such as project failures, bad acquisitions or poor financial control Severity of decline Stock returns ranking in the year of decline Size Log of market value of equity at the end of pre-decline year our 50 : 50 : 20 rule is assembled from a 3.4. Sample characteristics total of 3706 firms covered by Datastream over the period 1985–1993. Sampling 3.4.1. Performance decline excludes financials, utilities and firms with a market capitalisation of less than £10m. Table 22.3 provides the descriptive statistics Small firms are excluded for want of for the sample. From Panel A, the mean sufficient data on their restructuring. (median) annual log returns for the sample Data on the sample firms’ restructuring in the base and distress years are: 42% activities and on the explanatory variables (36%) (base year 2), 33% (28%) (base are collected from Datastream Inter- year 1) and 51% (38%) (decline national, company annual reports and year). The returns to the Financial Times Extel Annual News Summaries. Such data All Share (FTALL) Index in the same years were not available for all companies defined are: 16%, 15% and 16% respectively. as declining. The final sample consists of The sample firms clearly outperform the 297 poor performing firms. market in the base years and under perform it in the decline year. CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 579 Moreover, the decline in performance for unassociated block holding has a median the sample is also very steep. This pattern of 0% (but a mean of 7%). of steep decline is repeated for each of the Comparison with the only other study by sample decline years 1987 to 1993. The Ofek (1993) that examines the determinants sample median returns in the base years for of restructuring strategy choice during the sample range from 6% in 1990, a performance decline reveals interesting recession period, to 53% in 1986. In the differences in agency mechanism between decline years the median return ranges U.S. and U.K. firms. Ofek reports median from 12% in 1993 to 112% in 1990. leverage, managerial shareholding and Panel B of Table 22.3 gives the per- outside (non-managerial) shareholding of formance statistics based on accounting 31%, 22% and 6% respectively. This variables. Both profitability and cash flows compares with our sample’s 27%, 9.5% deteriorate significantly in the decline and 15.7% (not shown in Table 22.4). year. The median fall in the profitability U.S. firms have apparently higher levels of measures – operating margin, earnings managerial shareholding but lower levels per share, return on equity and return on of outside or non-managerial sharehold- asset – ranges from 18% to 22%, all ings. The difference may lie in difference significant at 1%. Median fall in operating between sampling criteria as firms studied cash flows – profit before interest, tax and by Ofek are in the bottom 10% in the depreciation deflated by total assets – is decline year. When we test for this differ- 15%, significant at 1%. The accounting- ence by selecting only firms that fall to the based performance measures thus reflect bottom 10% ie. 50 : 50 : 10 firms the sta- the stock return decline. Our sample thus tistics are respectively 27%, 11% and captures both operating performance and 16%. Therefore, the differences between stock return performance decline.7 Excess the two studies are more likely to stem volatility of stock returns does not appear from market differences than sampling to cause the performance decline in our criterion differences. sample firms. Indeed, the sample betas are As regards board composition, in 44% not unusual (mean and median values of sample firms one person plays the dual are both less than 1) and likely to cause the roles of Chairman and CEO. Non-executive stock return decline. Chairmen preside over the board in 24% of the companies. The median proportion of outside directors in the sample boards 3.4.2. Agency monitoring mechanisms is 22%. Table 22.4 provides descriptive statistics on the leverage, share ownership and 3.4.3. Frequency of turnaround governance structure in base year 1. The strategies median leverage, total book debt to book debt and equity, is 27% and median short- Panel A of Table 22.5 reports the frequen- term debt, bank debt and unsecured debt as cies of sample firms undertaking different proportions of book debt and equity are in turnaround strategies in the decline year the range 12% to 16%. The leverage and in the two post-decline years. We find variables are not entirely mutually exclu- that the most frequent form of restructur- sive. For example, there may be an overlap ing is operational with 59% of the sample between short term and bank leverage.This firms undertaking it in the decline year and implies that when all the leverage variables 47% and 52% of the firms in the two fol- are included in a regression, the empirical lowing years. This is comparable to Ofek’s result has to be interpreted cautiously. 53% for the decline year. Median directors’ shareholding is 9.5% Asset sales are carried out by between and associated block holding is negligible. 27% and 38% of the firms in those years. Institutional ownership amounts to a median This is much higher than the 15% rate value of 6.9%. Non-institutional but reported by Ofek (1993) for the decline year. 580 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 11.58 80.53 54.03 41.72 33.41 23.97 37.87 112.31 mputed by the Risk ncial All Share Times e beginning of the year e beginning profit before interest and interest before profit 20.40 95.70 59.20 10.3 50.20 39.70 30.10 51.12 131.00 profit before interest and tax/total interest before profit 6.19 7.00 10.3 18.9 xcess of 90% of stock market capitalisation in the London Stock in capitalisation xcess of 90% stock market ering in e ends but before extraordinary items/the average number of shares in issue in the year. issue in of shares in number items/the average extraordinary ends but before -test and the Mann-Whitney test. Wilcoxon t ivid 10.3 18.9 18.7 Returns in decline year-2 Returns in decline year-1 Returns in decline year profit before interest and tax plus depreciation (proxy for cash flows).Total debt is the cash flows).Total for (proxy and tax plus depreciation interest before profit total of all interest-bearing No. Mean Med. Mean Med. Mean Med. profit attributable to shareholders/share capital and reserves less intangibles. to shareholders/share attributable profit on assets Return profit after tax, after profit d and preference interests minority Descriptive statistics for stock returns and accounting performance statistics and accounting stock returns for Descriptive Decline year,Decline 1993 68 47.20 40.51 27.10 23.06 FT-All Share FT-All Decline year,Decline 1992 19 1.60 2.79 36.00 24.88 FT-All ShareFT-All 30.8 FT-All ShareFT-All year,Decline 1991 28 23.90 11 20.58 30.8 Decline year 1990 year Decline 32 38.40 35.14 32.70 25.98 FT-All ShareFT-All 7.7 11 30.8 Panel A:Panel of, the year in Annual stock returns to, prior one year to, prior and two years decline of decline. the are tested using B the mean and median Panel in Measurement Service of London Business School, to decline. of London Business Measurement Service prior at th of the year capitalisation is market Size at the beginning debt, i.e. short, debt. and subordinated long is the sum of book debt plus book equity. Capital employed co Beta is based on figures assets less current liabilities.assets less current PBITD Decline year,Decline 1989 54 44.10 39.46 28.70 23.61 FT-All ShareFT-All 24.3 7.7 11.0 Return on equity Return tax. EPS FT-All ShareFT-All year,Decline 1988 55 61.10 18.3 52.86 49.20 40.49 24.3 7.7 Index (FT-All),Index cov 800 firms based on around index a value-weighted Exchange, in comparison. based for performance accounting in changes B shows is provided Panel of decline. the year PBIT Decline year,Decline 1987 41 40.30 33.65 50.20 41.20 FT-All ShareFT-All 16.15 14.59 16.29 Total sampleTotal 297 42.20 35.56 32.63 28.07 Panel A shows the sample firms’ A shows of decline. the year Panel to and including prior the two years in stock on the Fina log returns Return Table 22.3 ShareFT-All 18.90 18.70 25.10 CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 581 a a a a a a * 7.17) 7.93) 8.85) 8.30) (7.52) 17.6 10.34) 19.6 21.6 20.9 14.7 26.8 ( ( ( ( ( a a a a a a 0.95 0.96 (9.42) (4.49) (7.21) (8.80) (7.30) (8.23) 19.1 25.6 23.5 22.4 15.3 27.6 Mean(t statistics) (Z statistics) Median (0.05)Min Max (1.49) PBIT/Sales Panel B:Panel of decline the year in and cash flows profitability in Changes share per Earnings on equity Return on asset Return PBITD/Capital employed debt PBITD/Total Risk (beta) a at 1%. significance Indicates 582 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 22.4 Descriptive statistics for agency and control variables

The table shows the descriptive statistics for the explanatory variables in the pre-decline year. For definitions of the variables see Tables 22.1 and 22.2. Chairman cum CEO, Non-executive Chairman and internal cause of decline are dummy variables coded as 1 when these are present and 0 if otherwise. Source: Hambro Company Guide and Corporate Register, Datastream International and Company Reports and Accounts

Agency variables Pre-decline year

Mean Median Capital structure 0.29 0.27 Leverage 0.15 0.12 Short leverage 0.19 0.16 Bank leverage 0.17 0.12 Unsecured leverage Ownership structure (%) Managerial shareholding 19.90 9.5 Associated block shareholding 0.71 0.0 Institutional block shareholding 12.2 6.9 Non-institutional unassociated 7.1 0.0 block shareholding Governance structure Chairman cum CEO 44.1% – Non-executive Chairman 24.2% – Proportion of outside directors 0.20 0.22 Control variables Internal cause of decline 0.30 – Severity of decline 10.8 10.9 Size (£M) 344.2 54.8

Decline year Decline year 1 Decline year 2

Mean Median Mean Median Mean Median

Economic condition 2.27 2.34 2.08 2.34 0.68 0.70 Industry condition 7.80 11.29 3.60 7.80 32.75 16.88

Surprisingly, acquisitions do not cease the decline year and 3% and 7% in the when firms hit trouble and they are carried following years respectively taking recourse out by nearly 50% of the sample firms in to it. In contrast, Ofek (1993) finds 11% the decline year and by 36% and 27% of of his sample firms restructure their debt in the firms in the post-decline years. Internal the year of decline. Debt restructuring capital expenditure, again surprisingly, appears to be more common among U.S. does not cease but is incurred by 62% of than U.K. firms (only 4% of firms declining firms in the decline year and by 50% and to the bottom 10%, i.e. 50 : 50 : 10 firms, 48% in the following years. adopt it). Equity issues are made by 20% Removal of top management is observed of sample firms in the decline year but by in 20% (in decline year) to 26% (in decline only about 10% in the following years.The year1) of the sample firms. Again, Ofek most frequently employed financial (1993) reports a similar 21% of top restructuring device is dividend cut or management replacement in the year of omission.The proportions of firms adopting decline. Debt restructuring is quite infre- this strategy in the three years are: 24%, quent with only 2% of sample firms in 27% and 34%. CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 583 Table 22.5 Descriptive statistics for restructuring strategies and frequency of restructuring strategies pursued by U.K. listed firms during 1989–1994

This table shows the distribution of corporate restructuring actions in the year of decline, one year, and two years post-decline. Frequency is the proportion of sample firms adopting the strategy. Sample size declines in post-decline years due to failure of firm, takeover or where no data is available, i.e. firms declining in 1993 (68 firms) are excluded from decline year 2 analysis. Source: Company press releases and Company Reports and Accounts. Panel A: Descriptive statistics for restructuring strategies in the year of, one and two years after decline

Decline year, Decline year1, Decline year2, Restructuring strategy frequency(%) frequency (%) frequency (%)

Operational restructuring Cost rationalisation, closures and integration of business units 58.6 46.7 51.6 Asset restructuring Asset reduction 26.6 37.8 35.6 Acquisition 50.2 35.9 27.1 Capital expenditure 61.6 50.4 47.9 Managerial restructuring Remove top management 19.5 25.9 21.8 Financial restructuring Dividend cut or omission 23.6 27.0 34.0 Equity issue 20.2 10.4 13.3 Debt restructuring 2.4 3.3 7.4 Cash generative actions 40.1 44.1 43.6 Sample size 297 270 188

Panel B: Frequency of restructuring strategies pursued by U.K. listed firms during 1989–1994 Asset sales cover divestments, management buy-outs and other asset sales. Acquisitions represent full and partial acquisitions. Dividend cut/omission refers to cut/omission in dividends per share over the previous year. Rights issues encompass rights issue, rights offer, offer for sale, open offer and placing of firm shares with institutions and financial intermediaries. Source: Financial Times Extel Company Research and Company Analysis. Average Asset sales Acquisitions Dividend cut/ Rights issue Cash generative no. of firms % % omission % % actions %

1521 19.6 34.5 20.8 15.2 31.6

An interesting question to ask is how Research CD-ROM, we compile the does the frequency of strategies followed by average number of firms in the population performance decline firms compare with adopting asset sales, acquisition, dividend that in the population of firms listed in the cut/omission, rights issue and cash genera- U.K. Do these frequencies differ from the tive actions, during the period 1989–1994. population benchmarks and why? Panel B The period broadly coincides with the of Table 22.5 provides answers to these period under study, and the year 1989 is questions where population data are the first year covered by the Company available. From an extensive and rigorous Research CD-ROM. search of financial news reported by firms Unsurprisingly, a higher proportion of and reported by FT Extel in their Company our sample firms (34%) sell their assets 584 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY than in the population (20%),8 from the between dominant and non-dominant year of decline to two years thereafter. In groups are significant. Sample firms are the case of acquisitions, sample firms lenders dominated when their leverage is in clearly overtake the population with 50% the top quartile of all the sample firms and versus 35% of firms in the population they are in severe decline (bottom 50% in making acquisitions in the year of decline. sample stock ranking in the year of However, with the onset of decline, sample decline). Lenders under such circumstances firms reduce their rate of acquisitions to are likely to have high stakes in recovery the population rate one year after decline. and to exercise their priority rights. Sample Two years post-decline, far fewer sample firms are manager-owner dominated when firms seek acquisitions compared to the they are not lender dominated according to population. A similar pattern is observed the above definition and the managerial with regard to equity issue. More sample and associated shareholdings are in the top firms tap the market than the population at quartile of all sample firms. large, 20% in the decline year,versus 15%. Where neither lenders nor manager- However, sample firms are less likely to be owners are dominant according to the successful in raising finance from equity above definitions and the unassociated investors subsequent to performance block shareholding is in the top quartile of decline than the firm in the population. all sample firms, the firms are deemed Finally, more sample firms resort to cash block shareholders dominated. Finally, the generative actions than the population with sample firms not dominated by lenders, an annual average of 43% of sample firms manager-owners and block shareholders, taking it compared to only 32% of firms they are deemed to be under the control of in the population. Overall, our sample of the board of directors. In turn, the board declining firms carries out various restruc- may be dominated by a dual CEO or turing activities more intensively than the collectively by the board members. firm population at large. Panel A of Table 22.6 shows the effect of lender dominance on strategy choice. To summarise, lender-dominated firms are 4. RESULTS more likely to opt for operational restructuring, cash generative actions, dividend cut/ 4.1. Impact of stakeholder omission and debt restructuring. They are dominance on turnaround strategy less likely to approve of cash-consuming choice strategy such as capital expenditure. They have little influence on management We divide our sample into two groups – one changes. stakeholder dominated and the other non- In Panel B, the effects of manager- dominated by that stakeholder. For each owner dominance are shown. Manager-owner stakeholder group – lenders, manager- dominated firms are more likely to under- owners, block shareholders, CEO and take capital expenditure and less inclined collective board of directors – we examine to pursue operational restructuring and the likelihood of a given strategy being acquisitions. They are also less likely to chosen. The difference in the proportions sack their top management! of sample firms in the dominated and When firms are dominated by block- non-dominated groups choosing a strategy holders, their influence is less pronounced is tested for statistical significance. Any and limited to two strategies as shown in significant difference reflects the influence Panel C. These shareholders make opera- of the dominant stakeholder. tional restructuring and capital expenditure Table 22.6 shows the proportions of less likely. CEO dominant board influence sample firms pursuing a given strategy in is also limited but stronger than with the decline and two post-decline years blockholder dominance. Declining firms when the differences in these proportions dominated by their CEOs prefer capital CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 585 expenditure but disfavour dividend cut/ restructuring strategies. They, however, omission. They understandably reduce the disfavour capital expenditure. chances of managerial restructuring. Associated shareholding significantly Collective board dominance again influences influences the choice of several strategies. only three strategies. With little conflict of It reduces the probability of the declining interests in the board, operational restructur- firm pursuing operational restructuring, ing, cash generative actions and acquisitions cash generative actions and acquisitions. are favoured by the board collectively. Unassociated shareholders, however,have no Having explored the impact of stakeholder influence whatsoever in the year of decline. dominance on restructuring strategy As regards the governance structure, choice, we now examine the individual and declining firms with CEO duality are more joint impact of agency monitoring mecha- likely to increase capital expenditure and nisms on strategy choice. less likely to take cash generative actions. Non-executive Chairmen and the propor- 4.2. Impact of individual agency tion of outside directors on the board have monitoring mechanisms on little influence in the choice of turnaround turnaround strategy choice strategy, at least in the decline year. The control variables have varying impact Tables 22.7–22.9 report the model on strategy choice. Where firm decline coefficients for the logistic regressions of coincides with an economic downturn, firms corporate restructuring strategy choices react with several strategies.They resort to on the agency and control variables. A more operational restructuring, managerial separate regression is run for each strategy restructuring and dividend cut/omission. and for each of the following years: year of However,cash generative actions and invest- performance decline (the decline year), the ments are less likely during an economic year after the decline year (decline year1) downturn. On the other hand, if the whole and the second year after the decline year of their industry suffers decline, the sample (decline year2). firms are more likely to increase their We model the strategy choices in each capital expenditure perhaps to gain a year,rather than over a single period covering competitive advantage, cut/omit their the three years, to examine whether there is dividends, and restructure their debts. a time lag in the impact of the agency and Where decline has resulted from firm control variables. It is plausible that certain specific internal problems, operational drastic strategies like top managerial restructuring is more likely. change or asset reduction may be under- The more severely declining firms (rep- taken after strategies such as operational resented by low ranking on stock returns in restructuring. the decline year) are more likely to go for In the decline year, in Table 22.7, the operational restructuring, top management logistic models are significant (based on replacement, dividend cut/omission and the Chi-square statistic) in all except debt restructuring. Finally, large companies where managerial restructuring and debt are more likely to avoid the need to cut/ restructuring are the dependent variables. omit dividends. Significance of the individual variables The strategy choices made in the second is tested for using the Wald statistic.9 year of decline are shown in Table 22.8. All The explanatory power of the models, logit models are significant at the 5% level measured by McFadden’s R3, ranges from or better. McFadden’s R2 ranges from 7% 4% to 24%. to 27% and for most of the models the It appears that in the decline year itself explanatory power is much higher than of significant restructuring begins to take their counterparts in the decline year in place and the impact of several agency and Table 22.7. It appears that agency and control variables is felt. Lenders increase control variables exercise their influence the probability of cash generative and debt more strongly in the second year of decline 586 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY * *** y the board.The owners are dom- owners 2 takes. If the firm’s tion of sample firms tion mple firms ranked on ranked mple firms ders, managers, block- uffered a severe decline a severe uffered t groups are shown.t groups Z is olders, are blockholders 0.470.530.330.37 0.530.10 0.42 0.51 0.34 0.59 0.070.50 1.17 0.25 1.73 0.23 0.33 0.23 0.58 0.52 0.49 0.28 0.22 0.24 2.84 0.75 0.17 *** *** ** *** *** * *** ** ** 1 Year of decline 0.690.51 0.430.17 0.23 2.78 0.36 0.010.220.22 3.48 4.87 0.49 0.50 0.39 1.80 3.74 2.42 ** *** *** *** *** ** rmined by the relative bargaining powers of various stakeholders in the firm.The compares the choice in table stakeholders of various powers bargaining by the relative rmined Yes No Yes No Yes No Year of decline Year of decline Stakeholder dominance and choice of restructuring strategy of restructuring and choice dominance Stakeholder end cut/omission 0.44 0.20 3.44 the absolute value of testthe absolute statistic the Mann-Whitney value for test. Wilcoxon board is deemed to be CEO dominated if Chairman is also the CEO and collectively dominated, is also the CEO and collectively if Chairman board is deemed to be CEO dominated the propor otherwise.The shows table and non-dominan between dominant difference is a significant there which the strategy. for choosing each group variables in Only deemed dominant.The remaining firms,deemed dominant.The remaining which are not lenders, dominated, or blockholders are deemed manager-owners to be controlled b such shareholding), influence. are deemed to be entrenched and possess dominant nor manager- lenders neither Where manager-owners inant, (i.e. are high shareh blockholding and unassociated to the total of all significant relative of sample firms) top quartile decision-making process is not dominated by lenders, is not dominated (i.e. process are high decision-making shareholding and associated and managerial of sa top quartile (bottom 50% of sample stock ranking in year of decline), year in the firm the type of actions (bottom 50% of sample stock ranking influencing in are deemed leaders dominant holders and board of directors. Where a firm is highly leveraged (i.e. and board of directors.holders leveraged and has s is highly on leverage) a firm ranked Where of sample firms top quartile of strategy between stakeholder dominated and non-dominated groups of declining firms. Four stakeholder groups are identified: firms.of groups stakeholder of declining Four len groups and non-dominated dominated between stakeholder strategy Operational restructuringOperational 0.73 0.56 1.94 Panel B:Panel dominance Manager-owner restructuringOperational 0.41 0.63 3.13 Panel A:Panel dominance Lender Cash generative actionsCash generative Capital expenditure 0.49Debt restructuringSample size 0.53 0.39Capital expenditure 0.09 0.63 restructuringManagerial 1.27Sample size 0.01 43 0.15 0.20 1.18 0.69 3.24 0.21 254 0.45 0.31 0.40 61 1.05 3.64 0.53 3.13 236 0.16 2.39 35 0.29 235 2.04 58 212 30 158 44 144 The choice of restructuring strategies is dete strategies of restructuring choice The Table 22.6 StrategiesDivid Mean Mean 0.36Acquisitions Z 0.54 Mean 2.47 Mean Z Mean Mean Z CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 587 * 0.430.53 0.50 0.480.560.26 1.00 0.24 0.46 0.210.51 1.19 0.26 0.65 0.41 0.28 1.18 0.28 ** * *** * * ** 0.400.67 0.530.19 1.81 0.460.50 0.290.53 2.58 0.47 1.32 0.46 0.42 0.32 0.38 0.62 1.68 2.15 0.33 0.64 0.65 0.48 1.93 ** *** ** *** ** ** indicate significance at 1%, significance indicate 5% and 10% respectively. end cut/omission 0.11 0.27 2.58 Panel D:Panel Non-CEO board dominance restructuringOperational 0.73 0.54 2.89 Panel D:Panel CEO board dominance Capital expenditure 0.81 0.57 3.29 Sample size 57 240 51 219 39 149 Capital expenditureSample size 0.51 restructuringManagerial Divid 0.18 0.65 65 0.20Acquisitions 2.03 232 0.42 0.16 0.61 0.28 60 0.47 1.85 2.00 210 28 160 Panel C:Panel dominance Blockholder restructuringOperational 0.55 0.51 0.59 actionsCash generative 0.33Sample size 0.52***, **, * 0.50 0.36 2.34 71 2.37 226 66 204 47 141 588 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY * * * 1.05 0.01 0.18 0.04 0.13 0.06 3.92 s consecutive years. r refers to the year in to the year r refers *** ** *** ** ** t variable equals one if a t variable 0.36 0.63 0.69 0.11 0.50 0.12 0.61 0.97 0.02 0.30 * ** 0.54 0.19 0.86 2.20 0.07 0.860.09 0.450.13 4.63 0.00 ** ** *** ** 0.01 0.01 0.01 0.01 0.01 0.13 0.71 1.76 0.00 0.000.22 0.01 0.04 0.14 ** ** 0.19 0.24 0.01 0.09 0.01 0.38 0.02 0.44 0.02 0.21 0.08 0.13 ** *** * * 0.21 0.21 1.78 0.58 0.28 0.01 0.00 0.01 0.02 0.37 0.13 0.24 0.02 0.01 0.02 0.05 0.07 ** ** *** ** ** 0.15 0.05 0.36 0.05 0.01 0.01 0.02 0.10 Operational Cash Capital Managerial Dividend Debt (Debt, ownership, variables) and control governance f -square 12.1% 13.0% 5.9% 12.0% 4.4% 23.9% 4.7% R -value 0.00 0.00 0.08 0.00 0.27 0.00 0.22 Logistic regression of restructuring strategies on agency and control variables (year of decline) (year variables on agency and control strategies of restructuring regression Logistic p Restructuring strategy Restructuring indicate significance at 1%, significance indicate 5% and 10% respectively. * , ** , Model: Coefficients are tested for significance using the Wald test statistic. Wald the using significance are tested for Coefficients which a firm declines to the bottom 20% ranking in stock in returns, to the bottom 20% ranking declines which a firm the market, in two previou the top 50% for been in having after Economic condition Economic Chi-square 38.2 41.2 18.0 38.1 13.3 81.2 14.2 strategy is taken, and zero if otherwise. The sample consists of 297 declining firms over the period 1987 to 1993. is taken,strategy yea the period decline over The if otherwise. firms and zero sample consists of 297 declining The Proportion of outside directorsProportion 1.45 Chairman cum CEO Chairman Severity of decline Severity Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. variables on agency and control dependen strategies The of restructuring regressions of the logistic Coefficients Unassociated shareholding Unassociated Industry condition ConstantMcFadden’s 1.69 Non-executive Chairman Non-executive Table 22.7 Explanatory variablesLeverage shareholding Associated restructuring generation Acquisition problemInternal expenditureSize 0.63 restructuring cut/omissionRegression restructuring *** 2.58 1.12 CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 589 ** ** ** ** ** ** * 1.09 7.86 7.36 0.55 0.50 0.10 0.10 10.47 cquired, post one year *** ** *** t variable equals one if a t variable 0.01 0.36 0.27 2.43 0.43 0.08 0.84 0.00 0.12 ** * 0.48 0.47 0.03 0.87 0.07 0.10 0.04 0.84 0.01 * *** *** *** 0.21 0.02 0.05 2.46 0.61 0.31 0.02 0.30 * *** * ** *** 0.04 0.60 0.47 3.20 0.01 0.01 0.81 -0.830.73 0.81 -0.50 0.57 1.41 2.47 10.83 *** ** ** *** 0.91 0.49 0.02 0.01 0.00 0.000.46 0.00 0.00 0.05 0.000.010.03 0.29 0.01 0.05 0.00 0.00 *** *** ** ** ** *** 0.06 0.52 0.14 3.35 0.02 0.03 0.11 0.03 0.22 Operational Cash Capital Managerial Dividend Debt (Debt, ownership, variables) and control governance f -square 26.5% 16.5% 14.0% 14.0% 6.8% 19.4% 17.7%. R -value 0.00 0.00 0.00 0.00 0.05 0.00 0.00. Logistic regression of restructuring strategies on agency and control variables (year 1 after decline) 1 after (year variables on agency and control strategies of restructuring regression Logistic p Restructuring strategy Restructuring indicate significance at 1%, significance indicate 5% and 10% respectively. * , ** , Model: decline, in the period 1987 to 1993. Coefficients are tested for significance using the Wald test statistic.decline,Wald 1987 to 1993. the using the period in significance are tested for Coefficients strategy is taken,strategy nor a (bankrupt) insolvent are neither which if otherwise. firms and zero sample consists of 270 declining The of decline Severity Chi-square 83.0 48.8 40.8 40.7 19.1 58.3 52.5. Non-executive ChairmanNon-executive 0.24 Chairman cum CEOChairman 0.07 Economic condition Economic Constant 2.60 Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. variables on agency and control dependen strategies The of restructuring regressions of the logistic Coefficients Table 22.8 Explanatory variablesLeverage shareholding Associated shareholding Unassociated restructuring of outside directors Proportion generationIndustry condition Acquisition problemInternal expenditureSize 0.85 restructuringMcFadden’s cut/omissionRegression 0.02 restructuring *** 2.27 0.38 0.47 590 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY * * *** 0.18 4.96 0.01 0.01 0.43 0.07 0.00 ecline strategies to the strategies ecline e test statistic. Wald cquired, post- two years *** *** * t variable equals one if a t variable 0.06 0.15 0.22 0.66 0.06 0.01 0.31 0.44 0.07 0.56 *** 0.02 0.01 0.24 1.190.240.08 3.36 0.57 0.05 1.86 0.01 0.02 0.00 ** * * *** 0.16 0.07 2.71 1.96 2.55 0.34 ** *** *** 3.03 0.44 0.000.003.13 0.00 0.00 0.00 0.00 * * * 1.48 0.02 0.25 0.37 0.68 0.02 0.02 0.22 0.000.41 0.00 0.00 *** * ** 0.04 0.03 0.02 0.04 0.51 0.01 0.54 0.01 0.20 1.32 0.71 Operational Cash Capital Managerial Dividend Debt (Debt, ownership, variables) and control governance f -square 22.2% 12.9% 9.7% 19.3% 7.0% 27.2% 5.4% R -value 0.00 0.00 0.05 0.00 0.25 0.00 0.47 Logistic regression of restructuring strategies on agency and control variables (year 2 after decline) 2 after (year variables on agency and control strategies of restructuring regression Logistic p Restructuring strategy Restructuring indicate significance at 1%, significance indicate 5% and 10% respectively. * , ** , end of the analysis period i.e. period end of the analysis 1994. December reduced. th is therefore using Sample size significance are tested for Coefficients Model: decline.The sample covers only firms in decline in the period 1987 to 1992, post-d one year 1993 have only the period in in decline in declining firms decline.The only as firms sample covers Proportion of outside directorsProportion 0.55 2.61 Constant 0.27 Severity of decline Severity strategy is taken,strategy nor a (bankrupt) insolvent are neither which firms if otherwise.The and zero sample consists of 188 declining Chi-square 47.1 25.9 19.2 40.2 13.6 59.8 10.6 Economic condition Economic Unassociated shareholding Unassociated cum CEO Chairman Nonexecutive ChairmanNonexecutive 0.07 0.19 0.46 0.55 0.06 0.15 Industry conditionSize 0.01 0.17 0.13 0.11 0.23 0.15 Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. variables on agency and control dependen strategies The of restructuring regressions of the logistic Coefficients Table 22.9 Explanatory variablesLeverage shareholding Associated restructuring generation Acquisition problemInternal expenditure restructuringMcFadden’s cut/omissionRegression restructuring *** 0.81 CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 591 suggesting delayed reaction to the onset The logit models of strategic choices of decline. made in the third year of decline (decline Unsurprisingly, lenders continue to press year 2) are shown in Table 22.9. In for cash generative actions. They are also contrast to the model for the previous year more likely to resort to debt restructuring. in table 8, only five of the third year models Ownership continues to influence strategy are significant at least at the 10% level. choices in the second year. Associated McFadden’s R2 ranges from 5% to 27%. shareholding decreases the probability of The third year models thus generally have operational restructuring and cash genera- less explanatory power than the models for tive actions. Thus, management-associated the first two decline years. It appears that shareholders’ resistance to these strategies the influence of the agency and control in the decline year is reinforced in the second variables on strategy choices is waning. year. Interestingly, associated shareholders Nevertheless, some of these variables resist debt restructuring. Unassociated continue to exert significant impact. shareholding begins to influence adoption Lenders continue to restrict capital of strategies in the second year. They expenditure. Debt restructuring is again are negatively related to the choice of made more likely by lenders. Associated operational restructuring. CEO- duality shareholders stubbornly resist cash genera- continues to increase the chances of capital tive strategies. Interestingly, unassociated expenditure but, unsurprisingly, reduces the shareholders join associated shareholders probability of managerial restructuring. in resisting cash generative strategies. Non-executive Chairmen make cash As regards governance structure, dual generative actions and debt restructuring CEOs are still more likely to go for capital less likely but capital expenditure more expenditure. Non-executive Chairmen, likely. More outside directors now means however, are quite inactive. With more greater chances of cash generative actions outside directors a poorly performing firm and managerial restructuring. Outside is now more likely to undertake not only directors’ activism in the second year is more cash generative actions but acquisi- in stark contrast to their passivity in the tions and capital expenditure too, perhaps decline year. to expand the firm after two years of The effects of economic downturn are restructuring. equally significant in the second year. It The impact of external environment is continues to increase the probability of still important in the third year. Economic operational restructuring, dividend cut/ downturn still increases the probability of omission and debt restructuring, but operational restructuring and dividend reduces the probability of investments both cut/omission, and reduces the probability of acquisitions and capital expenditure. acquisitions and capital expenditure. If the Industry downturn makes operational industry is depressed in the third year, and debt restructuring less likely and sample firms would be inclined to remove dividend cut/omission more likely. Severity top management, cut/omit dividends and of decline impacts further in the second restructure their debts. Operational year. It continues to make operational restructuring, however, is less likely to be restructuring, dividend cut/omission and needed. Sample firms are still constrained debt restructuring more likely. In addition, in their strategy by the existence of an the more severely declining firms are also internal cause of decline and severity of less likely to undertake acquisitions. the initial decline. Firms with an internal Internal cause of decline has little influence cause of decline are still more likely to on strategy choice in the second year of restructure their operations. Severe decline decline. Finally, large firms increase the firms are still more likely to cut/omit their probability of investments in the form dividends. Firm size, however, ceases to acquisitions and capital expenditure. have any impact in the third year. 592 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 4.3. Summary of the impact of similar ways to reduce or increase the agency and control variables probability of certain restructuring actions. Associate and unassociated shareholders In Table 22.10 the results of the logit make operational restructuring and cash models of turnaround strategy choice generative actions less likely. This coalition reported in Tables 22.7–22.9 are sum- in resisting cash-generative actions stems marised to highlight the impact of each from owners’ desire to avoid option value- agency or control variable on the probability destroying asset sales and stump up new of choosing or avoiding different strategies. cash in the form of equity injections. A striking feature of the results is that Similarly, Chairmen cum CEOs and certain stakeholder groups seem to act in non-executive Chairmen seem to prefer

Table 22.10 Summary of the effect of each explanatory variable on the choice of restructuring strategies

This table summarises the results in Tables 22.7–22.9 The multiple influences of each explanatory variable on the probability of various restructuring actions occurring are highlighted. Variables that are significantly positively/negatively related to particular strategies (i.e. increasing/decreasing the probability of those actions occurring) in the logistic regression models in Tables 22.7–22.9, are separately listed.

Probability of restructuring action

Explanatory variable Increased Decreased

Leverage Debt restructuring Capital expenditure Cash generation Associated shareholding Acquisitions Operational restructuring Debt restructuring Cash generative actions Unassociated shareholding Operational restructuring Cash generative action Chairman cum CEO Capital expenditure Managerial restructuring Cash generative actions Non-executive Chairman Acquisitions Debt restructuring Cash generative actions Proportion of outside directors Cash generative actions Acquisitions Capital expenditure Managerial restructuring Economic downturn Operational restructuring Cash generative actions Dividend cut/omission Acquisitions Debt restructuring Capital expenditure Industry downturn Capital expenditure Operational restructuring Dividend cut/omission Debt restructuring Managerial restructuring Debt restructuring Internal problem Operational restructuring Severe decline Operational restructuring Acquisitions Managerial restructuring Dividend cut/omission Debt restructuring Size Acquisitions Dividend cut/omission Capital expenditure CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 593 investments but dislike cash-generative None of the strategies is favoured by all actions. Outside directors appear to side stakeholders. Strategies such as cash- with lenders in pressing for cash-generative generative actions are favoured or opposed actions. However, only outside directors by different coalitions of interests. While do not disfavour any particular strategies, bank creditors’ push for cash-generative and favour both cash-generating and cash- actions, they are supported by outside consuming actions, and more importantly, directors but the coalition of associated and instigate managerial restructuring. This unassociated shareholders, Chairman cum lends support to the effectiveness of a CEO and non-executive Chairmen makes it governance structure characterised by a less probable. Associated and unassociated substantial independent director presence. shareholding jointly resist operational In contrast, lenders appear to be primarily restructuring. concerned with conserving or augmenting While capital expenditure is favoured by the cash position of declining firms and are Chairmen cum CEO and outside directors, neutral towards management change. lenders seem opposed to the strategy. Declining firms react differently to Similarly, Chairman cum CEO and outside deterioration of the business environment. directors’ preference for acquisitions is Faced with an economic downturn, they matched by associated shareholders resist- resort to operational restructuring, dividend ance to the same strategy. Managerial cut/omission and debt restructuring. Cash replacement is made more likely by outside generative actions, however, are also more directors, but it is, predictably, opposed by difficult in depressed economic climates. Chairmen cum CEOs. Our results thus Investments are also less likely in harsh reveal shifting coalitions of stakeholders economic conditions. In contrast, when vis-à-vis different turnaround strategies. their industry as a whole experiences a The results based on logit regression downturn, these firms pursue capital expen- models are consistent with those discussed diture, dividend cut/omission, managerial earlier under stakeholder dominance and and debt restructuring. Operational restruc- thus add to the robustness of our conclusions turing, however, is less needed during an about the impact of stakeholder dominance industry downturn. Firms with an internal on restructuring strategy choice. cause of decline are more likely to restructure their operations. The response to 4.4. Impact of lender and severe decline takes the form of operational ownership types on restructuring restructuring, dividend cut/omission, debt strategy choice restructuring and a reduction in acquisitions. More interestingly, management replacement We know that when lenders dominate is also more likely in such firms. This firms’ decision-making process, they suggests that top managers are able to demand operational restructuring, cash- fend off attempts to replace them until the generative actions, dividend cuts/omissions firm situation deteriorates perilously. Large and cut in cash-consuming capital firms, being more resourceful, are also less expenditure. We also know that when likely to resort to dividend cut/omission and manager-owners dominate firms’ decision- are more able to afford investments. making process, they refrain from adopting Table 22.11 summarises the joint impact operational restructuring, acquisitions of one or more agency or control variables and managerial restructuring strategies, on the probability of choosing or avoiding a and opt instead for capital expenditure. particular strategy. It answers the questions Similarly, when blockholders dominate the ‘which factors make a given restructuring firm’s decision-making process, they strategy more likely and which factors disfavour operational restructuring and make it less likely?’ and ‘what is the capital expenditure. The logit regression coalition of stakeholders bearing on the results on the effects of lender and share- adoption of a given strategy?’ holders monitoring on strategy choice 594 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Table 22.11 Joint impact of explanatory variables on individual restructuring strategy choice

As explanatory variables collectively influence the choice of restructuring strategies, the combined impact of explanatory variables on the choice of a specific restructuring strategy is summarised from the results reported in Tables 22.7–22.9. Explanatory variables that are significantly positively/ negatively related to a specific strategy, in the logistic regression models in Tables 22.7–22.9, increase/decrease the probability of that action occurring. Explanatory variables

Restructuring strategy Probability increasing Probability decreasing

Operational restructuring Industry condition Associated shareholding Internal problem Unassociated shareholding Economic condition Severity of decline Acquisitions Non-executive Chairman Associated shareholding Proportion of outside directors Severity of decline Economic condition Size Capital expenditure Chairman cum CEO Leverage Proportion of outside directors Industry condition Economic condition Size Managerial restructuring Proportion of outside directors Chairman cum CEO Economic condition Industry condition Severity of decline Dividend cut/omission Economic condition Industry condition Severity of decline Size Debt restructuring Leverage Associated shareholding Industry condition Non-executive Chairman Economic condition Industry condition Severity of decline Cash generation Leverage Associated shareholding Proportion of outside directors Unassociated shareholding Economic condition Chairman cum CEO Non-executive Chairman broadly confirm these tendencies. The The results show that short term and question then is which type of lenders and unsecured lenders are the parties behind owners favour which type of strategy. To lenders’ demand for operational restructur- test for these individual impacts we rerun ing.10 Bank and short term lenders jointly all the regressions in Tables 22.7–22.9 press for cash generative actions. However, with three types of lenders instead of one, only bank lenders have the clout to restrict short term lenders, bank lenders and firms’ capital expenditure. unsecured lenders, and four types of Manager shareholders alone resist shareholders instead of two, manager operational restructuring, cash generative shareholders, associated block shareholders, and acquisition strategies. Conversely, institutional block shareholders and associated block shareholders prefer non-institutional unassociated block acquisitions and surprisingly, in a show shareholders. of independence, managerial restructuring. CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 595 Institutional block shareholders and board as well as control variables to allow non-institutional unassociated block for the external economic and industry shareholders jointly resist operational environment, severity of decline and firm restructuring and cash-generative actions. size are included as explanatory variables. However, non-institutional block share- The impact of these variables on turnaround holders also make managerial restructuring strategy choices in the decline year and in more likely. They are also supportive of the two post-decline years is modelled with declining firms, in terms of dividend cut/ logit regressions. omission. Our results show that turnaround strategy choices are significantly influenced by both agency variables and control variables.While 5. SUMMARY AND CONCLUSIONS there is agreement among stakeholders on certain strategies there is also evidence of Firms which experience performance conflict of interests between lenders and decline may choose a variety of alternative managers and between managers and some methods of restructuring themselves to block shareholders. Lenders’ preference restore their financial health. These for cash-generative actions are in direct restructuring strategies for poorly perform- conflict with shareholders’ preference. ing companies include operational, asset, Weak governance structure helps entrench financial and managerial restructuring. managers and perpetuate their self-serving However, any restructuring strategy has behaviour resulting in less restructuring different, and often conflicting, welfare and top management replacement. Non- implications for the different stakeholders institutional rather than institutional in firms – shareholders, lenders and man- shareholders appear to be active monitors agers. Within the agency model of the firm and influential in instituting top manage- the strategy choices made by managers ment changes. However, all types of may benefit one group of stakeholders at shareholders disfavour any type of costly the expense of the other groups. However, strategy such as operational restructuring managerial choices are also constrained by or option value-destroying strategies such the agency monitoring embodied in the as asset sales. Boards of directors, however, firms. Agency monitoring may be embodied seem to be effective in their oversight in the rights of lenders, the power and of managers, as they intensify adoption of influence of large block shareholders or in generic turnaround strategies. There is the oversight function and independence evidence of shifting coalitions among of the board of directors. The choice of lenders, managers and directors in the recovery strategies is, therefore, determined choice of recovery strategies. Institutional by the complex interplay of the ownership shareholders generally seem to go along structure, corporate governance and lender with management shareholders. Response monitoring of the firms in decline. of non-executive Chairmen and CEO cum For a sample 297 U.K. firms which Chairman to turnaround is broadly similar. experienced stock return performance The results also show the effects of decline during the period 1987–1993, we dominance by certain stakeholder groups. examine the impact of agency monitoring Lenders’ tight financial reigns through on the turnaround strategies selected by wholesale bans on investments can cause them. Performance decline is defined as the an under investment problem. Lenders fall into the bottom 20% of the ranking of may not only be depriving firms of vital all U.K. listed firms on annual stock resources necessary to compete and market returns from being in the top 50% reverse decline but also side lining their in the preceding two years. A number of longterm health by favouring short-term variables proxying for different lender types, cash generative measures to facilitate debt for different types of block shareholders and repayment. It raises the question if banks for the strength and independence of the are too keen to pull the plug on ailing firms 596 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY which lack short-term cash generation The appointment of a receiver by secured ability in spite of their healthy long-term lenders, or an administrative receiver when a potential. Entrenched managers’ inaction floating charge is held, effectively bankrupts the to performance decline may lead a down- distressed firm. ward spiral to failure. Blockholders have 5 The literature on stock price overreaction (De Bondt and Thaler, 1985) raises the concern a weak influence on a limited range of turn- that a stock return based measure of perform- around strategies. Consequently, corporate ance decline may merely represent a correction failures can potentially be explained by poor for the earlier overreaction.The condition of two management of stakeholders’ interests consecutive years’ good performance preceding during decline, resulting in poor selection of the decline which we have applied in our turnaround strategies. sampling mitigates this problem. Further, anecdotal evidence suggests that stock market performance decline is not greeted with inertia and indifference by managers who smugly NOTES attribute such decline to the stock market whims such as overreaction. It appears that * This paper has benefited from the comments such performance decline causes managerial of Richard Brealey, Julian Franks, Michel Habib, concern and triggers remedial action including Mezzaine Lasfer, Paul Marsh, Colin Mayer, John corporate restructuring. McGinnes, Ayo Salami, Henri Servaes, Richard 6 Routine replacements proxied by sample Tafller, Dylan Thomas, seminar participants at firm annual depreciation charge average 6.5%. City University Business School and London 7 The similarity of performance decline in Business School, and conference participants at both stock return and accounting terms suggests the 1996 European Finance Association that the stock return decline is no freak caused Meeting and the 1996 Midwest Finance by correction of stock market overreaction in Association Meeting. We are grateful to the the base years unrelated to underlying operating Chartered Association of Certified Accountants, performance. U.K. for partially funding this research. 8 The population frequency is the average for 1 A potential explanation for this inconsis- the period 1989–1994. The same benchmark tency lies in the difference in length of distress population is used in the rest of this section. It examined in the two studies. Ofek (1993) studies should be noted that the population includes the the short-term restructuring actions in the year sample firms. Hence, comparison is not between of performance decline whereas Gilson et al. two independent groups. Managerial and opera- (1990) examine firm actions following three tional restructuring are not examined due to years of low performance. lack of details in the Extel CD-ROMs. 2 Pound (1988) provides counter-arguments 9 To simplify the tables, the Wald statistic is for a less effective monitoring role for institu- not reported and only its level of significance tional and large shareholders due to their being indicated when it is significant at least at passive investors or having other business deal- the 10% level. ings with the company which lead to a conflict 10 Due to space constraints the tables are not of interest detracting from effective monitoring. presented in the paper.The results are available Mallette and Fowler (1992) also report that from the authors. high levels of institutional shareholdings are more positively associated with the adoption of anti-takeover poison pills than lower levels of REFERENCES institutional shareholding. 3 There are impediments to effective moni- Agrawal, A. and Mandelker, G. (1990) Large toring by non-executive directors. Baysinger and shareholders and the monitoring of man- Hoskisson (1990) cite information asymmetry agers: The case of anti-takeover charter whereby outside directors do not possess all the amendments, J. Financ. Quantitat. Analysis information that executive directors have. Moreover, the insiders may have packed the 25(2), June. board with outside directors who are beholden Asquith, P. and Mullins, D. W., Jr. (1986) to them in some way and therefore subservient. Signalling with dividends, stock repurchases 4 In the U.K., secured creditors are frequently and equity issues, Financ. Managt Autumn, blamed for pulling the plug on firms too soon. 27–44. CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 597 Barclay, M. and Holderness, C. (1991) corporate ownership and control when firms Negotiated block trades and corporate default, J. Financ. Econom. 27, 355–387. control, J. Finance 46, 861–878. Gilson, S. C., John, K., and Lang, L. H. P.(1990) Baysinger,B. D. and Hoskisson, R. E. (1990) The Troubled debt restructuring: An empirical composition of boards of directors and strate- study of private reorganization of firms in gic control: Effects on corporate strategy, default, J. Financ. Econom. 27, 315–353. Academy of Managt Rev. 15, 72–87. Hermalin, B. and Weisbach, M. (1992) The Bethel, J. E. and Liebeskind, J. (1993) The effects effects of board composition and direct of ownership structure on corporate restructur- incentives on firm performance, Financ. ing, Strategic Managt J. 14, 15–31. Managt 20, 101–112. Boeker,W. and Goodstein, J. (1993) Performance Hill, C. W. L. and Snell, S. A. (1989) Effects of and successor choice: The moderating effects ownership structure on corporate productivity, of governance and ownership, Academy of Academy Managt J. 32, 25–46. Managt J. 36, 172–186. Hirschey, M., Slovin, M. B., and Zaima, J. K. Brickley, J. A., Lease, R. C., and Smith, C. W. (1990) Bank debt, insider trading and the (1988) Ownership structure and voting return to corporate sell-offs, J. Banking and on anti-takeover amendments, J. Financ. Finance 14, 85–98. Econom. 20, 267–292. Holderness, C. G. and Sheehan, D. P. (1985) Brown, D.T., James, C. M., and Mooradian, R. M. Raiders or saviours?: The Evidence on six (1994) Asset sales by financially distressed controversial investors, J. Financ. Econom. firms, J. Corporate Finance 1, 233–257. 14, 555–579. The Cadbury report on the Financial Aspects of James, C. (1987) Some evidence of the unique- Corporate Governance, Institute of Chartered ness of bank loans, J. Financ. Econom. 19, Accountants, England and Wales, 1992. 217–235. Choi, D. Toehold acquisitions, shareholder Jarrell, G. A. and Poulsen, A. B. (1987) Shark wealth and the market for corporate control, repellents and stock prices: The effects of J. Financ. Quantitative Anal. 26(3), Sep. anti-takeover amendments since 1980, De Angelo, H. and De Angelo, L. (1990) J. Financ. Econom. 18, 127–168. Dividend policy and financial distress: An Jensen, M. C. and Meckling, W. H. (1976) empirical investigation of troubled NYSE Theory of the firm: Managerial behaviour, firms, Journal of Finance 45, 1425–1431. agency costs and ownership structure, De Bondt, W. F. M. and Thaler, R. H. (1985) J. Financ. Econom. October, 305–360. Does the stock market overreact?, Journal of Jensen, M. (1989a) Active investors, LBO’s and Finance 40, 793–805. privatisation of bankruptcy, J. Appl. Demsetz, H. and Lehn, K. (1985) The structure Corporate Finance 2, 35–44. of corporate ownership: Causes and conse- Jensen, M. (1989b) Eclipse of the public quences, J. Polit. Econ. 93, 1155–1177. corporation, Harvard Bus. Rev. 61–74, Diamond, D. (1983) Seniority and maturity Sep./Oct. of debt contracts, J. Financ. Econom. 33, Lang, L., Poulsen, A., and Stulz, R. (1995) 341–368. Asset sales, firm performance, and the agency Fama, E. F. and Jensen, M. C. (1993) costs of managerial discretion, J. Financ. Separation of ownership and control, J. Law Econom. 37, 3–37. Econom. 26, 301–325. Mallette, P. and Fowler, K. L. (1992) Effects of Gertner, R. and Scharfstein, D. (1991) A theory board composition and stock ownership on of workouts and the effects of reorganization the adoption of poison pills, Academy Managt law, J. Finance XLVI(4), September, J. 35(5), 1010–1035. 1189–1222. Marris, R. (1964) The Economic Theory of Gilson, S. C. (1989) Management turnover and Managerial Capitalism, Free Press, New York. financial distress, J. Financ. Econom. 25, McConnell, J. J. and Servaes, H. (1990) 241–262. Additional evidence on equity ownership and Gilson, S. C. (1990) Bankruptcy, boards, banks, corporate value, J. Financ. Econom. 27, and bondholders – Evidence on changes in 595–612. 598 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Mikkleson, W. H. Ruback, R. S. (1985) An Schleifer, A. and Vishny, R. W. (1986) Large empirical analysis of the interfirm equity shareholders and corporate control, J. Polit. investment process, J. Financ. Econom. 14, Economy 94, 461–488. 523–553. Schipper,K. and Smith, A. (1986) A comparison Murphy, K. J. and Zimmerman, J. L. (1993) of equity carve-outs and seasoned equity ‘Financial performance surrounding CEO offerings: Share price effects and corporate turnover’, Journal of Accounting and restructuring, J. Financ. Econom. 15, Economics 16, 273–315. 153–186. Myers, S. C. (1977) Determinants of corporate Shivdasani, A. (1993) Board composition, borrowing, J. Financ. Econom. November, ownership structure, and hostile takeovers, 147–176. J. Account. Econom. 16, 167–198. Ofek, E. (1993) Capital structure and firm Slatter, S. (1984) Corporate Recovery: response to poor performance: An empirical Successful turnaround strategies and their analysis, J. Financ. Econom. 34, 3–30. implementation, Penguin. Pound, J. (1988) Proxy contests and the Storey, in Keasey and Watson (1987) J. Banking efficiency of shareholder oversight, J. Financ. and Finance 14(3), Autumn, 335–350. Econom. 20, 237–265. Sudarsanam, S. (1995) Large shareholders, Rajan, R. (1992) Insiders and outsiders: The corporate restructuring and target valua- choice between informed and arm’s-length tion, Paper presented to Financial debt, J. Finance 47, 1367–1400. Management Association Annual Meeting, Rechner, P. L. and Dalton, D. R. (1989) The New York. impact of CEO as board chairperson on Sudarsanam, S. (1996) Large shareholders, corporate performance: Evidence vs. rhetoric, takeovers and target valuation, J. Busin. Academy Managt Exec. 3, 141–143. Finance Account. 23(2), March. Rosenstein, S. and Wyatt, J. H. (1990) Outside Weisbach, M. (1989) Outside directors and directors, board independence, and shareholder CEO turnover, J. Financ. Econom. 20, wealth, J. Financ. Econom. 26, 175–191. 431–460. Chapter 23

Jorge Farinha* DIVIDEND POLICY, CORPORATE GOVERNANCE AND THE MANAGERIAL ENTRENCHMENT HYPOTHESIS: AN EMPIRICAL ANALYSIS

Source: Journal of Business Finance & Accounting, 30(9–10) (2003): 1173–1210.

1. INTRODUCTION that after a critical entrenchment level estimated in the region of 30%, the coeffi- THIS PAPER PROVIDES AN EMPIRICAL ANALYSIS cient of insider ownership changes from of the agency theory explanation for the negative to positive. In addition, the cross-sectional distribution of dividend analysis suggests that directors’ control payouts in the United Kingdom (UK). This over non-beneficial shares managed on perspective asserts that cash payments to behalf of other shareholders (typically shareholders may help to reduce agency the company’s pension fund, charity trusts problems either by increasing the frequency or employee stock ownership plans) can of external capital raising and associated also lead to managerial entrenchment. monitoring by investment bankers and The remainder of this paper is organised investors (Easterbrook, 1984), or by as follows. Section 2 presents a summary eliminating free cash-flow (Jensen, 1986). of previous literature and outlines the Although other theories have been pro- research motivation. Section 3 describes posed to explain cross-sectional dividend the research design, sampling procedures policy (notably those based on signalling and data characteristics. Main empirical and tax clienteles), the existing empirical results and several robustness checks are literature typically finds that the observed reported in Sections 4 and 5, respectively. dividend behaviour is consistent with more The final section summarises the paper and than a single theory, and therefore usually its main conclusions. fails to dismiss alternative explanations. However, the managerial entrenchment hypothesis taken from the agency literature 2. PREVIOUS LITERATURE AND offers a distinctive set of predictions that RESEARCH MOTIVATION cannot be found in other competing stories for the explanation of cross-sectional (i) Agency theory view of dividend policy behaviour. Consistent with dividend policy such hypothesis, this paper finds evidence of a strong U-shaped relationship between Easterbrook (1984) argues that dividends dividend payouts and insider ownership in play a role in controlling equity agency the UK. Specifically, these findings show problems by facilitating primary capital 600 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY market monitoring of the firm’s activities assets to obtain personal benefits, such as and performance.The reason is that higher shirking and perquisite consumption. As dividend payouts increase the likelihood insider ownership increases, agency costs that the firm will have to sell common may be reduced since managers bear a stock in primary capital markets. This in larger share of these costs. However, as turn leads to an investigation of manage- Demsetz (1983) and Fama and Jensen ment by investment banks, securities (1983) point out, managers holding a exchanges and capital suppliers. Studies by substantial portion of a firm’s equity may Baghat (1986), Smith (1986), Hansen and have enough voting power to ensure that Torregrosa (1992) and Jain and Kini their position inside the company is secure. (1999) have recognised the importance of As a result, they may become to a great monitoring by investment bankers in new extent insulated from external disciplining equity issues. Recent theoretical work by forces such as the takeover threat or the Fluck (1998), and Myers (2000) also managerial labour market. Stulz (1988) presents agency-theoretic models of presents a model where high ownership by dividend behaviour where managers pay managers can effectively preclude the dividends in order to avoid disciplining possibility of a takeover,in accordance with action by shareholders. an entrenchment hypothesis. Consistent Additionally, Jensen (1986) sees expected, with this,Weston (1979) reports that firms continuing dividend payments as helping where insiders held more than 30% have to dissipate cash which might otherwise never been acquired in hostile takeovers. have been wasted in non-value maximising Morck et al (1988) and McConnel and projects, therefore reducing the extent of Servaes (1990) find an inverted U-shaped overinvestment by managers. relationship between insider ownership and In Rozeff’s (1982) model, an optimal firm performance in accordance with the dividend policy is the outcome of a trade-off existence of managerial entrenchment between equity agency costs and transac- above a critical level of ownership. tion costs. Consistent with such trade-off This entrenchment hypothesis taken model, Rozeff reports evidence of a strong from the agency literature is particularly relationship between dividend payouts and interesting since it has consequences for a set of variables proxying for agency dividend policy which are distinct from and transaction costs in a large sample other competing theories of dividend composed of one thousand US firms for the behaviour. Specifically, the prediction is period 1974 to 1980. that below an entrenchment level insider A cross-sectional analysis of dividend ownership and dividend policies can be seen policy by Crutchley and Hansen (1989) as substitute corporate governance devices, also shows results consistent with dividend therefore leading to a negative relationship policy acting as a corporate monitoring between these two variables. After such vehicle and with substitution effects critical entrenchment level, however, between dividend payments and two other when insider ownership increases are control mechanisms, managerial ownership associated with additional, entrenchment- and leverage. related, agency costs, dividend policy may become a compensating monitoring (ii) The managerial entrenchment force and, accordingly, a positive relation- hypothesis ship with insider ownership would be observed. This prediction is a distinctive Following Jensen and Meckling (1976), one given that signalling, tax clienteles or when managers hold little equity and other competing theories for dividend shareholders are too dispersed to take behaviour do not predict such U-shaped action against non-value maximisation relationship between insider ownership behaviour, insiders may deploy corporate and dividend policy. DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 601 Schooley and Barney (1994), using US is the notion that dividend policy is set up data, document a U-shaped relationship optimally so as to minimise total agency between dividend yield and CEO ownership. and transaction costs.1 That study suffers from several limitations. Additionally, given that the presence of CEO ownership is not always the best other managerial monitoring devices is measure of insider ownership as in frequent likely to affect the usage of any particular cases board members other than the CEO mechanism for reducing agency costs hold significant amounts of a firm’s equity. (Hart, 1995), allowance is made for the Apart from using a relatively small sample presence of such variables when modelling size (235 firms against our sample in excess dividend policy. In line with previous of 600 firms) Schooley and Barney’s data research, it is possible that not just inverse is confined to large firms with a small relationships, or substitution effects, exist number of cases where the CEO holds between different governance mechanisms. substantial holdings (the average CEO In fact, Rediker and Seth (1995) argue ownership in their sample is 2.5% against that different control mechanisms may our mean insider ownership of 16% (in be complementary, and not just alternative, 1991) and 13% (in 1996). Finally, they do instruments for corporate governance. not control for alternative monitoring The other monitoring mechanisms consid- mechanisms on managers which have been ered in the analysis are debt, analysts recognised in the literature. following, institutional ownership, the Also, no studies have analysed, in the presence of outsiders in the board and context of dividend policy, the possibility compliance with Cadbury (1992) Code of that beneficial and non-beneficial insider Best Practice. holdings may be conducive to entrenchment. The formal basic model used, along with Such possibility has, however,been suggested the variables employed, is described in by previous research from Gordon and sub-section 3 (iii). Pound (1990), Chang and Mayers (1992) and Cole and Mehran (1998), who find (ii) Testable hypotheses evidence that manager’s voting control over Employees Stock Ownership Plans A particular concern of this paper is an (ESOPs) can contribute to managerial attempt to distinguish the agency per- entrenchment. spective of dividend policy from other competing explanations by focusing on the relationship between insider ownership and 3. RESEARCH DESIGN dividend payouts. Following the discussion above, it is likely that managers controlling (i) General model specification large holdings in the firm can to a significant extent insulate themselves from other The general model specified for this analysis disciplining mechanisms. If dividend policy is represented by a single-equation cross- and insider ownership both perform a sectional regression between dividend monitoring role, one might expect that payouts and a set of variables related to before a critical level of entrenchment, Rozeff’s (1982) trade-off argument. This dividend policy and insider ownership could trade-off is between the marginal benefits be substitute monitoring mechanisms. As of dividend payouts (a reduction in agency such, increasing insider ownership would costs) and respective marginal costs (an be accompanied by decreasing dividend increase in the so-called ‘transaction payouts. After a critical level of ownership, costs’). ‘Transaction costs’ here relate to however, larger dividend payouts may be the direct or indirect costs of external needed to compensate for entrenchment- equity financing and potential tax costs related agency costs arising from larger associated with dividend payouts. Implicit (above critical level) insider holdings. 602 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY This leads to the following testable for the U-shaped relationship, but also as a hypothesis: test for the proposition that non-beneficial holdings can be conducive to entrenchment H1: The cross-sectional distribution of by insiders. dividend payouts, all else constant, A final test is to analyse whether is negatively related to beneficial compliance with Cadbury (1992) Code of insider ownership below an entren- Best Practice affects dividend policy. Since chment level of ownership, and the Code was published in 1992, this is positively related above that level. done in the 1996 regression only. That document reviewed the role of corporate Hypothesis H1 can be modelled by using boards in corporate governance and pro- a second-degree polynomial specification vided a set of recommendations of best for the insider ownership variable and test- practice to enhance the accountability and ing the expectation of a negative sign for monitoring function of the directors of UK the first term and a positive one for the firms. After publication of the report, the second. However,the positive sign predicted London Stock Exchange required listed after the critical level of insider ownership firms to state their compliance, and rea- could alternatively be explained by a man- sons for not complying, with the Code’s agerial desire to diversify their wealth by prescriptions. The analysis of the relation- increasing liquidity (Beck and Zorn, 1982) ship between dividend policy and Cadbury when their shareholdings in the firm are (1992) compliance is a novel way of testing large, rather than by an entrenchment the agency explanation for dividend policy explanation.To control for this possibility, it given the Cadbury (1992) recognised role is hypothesised that in a setting where in corporate governance in the UK.2 If insiders control shares on non-beneficial dividend policy is a substitute (complemen- terms at the same time that their beneficial tary) disciplining mechanism to compliance stakes are too small to enable entrench- with the Cadbury (1992) Code, then a ment, a similar U-shaped relationship will negative (positive) association between still be observed between dividend payouts dividend payouts and Cadbury (1992) and total (beneficial or not) holdings compliance should be apparent. Since none with the turning point above the level of of these two alternative possibilities can be beneficial holdings. This is because both ruled out, the null hypothesis is that there categories of holdings can bring control is no association between Cadbury (1992) and, therefore, potential entrenchment. In compliance and dividend payouts. The other words, this hypothesis looks at the corresponding testable hypothesis is thus: possibility that managers may use not just beneficial but also non-beneficial holdings H3: Compliance with the Cadbury to entrench themselves. This leads to the (1992) Code of Best Practice has following testable hypothesis: a zero impact on dividend payouts, all else constant. H2: For low (below critical entrenchment point) levels of beneficial insider ownership, dividend payouts The cross-sectional analyses in this have a U-shaped relationship with paper are reported for 1991 and 1996. total (beneficial and non-beneficial) Using two periods provides a robustness insider ownership, all else constant, check on the results obtained for a single year. It also enables a control for possible as predicted in Hypothesis H1 for beneficial ownership, with a turning structural changes, particularly those poten- point above the level of beneficial tially arising from the intense academic holdings. and public debate on corporate governance in the UK since the early nineties. A five-year Hypothesis H2 can thus also be seen not interval between the two cross-sections just as a test for an alternative explanation was considered because the construction DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 603 of some variables, notably the dividend choice of a five-year period balanced the payout measure, required computations trade-off between the advantage of using a over five years. longer period to provide a more accurate measure of the long term dividend payout (iii) Formal model and ratio, and the costs associated to the variables employed survivorship bias problems arising from the requirement of longer series of data for The empirical mode model used in the each firm in the sample. analysis can be described as follows: Beneficial insider ownership, INSBEN, is defined as the percentage of the company’s MNPAYi 0 1INSBENi shares directly or indirectly controlled by 2 2INSBEN i the firm’s managers, their families or 3GROW1i family trusts (as disclosed in firm’s annual 4GROW2i 5DEBTi reports). Jensen and Meckling (1976) posit a negative relationship between 6VARIABi 7CASHi insider equity ownership and agency costs 8DISPERSi INSTIT while Morck et al. (1988) and McConnel 9 i and Servaes (1990) present evidence NONEXPCT 10 i consistent with that assertion. Within 11IACTi 12SIZEi certain ownership ranges, higher insider 13LANALYSTi ownership can reduce expected agency 14ROAi costs and hence dividend policy may 15ROAxDUMNEGi becomes less important as a monitoring 16CADBURYi vehicle.Therefore, the expected sign for the n coefficient of INSBEN in the regression is jINDUMMYj,i 1 j1 negative. A square term for insider ownership is included, as discussed above, to where: regression coefficients, account for the possibility of managerial iindex of the ith firm, entrenchment, which translates into the jsector index, expectation of a positive sign. This follows, nnumber of sector dummies in particular, arguments and evidence by (2-digit AIC codes), Morck et al (1988) and McConnel and Servaes (1990) that the effect of insider 1 error term. ownership in the reduction of agency costs The dependent variable, MNPAY is the may change its sign after a certain critical dividend payout ratio, constructed as a five- level of ownership. year mean ratio of total ordinary annual The remaining variables are control dividends declared (interim plus final) to factors that either (i) have been observed after-tax earnings (before extraordinary in the literature to influence dividend items).3 Similar to Rozeff (1982), a mean payments, (ii) can be seen as alternative payout ratio is preferred to annual payout or complementary managerial monitoring figures, to reduce the effects of transitory vehicles or (iii) can proxy for the presence and noisy components in short-term earn- of potential agency problems. ings. Thus, the focus is on a measure Past growth (GROW1), defined as the of long-term dividend payout, given the geometric mean rate of growth of the evidence, from a series of studies dating firm’s total assets for the last five years, is back to Lintner (1956), that firms typically included on the grounds that higher historic stabilise dividends around a long-term growth may render dividend policy less payout objective.4 Observations with mean relevant for inducing primary market dividend payout ratios in excess of one or monitoring vehicle given the likelihood that negative are excluded due to the lack of growth may already be inducing external economic significance of these values. The fund raising (and associated monitoring). 604 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Hence, a negative sign is expected.5 A expected sign for its regression coefficient similar argument applies to GROW2, a is positive. variable proxying for future growth oppor- Shareholder dispersion (DISPERS), tunities, measured as the ratio of market represents a measure of stock ownership to book value of equity). Consistent with diffusion. This variable is defined as 100% these assertions, Rozeff (1982) reports a minus the accumulated sum of the owner- negative association between dividend ship by individual entities with more than payouts and variables proxying for past or 3% of the firm’s stock in 1991 or 1996. future growth opportunities. The existence of a large number of (small) The inclusion of DEBT, the book value of shareholders (or a low level of ownership total debt deflated by the market value concentration) increases the potential of equity is mainly motivated by its poten- agency costs given the free-rider problem tial monitoring role on managers. In associated with higher ownership diffusion. particular, financial leverage has been The predicted sign for the coefficient of argued by Jensen and Meckling (1976), DISPERS is thus positive. Jensen (1986) and Stulz (1988), among INSTIT measures total institutional others, to play a role in reducing agency blockholder ownership of the firm’s shares. costs arising from the shareholder-manager Institutional blockholders may act as a conflict. Debt may also have an impact on monitoring device on the firm’s managers, dividends because of debt covenants and as argued by Demsetz and Lehn (1985) related restrictions imposed by debtholders.6 and Schleifer and Vishny (1986), thus Thus a negative sign is expected. dampening in principle the need for high The total variance of a firm’s stock dividend payouts. However, it is possible returns, VARIAB, is also included in the that institutions may influence higher divi- analysis. High fixed operating costs or dend payouts by a company to enhance business risk may affect the firm’s dividend managerial monitoring by external capital payout, all else constant, to the extent that markets, namely if they believe their own these will increase the frequency of costly direct monitoring efforts to be insufficient additional external financing.This is due to or too costly. In this case a complemen- the greater variability in earnings and tarity between these alternative governance funding needs that high operating leverage mechanisms could be apparent. Thus, the or business risk may induce in a firm. The expected sign for this coefficient may same reasoning applies to interest charges, be positive or negative. which are characterised by Rozeff (1982) Following Winter (1977), Fama (1980) as ‘quasi-fixed costs’. Both these operating and Weisbach (1988), the percentage of and financial risks should translate into a non-executives on the firm’s board, NON- high total risk (or variance) of the firm’s EXPCT, is also included to account for stock returns. In addition, as observed by the possibility that such outside directors Holder et al. (1998), transaction costs of may act as management monitors.Thus, the new issues in the form of underwriting fees expected sign for this coefficient is negative, are usually larger for riskier firms. The unless the same observations referred expected sign of the coefficient of VARIAB about INSTIT apply, in which case a in the regression is thus negative. positive relationship might emerge. To the extent that high figures of CASH, IACT is a control variable defined as the defined as a five-year average of cash and cumulative 5-year sum (1987–91 or cash equivalents as a percentage of a firm’s 1992–96) of the amounts shown in a assets represent, or are correlated with, firm’s accounts as irrecoverable Advance a firm’s free cash-flow in the Jensen Corporate Taxation (ACT), deflated by (1986) sense and associated agency costs, total assets.The usage of a cumulative sum expected dividend payouts will be higher. instead of single-year figures is mainly to Thus, greater payouts might be associated account for the often observed situation of with higher figures of CASH and so the a firm declaring in one year surplus ACT DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 605 which eventually is written off in following Jensen et al. (1992) find evidence of a years. If a company has had significant positive association between return on irrecoverable ACT in the past then it is assets and dividend payouts. To the extent, likely that this should translate into a however, that there are links between past higher perceived cost of paying dividends7 profitability and current or expected (or dividend ‘transaction costs’). A negative growth, such measures of profitability may sign is thus expected for this variable’s have a different impact on payouts. For coefficient. instance, past profitability may capture The control variable firm size (SIZE) is information on growth prospects missed by defined as the log of market capitalisation. other variables (namely GROW2), possibly Size may be an important factor not just as because more profitable firms may be more a proxy for agency costs (which can be (or less) likely to grow in the future. In expected to be higher in larger firms) but addition, higher profitability may be evi- also because transaction costs associated dence that agency problems are not very with the issue of securities are also (nega- relevant so that monitoring mechanisms tively) related to firm size as documented, such as dividend policy are less needed. among others, by Smith (1977). However, Therefore, the sign for this control variable as Smith and Watts (1992) point out, the can either be negative or positive. A theoretical basis for an impact of size on problem arises, however,because firms may dividend policy is not strong, and indeed face constraints to pay dividends when some negative relationships have been their earnings are negative. DeAngelo and observed (Allen and Michaely, 1995; and DeAngelo (1990) and DeAngelo et al. Keim, 1985). Therefore, the inclusion of (1992) document that a significant pro- size may be best regarded as a simple portion of firms having losses over a 5 year control variable, with no particular sign period tend to omit their dividends entirely. expectation. Similarly, Baker (1989) finds that an LANALYST is the log of the number important reason cited by firms for not of analysts (ANALYSTS) following a partic- paying dividends is ‘poor earnings’. ular firm (as taken from I/B/E/S). Former Therefore, small or zero dividend payouts research suggests that financial analysts could reflect not high levels of alternative may constitute a source of managerial monitoring mechanisms included in monitoring. Specifically, Moyer, Chatfield equation (1) but simply be the result of and Sisneros (1989) and Chung and Jo negative earnings. Given such possible non- (1996) present evidence consistent with linearities, ROA is included in the analysis the number of financial analysts following along with a dummy (DUMNEG) account- a firm having a negative impact on agency ing for the existence of any negative costs. The expected sign for the impact of earnings during the period used for the analysts following can either be positive or calculation of ROA, as well as an interac- negative, in accordance with the Rediker tive term between such dummy and the and Seth (1995) argument. A logarithmic ROA measure. transformation is used because it is likely The final independent variable is that the impact of an additional analyst CADBURY, which consists of a dummy may become smaller as the number of term taking the value of 1 if the firm states analysts following a firm increases. its full compliance, in the 1996 regression, Return on assets, ROA, is defined as the with the Cadbury (1992) Code of Best mean ratio between after-tax earnings Practice. Since the purpose for the intro- before extraordinary items and total assets duction of this Code was the improvement calculated over a 5-year period (1987–91 of firm’s corporate governance practices, or 1992–96). In general accordance with a an impact may be expected on firm’s signalling perspective (Miller and Rock, dividend payouts. The sign of this impact 1985), dividend payouts may be positively is, however, unclear, as discussed in the related with measures of profitability. statement of Hypothesis H3. 606 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY INDUMMY represents industry dummies sector distribution of LSPD’s non-financial using two-digit AIC–Actuaries Industry constituents (not shown) reveals no obvi- Classification codes published by the ous differences between this and our sam- London Stock Exchange and obtained from ple’s. Descriptive statistics are presented in LSPD. Michel (1979), among others, shows Table 23.2, while Table 23.3 provides a evidence that industry classification may detailed breakdown of insider ownership have an impact on dividend policy, an effect variables. The dependent variable shows which is usually attributed to industry- considerable cross-sectional variation, as related growth opportunities but that also does the insider ownership variables. In can be related to industry-specific level of particular, it should be noticed the large competition or takeover threat. number of firms with relatively high levels of ownership by insiders. For example, in more than 20% (or 152 firms) of the sam- 4. SAMPLE SELECTION AND ple in 1991 beneficial insider ownership DATA SOURCES exceeds 25%, while in 1996 such figure is around 15% (or 92 firms). The sample of firms used for the subsequent analysis was taken from Standard and (ii) Ordinary least-squares (OLS) Poor’s (S&P) Global Vantage Database. results Financial data was obtained from Global Vantage, Datastream and from companies’ Table 23.4 reports the results of cross- annual reports. Market statistics were drawn sectional OLS regressions of dividend pay- from LBS Risk Measurement Service. outs (MNPAY) on the set of variables Ownership data was compiled from compa- defined in sub-section 3(iii). Different spec- nies’ annual reports. Board data was ifications are considered (models 1, 2, 3, 4 obtained from Datastream and companies’ and 5). It can be seen in models 2 to 5, that reports. Information on the number of the insider ownership variable (INSBEN) analysts following a particular firm was and its square are signed as expected (in taken from the I/B/E/S database. models 4 and 5 the p-values are in the region The selection procedure can briefly be of 1% in 1996, and even lower in 1991). described as follows. In a first stage, all Overall, the regressions yield remarkably firms incorporated in the United Kingdom high adjusted R-squares (around 33% in and listed on the London Stock Exchange 1991 and 44% in 1996). Such results are with complete data were taken from the in accordance with the notion that an align- Industrial Active and Industrial Research ment of interests caused by increased levels (Dead) Global Vantage files. Firms with of insider ownership makes dividends less Sector Index Codes (SIC) between 6000 needed for monitoring purposes, but only and 6999 (financials) and between 4800 up to a certain point. Indeed, after a critical and 4941 (regulated utilities) were level of holdings by managers, companies excluded. Also excluded were firms that were feel the need to compensate potential involved in major mergers or demergers in managerial entrenchment with increased the period 1987–91 or 1992–96. The final dividend payouts to shareholders. In other number of firms in the sample are 693 in words, the results are consistent with the 1991 and 609 in 1996. Table 23.1 depicts expected U-shaped relationship between the sector distribution of the final samples. dividend payouts and the level of ownership by managers as predicted in our Hypothesis H1. 5. EMPIRICAL RESULTS Given the link between dispersion and associated potential agency costs, an (i) Descriptive statistics important result to the agency perspective of dividends is also the positive and signifi- The sector distribution of the sample is cant (p-value close to 1%) impact of shown in Table 23.1. Comparison with the shareholder dispersion variable on dividend DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 607 Table 23.1 Sector Distribution of Sample According to AIC-Actuaries Industry Classification Codes

1991 1996

AIC Sector Name Frequency % Frequency %

12 Extractive Industries 6 0.9 4 0.7 15 Oil, integrated 4 0.6 3 0.5 16 Oil exploration and production 14 2.0 9 1.5 21 Building and construction 37 5.3 39 6.4 22 Building materials and merchants 38 5.5 33 5.4 23 Chemicals 23 3.3 21 3.4 24 Diversified industrials 25 3.6 10 1.6 25 Electronic and electrical equipment 58 8.4 39 6.4 26 Engineering 89 12.8 79 13.0 27 Engineering, vehicles 10 1.4 10 1.6 28 Paper, packaging and printing 31 4.5 26 4.3 29 Textiles and apparel 47 6.8 0 0.0 32 Alcoholic beverages 8 1.2 4 0.7 33 Food producers 37 5.3 29 4.8 34 Household goods 21 3.0 58 9.5 36 Health care 15 2.2 15 2.5 37 Pharmaceuticals 6 0.9 8 1.3 41 Distributors 39 5.6 40 6.6 42 Leisure and hotels 25 3.6 16 2.6 43 Media 28 4.0 34 5.6 44 Retailers, food 19 2.7 14 2.3 45 Retailers, general 37 5.3 38 6.2 47 Breweries, pubs and restaurants 14 2.0 15 2.5 48 Support services 45 6.5 50 8.2 49 Transport 17 2.5 15 2.5

Total 693 100.0 609 100.0

payouts, either for 1991 or 1996. In should be mentioned that modelling the economic terms, an increase in dispersion impact of profitability with a ROA variable, of 10 p.p. increases on average the dividend a dummy for negative earnings and an payout ratio in about 1 p.p. interactive term provides a significantly Other variables included in the regres- better fit, either for 1991 or 1996, but sions are, in general terms, either signed as particularly so in 1991 (where adjusted expected or insignificant. An exception to R-square increases from 20.64% in model this is the Irrecoverable Advance Corporate 3 to 32.93% in model 4). Taxes variable, where its transaction cost Consistent with Allen and Michaely role for dividends generated the expecta- (1995) and Keim (1985), a significant tion of a positive sign but the results yield negative relationship between firm size and a positive coefficient (in the 1991 regres- dividend payouts is observed for 1991 and sion only). Such result is, however, consis- 1996. An interesting result also is that full tent with Adedeji (1998), who finds a compliance with the Cadbury (1992) similar association between irrecoverable Report has a positive impact on dividend ACT and dividend payouts and interprets payouts (with a p-value close to 1%) as that evidence as related to firms seeing well as economic significance (compliance irrecoverable ACT as a tax allowance that with Cadbury increases payouts in around can enhance distributable earnings. It also 4% of earnings).8 This result is consistent 608 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 1.792 0.072 1.085 0.567 0.765 0.305 2.003 9.650 146.947 0.059 0.048 0.260 0.576 0.411 0.246 1.171 0.268 3.661 14.081 276.736 0.374 3.575 25.633 217.692 1,034.400 10.744 143.768 693) 609) Summary Descriptive Statistics Summary Descriptive N N ( ( CLOSE 0.084 0.277 1.000 0.000 0.000 0.000 0.000 3.013 7.101 DUMNEG 0.388 0.488 1.000 1.000 0.000 0.000 0.000 0.463 ROA 3.844 12.960 31.414 8.424 5.643 2.199 SIZE2 4.646 1.707 10.391 5.700 4.401 3.457 SIZE1 4.560 1.933 10.585 5.817 4.466 3.178 LANALYST 1.509 0.948 3.332 2.303 1.386 0.693 0.000 MKCAP 720.350 3,055.271 39,544.000 336.000 87.000 24.000 0.310 9.679 105.307 SEQ 228.586 953.304 14,887.000 125.400 33.852 11.411 IACTATANALYSTS 0.120 5.814 559.758 1.119 1,961.070 5.945 32,572.000 22.722 27.000 298.800 0.000 81.517 9.000 31.730 0.000 3.000 0.000 1.000 0.562 9.635 0.000 129.445 1.086 0.167 NONEXPCT 0.424 0.144 0.857 0.500 0.429 0.333 0.000 NONEX 3.174 1.550 10.000 4.000 3.000 2.000 0.000 0.972 1.510 DIR 7.433 2.531 25.000 9.000 7.000 6.000 2.000 1.447 4.585 INSTIT 23.665 16.767 78.770 35.890 21.300 9.990 0.000 0.483 CASHDISPERS 55.273 11.102 21.348 10.981 99.985 70.209 71.247 15.294 54.175 8.025 38.840 3.155 5.838 0.000 0.062 1.886 4.557 VARIAB 37.036 16.394 110.000 46.000 32.000 25.000 16.000 1.264 1.577 GROW2DEBT 1.820 0.819 1.217 11.233 12.664 276.412 2.050 0.354 1.515 0.161 1.123 0.057 0.555 0.000 3.742 24.376 598.737 21.642 GROW1 0.108 0.188 1.921 0.147 0.078 0.018 INSIDER 13.137 17.861 82.861 18.916 4.410 0.417 0.001 1.574 1.565 INSNONB 2.333 7.163 67.489 0.813 0.000 0.000 0.000 5.050 30.115 INSBEN 14.412 19.487 79.570 21.370 5.290 0.410 0.000 1.511 1.179 INSBEN 10.804 15.747 75.243 14.000 3.585 0.347 0.001 1.848 2.692 MNPAY 0.375 0.194 0.997 0.481 0.355 0.247 0.000 0.537 0.440 Panel B: 1991 Sample MNPAY 0.373 0.228 0.999 0.517 0.379 0.205 0.000 0.295 Sample Panel A: 1996 Table 23.2 Variable Mean Std. Dev. Max. Q3 Med. Q1 Min.CADBURY Skew. 0.486 Kurt. 0.500 1.000 1.000 0.000 0.000 0.000 0.056 DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 609 Sum Sum 1.327 0.192 0.275 0.772 Natural Book value of Book value 5-year volatil- 5-year Percentage of the Percentage 3.282 21.119 0.176 rwise. ompany status, otherwise; and zero 0.616 0.374 0.007 5.135 1.519 23.079 0.260 3.220 18.224 UMNEG: Dummy of 1 if at least some 49.127 455.883 9.968 128.868 aua o fAAYT;SIZE1 log of ANALYSTS; Natural Percentage of the firm’s shares owned collectively shares owned of the firm’s Percentage Percentage of external directors on the board; of external Percentage IACT market to book value,market of equity capitalisation as market defined Total debt deflated by market capitalisation; by market debt deflated VARIAB Total Market capitalisation of a firms’s equity as of 31 December; of a firms’s capitalisation SEQ Market ends divided by after-tax earnings before extraordinary items; extraordinary before earnings INSBEN by after-tax ends divided ivid Percentage of firm’s shares owned collectively by institutions with 3% or more of the firm’s stock; with 3% or more of the firm’s by institutions collectively shares owned of firm’s Percentage Percentage of the firm’s shares controlled on non-beneficial terms by board directors; terms on non-beneficial shares controlled INSIDER of the firm’s Percentage book value of total assets;book value DEBT ivided by Number of external directors in the Board; directors in of external Number NONEXPCT 5-year geometric mean rate of growth in total assets; in of growth mean rate geometric 5-year GROW2 5-year mean of the ratio of interim plus final ordinary d ordinary plus final of interim mean of the ratio 5-year Book value of total assets; by analysts; forecasts Book value ahead earnings of one year Number ANALYSTS; LANALYST Number of directors in the board; of directors in Number NONEX DUMNEG 0.227 0.419 1.000 0.000 0.000 0.000 0.000 1.309 -0.286 ROA 6.472 6.949 24.885 9.520 6.749 4.362 SIZE2 4.423 1.714 10.367 5.387 4.228 3.225 0.305 0.510 0.112 SIZE1 3.988 1.882 10.152 5.170 3.829 2.708 LANALYST 1.336 0.999 3.091 2.197 1.386 0.693 0.000 0.055 ATANALYSTS 5.023 470.385 1,740.742 5.501 31,792.000 21.000 218.600 68.556 8.000 25.141 3.000 1.000 1.356 11.042 0.000 167.806 1.019 MKCAP 399.560 1,626.134 25,638.000 176.000 46.000 15.000 0.540 9.487 114.211 SEQ 199.929 757.814 11,561.000 96.790 30.442 10.886 IACT 0.093 0.826 6.926 0.000 0.000 0.000 of the earnings are negative in the 5-year period and 0 otherwise; period the 5-year in are negative of the earnings CLOSE: of 1 if the company has a close the value Dummy taking c full with the Cadbury (1992) Code in of Best Practice,CADBURY: of 1 the company complies the value Dummy taking othe and zero NONEXPCT 0.349 0.170 0.800 0.462 0.375 0.250 0.000 log of the firms’s market capitalisation; market log of the firms’s SIZE2: of total assets; log of the book value Natural ROA: on assets; mean return 5-year D NONEX 2.657 1.716 9.000 4.000 2.000 1.000 0.000 0.803 0.710 DIR 7.339 2.613 21.000 9.000 7.000 6.000 3.000 1.188 2.594 equity; AT INSTIT 22.667 16.415 90.430 32.900 20.360 9.600 0.000 0.746 0.350 DISPERS 50.361 22.333 99.900 65.810 49.160 33.000 0.470 0.163 of consecutive five years of irrecoverable advance tax deflated by total assets; tax deflated advance of irrecoverable years MKCAP of consecutive five CASH 10.022 10.373 70.994 13.282 6.973 2.825 0.000 2.073 5.704 DIR VARIAB 40.329 12.973 92.000 46.000 38.000 31.000 13.000 1.111 1.457 by entities (non-insiders) with less than 3% individual stakes; with less than 3% individual by entities (non-insiders) INSTIT DEBT 0.987 8.247 209.378 0.542 0.248 0.091 0.000 23.611 592.204 ity of stock returns; CASH: by total assets; deflated of cash plus equivalents mean of the ratio 5-year DISPERS GROW2 1.421 0.790 9.123 1.622 1.252 0.957 0.420 3.792 24.175 plus book value of assets minus book value of equity, book value of assets minus plus book value d of INSBEN and INSNONB; GROW1 GROW1 0.201 0.218 2.202 0.251 0.152 0.083 INSIDER 16.578 20.656 79.570 26.640 6.760 0.530 0.000 1.287 0.504 firms’s shares controlled beneficially by board directors; beneficially shares controlled firms’s INSNONB INSNONB 2.166 5.848 50.910 1.150 0.000 0.000CLOSENotes: MNPAY 0.000 4.464 0.121 24.944 0.327 1.000 0.000 0.000 0.000 0.000 2.326 3.421 610 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 1996 1991 1996 Beneficial and Non-beneficial Insider Ownership Statistics Ownership Insider and Non-beneficial Beneficial 75% 2 0.3 1 0.2 0 0.0 0 0.0 70%–75% 6 0.9 1 0.2 0 0.0 0 0.0 65%–70% 14 2.0 2 0.3 0 0.0 1 0.2 60%–65% 12 1.7 6 1.0 0 0.0 0 0.0 50%–55%55%–60% 18 21 2.6 3.0 12 7 2.0 1.1 1 0 0.1 0.0 2 0 0.3 0.0 40%–45%45%–50% 12 8 1.7 1.2 11 14 1.8 2.3 0 2 0.0 0.3 3 2 0.5 0.3 5%–10%10%–15%15%–20%20%–25% 9125%–30% 5730%–35% 2935%–40% 26 25 21 13 13.1 8.2 4.2 3.8 79 3.6 48 3.0 27 1.9 26 18 11 13.0 9 7.9 4.4 36 4.3 3.0 19 1.8 13 1.5 7 4 2 1 5.2 2.7 1.9 33 1.0 13 0.6 0.3 7 0.1 7 2 1 5.4 1 2.1 1.1 1.1 0.3 0.2 0.2 0%–5% 335 48.3 337 55.3 216 31.2 206 33.8 Ownership 1991 Beneficial Beneficial Non-Beneficial Non-Beneficial Table 23.3 Range0% No.Firms 3 % No.Firms 0.4 %Total 0 No.Firms 693 % 0.0 392 No.Firms 100.0 % 609 56.6 331 100.0 693 54.4 100.0 609 100.0 DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 611 with the idea that firms with better internal disciplinary force could eventually be corporate governance rules are also those possible with a smaller share ownership by that use dividend payouts more intensely, insiders. Accordingly, and using a proce- suggesting that these two monitoring forces dure employed by Peasnell et al. (1998), a act as complements rather than substi- dummy variable was created taking the tutes. This is similar in spirit to findings by value of 1 if a company’s measure of size is Laporta et al. (2000), who observe that in above the sample median and zero other- countries where investor protection is wise.This binary variable was then made to greater, dividend payouts tend to be higher interact with the insider ownership vari- as well, suggesting that the legal environ- ables as for 1991 and 1996. Results are ment and dividend policy may complement reported in Table 23.5 (Panel A). Contrary each other in terms of their disciplining to the hypothesis that critical entrench- effects on managers. Thus, Hypothesis H3 ment levels vary according to firm size, of no impact of compliance with Cadbury small and large firms have virtually identi- on dividend policy can be rejected. cal estimated critical entrenchment levels in 1991 (around 32%). In 1996, however, it can be observed that the U-shaped rela- (iii) Critical entrenchment levels tionship between dividend payouts and From the results in Table 23.4, critical insider ownership is confined to larger entrenchment levels can be derived as the firms (above the size median), where it is turning points in the U-shaped relationship significant at the 1% level, with an esti- between dividend payouts and beneficial mated entrenchment level almost identical insider ownership. The estimated critical to that of 1991. The number of firms in entrenchment levels for beneficial insider 1996 where insider ownership levels are ownership are approximately 32% in 1991 higher than 32% is 66 (about 11% of the (model 4) and 25% in 1996 (model 5). sample). Thus, results suggest that an These numbers are intuitively plausible and entrenchment level slightly above 30% is a in line with Weston’s (1979) observation of consistent feature for all firms in 1991 and no hostile takeovers occurring in firms large firms in 1996. From the findings where insiders hold 30% or more of the above it can be inferred that either some equity. structural change affected the U-shaped An important question that could be relationship between insider ownership and asked is whether the number of firms above dividend policy for small firms or that some the estimated critical entrenchment level is empirical problems might be affecting the sufficiently significant to make the results significance of the estimated coefficients reliable. In 1991, 120 firms have beneficial for small firms in 1996. insider ownership in excess of the estimated To analyse this issue in a somewhat dif- critical level of 32%, which corresponds to ferent way, the sample was split in two about 17% of the firms in the sample. In according to size. Regressions were then 1996, 92 firms (15% of the sample) have re-run separately for firms below and above ownership above the 25% threshold. the size median in 1991 and 1996. Results Overall, this suggests that the estimated are reported in Panel B of Table 23.5 and turning points are driven by a non-negligible it can be seen they are very close to those number of observations. presented in Panel A. The hypothesis that size may affect the critical level of entrenchment was investi- (iv) Entrenchment versus liquidity gated as one might expect that in larger firms less ownership would be needed to hypothesis achieve entrenchment. This could happen, A possible alternative explanation to the for instance, because larger firms may be entrenchment hypothesis for the U-shaped more difficult to acquire by means of hos- relationship between dividend payouts is tile takeovers, so that insulation from such the possibility of liquidity motivations. 612 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 0.0002 0.0431 0.0039 0.0780 0.0036 0.7129 0.0001 0.0013 0.0014 0.0340 ** ** ** *** *** ** ** *** 0.0003 0.0087 0.7033 0.0001 0.0014 0.0014 0.0376 0.0042 0.0859 0.0037 ** ** ** *** *** ** ** *** 0.0008 0.0071 0.0080 0.0327 0.0229 0.0031 0.0072 0.6747 0.0001 0.0015 0.0326 * ** ** *** * *** 0.0009 0.0360 0.0190 0.0031 0.0077 0.6942 0.0001 0.0014 0.0340 ** ** ** *** * *** 0.0174 0.0316 0.0076 0.6433 0.0014 0.0335 ** ** *** *** 0.0006 0.0001 0.0000 0.0001 0.0005 0.0001 0.0001 0.0001 0.0007 0.0007 0.6052 0.0001 0.01970.0010 0.0000 0.0049 0.0004 0.1443 0.0012 ** ** *** *** * end Payout Ratio (MNPAY) Ratio end Payout *** *** ivid 0.0010 0.0011 0.5362 0.0001 0.0163 0.0009 0.0044 0.1271 0.0273 0.00460.0028 0.0027 0.0036 0.0024 0.0097 0.0104 * * *** *** *** *** *** *** 0.0006 0.0041 0.1182 0.0237 0.0039 0.5893 0.0001 0.0173 0.0008 0.0018 0.0012 0.0001 0.0368 0.0404 0.0028 0.0045 * ** *** *** ** *** *** *** 0.4217 0.3574 0.1178 0.4111 0.9143 0.9042 0.9454 0.3659 0.3295 0.0006 0.0010 -0.0011 0.0280 0.0394 0.0514 0.0105 0.9980 0.8971 0.9114 0.2784 0.2176 0.2571 0.1878 0.2347 0.4532 0.8703 0.9629 0.9069 0.6390 0.7156 0.1318 0.0252 0.0038 0.0169 0.0240 0.0093 0.6801 0.7366 0.6545 0.7638 0.2237 0.1923 0.0003 0.1856 0.2907 0.2265 0.2278 0.7707 0.5492 0.5852 0.8717 0.4362 0.5485 0.0186 0.5726 0.00190.0001 0.0009 0.0005 0.00010.0001 0.0002 0.5445 0.0001 0.0001 0.0926 0.0001 0.4792 0.0930 0.0730 0.6729 0.0107 0.6092 0.0001 0.0157 0.0001 0.0354 0.0560 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.00010.1021 0.00010.3758 0.0936 0.0001 0.4852 0.0001 0.1237 0.0682 0.5605 0.1174 0.0192 0.2619 0.2239 0.0200 0.1589 0.5419 0.0216 0.2300 0.0166 0.0478 0.2143 0.0445 0.0109 0.0116 0.0146 0.0534 0.0097 0.0129 0.0246 ** *** ** *** *** (1) (2) (3) (4) (1) (2) (3) (4) (5) 1991 1996 ) ) 0.0208 0.0211 0.0163 0.0076 )) 0.0005 0.0566 0.0004 0.0451 0.0003 0.0513 0.0005 0.0470 0.0009 0.0007 0.0008 ) 0.0008 ) ) )/( ) ) )/( )/( ) )/( ) ) Expected Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient ( OLS Regression Results:OLS Regression Dependent All Firms Variable: D 2 IACT ( DISPERS ( DEBT ( VARIAB ( GROW1 ( LANALYST ( INSBEN GROW2 ( CASH ( INSBEN ( INSTITNONEXPCT ( ( Table 23.4 Variable Sign p-value p-value p-value p-value p-value p-value p-value p-value p-value INTERCEPT DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 613 indicate 0.0120 0.0423 0.0163 0.0381 0.0111 0.3086 *** ** *** Percentage of Percentage *** *** *** and *, ** . 2 5-year volatility of stock volatility 5-year 0.0120 0.0357 0.0112 0.3094 *** *** *** *** 0.0331 0.0015 ** Percentage of external directors on the board; of external Percentage *** market to book value,market capitalisa- as market defined Natural log of the firms’s market capitalisation; market log of the firms’s Natural ROA: ients for INSBEN and ients for se; CADBURY: of 1 the company the value Dummy taking 0.0329 s computed as the turning pont where the relationship between the the relationship pont where s computed as the turning *** 0.0290 Total debt deflated by market capitalisation; by market debt deflated VARIAB Total *** Percentage of the firm’s shares owned collectively by entities (non-insiders) with less than by entities (non-insiders) collectively shares owned of the firm’s Percentage aua o fAAYT;SIZE log of ANALYSTS; Natural 0.0190 0.0116 0.2762 0.0158 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 *** after-tax earnings before extraordinary items; extraordinary before earnings INTERCEP:after-tax term; Intercept INSBEN ** *** *** ivided by 0.0142 0.0039 0.0013 0.0001 0.0420 0.0001 0.0001 5-year geometric mean rate of growth in total assets; in of growth mean rate geometric 5-year GROW2 ends d book value of total assets;book value DEBT *** * ivid ivided by 0.0146 * 0.0103 0.2181 0.0833 0.0899 0.0142 0.0022 0.0007 0.0006 0.0001 0.0001 693 693 693 693 609 609 609 609 609 5.462 5.654 5.864 9.711 8.024 7.969 7.909 13.047 12.973 22.54%18.41% 23.68% 19.49% 24.88% 20.64% 36.71% 32.93% 32.89% 28.79% 33.40% 29.21% 33.88% 29.60% 47.21% 43.59% 47.74% 44.06% ) ) ) ) ) Percentage of firm’s shares owned collectively by institutions with 3% or more of the firm’s stock; with 3% or more of the firm’s by institutions collectively shares owned of firm’s Percentage NONEXPCT )/( )/( )/( )/( )/( 5-year mean of the ratio of interim plus final ordinary d ordinary plus final of interim mean of the ratio 5-year 2 R Sum of consecutive 5-years of irrecoverable advance tax deflated by total assets; tax deflated advance of irrecoverable Sum of consecutive 5-years LANALYST end payout ratio (MNPAY) and beneficial insider ownership changes from negative to positive, negative from changes ownership insider coeffic and beneficial by the estimated as implied (MNPAY) ratio end payout test for signif.test for of ind.dummies 2.7495 2.7519 2.2381 2.2657 1.6349 1.7009 1.5779 2.5958 2.6132 -value) 0.0001 0.0001 0.0010 0.0009 0.0344 0.0244 0.0460 0.0001 0.0001 2 p divid complies in full with the Cadbury (1992) Code in of Best Practice,complies otherwise. i and zero (last line) entrenchment level critical The 5-year mean return on assets; mean return 5-year and 0 otherwi DUMNEG: period the 5-year in are negative Dummy of 1 if at least some of the earnings IACT 3% individual stakes;3% individual INSTIT returns; CASH: by total assets; deflated of cash plus equivalents mean of the ratio 5-year DISPERS two-tailed significance at the 10%, significance two-tailed 5% and 1% levels, respectively. Adjusted N the firms’s shares controlled on beneficial terms by board directors; terms on beneficial shares controlled the firms’s GROW1 tion of equity plus book value of assets minus book value of equity, book value of assets minus of equity plus book value tion d F CADBURY ( SIZE ( F ( (No. above Firms level)critical Notes: MNPAY (111) (107) (120) (98) (98) (90) (92) ROA ( ROA x DUMNEG ( R DUMNEG ( Estimated critical Estimated entrenchment level – 33.14% 34.26% 31.99% – 23.93% 24.34% 25.37% 25.25% 614 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY 2 ables control variables,control where Critical Level INSBEN INSBEN ) j i DUMMYLARGE 2 2 COV( 2 INBEN 4 ) j 0. VAR( ) i VAR( DUMMYLARGE -statistics: t ) j i INSBEN 3 STD( 2 2 ) median SIZE.median DUMMYLARGE else 4 4 INSBEN 2 2 2 3 4 1991 EstimatedCritical Level p-value of INSBEN p-value INSBEN 1996 Estimated p-value of p-value of 1 if SIZE STD( 1 2 1 2 and standard deviation, and INSBEN ) Critical Entrenchment Levels:Critical Firms size) median Large (above versus size) median Small (below 1 3 3 variance covariance 1 1 -values are obtained from computing the following STD( Expected coefficient on INSBEN: Expected coefficient small firms: – for Panel B:Panel large or small firms to either the regression restricting from estimates DUMMYLARGE Panel A: estimates from the inclusion of an interactive term between a size dummy and beneficial insider ownership (INSBEN) vari ownership insider A: dummy and beneficial between a size Panel term of an interactive the inclusion from estimates where STD where VAR COV Table 23.5 Panel A All firms 31.99%Large A is: Notes:The Panel used in model MNPAY 31.07% 0.0001 large firms: – for on INSBEN: Expected coefficient small firms: – for large firms: – for P 0.0001 0.0030 25.25% 0.0037 33.40% 0.0157 0.0072 0.0045 0.0251 Small 32.03% 0.0017 0.0011 16.48% 0.4688 0.1989 Large 31.45% 0.0014 0.0044 32.08% 0.0003 0.0181 Panel B Small 30.92% 0.0142 0.0062 20.44% 0.1707 0.0406 DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 615 In firms where insider holdings are relatively without the contribution of non-beneficial high, managers could be tempted to increase holdings.The number of firms for which, in dividend payouts to obtain liquidity in order regressions 3 and 4, total insider ownership to diversify their personal wealth by invest- exceeds the critical turning is around 3% ing elsewhere the cash received without of the sample in either 1991 (16 firms) or reducing their share of the firm. However, 1996 (18 firms). Although the number this liquidity argument would lose most of observations above the turning points is of its power if one could demonstrate a not large, the consistent results across the similar upward swing driven by shareholdings two cross-sections and the significance that were controlled (in terms of voting of the coefficients lend some support to power), but not beneficially owned, by Hypothesis H2, i.e., the proposition that insiders. This is the basis for the statement liquidity is not behind the upward of Hypothesis H2. movement in dividend policy after the To test for this hypothesis, models 4 in turning points. 1991 and 5 in 1996 were re-run with the The results above also offer an interesting restriction that beneficial insider ownership insight on the role of non-beneficial insider is below the estimated entrenchment level ownership that has been little addressed in and by substituting beneficial insider the literature. Specifically, they show that ownership with total (beneficial and non- non-beneficial holdings where insiders can beneficial) holdings by directors. If again a control voting rights (but not cash-flow U-shaped relationship is observed with the rights) can be used as an entrenchment turning point above the maximum level tool, along with their beneficial holdings. allowed for beneficial holdings, one would Such findings are consistent with Gordon conclude that the upward swing in the and Pound’s (1990) and Cole and Mehran’s dividend/insider ownership curve can only (1998) results that managers can use their be driven with the contribution of nonvoting control over ESOPs (Employee beneficial holdings. This would then Stock Ownership Plans) as a management contradict the liquidity hypothesis. entrenchment device against takeovers. Table 23.6 reports the results. They are also in accordance with Chang Regressions from Table 23.4 are re-run and Mayers’ (1992) finding that the usage with the substitution of beneficial insider of ESOPs is especially prone to provoke ownership for total (beneficial and non- entrenchment when insiders already have beneficial) holdings by insiders. The results substantial voting rights. However, our in model 1 (1991) and 2 (1996) show that results are more general than these given the U-shaped relationship between dividend that our data on non-beneficial holdings payouts and insider ownership is still includes holdings owned not just by ESOPs strong for the new definition of insider but also by a number of other entities (like ownership. The new critical entrenchment charity trusts, founder trusts and company points are now 36.43% and 30.47% for pension funds). 1991 and 1996, respectively. Next, in regres- Since our results on the liquidity hypoth- sions 3 and 4, the calculations are repeated esis, although consistent along the two with the exclusion of firms for which benefi- cross-sections, rely on a relatively small cial ownership by insiders (INSBEN) is number of observations above the turning below estimated entrenchment levels (given points, an additional test was made to test that these levels are estimated with a the robustness of the findings. Specifically, degree of error, the restriction on INSBEN if diversification driven-liquidity needs are was arbitrarily set at 1.5 p.p. below the deemed to increase dividend payouts when estimated entrenchment levels).9 One can holdings by insiders are large, then one see that the U-shaped relation is still should expect that the larger the market apparent, with the crucial difference that value of insider holdings, the larger dividend the upward movement in the dividend-insider payouts should be, all else constant. ownership curve cannot now be driven Therefore, a positive relationship between 616 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY striction 0.0001 0.0002 0.0274 0.0615 0.0043 0.0041 0.7419 0.0001 0.0159 0.0011 0.0011 0.0431 * * * * 29% *** *** ** *** 0.0034 0.0008 0.0006 0.0068 0.1376 0.0019 0.6993 0.0001 0.02050.0008 0.0085 * * ** *** 35% ** *** *** 0.0373 0.0363 0.0003 0.0036 0.0804 0.0036 0.7282 0.0001 0.0012 0.00120.0351 0.0001 0.0045 ** ** ** *** *** ** ** *** 0.4541 0.3601 0.2748 0.8826 0.3970 0.7047 0.4165 0.7472 0.2776 0.5005 0.3982 0.6513 0.0005 0.0007 0.0006 0.0001 0.0482 0.0001 0.1562 0.0001 0.01400.0417 0.0005 0.0001 0.0169 0.0159 0.0001 0.0101 0.2189 0.0169 0.1939 0.00010.0001 0.00010.7328 0.0097 0.0001 0.17950.1010 0.05690.6743 0.0001 0.78990.5741 0.0486 0.0751 0.0391 0.0687 0.0202 0.0931 0.8255 0.7416 0.0683 0.0774 0.0083 0.0053 0.1439 0.0014 0.6509 0.0001 0.0197 0.0080 ** *** *** ** *** *** (1) Estim. (2) Estim. (3) Estim. (4) Estim. 1991 1996 1991 1996 ) 0.0069 ) 0.0038 0.0107 ) 0.0002 ) 0.0421 ) )/( ) )/( ) ) ) 0.0008 ) ) )/( )/ ) ( Turning Points (models 3 and 4) Dependent (models Points Variable:Turning MNPAY that INSBEN OLS Regression Results with Insider Ownership Defined as Total (beneficial and non-beneficial) (models 1 and 2) with the Re (models and non-beneficial) (beneficial Total as Defined Ownership Results with Insider OLS Regression 2 IACT ( VARIAB ( LANALYST ( GROW2 ( INSIDER CASH ( DISPERS ( INSIDERGROW1 ( DEBT ( ( INSTITNONEXPCT ( ( Table 23.6 VariableRestriction on INSBEN Expected Sign Coefficient p-value None Coefficient p-value Coefficient p-value Coefficient p-value None INTERCEP DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 617 0.0416 0.0098 0.2851 0.0105 0.0424 ** *** Percentage of firm’s Percentage *** *** *** Sum of consecutive five book value of total assets; book value URY: of 1 the value Dummy taking ivided by l (last line) is computed as the turning l (last line) 0.0218 0.0113 0.2787 0.0147 5-year geometric mean rate of growth in total in of growth mean rate geometric 5-year *** ** *** *** Natural log of the firms’s market capitalisation; market log of the firms’s Natural ROA: 5-year 0.0243 0.0253 0.0383 0.0110 0.3080 0.0120 0.0399 Percentage of external directors on the board; of external Percentage IACT ** *** *** *** *** ends divided by after-tax earnings before extraordinary items; extraordinary before earnings by after-tax INTERCEP:ends divided term; Intercept divid aua o fAAYT;SIZE log of ANALYSTS; Natural 5-year volatility of stock returns; volatility 5-year CASH: deflated of cash plus equivalents mean of the ratio 5-year 0.01030.00010.0001 0.0001 0.0001 0.0001 0.0118 0.0001 0.0001 0.0001 0.0014 0.0001 0.0001 0.0001 0.0001 0.0009 0.0198 0.0114 0.2752 0.0154 9.895 12.899 8.967 11.555 693 609 587 533 37.14%33.39% 47.60% 43.91% 39.00% 34.65% 48.44% 44.25% *** ** *** *** end payout ratio (MNPAY) and insider ownership changes from negative to positive, for negative from changes coefficients by the estimated ownership as implied and insider (MNPAY) ratio end payout ) ) ) ) ) ivid )/( )/( )/( )/( )/( Percentage of the firm’s shares owned collectively by entities (non-insiders) with less than 3% individual stakes; with less than 3% individual by entities (non-insiders) collectively INSTIT shares owned of the firm’s Percentage . 2 5-year mean of the ratio of interim plus final ordinary ordinary plus final of interim mean of the ratio 5-year market to book value, defined as market capitalisation of equity plus book value of assets minus book value of equity, book value of assets minus of equity plus book value to book value, d market capitalisation as market defined Percentage of the firms’s shares controlled on beneficial and non-beneficial terms by board directors; terms and non-beneficial on beneficial shares controlled GROW1 of the firms’s Percentage 2 indicate two-tailed significance at the 10%, significance two-tailed indicate 5% and 1% levels, respectively. R Total debt deflated by market capitalisation; by market debt deflated VARIAB Total *** and 2 ** , INSIDER and INSIDER INSIDER point where the relationship between the d the relationship where point of the company complies in full with the Cadbury (1992) Code in of Best Practice,of the company complies otherwise. and zero entrenchment leve critical The mean return on assets;mean return and 0 otherwise; DUMNEG: period the 5-year in are negative Dummy of 1 if at least CADB some of the earnings years of irrecoverable advance tax deflated by total assets; tax deflated advance of irrecoverable years LANALYST DEBT shares owned collectively by institutions with 3% or more of the firm’s stock; with 3% or more of the firm’s by institutions collectively shares owned NONEXPCT by total assets; DISPERS ROADUMNEGROAxDUMNEG ( ( ( R assets; GROW2 INSIDER Adjusted N SIZE1CADBURY ( F ( level)critical Notes: MNPAY * (124) (99) (16) (18) Estimated critical Estimated entrenchment level(No. above Firms 36.43% 30.47% 37.90% 33.76% 618 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY dividend payouts and the market value of entered in the regressions and this insider holdings should emerge. variable, when significant, is negatively, Accordingly, regressions were re-run not positively, signed. with the inclusion of MKVINS, a variable constructed as the product between SIZE (v) Alternative specifications (the market value of the company) and INSBEN, while keeping in the regression The general reasoning behind the way the the ownership variables INSBEN and variables were defined above was to con- INSBEN2. The null hypothesis to test is sider contemporaneous variables (at 1991 whether MKVINS enters the regression or 1996) in both sides of equation (1). with a significant positive slope. Also, if However, in the definition of the dividend the documented positive coefficient of payout and cash variables, the annual INSBEN2 is due to liquidity rather than figures were observed to be remarkably entrenchment, one would expect that the unstable so in those cases a mean was inclusion of MKVINS would alter the judged, as referred above, to be the best significance of INSBEN2. Unreported estimate for the value of each of those results show that MKVINS has either an variables in 1991 or 1996. Given that the insignificantly different from zero coefficient dividend payout variable was thus defined (in 1991) or a significantly negative one (in as a mean over a five-year period (1987–91 1996), and in this last case the slope of or 1992–96), while some of the variables INSBEN is no longer significant (most (e.g., debt, the number of analysts or own- likely as a result of the correlation between ership variables) relate to either 1991 or MKVINS and INSBEN). As for INSBEN2, 1996, a degree of look-ahead bias could its slope remains positive and significant occur,although the direction of such poten- throughout. Also, an attempt was made to tial bias is undetermined. However, when see if results would change with the alternative definitions of dividend payout omission of INSBEN2, so as to analyse if were attempted (defining it, in the 1991 potential correlation problems between cross-section, as either the mean over the INSBEN2 and MKVINS are affecting the 1989–93 or the 1991–95 period), the significance of MKVINS. Results show, results were similar so any possible look- however, that MKVINS remains insignifi- ahead bias was dismissed as not serious. cantly different from zero in 1991 and Several other robustness checks were significantly negative in 1996.The slope on made to see if the conclusions above were INSBEN in 1996 becomes significantly sensitive to the usage of other specifica- positive but this is most likely biased due to tions or when considering other potentially the misspecification arising from the omis- relevant factors. Regressions were thus sion of INSBEN2. One can thus conclude repeated by excluding firms having close that, once again, the liquidity hypothesis company tax status10 which could affect stated above is contradicted by the data. the relative transaction costs of dividends. To summarise, our tests offer some Results were, however, unchanged. support for the notion that liquidity is A piecewise linear regression was also not behind the U-shaped relationship performed allowing for one turning point in documented between dividend payouts and the vicinity of the estimated critical insider ownership. Results show in fact that entrenchment levels in Table 23.4. The the same relationship can be observed for a hypothesis of changing slopes under this restricted sample where non-beneficial new specification was confirmed with holdings are essential to achieve total significance levels and R-squares very close insider holdings above the critical turning to those under the quadratic specification. points. Secondly, the quadratic term for The usage of switching regimes to analyse INSBEN2 remains significantly positive alternative critical entrenchment points when a variable controlling for the market under the piecewise linear regression did value of insider holdings (MKVINS) is not produce results that could contradict DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 619 the existence of a negative slope followed year periods (1987–91 and 1992–96) and by a positive one after a critical level. Such a considerably large sample (in excess of alternative specifications did not yield, 600 firms), it tests the hypothesis that however, higher R-squares than the insider ownership affects dividend policies quadratic specification used before. Since in a manner consistent with a managerial it is reasonable to think that in general entrenchment perspective, drawn from the terms a piecewise linear regression imposes agency literature. a much stricter structure than the quadratic In line with predictions, and controlling specification used above, our preference for other factors, strong evidence is found goes to this one. that after a critical entrenchment level of Collinearity diagnostics prescribed by insider ownership estimated in the region of Belsley et al. (1980) and VIF (variance 30%, the coefficient on insider ownership inflation factors) analysis were used to changes from negative to positive. observe if multicollinearity problems The hypothesis that liquidity needs on could be obscuring some of the results. the part of insiders are responsible for the The conclusion was that the statistically positive association between dividend pay- insignificant variables in Table 23.4 were outs and insider ownership after the critical not significantly affected by collinearity turning point was also investigated. The problems. conclusion was the rejection of the liquidity In addition, log transformations was explanation given that a similar relation- used in all variables for which skewness ship is also observed when insiders hold was seen as relatively high but again no non-beneficial holdings in addition to relevant departures from former results beneficial holdings that alone are below were observed. Also, since some evidence of the critical turning point. This point was non-normality in the residuals was observed also reinforced when no positive association that could be a symptom of misspecification, was observed between dividend payouts a robust estimation analysis was performed. and the market value of beneficial insider Specifically, a rank regression procedure by holdings. The analysis suggested that hold- which all variables (except dummies) were ings over which insiders have control, but converted into their respective ranks (and not cash-flow rights, can be conducive to INSBEN2 was redefined as the square of entrenchment. the rank of INSBEN). The results of using Consistent with the existence of links this procedure revealed that the insider between corporate governance and dividend ownership variables were still highly signif- policy, compliance with the Cadbury (1992) icant at the 1% level or very close to it. Code of Best Practice was observed to have Finally, no evidence of significant a statistically and economically significant heteroskedasticity was found when using a impact on dividend payouts. Also in accor- White (1980) test, which is also a test for dance with an agency perspective, strong misspecification. In spite of this, White evidence was produced that shareholder (1980) adjusted t-statistics were used but, dispersion has a significant positive impact as expected, did not reveal any significant on dividend policy. departures from the significance levels The main results presented in this paper observed before. vindicate the agency explanation for cross- sectional dividend policy. Some limitations of the analysis should, however, be kept in 7. SUMMARY AND DISCUSSION mind. First, the non-insider ownership data OF FINDINGS does not include ownership levels below 3%. In principle, it is possible that allowing This paper provides an empirical examina- for a finer partition could alter the signifi- tion of the agency theory explanation for cance of some of the ownership variables. the cross-sectional distribution of dividend In addition, the results may not be easily policies in the UK. Using data for two five extrapolated to the smallest firms. Indeed, 620 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY the analysis of interactions between insider 1 In the UK, dividends are proposed by ownership and size for the 1996 sample directors and no dividends can be approved by suggested that in smaller firms a U-shaped shareholders if they exceed the amount proposed relationship between dividend policy and (see Companies Act 1985,Table A, article 102). insider holdings might not hold.11 Finally, 2 For an analysis of the Cadbury (1992) report in the context of UK corporate gover- the results on the usage by managers of nance see Sheikh and Rees (1995). non-beneficial holdings as an entrenchment 3 DeAngelo et al. (1992) point out that, con- vehicle rely on a relatively small number sistent with arguments by Modigliani and Miller of observations above the critical levels of (1959), discarding unusual income items pro- insider ownership, suggesting therefore a vides a better explanation for firm’s dividend degree of caution in the interpretation of decisions. these particular findings. 4 An attempt was also made to use annual A final issue regards the existing literature cross-sections as an alternative to the specifica- on the simultaneous determination, or tion above. As predicted, the noisiness of divi- endogeneity, of several alternative or dend payout ratios in the short term reduced dramatically the overall significance of the complementary corporate governance equation. Results became, however, closer to mechanisms (see for instance Agrawal and those presented in the text (albeit with much Knoeber, 1996). Specifically, and in the lower significance levels), when annual regres- spirit of Jensen et al. (1992), a simultane- sions were restricted so as to exclude firms ous specification for the joint determination whose short-term earnings were more volatile. of dividend policy and other monitoring 5 In a questionnaire survey of companies’ devices with allowance for the entrenchment reasons for not paying dividends, Baker (1989) effects suggested in this paper, might yield observes that growth and expansion through some incremental explanatory power. investment is a reason listed by 76% of the respondents. 6 Baker (1989) documents that 22% of companies inquired on the reasons for paying no NOTES dividends cite debt covenants and restrictions. 7 Surplus ACT can arise when dividends are * The author is from CETE-Centro de Estudos paid in excess of the maximum amount of tax- de Economia Industrial, do Trabalho e da able profits that the UK tax system allowed Empresa, Faculdade de Economia, Universidade ACT to be set against. This surplus can be the do Porto, Portugal. This paper is based on part result of a variety of situations, namely when the of his PhD dissertation at Lancaster University, company pays dividends out of reserves, when UK. He is particularly grateful to his supervi- the tax system allows capital to be written off at sors, Ken Peasnell and Peter Pope, and also to a different rate than that used in the accounts, Sudi Sudarsanam, Steve Young, Sanjay Unni, or when dividends are paid out of foreign Massoud Mussavian, Robin Limmack, Abe de income. Although this surplus can, with some Jong, participants at the 1998 EFA-European limitations, be relieved by carrying it back or Finance Association Meeting, the 1998 forward, permanent differences between divi- BAAICAEW Doctoral Colloquium, the dends paid and taxable profits can occur that Accounting and Finance Seminar Series at lead to structural irrecoverable ACT surplus. See Lancaster University, the 2002 PFN-Portuguese Freeman and Griffith (1993) for a description Finance Network Meeting, the 2002 EFMA- of the mechanics of ACT. European Financial Management Association 8 It should be noted, however,that one cannot Conference, and the 2002 1st International reject the possibility of inverse causality. Conference on Corporate Governance at the 9 Using or benchmarks set at 0.5, 1, or 2 University of Birmingham, for all the helpful p.p. below estimated entrenchment levels comments received. He is thankful to I/B/E/S yielded similar results. International Inc. for allowing the usage of the 10 For a description of this tax condition I/B/E/S database and also gratefully acknowl- see, for instance, Whitehouse et al. (1993, edges the generous financial support received pp. 514–18). from Fundação para a Ciência e Tecnologia/ 11 Also, in a (unreported) more detailed Praxis XXI Programme and from Faculdade de analysis of the sample, comparison between Economia, Universidade do Porto. firms in the LSPD-London Share Price DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 621 Database (excluding utilities and financials) Corporate Dividends’, Financial Management, suggests that the proportion of small firms Vol. 18, pp. 36–76. in the sample is lower than in the LSPD, DeAngelo, H. and L. DeAngelo (1990), although firms from basically all size categories ‘Dividend Policy and Financial Distress: An are present. Empirical Investigation of Troubled NYSE Firms’, Journal of Finance, Vol. 45, pp. 1415–31. REFERENCES Demsetz, H. (1983), ‘The Structure of Ownership and the Theory of the Firm’, Adedeji, A. (1998), ‘Does the Pecking Order Journal of Law and Economics, Vol. 26, Hypothesis Explain the Dividend Payout pp. 301–25. Ratios of Firms in the UK?’, Journal of –––– and K. Lehn (1985), ‘The Structure of Business Finance & Accounting, Vol. 25, Corporate Ownership: Causes and pp. 1127–55. Consequences’, Journal of Political Agrawal, A. and C.R. Knoeber (1996), ‘Firm Economy, Vol. 93, pp. 1155–77. Performance and Mechanisms to Control Easterbrook, F.H. (1984), ‘Two Agency-Cost Agency Problems Between Managers and Explanations of Dividends’, American Shareholders’, Journal of Financial and Economic Review, Vol. 74, pp. 650–59. Quantitative Analysis, Vol. 31, pp. 377–97. Fama, E. and M. Jensen (1983), ‘Separation of Allen, F. and R. Michaely (1995), ‘Dividend Ownership and Control’, Journal of Law and Policy’, in R. Jarrow et al. (eds.), Finance, Economics, Vol. 26, pp. 327–49. Vol. 9 of Handbooks in Operations Research Fluck, Z. (1998), ‘Optimal Financial and Management Science, (North Holland), Contracting: Debt Versus Outside Equity’, pp. 793–837. Review of Financial Studies, Vol. 11, Baker, H. (1989), ‘Why Companies Pay No pp. 383–418. Dividends’, Akron Business and Economic Freeman, H. and R. Griffith (1993), ‘Surplus Review, Vol. 20, pp. 48–61. ACT – A Solution in Sight?’, Fiscal Studies, Beck, P. and T. Zorn (1982), ‘Managerial Vol. 14, pp. 58–73. Incentives in a Stock Market Economy’, Gordon, L. and J. Pound (1990), ‘ESOPs and Journal of Finance, Vol. 37, pp. 1151–67. Corporate Control’, Journal of Financial Belsley, D., E. Kuh and R. Welsh (1980), Economics, Vol. 27, pp. 525–55. Regression Diagnostics (New York, Wiley). Hansen, R.S. and P. Torregrosa (1992), Cadbury Committee (1992), ‘The Financial ‘Underwriter Compensation and Corporate Aspects of Corporate Governance’. Monitoring’, Journal of Finance, Vol. 47, Chang, S. and D. Mayers (1992), ‘Managerial pp. 1537–56. Vote Ownership and Shareholder Wealth: Hart, O. (1983), ‘The Market Mechanism as an Evidence from Employee Stock Ownership Incentive Scheme’, Bell Journal of Plans’, Journal of Financial Economics, Economics, Vol. 74, pp. 42–64. Vol. 32, pp. 103–31. –––– (1995), ‘Corporate Governance, Some Chung, K. and H. Jo (1996), ‘The Impact of Theory and Applications’, The Economic Security Analyst’s Monitoring and Marketing Journal, Vol. 105, pp. 687–89. Functions on the Market Value of Firms’, Holder, M., F. Langrehr and J. Hexter (1998), Journal of Financial and Quantitative ‘Dividend Policy Determinants: An Analysis, Vol. 31, pp. 493–512. Investigation of the Influences of Stakeholder Cole, R. and H. Mehran (1998), ‘The Effect of Theory’, Financial Management, Vol. 3, Changes in Ownership Structure on pp. 73–82. Performance: Evidence from the Thrift Jain, B. and O. Kini (1999), ‘On Investment Industry’, Journal of Financial Economics, Banker Monitoring in the New Issues Vol. 50, pp. 291–317. Market’, Journal of Banking and Finance, Crutchley, C. and R. Hansen (1989), ‘A Test Vol. 23, pp. 49–84. of the Agency Theory of Managerial Jensen, G., D. Solberg and T. Zorn (1992), Ownership, Corporate Leverage and ‘Simultaneous Determination of Insider 622 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY Ownership, Debt and Dividend Policies’, Outside Directors’, Working Paper (Lancaster Journal of Financial and Quantitative University). Analysis, Vol. 27, pp. 247–63. Rediker, K.J. and A. Seth (1995), ‘Board of Jensen, M. (1986), ‘Agency Costs of Free Cash- Directors and Substitution Effects of flow, Corporate Finance and Takeovers’, Alternative Governance Mechanisms’, American Economic Review, Vol. 76, Strategic Management Journal, Vol. 16, pp. 323–29. pp. 85–99. –––– (1993), ‘The Modern Industrial Rozeff, M. (1982), ‘Growth, Beta and Agency Revolution, Exit and the Failure of Internal Costs as Determinants of Dividend Payout Control Systems’, Journal of Finance, Ratios’, Journal of Financial Research, Vol. 48, pp. 831–80. Vol. 5, pp. 249–59. –––– and W. Meckling (1976), ‘Theory of the Schooley, D.K. and L.D. Barney, Jr. (1994), Firm: Managerial Behavior,Agency Costs and ‘Using Dividend Policy and Managerial Ownership Structure’, Journal of Financial Ownership to Reduce Agency Costs’, Journal Economics, Vol. 3, pp. 305–60. of Financial Research, Vol. 17, pp. 363–73. Keim, D.B. (1985), ‘Dividend Yields and Stock Sheikh, S. and W. Rees (eds.) (1995), Corporate Returns: Implications of Abnormal January Governance and Corporate Control (London, Returns’, Journal of Financial Economics, Cavendish Publishing). Vol. 14, pp. 473–89. Shleifer, A. and R. Vishny (1986), ‘Large Laporta, R., F.Lopez-de-Silanes, A. Shleifer and Shareholders and Corporate Control’, R. W. Vishny (2000), ‘Agency Problems and Journal of Political Economy, Vol. 94, Dividend Policies Around the World’, Journal pp. 461–88. of Finance, Vol. 55, pp. 1–33. Smith, C.W., Jr. (1977), ‘Alternative Methods Lintner, J. (1956), ‘Distribution of Incomes of for Raising Capital: Rights Versus Corporations Among Dividends, Retained Underwritten Offerings’, Journal of Financial Earnings and Taxes’, American Economic Economics, Vol. 5, pp. 273–307. Review, Vol. 46, pp. 97–113. –––– and R. Watts (1992), ‘The Investment McConnel, J.J. and H. Servaes (1990), Opportunity Set and Corporate Financing, ‘Additional Evidence on Equity Ownership Dividend and Compensation Policies’, Journal and Corporate Value’, Journal of Financial of Financial Economics, Vol. 32, pp. 263–92. Economics, Vol. 27, pp. 595–612. Stulz, R.M. (1988), ‘Managerial Control of Michel, A. (1979), ‘Industry Influence on Voting Rights: Financing Policies and the Dividend Policy’, Financial Management, Market for Corporate Control’, Journal of Vol. 8, pp. 22–26. Financial Economics, Vol. 20, pp. 25–54. Miller, M. and K. Rock (1985), ‘Dividend Policy Weisbach, M. S. (1988), ‘Outside Directors and Under Asymmetric Information’, Journal of CEO Turnover’, Journal of Accounting and Finance, Vol. 40, pp. 1031–51. Economics, Vol. 20, pp. 431–60. Morck, R., A. Schleifer and R. Vishny (1988), Weston, J. (1979), ‘The Tender Takeover’, ‘Managerial Ownership and Market Mergers and Acquisitions, pp. 74–82. Valuation’, Journal of Financial Economics, White, H. (1980), ‘A Heteroskedasticity- Vol. 20, pp. 293–315. Consistent Covariance Matrix Estimator Moyer, R., R. Chatfield and P. Sisneros (1989), and a Direct Test for Heteroskedasticity’, ‘Security Analyst Monitoring Activity: Econometrica, Vol. 48, pp. 817–38. Agency Costs and Information Demands’, Whitehouse, C., L. Watson and N. Lee (1993), Journal of Financial and Quantitative Revenue Law: Principles and Practices Analysis, Vol. 24, pp. 503–12. (11th ed., Butterworths). Myers, S. (2000), ‘Outside Equity’, Journal of Winter, R. (1977), ‘State Law, Shareholder Finance, Vol. 55, pp. 1005–37. Protection and the Theory of the Peasnell, K., P. Pope and S. Young (1998), Corporation’, Journal of Legal Studies, ‘Managerial Ownership and the Demand for Vol. 6, pp. 251–92. Chapter 24

Christian Leuz†, Dhananjay Nanda and Peter D.Wysocki EARNINGS MANAGEMENT AND INVESTOR PROTECTION: AN INTERNATIONAL COMPARISON*

Source: Journal of Financial Economics, 69(3) (2003): 505–527.

ABSTRACT This paper examines systematic differences in earnings management across 31 countries. We propose an explanation for these differences based on the notion that insiders, in an attempt to protect their private control benefits, use earnings management to conceal firm performance from outsiders. Thus, earnings management is expected to decrease in investor protection because strong protection limits insiders’ ability to acquire private control benefits, which reduces their incentives to mask firm performance. Our findings are consistent with this prediction and suggest an endogenous link between corporate governance and the quality of reported earnings.

1. INTRODUCTION We argue that strong and well-enforced outsider rights limit insiders’ acquisition of THIS PAPER PROVIDES COMPARATIVE EVIDENCE private control benefits, and consequently, on corporate earnings management across mitigate insiders’ incentives to manage 31 countries. At a descriptive level, we find accounting earnings because they have large international differences across several little to conceal from outsiders.This insight earnings management measures, including suggests that the pervasiveness of earnings loss avoidance and earnings smoothing. Our management is increasing in private control descriptive evidence suggests that firms in benefits and decreasing in outside investor countries with developed equity markets, protection. Our empirical results are dispersed ownership structures, strong consistent with this prediction and suggest investor rights, and legal enforcement that investor protection plays an important engage in less earnings management. We role in influencing international differences then delve deeper and present an incentives- in corporate earnings management. based explanation for these patterns. Following Healy and Wahlen (1999), we Based on prior research that identifies define earnings management as the alter- investor protection as a key institutional ation of firms’ reported economic perform- factor affecting corporate policy choices ance by insiders to either mislead some (see Shleifer and Vishny, 1997; La Porta et stakeholders or to influence contractual al., 2000), we focus on investor protection outcomes. We argue that incentives to mis- as a significant determinant of earnings represent firm performance through earnings management activity around the world.1 management arise, in part, from a conflict 624 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY of interest between firms’ insiders and enjoy greater private control benefits and outsiders. Insiders, such as controlling own- hence have stronger incentives to obfuscate ers or managers, can use their control over firm performance. the firm to benefit themselves at the Our analysis is based on financial expense of other stakeholders. Examples of accounting data from 1990 to 1999 for such private control benefits range from over 8,000 firms from 31 countries. To perquisite consumption to the transfer of measure the pervasiveness of earnings firm assets to other firms owned by insiders management in a country, we create four or their families. The common theme, proxies that capture the extent to which however, is that some value is enjoyed corporate insiders use their accounting exclusively by insiders and thus not shared discretion to mask their firm’s economic with non-controlling outsiders. performance. As it is difficult to specify ex Insiders have incentives to conceal their ante which techniques firms use to obfus- private control benefits from outsiders cate firm performance, our earnings because, if these benefits are detected, management proxies are designed to outsiders will likely take disciplinary action capture a variety of earnings management against them (see, e.g., Zingales, 1994; practices such as earnings smoothing and Shleifer and Vishny, 1997). Accordingly, we accrual manipulations. argue that managers and controlling own- We begin with a descriptive country ers have incentives to manage reported cluster analysis, which groups countries earnings in order to mask true firm with similar legal and institutional charac- performance and to conceal their private teristics.Three distinct country clusters are control benefits from outsiders. For exam- identified: (1) outsider economies with ple, insiders can use their financial report- large stock markets, dispersed ownership, ing discretion to overstate earnings and strong investor rights, and strong legal conceal unfavorable earnings realizations enforcement (e.g., United Kingdom and (i.e., losses) that would prompt outsider United States); (2) insider economies with interference. Insiders can also use their less-developed stock markets, concentrated accounting discretion to create reserves for ownership, weak investor rights, but strong future periods by understating earnings in legal enforcement (e.g. Germany and years of good performance, effectively Sweden); and, (3) insider economies with making reported earnings less variable weak legal enforcement (e.g., Italy and than the firm’s true economic perform- India). These clusters closely parallel ance. In essence, insiders mask their private simple code-law and common-law as well control benefits and hence reduce the as regional characterizations used in prior likelihood of outside intervention by work (e.g., La Porta et al., 1997; Ball et al., managing the level and variability of 2000). We find significant differences in reported earnings. earnings management across these three Legal systems protect investors by institutional clusters. Outsider economies conferring on them rights to discipline with strong enforcement display the lowest insiders (e.g., to replace managers), as well level of earnings management and insider as by enforcing contracts designed to limit economies with weak enforcement the insiders’ private control benefits (e.g., La highest level of earnings management.That Porta et al., 1998; Nenova, 2000; is, earnings management appears to be Claessens et al., 2002; Dyck and Zingales, lower in economies with large stock mar- 2002).2 As a result, legal systems that kets, dispersed ownership, strong investor effectively protect outside investors reduce rights, and strong legal enforcement. insiders’ need to conceal their activities.We To examine more explicitly whether therefore propose that earnings manage- differences in earnings management are ment is more pervasive in countries where related to private control benefits and the legal protection of outside investors is investor protection, we undertake a multi- weak, because in these countries insiders ple regression analysis.We measure outside EARNINGS MANAGEMENT AND INVESTOR PROTECTION 625 investor protection by both the extent of differences in institutional factors (e.g., minority shareholder rights as well as the Alford et al., 1993; Joos and Lang, 1994; quality of legal enforcement. Our results Land and Lang, 2002). Our results suggest show that earnings management is negatively that a country’s legal and institutional related to outsider rights and legal enforce- environment influences the properties of ment.These results remain significant after reported earnings. In this regard, our study we control for the endogeneity of investor complements recent work by Ali and protection as well as for differences in Hwang (2000), Ball et al. (2000), Fan and economic development, macroeconomic Wong (2001), and Hung (2001), which stability, industry composition, and firm documents that various institutional characteristics. We also provide direct factors explain differences in the price- evidence that earnings management is earnings association across countries.3 positively associated with the level of However, the price-earnings relation of a private control benefits enjoyed by insiders. country reflects both its prevailing pricing While these results highlight insiders’ mechanism and earnings quality. incentives to manage earnings as a way to Consequently, it is important to understand conceal their private control benefits, we the effect of institutional factors on acknowledge that accounting rules may reported earnings when examining the limit insiders’ ability to manage earnings. relation between stock prices and managed We therefore attempt to control for cross- earnings. country differences in accounting rules that Our empirical findings are subject to potentially affect insiders’ ability to man- several caveats. First, earnings manage- age earnings and find that our results are ment is difficult to measure, especially as it robust to the inclusion of this control. manifests itself in different forms. We Finally, we demonstrate that our results are attempt to address this issue by computing not sensitive to the inclusion or exclusion of several proxies for earnings management any particular country (in particular, the and we obtain consistent results across all U.S.) in our sample. measures. However,our findings are contin- This study builds on recent advances in gent on the ability of these measures to the corporate governance literature on the appropriately and consistently capture role of legal protection for financial market earnings management activities around the development, ownership structure, and world. Second, we acknowledge that other private control benefits (e.g., Shleifer and institutional factors correlated with Vishny, 1997; La Porta et al., 2000). We investor protection may also affect insid- extend this literature by presenting ers’ earnings management incentives. Since evidence that the level of outside investor institutional factors are often complemen- protection endogenously determines the tary, it is difficult to fully control for the quality of financial information reported to potential impact of other factors and to outsiders. These results add to our under- disentangle them from the direct effect of standing of how legal protection influences investor protection. Moreover,the existence the agency conflict between outside of complementarities raises concerns about investors and controlling insiders. Weak endogeneity bias. We attempt to address legal protection appears to result in poor- these concerns with two-stage least quality financial reporting, which likely squares (2SLS) estimation. However, as undermines the development of arm’s the relations among the institutional fac- length financial markets. tors are difficult to model, we acknowledge Our work also contributes to a growing that other endogenous interactions may literature on international differences in still exist. Finally, we note that, holding pri- firms’ financial reporting. Prior research vate control benefits constant, strong has analyzed the relation between earnings investor protection potentially encourages and stock prices around the world, only earnings management because insiders implicitly accounting for international have greater incentives to hide their control 626 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY benefits when faced with higher penalties. median ratio of the firm-level standard While we acknowledge the potential deviation of operating earnings divided by existence of such a penalty effect, the the firm-level standard deviation of cash empirical evidence suggests that it is flow from operations. Scaling by the cash dominated by international differences in flow from operations controls for differ- private control benefits, and thus the ences in the variability of economic per- negative relation between investor protection formance across firms. Low values of this and earnings management prevails. measure indicate that, ceteris paribus, The remainder of the paper is organized insiders exercise accounting discretion to as follows. Section 2 describes the con- smooth reported earnings. struction of our earnings management Cash flow from operations is computed measures. In Section 3, we describe the indirectly by subtracting the accrual sample and provide descriptive statistics. component from earnings because direct Empirical tests and results are presented in information on firms’ cash flows is not Section 4. Section 5 concludes. widely available in many countries. Following Dechow et al. (1995), we compute the accrual component of earnings as 2. EARNINGS MANAGEMENT MEASURES Accrualsit ( CAit Cashit) (CLit STDit This section describes the earnings man- TPit) Depit, (1) agement measures used in our empirical analysis. Drawing on the existing earnings where CAit change in total current management literature (see Healy and assets, Cashit change in cash/cash Wahlen, 1999; Dechow and Skinner, equivalents, CLit change in total cur- 2000), we develop four different country- rent liabilities, STDit change in short- level measures of earnings management term debt included in current liabilities, that capture various dimensions along TPit change in income taxes payable, which insiders can exercise their discretion and Depit depreciation and amortization to manage reported earnings. The four expense for firm i in year t. Changes in measures capture outcomes of insiders’ short-term debt are excluded from accruals earnings management activities and avoid because they relate to financing transac- the problem that stated accounting rules tions as opposed to operating activities. If can be (and often are) circumvented by a firm does not report information on taxes insiders and hence do not reflect firms’ payable or short-term debt, then the change actual reporting practices (see also Ball in both variables is assumed to be zero. et al., 2003). 2.2. Smoothing and the correlation 2.1. Smoothing reported operating between changes in accounting earnings using accruals accruals and operating cash flows Insiders can conceal changes in their firm’s Insiders can also use their accounting dis- economic performance using both real cretion to conceal economic shocks to the operating decisions and financial reporting firm’s operating cash flow. For example, choices. Focusing on insiders’ reporting they may accelerate the reporting of future choices, our first earnings management revenues or delay the reporting of current measure captures the degree to which costs to hide poor current performance. insiders “smooth”, i.e., reduce the variabil- Conversely, insiders underreport strong ity of reported earnings by altering the current performance to create reserves for accounting component of earnings, namely the future. In either case, accounting accru- accruals. The measure is a country’s als buffer cash flow shocks and result in a EARNINGS MANAGEMENT AND INVESTOR PROTECTION 627 negative correlation between changes in 2.4. Discretion in reported earnings: accruals and operating cash flows. A small loss avoidance negative correlation is a natural result of accrual accounting (see, e.g., Dechow, Degeorge et al. (1999) and Burgstahler 1994). However, larger magnitudes of this and Dichev (1997) present evidence that correlation indicate, ceteris paribus, U.S. managers use accounting discretion to smoothing of reported earnings that does avoid reporting small losses. While one not reflect a firm’s underlying economic may argue that managers have incentives performance (see Skinner and Myers, to avoid losses of any magnitude, they only 1999).4 Consequently, the contemporaneous have limited reporting discretion and are correlation between changes in accounting consequently unable to report profits in the accruals and changes in operating cash presence of large losses. Small losses, how- flows is our second measure of earnings ever,are more likely to lie within the bounds smoothing. The accrual and operating cash of insiders’ reporting discretion. Thus, in flow components of earnings are computed each country, the ratio of small reported as in equation (1) and the correlation is profits to small reported losses reflects the computed over the pooled set of firms in extent to which insiders manage earnings each country. to avoid reporting losses. Following Burgstahler and Dichev (1997), the ratio of “small profits” to 2.3. Discretion in reported earnings: “small losses” is computed, for each The magnitude of accruals country, using after-tax earnings scaled by total assets. Small losses are defined to be Apart from dampening fluctuations in firm in the range [ 0.01, 0.00) and small performance, insiders can use their reporting profits are defined to be in the range [0.00, discretion to misstate their firm’s economic 0.01]. In order to reliably compute this performance. For instance, insiders can ratio, we require at least five observations overstate reported earnings to achieve of small losses for a country to be included certain earnings targets or report extraor- in the sample. dinary performance in specific instances, such as an equity issuance (see, e.g., 2.5. Aggregate measure of Dechow and Skinner, 2000). Accordingly, our third earnings management measure earnings management uses the magnitude of accruals as a proxy Finally, to mitigate potential measurement for the extent to which insiders exercise error, we construct an overall summary discretion in reporting earnings. It is com- measure of earnings management for each puted as a country’s median of the absolute country. For each of the four earnings value of firms’ accruals scaled by the management measures, countries are absolute value of firms’ cash flow from ranked such that a higher score suggests a operations. The scaling controls for differ- higher level of earnings management. The ences in firm size and performance. It aggregate earnings management score is should be noted that managers can some- computed by averaging the country times use discretionary accruals to rankings for the four individual earnings increase the informativeness of financial management measures. reports. In fact, the evidence for the U.S. suggests that, on average, managers use their discretion in a way that increases the 3. SAMPLE SELECTION AND informativeness of earnings (e.g., Watts DESCRIPTIVE STATISTICS and Zimmerman, 1986). These findings, however, may be the result of effective out- Our data are obtained from the Worldscope side investor protection and therefore may Database, which contains up to ten years of not extend to countries with weak investor historical financial data from annual protection. reports of publicly traded companies 628 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY around the world. Banks and financial earnings management measures as well as institutions are excluded from the empiri- the aggregate earnings management score. cal analysis.To be included in the sample, a The countries are sorted in descending country must have at least 300 firm-year order based on their aggregate score. observations for a number of accounting The four individual earnings management variables, including total assets, sales, net measures exhibit striking differences across income, and operating income. Each firm countries, but similar patterns in terms of must have income statement and balance their relative magnitudes. The statistics sheet information for at least three consec- of the first measure (EM1) show that utive years. Finally, Argentina, Brazil, and earnings are smoother in Continental Mexico experienced hyperinflation over the Europe and Asia than in Anglo-American sample period and are excluded from the countries, after controlling for the volatility main sample because high inflation may of cash flows. Similarly, large negative unduly affect our earnings management correlations between changes in firms’ measures. However, the results are qualita- accruals and cash flows (EM2) indicate tively unchanged if these countries remain that earnings smoothing is more pervasive in the sample. The final sample consists in, for instance, Greece and Japan than in of 70,955 firm-year observations, across Canada and the U.S. With regard to 31 countries and 8,616 non-financial firms accounting discretion, the third measure for the fiscal years 1990 to 1999. (EM3) shows that the magnitude of firms’ Table 24.1 presents the number of firm- accruals, relative to the magnitude of their year observations per country as well as operating cash flows, is small in the U.K. descriptive statistics for the sample firms and the U.S. compared to Austria, and countries.There is significant variation Germany, and South Korea. Similarly, the in the number of firm-year observations fourth measure (EM4) reveals that across countries due to differences in European and Asian firms exhibit a greater capital market development, country size, degree of loss avoidance than Anglo- and the availability of complete financial American firms.5 accounting data. Note that the U.S. version The earnings management measures of the Worldscope Database includes only are highly correlated and the rankings U.S. firms belonging to the S&P 500 index. corresponding to the four individual However,our results are not sensitive to the measures and the aggregate earnings inclusion of the U.S. (or any particular management score are similar. Factor country). To allow for direct firm size analysis suggests that a single factor comparisons across countries, the median represents the four individual measures. firm’s sales in US$ is reported for each Thus, it seems appropriate to combine country. Based on the large differences in the four measures into a single summary the median firm size across countries, measure of earnings management. Our we scale all financial variables by the results hold for the smoothing and discre- lagged value of total assets. Scaling by tion measures separately, as well as for the other variables such as lagged sales or single factor identified by factor analysis. market value of equity does not affect the The last column of Table 24.2 Panel A results.Table 24.1 also shows a substantial presents a country ranking based on this cross-country variation in capital intensity, aggregate earnings management score, the fraction of manufacturing firms, per showing high ranks for countries such as capita GDP, inflation and volatility Austria, Italy, and South Korea, and low of growth. We address the potentially ranks for countries such as Australia, the confounding effects of cross-country U.K. and the U.S. differences in these variables in subsequent Panel B of Table 24.2 provides descriptive multiple regressions. statistics on the institutional characteris- Panel A of Table 24.2 provides tics of each country in the sample and is descriptive statistics for the four individual sorted based on countries’ aggregate EARNINGS MANAGEMENT AND INVESTOR PROTECTION 629

Table 24.1 Descriptive statistics of sample firms and countries

The full sample consists of 70,955 firm-year observations for the fiscal years 1990 to 1999 across 31 countries and 8,616 non-financial firms. Financial accounting information is obtained from the November 2000 version of the Worldscope Database.To be included in our sample, countries must have at least 300 firm-year observations for a number of accounting variables, including total assets, sales, net income, and operating income. For each firm, we require income statement and balance sheet information for at least three consecutive years.We discard three countries (Chile, New Zealand,Turkey) because of an insufficient number of observations to compute the loss avoidance measure, and three countries (Argentina, Brazil, Mexico) due to hyperinflation. Firm size is measured as total US$ sales (in thousands). Capital intensity is measured as the ratio of long-term assets over total assets.The fraction of manufacturing firms is the percentage of firm-year observations with SIC 2000 to 3999. Average per capita GDP in constant 1995 US$ is computed from 1990 to 1999. Inflation is measured as the average percentage change in consumer prices from 1990 to 1998.Volatility of GDP growth is measured as the standard deviation of the growth rate in real per capita GDP from 1990 to 1998.

Median Median Fraction Per-capita Volatility # Firm- firm size capital of mfg. GDP in Inflation of GDP Country years in US$ intensity firms US$ (%) growth (%)

AUSTRALIA 1,483 233,344 0.425 0.319 20,642 2.62 2.01 AUSTRIA 564 213,101 0.313 0.710 29,287 2.62 1.22 BELGIUM 727 277,510 0.280 0.563 27,357 2.26 1.45 CANADA 3,322 271,287 0.465 0.381 19,687 2.25 1.92 DENMARK 1,235 119,113 0.344 0.573 34,163 2.07 1.23 FINLAND 854 308,974 0.345 0.618 26,296 2.25 4.69 FRANCE 4,404 178,163 0.187 0.548 26,960 2.04 1.42 GERMANY 4,440 336,894 0.282 0.637 30,166 2.51 1.46 GREECE 858 38,305 0.295 0.568 11,393 12.06 1.48 HONG KONG 1,483 167,754 0.376 0.513 21,610 4.10 3.89 INDIA 2,064 63,027 0.409 0.859 374 10.09 2.32 INDONESIA 787 75,502 0.361 0.694 961 13.86 7.26 IRELAND 436 124,021 0.386 0.438 18,707 2.38 3.03 ITALY 1,213 350,380 0.280 0.721 19,025 4.40 1.25 JAPAN 16,475 463,191 0.289 0.583 41,200 1.38 2.29 KOREA (SOUTH) 1,692 452,349 0.382 0.724 10,250 6.28 4.64 MALAYSIA 2,036 81,407 0.403 0.557 4,043 3.97 4.35 NETHERLANDS 1,561 349,909 0.333 0.503 27,037 2.48 1.07 NORWAY 988 104,483 0.356 0.410 33,189 2.46 1.28 PAKISTAN 508 24,907 0.432 0.913 488 10.34 2.25 PHILIPPINES 429 60,814 0.460 0.500 1,093 9.80 2.42 PORTUGAL 460 97,229 0.412 0.545 10,942 6.40 1.68 SINGAPORE 1,100 104,187 0.377 0.472 22,721 2.15 2.66 SOUTH AFRICA 1,043 380,644 0.327 0.445 3,914 10.41 1.92 SPAIN 1,082 333,207 0.424 0.492 15,092 4.43 1.64 SWEDEN 1,384 261,343 0.295 0.505 27,350 3.59 2.29 SWITZERLAND 1,320 377,488 0.394 0.626 44,485 2.51 1.65 TAIWAN 1,001 208,798 0.357 0.809 11,893 3.37 0.80 THAILAND 1,529 55,344 0.433 0.578 2,570 5.50 3.28 UNITED KINGDOM 10,685 109,337 0.335 0.430 19,126 3.95 2.03 UNITED STATES 3,792 3,597,429 0.333 0.556 27,836 3.09 1.64

Mean 2,289 316,756 0.358 0.574 19,028 4.76 2.34 Median 1,235 208,798 0.357 0.557 19,687 3.37 1.92 Min 429 24,907 0.187 0.319 374 1.38 0.80 Max 16,475 3,597,429 0.465 0.913 44,485 13.86 7.26 630 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY inancial her item. her depreciation icates whether oth scaled by lagged rd deviations of oper- rd deviations ) management score operations. EM4 is the num- by lagged total assets) are in by lagged taxes payable) taxes ). short-term debt short-term / EM4 # of SmProfit/ Aggregate earnings | ) of SmLoss ( # earnings management ( earnings Acc | less flow from operations is equal to operating income minus accruals, minus income is equal to operating operations from flow ) or cash total current liabilities total current ) |CFO| ( ( Acc, EM3 ( cash) 0.905 0.567 3.996 20.5 0.867 0.848 3.006 21.5 0.882 0.627 3.000 21.6 0.873 0.547 5.591 22.0 0.898 0.646 2.765 22.5 0.912 0.630 4.154 24.8 0.911 0.745 3.000 25.1 0.922 0.685 3.295 26.8 0.928 0.721 4.077 28.3 0.921 0.783 3.563 28.3 CFO)( total assets). The earnings management ( earnings lagged more ) (OpInc)/ EM2 (CFO)( Earnings smoothing measuresEM1 Earnings discretion measures total current assets total current the number of “small losses” for each country. A firm-year observation is classified as a small profit if net earnings (scaled “small losses” if net earnings of the number as a small profit is classified each country. for observation A firm-year divided by 0.01,0]. interest, reported after income are bottom-line Net earnings taxes, items, special items, extraordinary reserves, and any ot The variables are computed from 70,955 firm-year observations for fiscal years 1990 to 1999 across 31 countries and 8,616 non-f 31 countries 1990 to 1999 across years fiscal for observations 70,955 firm-year are computed from variables The scores for the respective EM measure imply the respective scores for Panel A: Country scores for earnings management measures (Sorted by aggregate management) JAPAN 0.560 higher The aggregate earnings management score is the average rank across all four measures, all four across rank management score is the average earnings aggregate EM1-EM4.The ind the column heading in sign The GERMANY 0.510 the range [ the range SINGAPORE 0.455 by lagged total assets) are in the range [0,0.01]. A firm-year observation is classified as a small loss if net earnings (scaled [0,0.01]. the range total assets) are in as a small loss if net earnings by lagged is classified observation A firm-year ber of “small profits” of ber SWITZERLAND 0.473 total assets). from of the cash flow and the absolute of accruals value of the absolute value ratio EM3 is the country’s median TAIWAN 0.431 expense. EM2 is the country’s Spearman correlation between the change in accruals and the change in cash flow from operations (b operations from expense. cash flow in and the change accruals in between the change correlation EM2 is the country’s Spearman ITALY 0.488 PORTUGAL 0.402 where accruals are calculated as: accruals where ( KOREA (SOUTH)KOREA 0.399 ating income and operating cash flow (both scaled by cash flow and operating income ating GREECE 0.415 AUSTRIA 0.345 firms. Data are obtained from the Worldscope Database (November 2000). Database standa (November of the firm-level firms.Worldscope ratio EM1 is the country’s median the from Data are obtained Table 24.2 EARNINGS MANAGEMENT AND INVESTOR PROTECTION 631 (Table 24.2 continued) 0.928 0.297 1.235 0.861 0.552 3.000 0.849 0.558 3.196 0.740 0.311 1.631 2.0 0.790 0.450 1.486 4.8 0.788 0.371 1.667 5.1 0.759 0.478 2.338 5.3 0.840 0.297 1.667 5.6 0.722 0.556 1.235 5.8 0.764 0.466 2.568 6.8 0.807 0.397 1.802 7.0 0.804 0.555 2.455 8.8 0.818 0.517 2.633 12.0 0.845 0.579 2.370 13.5 0.857 0.578 2.658 14.8 0.875 0.526 2.708 16.0 0.861 0.480 3.313 16.5 0.913 0.513 2.643 17.8 0.868 0.671 3.136 18.3 0.825 0.506 7.200 18.3 0.865 0.514 6.000 18.6 0.867 0.509 6.000 19.1 0.850 0.552 3.545 19.5 0.831 0.677 3.571 19.5 0.722 0.848 7.200 Min 0.345 Standard Deviation 0.100 0.056 0.128 1.413 Median 0.539 Mean 0.541 UNITED STATESUNITED 0.765 AUSTRALIA 0.625 IRELAND 0.607 CANADA 0.649 SOUTH AFRICA 0.643 NORWAY 0.713 SWEDEN 0.621 UNITED KINGDOMUNITED 0.574 PHILIPPINES 0.722 FINLAND 0.555 FRANCE 0.561 MALAYSIA 0.569 DENMARK 0.559 NETHERLANDS 0.491 PAKISTAN 0.508 THAILAND 0.602 INDONESIA 0.481 SPAIN 0.539 INDIA 0.523 HONG KONG 0.451 BELGIUM 0.526 Max 0.765 632 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY s variable is the s variable y minorities to gross y minorities hree countries in our in hree countries et is measured by the median percentage of percentage median judicial system,judicial (2) an the Legal Origin and the Origin the Legal s from zero to five. zero s from Legal population. Each variable 998).The Disclosure Index Legal Outside Investor Legal Important of Ownership Disclosure Continued FRANCE French CD 3 8.7 9.3 0.24 69 MALAYSIA English CM 4 7.7 25.3 0.52 76 DENMARK Scandinavian CD 2 10.0 20.0 0.40 62 PAKISTANNETHERLANDS French English CD CM 2 5 10.0 3.7 19.3 7.5 0.31 0.41 64 NA THAILAND English CM 2 4.9 14.3 0.48 64 sample. measures the inclusion or omission of 90 items in the 1990 annual reports (La Porta et al., the 1990 annual reports of 90 items in (La Porta or omission measures the inclusion 1998); t (NA) for it is not available SPAININDONESIA French French CD CD 2 4 2.9 7.1 4.7 7.2 0.62 0.50 NA 64 INDIA English CM 5 5.6 14.0 0.43 57 HONG KONG English CM 5 8.9 28.8 0.54 69 common shares owned by the largest three shareholders in the ten largest privately owned non-financial firms (La Porta et al., (La Porta firms non-financial owned 1 the ten largest privately in by the largest three shareholders common shares owned BELGIUM French CD 0 9.4 11.3 0.62 61 is ranked such that higher scores indicate a greater importance of the stock importance market. a greater scores indicate such that higher is measured as the is ranked Concentration Ownership JAPAN German CD 4 9.2 16.8 0.13 65 GERMANY German CD 1 9.1 5.0 0.50 62 national product,national to the population, relative domestic of listed firms (2) the number to the of IPOs relative and (3) the number SINGAPORE English CM 4 8.9 28.8 0.53 78 mean rank across three variables used in La Porta et al. La Porta used in three variables across mean rank (1997): held capitalization stock market of the aggregate (1) the ratio b SWITZERLAND German CD 2 10.0 24.8 0.48 68 assessment of rule of law,assessment of rule index. to ten. and (3) the corruption zero from Importance range of Equity Mark The All three variables ITALYTAIWAN French German CD CD 3 1 7.4 7.1 13.3 6.5 0.14 0.60 65 62 Enforcement is measured as the mean score across three legal variables used in La Porta et al. La Porta used in variables three legal is measured as the mean score across Enforcement (1998): of the (1) the efficiency PORTUGAL French CD 3 7.2 11.8 0.59 36 anti-director rights index created by La Porta et al. created by La Porta index rights anti-director (1998); and range rights shareholder measure of minority it is an aggregate KOREA (SOUTH)KOREA German CD 2 5.6 11.7 0.20 62 Panel B: Institutional characteristics of the sample countries (Sorted by aggregate earnings management) of classification The 24.2 Table A of Panel management score tabulated in earnings are sortedCountries based on the aggregate GREECE French CD 2 6.8 11.5 0.68 55 Legal et al., are based on La Porta Tradition (1998). a code-law (common-law) country. CD (CM) indicates Outside Right Investor The Table 24.2 CountryAUSTRIA Legal Origin German Tradition Rights CD 2 Enforcement Equity Market Concentration Index 9.4 7.0 0.51 54 EARNINGS MANAGEMENT AND INVESTOR PROTECTION 633 0.398 0.686 (0.036) variables used in La used in variables index from La Porta from index st three shareholders arnings management arnings ission of 90 items in the of 90 items in ission he number of listed domes- of listed he number f law, and (3) the corruption gher scores indicate a greater scores indicate gher 0.315 0.647 0.396 0.393 0.344 0.568 (0.084) (0.000) six. is measured as the mean score Enforcement Legal 0.522 0.418 0.434 (0.003) (0.028) (0.038) zero to 0.026 0.515 0.291 (0.888) (0.003) (0.058) (0.002) 0.538 (0.002) (0.112) (0.019) (0.015) (0.000) Outside InvestorRights Legal Enforcement Importance of Equity Market Ownership Concentration Disclosure Index 1990 annual reports (La Porta et al.,1990 annual reports (La Porta 1998). AUSTRALIA English CM 4 9.5 24.0 0.28 75 in the ten largest privately owned non-financial firms (La Porta et al., (La Porta firms non-financial owned or om 1998).The measures the inclusion the ten largest privately in Disclosure Index IRELAND English CM 4 8.4 17.3 0.36 NA importance of the stock market. Ownership Concentration is measured as the median percentage of common shares owned by the large of common shares owned of the stockimportance market. percentage is measured as the median Concentration Ownership CANADA English CM 5 9.8 23.3 0.24 74 SOUTH AFRICA English CM 5 6.4 16.3 0.52 70 tic firms relative to the population, relative to the population.tic firms such that hi of IPOs relative and (3) the number is ranked Each variable Ownership Concentration Ownership Porta et al. (1997): (1) the ratio of the aggregate stock market capitalization held by minorities to gross national product, et al. national Porta to gross by minorities held (2) (1997): capitalization stock market of the aggregate (1) the ratio t NORWAY Scandinavian CD 4 10.0 20.3 0.31 74 across three legal variables used in La Porta et al. La Porta used in variables three legal across (1998): system, of the judicial (1) the efficiency (2) an assessment of rule o index. to ten. three zero across from is measured by the mean rank range Importance of the Equity Market The All three variables Importance of Stock Market SWEDEN Scandinavian CDet al. from (1998). and ranges rights shareholder measure of (minority) It is an aggregate 3 10.0 16.7 0.28 83 Legal Enforcement Legal UNITED KINGDOMUNITED English CM 5 9.2 25.0 0.15 78 score is the average rank of all four earnings management measures, earnings of all four rank score is the average EM1–EM4. rights is the anti-director Outside Rights Investor Panel C: Correlation between earnings management and institutional characteristics measures. parentheses) between the following (in levels aggregate The and significance correlations presents Spearman table The e PHILIPPINES French CD 3 3.5 5.7 0.51 65 Outside Rights Investor Aggregate Earnings Management Earnings Aggregate FINLAND Scandinavian CD STATESUNITED English 3 CM 5 10.0 13.7 9.5 0.34 23.3 77 0.12 71 634 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY earnings management scores presented in 4. EMPIRICAL RESULTS Panel A. The institutional variables are drawn from La Porta et al. (1997, 1998). 4.1. Descriptive cluster analysis The Legal Origin and Legal Tradition assignments are presented in columns 2 To provide descriptive evidence on the and 3 of Panel B. The proxy for Outside systematic patterns in earnings management Investor Rights is an anti-director rights across groups of countries with similar insti- index that captures the voting rights of tutional characteristics, we begin with a minority shareholders. The Legal cluster analysis. Our aim is to first identify Enforcement measure for each country is country clusters with similar institutional the average score across three variables: features such as the level of investor protec- (1) an index of the legal system’s effi- tion, stock market development, and owner- ciency; (2) an index of the rule of law; and, ship concentration, and then to examine (3) the level of corruption.The Importance whether earnings management varies across of Equity Markets is measured by a coun- these clusters. This approach, while descrip- try’s average rank based on: (1) the ratio tive in nature, captures interactions among of the aggregate stock market held by institutional factors and documents system- minorities to gross national product; atic patterns in earnings management (2) the number of listed domestic stocks without relying on specific hypotheses. relative to the population; and, (3) the The cluster analysis is based on nine number of IPOs relative to the population. institutional variables from La Porta et al. Ownership Concentration is measured as (1997, 1998). We use those variables prior the median percentage of common shares to the aggregation presented in Table 24.2 owned by the largest three shareholders, in because it is preferable for cluster analysis the ten largest privately owned non-financial to have a large set of variables. However, firms. Finally, the Disclosure Index measures the results are similar if only the five the inclusion or omission of 90 accounting variables from Table 24.2 are used. The items in firms’ 1990 annual reports, and variables are standardized to z-scores, hence captures firms’ disclosure policies at and a k-means cluster analysis with three the country level. distinct country clusters is conducted. Simple correlations among institutional Panel A of Table 24.3 reports the means variables and the aggregate earnings of each institutional variable for each of management score for each country are the three clusters. The first cluster is presented in Panel C of Table 24.2 characterized by large stock markets, low Consistent with our hypothesis, there is a ownership concentration, extensive outsider strong negative correlation between the rights, high disclosure, and strong legal aggregate earnings management measure enforcement. The second and third clusters and both the outside investor rights and show markedly smaller stock markets, enforcement proxies. However, there are higher ownership concentration, weaker also significant correlations between the investor protection, lower disclosure levels, earnings management measure and other and weaker enforcement. Based on institu- institutional factors, suggesting that tional characteristics, we refer to countries earnings management is more pervasive in in the first cluster as “outsider economies.” countries characterized by less developed The countries in the second and third stock markets, more concentrated owner- clusters are referred to as “insider ship and lower disclosure levels. The latter economies,” with the distinction that correlation suggests that firms engaging in countries in the second cluster have signif- earnings management also provide fewer icantly better legal enforcement than disclosures. This finding questions the use countries in the third cluster.While cluster 2 of disclosure indices as exogenous variables seems “in-between” cluster 1 and 3, in prior research. a comparison of the Euclidean distances EARNINGS MANAGEMENT AND INVESTOR PROTECTION 635

Table 24.3 Earnings management and institutional clusters

The table presents results from a k-means cluster analysis using three distinct clusters and nine institutional variables from La Porta et al., (1997, 1998). See Panel B of Table 24.2 for details. The variables are standardized to z-scores. Panel A reports the means of the institutional variables by cluster. Panel B reports the cluster membership for the 31 sample countries based on the cluster analysis performed on the variables in panel A. Countries in each cluster are sorted by the aggregate earnings management score from Panel A in Table 24.2. CD (CM) indicates a code-law (common-law) tradition.This variable is not used in the cluster analysis. Panel C reports the mean aggregate earnings management score for each cluster.The last row reports one-sided p-values for differences in the means of the aggregate earnings management across clusters using a t-test. Panel A: Mean values of institutional characteristics by cluster

Institutional Variables Cluster 1 Cluster 2 Cluster 3

Stock Market Capitalization 0.82 0.46 0.21 Listed Firms 49.56 18.58 9.50 IPOs 4.04 0.55 0.37 Ownership Concentration 0.34 0.37 0.50 Anti-Director Rights 4.50 2.62 2.90 Disclosure Index 74.38 66.67 58.13 Efficiency of Judicial System 9.78 9.04 5.50 Rule of law 9.02 9.07 5.65 Corruption Index 8.80 9.09 5.13 Outsider features ↔ Insider Features

Panel B: Cluster membership of countries

Institutional variables Cluster 1 Cluster 2 Cluster 3

Countries Sorted by Aggregate Earnings Singapore (CM) Austria (CD) Greece (CD) Management Score Hong Kong (CM) Taiwan (CD) Korea (CD) Malaysia (CM) Switzerland (CD) Portugal (CD) UK (CM) Germany (CD) Italy (CD) Norway (CD) Japan (CD) India (CM) Canada (CM) Belgium (CD) Spain (CD) Australia (CM) Netherlands (CD) Indonesia (CD) USA (CM) Denmark (CD) Thailand (CM) France (CD) Pakistan (CM) Finland (CD) Philippines (CD) Sweden (CD) South Africa (CM) Ireland (CM)

Panel C: Pervasiveness of earnings management by cluster

Cluster 1 Cluster 2 Cluster 3

Mean Aggregate Earnings Management Score 10.1 16.1 20.6 Tests of EM differences between clusters C1 vs. C2 C2 vs. C3 C1 vs. C3 (p-values) (0.044) (0.059) (0.003) 636 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY between the cluster centers supports our protection, large stock markets and dispersed interpretation that clusters 2 and 3 are ownership.The third cluster exhibits signif- closer to each other than clusters 1 and 2. icantly higher earnings management than Overall, the results in Table 24.3, Panel A the second cluster, highlighting the salient are consistent with the existence of institu- importance of legal enforcement. tional complementarities. Table 24.3, Panel B shows the cluster membership of the sample countries. 4.2. The role of investor protection: Groupings are consistent with the common- multiple regression analysis and code-law as well as regional distinc- The previous analyses suggest that the tions used in prior research to classify pervasiveness of earnings management is countries (see, e.g., Ball et al., 2000; Ball systematically related to a country’s insti- et al., 2003). As indicated in Panel B, all tutional characteristics. A key question, countries in the first cluster with the excep- however, is: Which institutional factors are tion of Norway have a common-law tradi- primary determinants of earnings manage- tion. The three Southeast Asian countries ment and which are correlated outcomes? (Hong Kong, Malaysia, and Singapore) in We posit that better investor protection this cluster were formerly under British results in less earnings management rule and have inherited parts of the Anglo- because insiders enjoy fewer private control Saxon institutional framework. The fact benefits and hence have lower incentives to that the three East Asian countries have by conceal firm performance from outside far the worst earnings management ratings investors. This hypothesis ties in closely in this group is consistent with Ball et al. with findings in Nenova (2000) and Dyck (2003) who argue that, despite the com- and Zingales (2002), suggesting that mon-law influence, reported earnings do private control benefits decrease in the not exhibit common-law properties (i.e., level of investor protection. The notion of asymmetric timeliness). Fan and Wong investor protection as a key primitive is (2001) present similar findings. In the sec- also reinforced by recent work relating to ond cluster,all countries except Ireland and capital market development (e.g., Beck et South Africa have a code-law tradition. al., 2003), corporate policy choices around This cluster contains most of the Northern the world (e.g., La Porta et al., 2000), and European and Scandinavian countries. The cross-listing in the U.S. (e.g., Doidge et al., third cluster consists of several Asian and 2003; Lang et al., 2003). Consistent with Southern European countries with both this literature, we view low earnings common- and code-law traditions.Thus, the management, large equity markets, and cluster approach suggests that the com- dispersed ownership patterns as comple- mon- and code-law distinction matters only ments and joint outcomes of strong investor when legal enforcement is relatively high, protection. This view is in contrast to La as in the first and second clusters. In the Porta et al. (1997, 1999) who treat the third cluster, for which the quality of legal level of disclosure as an exogenous factor enforcement is low, legal tradition seems in explaining financing and ownership unrelated to cluster membership. patterns. Our results suggest, however, that Panel C of Table 24.3 shows that differ- the quality of reported earnings and finan- ences between the clusters’ average earn- cial disclosure is endogenous and hence a ings management scores are statistically joint outcome. significant. Outsider economies (cluster 1) Our multiple regressions examine the exhibit lower levels of earnings manage- relation between earnings management and ment than insider economies (clusters 2 investor protection. Column 1 of Table 24. 4 and 3). Thus, even after controlling for reports a rank regression using the aggre- interactions among various institutional gate earnings management measure as factors, earnings management appears to the dependent variable. Results show that be lower in economies with strong investor outside investor protection explains a EARNINGS MANAGEMENT AND INVESTOR PROTECTION 637 substantial portion (39%) of the variation is to directly estimate the relation between in earnings management. Outsider rights and earnings management and private control legal enforcement both exhibit a significant benefits, explicitly accounting for the effect negative association with earnings of investor protection on the level of private management. Ordinary least squares (OLS) control benefits.We use a country’s average regressions of the aggregate earnings block premium estimated by Dyck and management score on the unranked Zingales (2002) as a proxy for the level variables yield similar results in this and in of private control benefits. We estimate a subsequent regressions. 2SLS regression of the aggregate earnings The multiple regressions assume, however, management score on the control benefits that outside investor rights and legal proxy using the level of outsider rights and enforcement are exogenous variables. If, on legal enforcement as instruments. The the other hand, outsider protection and results presented in column 3 of Table 24.4 earnings management are simultaneously show that earnings management and private determined, our results suffer from an control benefits exhibit a significantly endogeneity bias. We address this concern positive association as predicted by our by using countries’ legal origins and wealth hypothesis. Similar results are obtained if as instruments for the investor protection the legal origins and per capita GDP are variables as suggested by Levine (1999). used as instruments (as in column 2). While related to the level of investor protection (see La Porta et al., 1998), a country’s legal origin can be considered 4.3. Robustness checks as predetermined and exogenous to our Prior work shows that per capita GDP analysis because the origins of most legal explains differences in financing, ownership, systems are several centuries old and many and payout policies across countries. countries obtained their legal system Consequently, we re-estimate our primary through occupation and colonization. We regressions using contemporaneous per use three dummy variables, indicating capita GDP as an additional explanatory English, French, German, and Scandinavian variable (not reported). While GDP is legal origins, as instrumental variables. In marginally significant in this regression addition, we use a country’s average per (p 0.140), the negative relation between capita GDP — measured prior to our investor protection and earnings manage- sample period, 1980 to 1989 — as an ment is robust to the inclusion of this proxy. instrument because an effective legal infra- Another potential concern is that our structure is costly to create and maintain, results are driven by economic heterogeneity and hence a country’s wealth potentially across countries. Although we control for influences the level of legal enforcement. economic differences across firms by scaling Column 2 of Table 24.4 reports results our earnings management measures by of a 2SLS regression using ranked firms’ operating cash flows, variation in variables. The regression results support industry composition and firm size across our hypothesis that the pervasiveness countries can potentially affect our results. of earnings management decreases in the Since Table 24.1 shows that the fraction of level of investor protection, and suggest manufacturing firms and median firm size that this relation is not driven by the poten- vary considerably across countries, the tial endogeneity of investor protection. regressions are re-estimated using two Finally, we attempt to provide more subsamples comprised exclusively of direct evidence on the hypothesis that manufacturing firms (SIC 2000–3999) insiders’ private control benefits are and medium-size firms from each country, positively related to earnings management. respectively.The medium-size firm subsample In the previous regressions, we employ an also eliminates many multinationals indirect approach by using the investor operating in several institutional settings. protection variables. An alternative approach The regression results for these subsamples 638 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY

Table 24.4 Earnings management, outside investor protection and private control benefits

The table presents coefficients and two-sided p-values (in parentheses) from rank regressions with the Aggregate Earnings Management Measure as the dependent variable, which is created by averaging the ranks of all four earnings management measures, EM1–EM4 (see Table 24.2). Outside Investor Rights are measured by the anti-director rights index from La Porta et al., (1998), which ranges from zero to five. Legal Enforcement is measured as the average score across three legal variables used in La Porta et al., (1998): (1) the efficiency of the judicial system, (2) an assessment of rule of law, and (3) the corruption index. All three variables range from zero to ten. Private Control Benefits are measured at the country level as the average block premium estimated by Dyck and Zingales (2002) based on transfers of controlling blocks of shares. The first column presents a simple rank regression.The second regression is estimated using two-stage least squares. Instrumental variables are the rank of the country’s real per capita GDP averaged from 1980 to 1989, and three binary variables indicating an English, German, French, or Scandinavian legal origin based on the classification in La Porta et al., (1998). The third regression is also estimated using two-stage least squares. The instrumental variables are the Outsider Rights Index and the Legal Enforcement.

Aggregate Earnings Aggregate Earnings Aggregate Earnings Management Measure Management Measure Management Measure -2SLS- -2SLS-

Constant 28.605 31.421 3.128 (0.001) (0.001) (0.463) Outside Investor Rights 0.499 0.641 — (0.001) (0.001) Legal Enforcement 0.289 0.322 — (0.025) (0.025) Private Control Benefits — — 0.931 (0.004) Adjusted R2 0.389 0.359 0.272 Number of Observations 31 31 26

(not reported) are essentially the same investor rights and legal enforcement as those presented in Table 24.4, alleviat- continue to have a significantly negative ing concerns that international differences relation with earnings management. in firm size and industry composition drive our findings. 4.4. The role of other Finally, we are concerned that differences institutional factors in firm characteristics and macroeconomic stability affect our inferences. For instance, While the robustness checks in the previous larger firms have smoother earnings, and section suggest that our findings are not operating leverage is positively related to driven by economic heterogeneity across earnings volatility. Similarly, inflation rates countries, we must still address the concern and growth rate volatility influence the vari- that other institutional variables, which ability of accounting earnings. Consequently, are correlated with investor protection, we re-estimate the regressions using median are responsible for our main findings. firm size, median capital intensity, a coun- In particular, we are concerned about try’s average yearly inflation rate, and the the influence of accounting rules and standard deviation of the real GDP growth firms’ ownership structures on earnings rate as additional controls.The results (not management. reported) are consistent with our original First, accounting rules can both limit a findings in Table 24.4. In particular,outside manager’s ability to distort reported EARNINGS MANAGEMENT AND INVESTOR PROTECTION 639 earnings, and affect the properties of Germany or Italy. Similarly, d’ Arcy (2000) reported earnings. But the extent to which shows that Anglo-American countries have accounting rules influence reported earn- stricter accounting rules with respect to ings and curb earnings management explicit accounting choices than Continental depends on how well these rules are European countries with less effective enforced. Moreover, accounting rules likely investor protection. reflect the influence of a country’s legal Ultimately, however, it is an empirical and institutional framework and are there- matter whether our results are robust to the fore endogenous in our analysis. Countries inclusion of controls for countries’ stated with strong outsider protection are expected accounting rules. To address this issue, we to enact and enforce accounting and secure-estimate the main regression and include rities laws that limit the manipulation an accrual rules index constructed by Hung of accounting information reported to out- (2001) as a control variable. This index siders. Consistent with this view, Enriques captures the use of accrual rules to acceler- (2000) argues that U.K. and the U.S. laws ate the recognition of economic transactions on director self-dealing are stricter and are (e.g., R&D activities or pension plans) in more reliant on disclosure than those in accounting, and it proxies for the extent

Table 24.5 Earnings management and outside investor protection: Controlling for differences in the accounting rules and ownership concentration

The table presents coefficients and two-sided p-values (in parentheses) from rank regressions of the Aggregate Earnings Management Measure on Outside Investor Rights and Legal Enforcement controlling for other institutional factors. Outside Investor Rights is the anti-director rights index from La Porta et al., (1998), which ranges from zero to five. Legal Enforcement is measured as the average score across three legal variables used in La Porta et al. (1998): (1) the efficiency of the judicial system, (2) an assessment of rule of law, and (3) the corruption index. All three variables range from zero to ten.The Accrual Rules variable captures the extent to which accrual rules accelerate the recognition of economic transactions (e.g., R&D activities or pension obligations) in accounting. It is constructed by Hung (2001). Ownership concentration is measured as the median percentage of common shares owned by the largest three shareholders in the ten largest privately owned non-financial firms (La Porta et al., 1998). The regressions in columns 2 and 4 are estimated using two-stage least squares. Instrumental variables are the rank of the country’s real per capita GDP averaged from 1980 to 1989, and three binary variables indicating an English, German, French, or Scandinavian legal origin based on the classification in La Porta et al. (1998).

Aggregate Aggregate Aggregate Earnings Aggregate Earnings Earnings Management Earnings Management Management Controlling for Management Controlling for Controlling for Accounting Rules Controlling for Ownership Accounting Rules -2SLS- Ownership -2SLS-

Constant 30.974 34.591 24.333 47.261 (0.001) (0.001) (0.001) (0.002) Outside Investor 0.285 0.501 0.444 0.774 Rights (0.079) (0.044) (0.003) (0.007) Legal Enforcement 0.297 0.420 0.228 0.571 (0.080) (0.048) (0.101) (0.048) Accrual Rules 0.689 0.425 — — (0.016) (0.313) Ownership Concentration — — 0.151 0.609 (0.302) (0.225) Adjusted R2 0.584 0.468 0.392 0.214 Number of Observations 20 20 31 31 640 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY to which a country’s stated accounting In summary, the regression results are rules are intended to produce timely and consistent with the hypothesis that weak informative reported earnings. outsider protection and private control The results presented in Table 24.5, benefits create incentives to manage column 1, show that the coefficients on the earnings. We acknowledge, however, that accounting rules variable and the outsider institutional factors are complementary rights and legal enforcement variables are and hence difficult to isolate. significant. However,as shown in column 2, the coefficient on the accounting rules variable is insignificant in the 2SLS 5. CONCLUSION regression specification, whereas the investor protection variables remain signifi- This paper documents systematic differences cant. These results support our view that in the level of earnings management across accounting rules are endogenous and 31 countries. We perform a descriptive suggest that investor protection is a more cluster analysis to identify groupings of fundamental determinant of earnings countries with similar institutional charac- management across countries. A related teristics and then show that earnings concern is that the use of earnings for tax management varies systematically across and financial accounting purposes may these institutional clusters. The analysis introduce earnings management and in suggests that outsider economies with particular smoothing incentives unrelated relatively dispersed ownership, strong to investor protection. We therefore re-run investor protection, and large stock the main regression including a proxy for the markets exhibit lower levels of earnings degree of a country’s tax-book conformity management than insider countries with (e.g., Alford et al., 1993; Hung, 2001). In relatively concentrated ownership, weak this regression (not reported), the tax investor protection, and less developed variable is not significant while the results stock markets. for the investor protection variables are As prior work shows that investor similar to those reported in Table 24.4. protection is a key primitive driving corpo- Finally, we examine the incremental rate choices such as firms’ financing and impact of ownership concentration on dividend policies as well as ownership insiders’ earnings management incentives structures, we explore the relation of legal since prior research highlights the relation investor protection and firms’ earnings between firms’ ownership structures and the management practices.The analysis is based properties of reported earnings (e.g., Fan on the notion that insiders, i.e., managers and Wong, 2001; Ball et al., 2003). We and controlling shareholders, have incentives re-estimate our main regressions using to acquire private control benefits. However, a proxy for ownership concentration the ability of insiders to divert resources for constructed by La Porta et al. (1998) as their own benefit is limited by legal systems an additional control variable. Neither the that protect the rights of outside investors. rank regression nor the 2SLS regression As outsiders can only take disciplinary presented in columns 3 and 4 of Table 24.5 actions against insiders if outsiders detect indicate any incremental explanatory the private benefits, insiders have an incen- power of the ownership variable. Thus, tive to manipulate accounting reports in while differences in ownership concentra- order to conceal their diversion activities. tion may be related to cross-sectional Thus, we expect that earnings management variation in earnings management within a decreases in legal protection because, country, our country-level tests suggest when investor protection is strong, insiders that average ownership patterns are not a enjoy fewer private control benefits and primary determinant of systematic earnings consequently incentives to mask firm management across countries. performance are moderated. EARNINGS MANAGEMENT AND INVESTOR PROTECTION 641 Consistent with this hypothesis, the 1 While the investor protection literature regression results show that earnings acknowledges the importance of accounting management is negatively associated with information, it typically treats the quality of the quality of minority shareholder rights this information as exogenous and does not and legal enforcement. The findings high- distinguish between stated accounting rules and firms’ actual reporting practices (e.g., La Porta light an important link between investor et al., 1998). protection and the quality of accounting 2 Outsiders are also expected to price protect earnings reported to market participants, themselves, leading to more internal financing, and complement both finance research that smaller arm’s length financial markets and treats the quality of corporate reporting higher cost of outside capital (see, for example, as exogenous and accounting research that La Porta et al., 1997). Bhattacharya et al. documents systematic patterns in the (2002) replicate our earnings management relation between stock returns and measures and provide evidence that firms’ accounting numbers. earnings management activities appear to be Our findings are robust to the inclusion priced in capital markets. 3 See also Basu et al. (1998) and Hope of controls for country wealth, economic (2003) relating the properties of analyst heterogeneity across countries, and inter- forecasts to institutional factors. national differences in accounting rules and 4 As accounting systems likely underreact to ownership concentration. They should economic shocks, insiders using accruals to nevertheless be interpreted cautiously as signal firm performance induce on average a earnings management is difficult to measure less negative (and in specific cases even and the theoretical relations among institu- positive) correlation with cash flows. tional factors are not yet well understood 5 Our loss avoidance results may appear to and hence difficult to disentangle. contradict the finding of Brown and Higgins (2001) that earnings surprise management is more pronounced in the US than in other countries. However, the two findings are compatible. NOTES Brown and Higgins (2002) show that US firms engage in more expectations management, * This paper was previously circulated and i.e., downward guidance of analysts, to meet or presented under the title “Earnings beat analysts’ earnings forecasts, rather than Management Around the World’’ at the 2000 earnings management. European Accounting Association meetings in Munich. It has also benefited from presentations at the Duke/UNC Fall Camp, the EAA meetings in Athens, the European Finance Association REFERENCES meeting in Berlin, the Financial Economics and Accounting conference at the University of Alford, A., Jones, J., Leftwich, R., Zmijewski, M., Michigan, Goethe University Frankfurt, MIT, London Business School, New York University, 1993. The relative informativesness of University of Vasa, University of Virginia, the accounting disclosures in different countries. Wharton School, and the College of William Journal of Accounting Research 31 (Suppl.), and Mary. We are grateful to Stan Baiman, 183–221. Sudipta Basu, Phil Berger,Larry Brown,Willem Ali, A., Hwang, L., 2000. Country-specific Buijink, Elroy Dimson, Simeon Djankov, factors related to financial reporting and Richard Frankel, Wayne Guay, Juha Kinnunen, the value relevance of accounting data. David Larcker, Christian Laux, Martien Journal of Accounting Research 38, 1–23. Lubberink, Peter Pope, Bill Schwert, Weiling Ball, R., Kothari, S., Robin, A., 2000.The effect Song, Ross Watts, and especially Bob of international institutional factors on Holthausen and an anonymous referee for helpful comments. properties of accounting earnings. Journal of † Corresponding author. Tel.: 1-215-898- Accounting and Economics 29, 1–52. 2610; fax: 1-215-573-2054. E-mail address: Ball, R., Robin, A., Wu, J., 2003. Incentives [email protected] (C. Leuz). versus standards: properties of accounting 642 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY income in four East Asian countries. Journal Enriques, L., 2000. The law on company of Accounting and Economics, forthcoming. directors’ self-dealing: a comparative analysis. Basu, S., Hwang, L., Jan, C., 1998. International International and Comparative Corporate variation in accounting measurement rules Law Journal 2, 297–314. and analysts’ earnings forecast errors. Journal Fan, J., Wong, T., 2001. Corporate ownership of Business, Finance and Accounting 25, structure and the informativeness of account- 1207–1247. ing earnings in east asia. Journal of Beck, T., Demirgüe-Kunt, A., Levine, R., 2003. Accounting and Economics 33, 401–426. Law, politics, and finance. Journal of Healy, P., Wahlen, J., 1999. A review of the Financial Economics, forthcoming. earnings management literature and its Bhattacharya, U., Daouk, H., Welker, M., 2002. implications for standard setting. Accounting The world price of earnings opacity. Unpub- Horizons 13, 365–383. lished working paper, Indiana University. Hope, O., 2003. Disclosure practices, enforcement Brown, L., Higgins, H., 2001. Managing of accounting standards, and analysts’ earnings surprises in the U.S. versus 12 other forecasts: an international study. Journal of countries. Journal of Accounting and Public Accounting Research 41, 235–272. Policy 20, 373–398. Hung, M., 2001. Accounting standards and Brown, L., Higgins, H., 2002. Managers’ value relevance of financial statements: an guidance of analysts: international evidence. international analysis. Journal of Accounting Unpublished working paper, Georgia State and Economics 30, 401–420. University and Worcester Polytechnic Institute. Joos, P., Lang, M., 1994. The effects of Burgstahler, D., Dichev, I., 1997. Earnings accounting diversity: evidence from the management to avoid earnings decreases and European Union. Journal of Accounting losses. Journal of Accounting and Economics Research 34(Suppl.), 141–168. 24, 99–129. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Claessens, S., Djankov, S., Fan, J., Lang, L., Vishny,R., 1997. Legal determinants of external 2002. Disentangling the incentive and finance. Journal of Finance 52, 1131–1150. entrenchment effects of large shareholdings. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Journal of Finance 57, 2741–2772. Vishny, R., 1998. Law and finance. Journal of d’Arcy, A., 2000.The degree of determination of Political Economy 106, 1113–1155. national accounting systems — an empirical La Porta, R., Lopez-de-Silanes, F., Shleifer, A., investigation. Schmalenbach Business Review Vishny, R., 1999. Corporate ownership 52, 45–67. around the world. Journal of Finance 54, Dechow, P., 1994. Accounting earnings and 471–517. cashflows as measures of firm performance: La Porta, R., Lopez-de-Silanes, F., Shleifer, A., the role of accounting accruals. Journal of Vishny, R., 2000. Investor protection and Accounting and Economics 18, 3–42. corporate governance. Journal of Financial Dechow, P., Skinner, D., 2000. Earnings man- Economics 58, 3–27. agement: reconciling the views of accounting Land, J., Lang, M., 2002. Empirical evidence on academics, practitioners, and regulators. the evolution of global accounting. Accounting Accounting Horizons 14, 235–250. Review 77(Suppl.), 115–133. Dechow, P., Sloan, R., Sweeney, A., 1995. Lang, M., Raedy, J., Yetman, M., 2003. How Detecting earnings management. The representative are cross-listed firms? An Accounting Review 70, 193–225. analysis of firm performance and accounting Degeorge, F., Patel, J., Zeckhauser, R., 1999. quality. Journal of Accounting Research 41, Earnings manipulation to exceed thresholds. 363–386. Journal of Business 72, 1–33. Levine, R., 1999. Law, finance and economic Doidge, C., Karolyi, A., Stulz, R., 2003.Why are growth. Journal of Financial Intermediation firms listed in the U.S. worth more? Journal 8, 8–35. of Financial Economics, forthcoming. Nenova, T., 2000. The value of corporate votes Dyck, A., Zingales, L., 2002. Private benefits and control benefits: a cross-country analysis. of control: an international comparison. Unpublished working paper, Harvard Unpublished NBER working paper (8711). University. EARNINGS MANAGEMENT AND INVESTOR PROTECTION 643 Shleifer, A., Vishny, R., 1997. A survey of Watts, R., Zimmerman, J., 1986. Positive corporate governance. Journal of Finance 52, Accounting Theory. Prentice-Hall, Englewood 737–783. Cliffs, NJ. Skinner,D., Myers, L., 1999. Earnings momentum Zingales, L., 1994.The value of the voting right: and earnings management. Unpublished a study of the Milan stock exchange experi- working paper, University of Michigan and ence. Review of Financial Studies 7, University of Illinois. 125–148.

Part 6 On takeover as disciplinary mechanism

INTRODUCTION

N PART 3 WE COLLECTED PAPERS THAT EXAMINE the relationship between Idifferent corporate control devices and top management turnover as a manifestation of their disciplinary effect. The control devices are generally internal to the control or ownership structure of the underperforming firms. In this part, we present papers that deal with an external control device.This is the market for control in which management teams compete for control of corporate assets. Manne (1965) was the first to conceptualise the market for corporate control as a managerial disciplinary device with underperforming firms becoming targets of bidders with presumably greater ability to correct underperformance and create greater value for target shareholders. Hostile takeover is the means by which corporate control is wrested from the underperforming target managers. In theory such an external disciplinary device may be considered redundant if the internal control mechanisms are effective.The incidence of a hostile takeover may thus be an indictment of the failure of the internal controls. This argument pre-supposes that internal controls and hostile takeover are substitutes. A contrary perspective is that efficient internal controls facilitate hostile takeovers by preventing the entrenched incumbent management at the target firm from raising the barricades against managerial change. In this view hostile takeover and robust internal controls are complementary tools serving the same purpose. The papers in this part deal with the role of hostile takeovers as a disciplinary device and how governance regimes in certain countries substitute for them or facilitate them. Rezaul Kabir, Dolph Cantrijn and Andreas Jeunink (Ch.25) empirically examine the relationships between a firm’s takeover defences, its ownership structure and stock returns in Holland. Analysing data of Dutch listed companies, they find that these firms increasingly adopt multiple anti-takeover defences when such firms are characterised by relatively lower ownership concentration.The evidence supports the hypothesis that more concentrated ownership of shares provides more effective monitoring of managers. Issue of preferred shares has recently been the most widely adopted anti-takeover defence mechanism in the Netherlands and its impact on shareholders’ wealth is also analysed. The adoption of stringent anti-takeover measures (e.g. the issue of preference shares to friendly investors) leads to a mixed reaction from the stock market. While in the period immediately prior to the issue the reaction is favourable, in the post-issue period it is negative. It appears that the move to adopt the defensive measure signals an impending takeover bid thereby raising the value of the issuing firm, but the actual adoption reduces the chances of such a bid thereby leading to value decline. Another European country with a similar governance and ownership structure and where hostile takeovers are rare is Germany. However, Tim Jenkinson and Alexander 646 ON TAKEOVER AS DISCIPLINARY MECHANISM Ljungqvist (Ch.26) provide clinical evidence to show that the German market for corporate control and the governance system are characterised by both more active and more hostile deals than previously believed. Their study provides a complete breakdown of ownership and takeover defence patterns in German listed companies and finds highly fragmented (but not dispersed) ownership in non-majority controlled firms. The paper documents how the accumulation of hostile stakes can be used to gain control of target companies given these ownership patterns.The article also suggests an important role for banks in helping predators accumulate, and avoid the disclosure of, large stakes. Jens Köke (Ch.27) also focuses on Germany and examines whether changes in ultimate firm ownership (control) play a disciplinary role in a bank-based economy. Germany is the prototype of a bank-based system and the study finds that poor performance makes a change in control more likely. This suggests a disciplinary role for managerial control. Tight shareholder control acts as a substitute for control changes beside strong creditor control as a complement for shareholder control. Following a change in control, management turnover increases, but not as a consequence of poor performance, and performance does not improve significantly. These findings are inconsistent with a disciplinary role of the market for corporate control in bank-based Germany. Julian Franks and Colin Mayer (Ch.28) test an important assumption about the motivation for hostile takeovers (i.e. a superior management team seeks to take control of poorly performing target firms and replace the inefficient incumbent management). Their paper examines the disciplining function of hostile takeovers in the UK in 1985 and 1986. It reports evidence of high board turnover and significant levels of post-takeover restructuring. Large gains are anticipated in hostile bids as reflected in high bid premiums. However, there is little evidence of poor performance prior to bids, suggesting that the high board turnover does not derive from past managerial failure. Hostile takeovers do not therefore perform a disciplining function. Instead, rejection of bids appears to derive from opposition to post-takeover redeployment of assets and renegotiation over the terms of bids. Overall there is some evidence that market for corporate control is active in both Germany and the UK but in Germany it is minority stakes that are traded, whereas in the UK trading it is majority control of target firms. There is mixed evidence regarding the disciplinary motivation for hostile bids. There is also mixed evidence concerning the substitutability or complementarity of internal and external control devices.

REFERENCE

Manne, Henry G. (1965), ‘Mergers and the market for corporate control’ Journal of Political Economy, 73: 110–120. Chapter 25

Rezaul Kabir, Dolph Cantrijn and Andreas Jeunink TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS IN THE NETHERLANDS: AN EMPIRICAL ANALYSIS

Source: Strategic Management Journal, 18(2) (1997): 97–109.

ABSTRACT This study empirically examines the relationships between a firm’s takeover defenses and its ownership structure and stock returns. Analyzing data of Dutch listed companies, we find that multiple antitakeover defenses are increasingly adopted when firms are characterized by relatively lower ownership concentration. The evidence supports the hypothesis that more concentrated ownership of shares provides more effective monitoring of managers. As defense by issuing preferred share has recently been the most widely adopted mechanism in the Netherlands, its impact on shareholders’ wealth is also analyzed. We observe the presence of two opposing effects of this antitakeover measure.

INTRODUCTION toward developing diverse tactics to defend against unfriendly takeovers. As a result, HOSTILE TAKEOVER BIDS ARE RARE IN THE the external market for corporate control Netherlands, and were successful, at most, plays a diminished disciplinary role in the on a few occasions.The reason is that stock Netherlands. An issue deserving investigation exchange listed companies are protected is under what circumstances this discipli- by multiple takeover defenses. Around the nary mechanism becomes ineffective. To turn of the twentieth century, defense address this issue, we investigate virtually the mechanisms started to be used to protect whole population of Dutch listed industrial Dutch corporations from foreign influences. companies which have adopted multiple Later on, they were applied to restrict the defense mechanisms. power of common shareholders. The use of The issue of corporate governance is also defense measures to repel corporate raids interesting in an international setting and unfriendly takeovers has become because it differs from country to country. more important since the 1960s, and has For example, there is an active takeover received both criticism and support from market in the U.S.A. and the U.K., but this various interest groups. Public corporations is not so in many other countries.There, as have been devoting time and resources for example, shareholders are considered to 648 ON TAKEOVER AS DISCIPLINARY MECHANISM be one group of stakeholders in a firm next also suggest that differences in firms’ to employees, suppliers and customers. The ownership structure (internal control aspect) equity ownership is also concentrated in can explain observed variations in anti- the hands of a few investors. Although the takeover defenses (external control aspect). pattern of cross-shareholdings in German The notion can be illustrated in the and Japanese companies may look similar, following way. the governance structures are quite dissim- Shareholders with large stakes are ilar. German firms have close relationships expected to participate actively in manage- with banks which supply both equity capital rial decision making (Demsetz, 1983; and debt. In contrast, Japanese firms are Shleifer and Vishny, 1986).They will not in characterized by large industrial groups their own interest allow managers to adopt with interlocking directorships. Hostile defensive measures, as disciplining will be takeovers are virtually nonexistent in more difficult. The same is true for large Germany and the Netherlands, but due to but passive shareholders who will also two different reasons. Extensive cross- try to resist any attempt by managers to shareholdings provide German companies adopt defenses. This is because any future with a strong defense, while Dutch compa- possibility of gain through facilitating a nies are protected by multiple antitakeover third-party takeover will then be reduced. devices. These and other differences imply Shareholders with small holdings, on the that the influence of various disciplinary other hand, may not take an active interest mechanisms will vary from country to in monitoring management, perhaps country.The takeover market is a relatively because of the ‘free-rider’ problem. Defense more important disciplinary mechanism in measures are then relatively easily adopted the U.S.A. and the U.K. But, for Germany by managers because there are no large and the Netherlands, concentrated ownership shareholders to counteract management’s and supervisory boards exert a relatively attempt.The purpose of the study is, there- more important role. Various antitakeover fore, to test empirically this theoretically measures are adopted in the U.S.A. to predicted relationship between firms’ own- protect the interests of shareholders during ership concentration and takeover defense takeover bids. But, in the Netherlands these measures.1 measures are primarily directed to limit the Incentives as well as the degree of power of common shareholders. monitoring can vary depending on the stakes A vast literature addresses the inter- and the types of shareholders. One may relationship between ownership structure be interested to know how institutional and different corporate governance devices. shareholders, as a separate group, affect Walsh and Seward (1990) examine different corporate decision making.These investors— internal and external mechanisms of cor- usually banks, insurance companies, pension porate control used in aligning the diverse funds and mutual funds—are expected to interests of managers and shareholders. play a more active role in the affairs of a Important internal control mechanisms company. They are in a better position to include the control function of the board invest resources for increased monitoring of directors, competition within the mana- so that management’s inclination to adopt gerial team, and the monitoring role of large defense mechanisms decreases. On the shareholders. The external control mecha- other hand, some institutional investors nisms, on the other hand, are the market may align with management because of for corporate control and the competition commercial ties and profitable business in the product market. Walsh and Seward opportunities. The role actually played (1990) argue that the failure of one by institutional shareholders, therefore, control mechanism triggers the presence of becomes an empirical issue.2 another mechanism. Studies by Jarrell and Although it has been argued that large Poulsen (1987), Ambrose and Megginson shareholders who are effective monitors (1992), and Gordon and Pound (1993) will prevent managers from adopting TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 649 defensive measures, one can not be sure if in target shareholder returns. Given these shareholders in general are harmed by such findings, we examine if the shareholders adoptions. In fact, adoption of takeover wealth effect of takeover defenses is defenses is usually explained under two related to ownership structure. competing hypotheses (DeAngelo and Rice, The remainder of the paper is organized 1983; Mahoney and Mahoney, 1993). as follows. Important takeover defense According to the managerial entrenchment measures are first discussed with particular hypothesis, defense measures primarily emphasis on those prevailing in the protect poorly functioning management Netherlands. The following two sections by reducing the probability of potential describe the sample and the methodology. takeover. These measures help incumbent The results are presented in the next management to abuse their power by acting section. A brief summary of the study and in their own interest at the expense of the research implications are presented in shareholders. On the other hand, the share- the final section. holder interest hypothesis postulates that adoption of defense measures allows current management to focus on long-term TAKEOVER DEFENSE MEASURES strategies of the firm while remaining protected from the worry of hostile Takeover defense measures help to make takeovers. Through a strong negotiating acquisition of a company more difficult, position, managers can also help share- if not impossible, and thereby serve to holders to obtain a fairer/higher premium if insulate managers from the free market a takeover does take place. for corporate control. These measures vary Empirical studies from the U.S.A. from country to country depending on document that while some defense mecha- institutional features and corporate gover- nisms are harmful for shareholders, others nance systems (Franks and Mayer, 1990). are not.3 This study, therefore, reexamines Moerland (1995) distinguishes two basic the valuation impact of defense measures, types of corporate systems: the market- using the Dutch data. If shareholders of oriented system (prevailing in the U.S.A. Dutch companies interpret the adoption and the U.K.) and the network-oriented of defense measures as managerial system (prevailing in, for example, the entrenchment, stock prices should decline. Netherlands, Germany, France and Japan). Alternatively, if these measures allow The former is characterized by relatively management to bargain for a higher developed financial markets, large-scale takeover premium, share prices should presence of corporations with widely increase. dispersed ownership, and active markets for The wealth effect of defense measures corporate control.The latter system, on the needs to be examined in conjunction with other hand, features closely held corpora- the ownership structure of firms. Jarrell tions, group membership of corporations, and Poulsen (1987) document that value- and substantial involvement of banks in reducing takeover defenses are adopted by corporate financing and corporate control. firms with larger insider holdings and These differences in governance systems smaller institutional holdings. McWilliams are also reflected in differences in adoption (1990) finds that defense measures induce of specific defense devices. positive effects on shareholder wealth for There exist a variety of ways to classify firms with low insider share ownership. takeover defenses. These can be either Agrawal and Mandelker (1990) report structural or technical.The first type arises that the effect is more favorable the larger from prevailing structures of stock market the level of institutional ownership. Song and equity ownership (e.g., relative impor- and Walkling (1993) find that managerial tance of debt financing, crossholdings).The ownership is related both to the probability second type of defenses are specifically of being a takeover target and to increments directed to impede hostile takeover attempts 650 ON TAKEOVER AS DISCIPLINARY MECHANISM (e.g., issuing preferred defense shares, of a company’s charter. Therefore, the limiting voting power). According to one power of the general meeting of common study,4 structural barriers to takeovers are shareholders is restricted. The approval of relatively strong in Italy, France, Germany priority shareholders is also needed for and Switzerland, and of medium strength decisions such as hiring or firing of in Spain and Sweden, but weak in the company directors and issuing new Netherlands and the U.K. Technical common shares. Depository receipts are measures, on the other hand, are relatively issued by an administrative office to strong in the Netherlands, Germany and investors after detaching the voting rights Switzerland, of medium strength in Italy, from ordinary shares. The holder of France, Spain and Sweden, and weak in depository receipts has all economic rights the U.K. attached to common shares, except for Defense mechanisms are also classified the voting right (which rests with the according to shareholders’ approval administrative office). Binding appoint- (Ruback, 1988). Some defenses require ments of new directors are made by the shareholders’ approval before adoption. management board, thereby strengthening These include super-majority provisions, their own control. Ordinary shareholders fair-price amendments and classified are, thus, deprived of the possibility to boards. Other measures may be adopted appoint their own directors. Only a two-third by management without requiring share- majority at the shareholders meeting can holders’ approval. Examples include poison overrule the binding appointment. Limited pills and targeted share repurchases. voting power mechanism restricts the The Dutch situation offers companies maximum number of votes that can be cast numerous possibilities of defense mecha- by one shareholder, regardless of the nisms, many of which do not exist in the number of shares actually held. U.S.A. These include (a) legal measures Besides the above-mentioned takeover such as the creation of structure companies defenses, the issue of preferred defense (‘structuur vennootschappen’); (b) statutory shares is the most widely adopted defense measures such as issuing preferred defense mechanism in the Netherlands.These shares shares, issuing priority shares, making are issued in the name of the holder (usually binding appointments of directors and friendly parties) because of their control limiting voting power per shareholder; function, with only the statutory minimum and (c) nonstatutory measures such as the of 25 percent of par value to be paid issue of depository receipts of shares up. Even though they are not fully paid up, (‘certificaten van aandelen’). Some impor- preferred shares have the same voting tant features of these antitakeover devices rights as common shares. In order to resist are explained below. any unfriendly takeover attempt, common The law on ‘structure companies’ compels stockholders authorize company manage- a large firm to establish a ‘supervisory ment to issue preferred shares whenever board’ (consisting of outsiders and differ- necessary and thus, grant substantial ent interest group representatives). This voting power to another entity. board (thus, not the shareholders of the The procedure of defense with preferred company) in turn appoints a ‘management shares takes place in three consecutive board’ to run day-to-day affairs of the firm. steps. First, common shareholders approve Many decisions of the ‘management the necessary charter amendment to create board’, such as adoption of annual the possibility of issuing preferred shares. accounts, investment plans and company Second, company management grants the restructuring, require approval of the option to a friendly party—usually a ‘supervisory board’, which meets on a few specially created foundation and/or an occasions per year. Priority shares are institutional investor. Third, management issued to a friendly foundation which decides to issue preferred share. This reserves the right to approve any amendment usually happens when there is a fear of TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 651 unfriendly takeover attempt. These three Table 25.2 Distribution of different takeover steps follow one after another, but do not defenses necessarily take place simultaneously. A company may create the possibility to issue Official Parallel marketa marketa Total preferred defense shares at a certain point of time, while the shares are actually issued Total number of firms 141 36 177 several years later (depending on any threat of hostile takeover). Priority shares 63 16 79 Preferred shares 89 16 105 Binding appointment 49 15 64 Limited voting power 7 0 7 DATA Depository receipts 41 19 70

Inspired by the European Community a The official market is the first-tier market for larger companies, while the parallel market is for smaller initiative, shareholders with holdings of companies. 5 percent or more in Dutch listed companies have been required to disclose their holdings publicly since February 1992. Before that, firms are protected by three or more there was no mandatory disclosure of share defense devices. Table 25.2 presents a list ownership, and no way existed even to of widely used antitakeover measures in the identify shareholders.5 The data on block- Netherlands. We find that 105 (32%) holdings are collected from the Dutch firms have adopted defense mechanism financial daily Het Financieele Dagblad. with preferred shares, 79 (24%) firms In total, we obtained a sample of 177 have issued priority shares, 70 (22%) companies listed on the Amsterdam Stock companies have issued depository receipts, Exchange.These companies represent more 64 (20%) firms have made binding than 90 percent of the Dutch stock market appointments of directors, and seven (2%) capitalization. Data on takeover defense companies have imposed restrictions on measures associated with these companies voting rights. are collected from Voogd (1989) and other After searching sources like the stock publications. Our findings are presented in exchange publication Beursplein 5 and the Tables 25.1 and 25.2. financial daily Het Financieele Dagblad, Table 25.1 shows that more than we find that 79 new defense mechanisms 90 percent of Dutch companies are were announced by Dutch companies dur- protected by at least one defense measure. ing 1984–90. Defense with preferred share We find that while only 16 (9%) companies was the most frequently announced mecha- are without any of these defenses, 52 (29%) nism—on 52 occasions, which represents companies have one defense mechanism, 66% of the total.The next most important 62 (35%) firms have two defense antitakeover devices were the issues of mechanisms, and as many as 47 (27%) priority shares and depository receipts. Both were adopted on seven occasions each. No new defense measure was announced during Table 25.1 Number of takeover defenses 1991–92 because of restrictions imposed adopted by Dutch companies by the Amsterdam Stock Exchange. Many companies accelerated the adoption of new No. of measures No. of firms % antitakover devices before the restriction took effect. 0 16 9.1 Out of the 52 announcements of defense 1 52 29.4 2 62 35.0 with preferred shares during 1984–90, 3 39 22.0 we select a sample of 47 to analyze share- 6 4 8 4.5 holder wealth effect. The sample is further Total 177 100 divided into three groups based on the announcement of three steps followed 652 ON TAKEOVER AS DISCIPLINARY MECHANISM during the issuing process. The statutory used to calculate ownership concentration. possibility to defend with preferred shares In addition to the four concentration meas- was created for the first time during ures defined earlier, we use the logarithmic 1984–90 by 17 companies. During the transformation of these variables as well same period, the announcement of granting as the Herfindahl measure of concentration an option allowing friendly parties to own in the regression analysis.9 preferred shares was made by 12 companies To examine whether shareholders (these 12 companies have taken the first experience any change in their wealth step either during 1984–90 or earlier). when new takeover defense measures are Finally, during the period of our investiga- announced, we follow the conventional tion, there were 18 companies which actually event study methodology.This methodology announced the issue of preferred shares. has been widely used in the financial We collect daily share price data from economics literature (e.g., DeAngelo and Datastream. These are adjusted for stock Rice, 1983; Linn and McConnell, 1983). splits and other capital changes. We also Recently, it has also become popular in adjust for cash dividends and then compute the strategic management literature continuously compounded stock returns for (e.g., Mahoney and Mahoney, 1993). The the analysis. purpose of this method is to estimate the deviation of actual stock returns (consequent upon the announcement of a METHODOLOGY specified event) from expected stock returns. We employ the Market Model and We first divide the aggregate sample into the Market Adjusted Returns Model to groups with cumulative takeover defense estimate these deviations for each stock. measures, and then determine the average The Market Model supposes that the ownership concentration for each group. return on an individual stock is linearly We calculate ownership concentration of related to the market return.The relationship a firm in several ways: the percentage is written as follows: of shares held by the largest block- R R e 7 jt j j mt jt holder (C1), the share of the three largest blockholders (C3), and the share of where C all blockholders ( block).We also separately R the continuously compounded calculate a concentration measure (C ) jt inst. j t to represent institutional ownership (esti- return of stock in period ; mated by blockholdings held by major Rmt the continuously compounded Dutch banks and insurance companies). On market return in period t; the basis of a t-test we then find out jj security specific and time whether average ownership concentration independent parameters; significantly varies among groups of com- ejt the error term of stock j in panies with different takeover defenses. period t. The above analysis is performed by comparing two sample averages at a time. The period to estimate the Market Model In order to examine the effect of firms’ parameters is selected as the period of ownership structure on the likelihood of 100 days before the start of the event (or adopting individual takeover defense, we announcement) period. We also estimate estimate the following logistic regression:8 the parameters using 100 days of data from the postevent period. A period p (defense measure) f (ownership of 20 days before the announcement until concentration) 20 days after the announcement is selected as the event period.The impact of takeover Here the dependent variable is equal to 1 defense announcements on stock returns is if a firm has a particular defense measure measured over this period. The parameters and 0 otherwise. Several new proxies are are estimated by using the ordinary least TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 653 squares method. We use the ‘CBS-Total The average abnormal returns and Return Index’ to calculate the market 0cumulative average abnormal returns returns used in the model.10 The abnormal are then computed as described previously. return (also called excess return or prediction error) is the difference between the actual return during the event period (20, EMPIRICAL RESULTS 20) and the return predicted from the estimation period: Ownership structure AR R ˆ R jt jt ˆ j j mt A descriptive analysis on Dutch ownership The abnormal returns for individual stocks structure is presented in Table 25.3. We are then averaged across all stocks to find that blockholders hold more than half obtain average abnormal returns for each of all shares in Dutch companies. The day. The excess returns for each stock are average share of the largest blockholder is also compounded over different time inter- 31 percent, that of the three largest block- vals around announcement date to calculate holders is 45 percent, and the average share cumulative abnormal returns. A t-test is of all blockholders together is 51 percent. performed to test whether the average It appears that the group with the three abnormal returns are significantly different largest shareholders dominates ownership from zero. The t-value is obtained by divid- concentration of Dutch firms. The correla- ing average daily abnormal returns by its tions between these variables are, as standard deviation calculated from the expected, very high. Our results show that estimation period. ownership concentration in the Netherlands In order to check the robustness of our is higher than in the U.S.A., the U.K. results, we also perform the stock return and Japan, but lower than in Sweden.11 analysis using the Market Adjusted Returns The variation within each measure of owner- Model. The model predicts individual stock ship concentration is also higher in the return to be equal to the corresponding Netherlands. The standard deviation of market return, or in other words, percentage of shares held by the top five blockholders in our sample is 26 percent R R jt mt compared with Prowse’s (1995) findings This model is distinct from the Market of 16 percent in the U.S.A. and the U.K. Model in the sense that here all stocks are and 14 percent in Japan. assumed to be of average risk. The abnor- Analyzing the distribution of sharehold- mal returns are calculated as the difference ings, we find that the largest shareholder between the actual stock return and the has more than 25 percent of shares in corresponding market return: 52 percent of the firms in the sample, and more than 50 percent of shares in 22 percent ARjt Rjt Rmt of the firms. A majority of the companies

Table 25.3 Means, medians, standard deviations, and correlations

Variables Mean Median S.D. C1 C3 C5 Cblock Cinst.

C1 30.8 25.5 1.68 1 C3 45.1 42.5 24.9 0.89 1 C5 49.2 49.8 25.8 0.81 0.97 1 Cblock 50.9 55.1 26.5 0.75 0.93 0.98 1 Cinst: 9.9 6.0 12.5 0.16 0.02 0.06 0.08 1

The table reports the results of different ownership concentration variables: C1, C3, C5, Cblock and Cinst. represent the percentage of shares held by the largest blockholder, the three largest blockholders, the five largest blockholders, all blockholders, and institutional blockholders, respectively.The sample consists of 177 industrial companies listed on the Amsterdam Stock Exchange in 1992. 654 ON TAKEOVER AS DISCIPLINARY MECHANISM has a blockholding in excess of 50 percent. 11 percent age points higher than that for After searching the identity of these firms with two devices. Both differences in blockholders, we find that the average concentration are statistically significant. shares of management and family members, In general, we find that the lower the companies, and individual blockholders ownership concentrations are, the more are 8 percent, 20 percent and 5 percent, takeover defenses companies adopt. This respectively.The average share of financial phenomenon is valid for all three measures institutions (banks and insurance compa- of ownership concentration. Our evidence is nies) in our sample is almost 10 percent. consistent with Bergström and Rydqvist The combined share of these investors is (1990), who observe that Swedish firms with less than 25 percent for 90 percent of high concentration of equity ownership rarely the companies. The sample contains adopt antitake over devices. These findings 18 companies in which banks and insurance suggest that firms adopt multiple takeover companies are the only blockholders. defenses when shareholdings are diffuse. The average share of other institutional Table 25.4 also reports the results blockholders is 6 percent. for institutional blockholders (banks and In Table 25.4 we present the average insurance companies).The concentration of ownership concentrations of companies these institutional shareholders does not with cumulative defense mechanisms. show any particular relationship with We also report in the lower panel corre- multiple takeover defenses. The share of sponding t-values testing the difference in these investors in firms with one defense average ownership concentrations. Our mechanism is five percentage points results show that the concentration of the higher than in firms without any defense. largest shareholder for firms without any Afterwards, as institutional ownership defense measure is almost 13 percentage concentration declines, firms adopt a higher points higher than that for firms with only number of defenses. These differences are one measure. Similarly, for companies not statistically significant. with one takeover defense device, the Next, we examine if the general finding concentration of the largest shareholder is on the negative relationship between

Table 25.4 The difference in ownership concentration of firms with cumulative takeover defenses

Measures of concentration

No. of defenses No. of firms C1 C3 Cblock Cinst.

0 16 48.51 65.21 73.08 7.83 1 52 35.88 53.01 59.78 12.16 2 62 24.99 37.59 41.48 9.36 3 39 28.99 42.24 49.93 9.01 4 8 20.33 27.04 29.99 7.45 t (0, 1) 2.00** 1.91* 2.17** 1.27 t (1, 2) 2.78** 3.46** 3.92** 1.11 t (2, 3) 0.78 0.96 1.59 0.13 t (3, 4) 0.95 1.72* 2.02** 0.35

** Statistically significant at the 5% level. * Statistically significant at the 10% level. The table reports results of four concentration variables—C1, C3, Cblock and Cinst. —representing the percentage of shares held by the largest blockholder,the three largest blockholders, all blockholders, and institutional blockholders, respectively. The t-statistic reported in the lower panel tests for the difference of means between two measures of ownership concentrations. TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 655 ownership concentration and defense of the Herfindahl measures. The results mechanisms also holds for individual with institutional concentration variables takeover defenses. The analysis is carried alone are, however, once again mixed. out by performing a logit regression. The estimated regression coefficient expresses Wealth effects the relationship between the likelihood of choosing one particular defense mecha- The sample here consists of 44 new nism and a measure of firms’ ownership preferred defenses announced during concentration. The results are presented in 1984–90.13 Table 6 presents the cumulated Table 25.5.12 The reported coefficient average abnormal returns based on the estimates are obtained from running Market Model for several intervals in the regressions with one explanatory variable event period.The results from the aggregate at a time. We find that the results are sample indicate that the announcement generally consistent with earlier findings. of the preferred share defense mechanism The probability of a firm adopting any one is, on average, associated with a decline in takeover defense mechanism is negatively common share price. During the 2-day related to ownership concentration. The announcement period [0, 1], shareholders finding is robust to all variables used suffer a statistically significant return in computing ownership concentration, decline of 1.18 percent. The result is not including the logarithmic transformations driven by a few outliers as the number of

Table 25.5 Estimates of logistic regressions relating the likelihood of adopting a specific takeover defense mechanism to ownership concentration

Priority share Preferred share Depository receipts Measures of concentration Intercept Coefficient Intercept Coefficient Intercept Coefficient

** ** ** C1 0.06 0.01 0.94 0.02 0.40 0.03 (0.24) (1.32) (3.45) (2.54) (1.47) (3.46) * ** ** ** C3 0.30 0.01 1.40 0.02 0.50 0.02 (0.97) (1.87) (3.99) (3.33) (1.55) (3.16) * ** ** ** Cblock 0.31 0.01 1.65 0.02 0.37 0.02 (0.93) (1.78) (4.26) (3.72) (1.12) (2.67) ** ** * Cinst. 0.08 0.03 0.23 0.01 0.67 0.02 (0.41) (2.24) (1.18) (1.16) (3.34) (1.95) ** ** ** ** LN(C1)0.630.27 1.89 0.48 1.26 0.55 (1.17) (1.63) (3.06) (2.58) (2.19) (3.04) * ** ** ** LN(C3)0.970.33 2.48 0.58 0.94 0.39 (1.53) (1.93) (3.06) (2.69) (1.48) (2.21) * ** ** * LN(Cblock)0.930.31 2.66 0.61 0.73 0.32 (1.48) (1.88) (3.16) (2.82) (1.18) (1.93) ** ** ** LN(Cinst.) 0.19 0.27 0.15 0.15 0.85 0.26 (0.85) (2.36) (0.66) (1.31) (3.43) (2.28) ** ** ** Hblock 0.06 0.90 0.76 2.22 0.11 3.51 (0.31) (1.09) (3.54) (2.58) (0.50) (3.15) * ** Hinst. 0.14 5.41 0.29 6.34 0.45 1.55 (0.87) (1.04) (1.77) (1.13) (2.68) (0.35)

** Statistically significant at the 5% level. * Statistically significant at the 10% level. The concentration variables are defined as follows.The variables C1, C3, Cblock, and Cinst. represent the percentage of shares held by the largest blockholder, three largest blockholders, all blockholders, and institutional blockholders, respectively.The LN and H variables are logarithmic transformed and Herfindahl concentration measures, respectively. Absolute t-values are shown in parentheses beneath each coefficient. 656 ON TAKEOVER AS DISCIPLINARY MECHANISM negative abnormal returns dominates the The almost negligible stock price impact sample.This is also found to be statistically with respect to the second step announce- significant at the 5 percent level after ment is not surprising, since granting a conducting a sign test (Z-statistic purchase option to a friendly party is an 2.34). Over the 6-day postannouncement obvious outcome of the charter amendment. period, the cumulative abnormal return is Another interesting finding is obtained 2.27 percent (with a t-value of 2.97).14 when we look at the third step of the defense Although the above result tends to process. The negative announcement effect support the managerial entrenchment of the aggregate sample is in fact deter- hypothesis, further analysis of the sample mined by the issue of the preferred share reveals some interesting findings. We split itself. We find that the announcement of the aggregate sample into three subsamples a preferred share issue is associated with a based on the three steps followed in the strong excess decline in stock returns issuing process. With the announcement of (4.09%), which is statistically significant the first step towards defense (creating the (with a t-value of 4.94). For the 6-day possibility of preferred share issue), a period [0, 5], the excess decline in share- positive and statistically significant stock holders’ wealth amounts to 6.40 percent.16 price effect is observed. This evidence does However,we find a significant price increase not support the managerial entrenchment before the announcement of the preferred hypothesis. Shareholders do not experience share issue.This increase in share price could any wealth decline from the charter be an indication of a takeover attempt amendment leading to takeover defense. On that eventually led managers to issue the contrary, they appear to benefit as there the preferred shares.17 Interestingly, the is a signficant increase in stock returns post-announcement periods indicate a (1.23% in 2 days).15 All other postan- significant decline in shareholders’ wealth. nouncement intervals also reveal positive This decline might provide an estimate of (but not significant) price increases. This the lost premium incurred by common result indicates that defense measures are shareholders—since the chance of eventual indeed adopted allowing shareholders to takeover was eliminated by actually issuing benefit from increased takeover premiums. preferred shares.18 Our finding is consistent

Table 25.6 Cumulative abnormal returns around the announcement of defense with preferred share issue

Return intervals (20, 1) (0, 1) (1, 1) (0, 5) (20, 5) (0, 20)

Full sample 1.06 1.18** 1.17** 2.27** 1.52 3.56** (n 44) (0.76) (2.67) (2.18) (2.97) (0.96) (2.49) Create possibility 3.11 1.23* 0.71 0.60 2.51 1.43 (n 17) (1.41) (1.77) (0.83) (0.01) (1.00) (0.64) Grant option 0.50 0.41 0.10 0.42 0.08 2.81 (n 10) (0.22) (0.57) (0.12) (0.33) (0.03) (1.20) Actual issue 4.93* 4.09** 1.70* 6.40** 1.21 9.64** (n 17) (1.88) (4.94) (1.68) (4.46) (0.41) (3.59) First issue 4.36 4.18** 2.20* 7.03** 1.42 11.73** (n 13) (1.32) (4.00) (1.72) (3.88) (0.38) (3.46) Second issue 6.78** 3.79** 0.09 4.37** 0.52 2.85 (n 4) (2.30) (4.06) (0.08) (2.71) (0.16) (0.94)

** Statistically significant at the 5% level. * Statistically significant at the 10% level. Abnormal returns are computed employing the Market Model, and are shown as a percentage. Results are presented for six different intervals.The numbers in parentheses below the coefficients are absolute t-values. TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 657 with prior studies showing that stock prices when a firm is characterized by diffuse increase with takeover bids but then decline shareholdings.We do not find any significant if they do not materialize. In sum, the relationship associated with institutional evidence provided here suggests that, stock ownership. The evidence provided although defense measures are beneficial here, therefore, does not strongly support to a certain extent, the benefits do not the hypothesis that institutional share- remain when they are used to fend off holders provide better monitoring than takeover attempts. other blockholders. We also examine whether there is a We also conduct a stock return analysis in difference in the results between the first the case of defense with preferred share— preferred share issue and a subsequent the most widely used takeover defense issue. The issue sample is further divided device in recent years in the Netherlands. into a subsample of 13 companies that Our results indicate two opposing effects issued preferred shares for the first time of defense on shareholders’ wealth: in one and a subsample of four companies with a situation, the stock market reacts subsequent issue. We find that the first positively, seemingly to allow managers to issue is more damaging for shareholders. bargain for a higher premium in takeover The abnormal return in the 5-day post- bids. In another situation, the stock market announcement period is 7.03 percent reacts negatively as potential takeover (t-value –3.88) in case of the first-time attempt appears to be eliminated. issue, compared to 4.37 percent (t-value Alternative disciplinary mechanisms 2.71) in case of a subsequent issue. This have been an area of extensive scrutiny. In difference is statistically very significant this paper, we document that low (high) (with a t-value of 6.24).19 Other post- ownership concentration is associated with announcement return intervals show greater (smaller) use of antitakeover devices similar results. which affect the functioning of the market for corporate control. We also provide evidence on the existence of positive and CONCLUSIONS AND negative share price effects of takeover IMPLICATIONS defense measures. Some implications of our findings are discussed below. This paper empirically analyzes the The empirical results obtained in this relationship of takeover defenses with study reconfirm the need of analyzing firms’ ownership structure and shareholders’ firms’ ownership structure as a mechanism wealth. A sample of Dutch industrial to control the agency conflict between companies is selected for the study. The shareholders and managers. Other such Dutch scenario is particularly interesting mechanisms include the capital market, the because almost all listed companies have market for corporate control, the manage- adopted multiple takeover defenses. rial labor market and the product market. We find that firms with a relatively lower Similar to Walsh and Seward (1990), we ownership concentration are the ones with believe that much can be learned about one a larger number of defense measures. Our control mechanism when it is analyzed in analysis suggests that firms with disperse and around another mechanism. Managers ownership adopt more defense tactics.The in the Netherlands seem to be immune analysis also shows that the likelihood for from the disciplinary threat of the market a firm to adopt takeover defenses is inversely for corporate control. Since the takeover and significantly related to ownership con- market is just one disciplinary mechanism, centration. The result is robust to different we would expect other control mechanisms ways of measuring ownership concentration. to be at work too. Our results in this Overall, our evidence is consistent with the paper demonstrate this—for example, hypothesis that company management is monitoring by concentrated ownership. As more likely to adopt defensive measures Prowse (1995) points out, concentrated 658 ON TAKEOVER AS DISCIPLINARY MECHANISM shareholdings are important because they substituted by another, these do not imply provide investors with both the incentive one mechanism being better than another. and the ability to monitor and influence the Analyzing the cost—benefit aspects of management. Without such concentration, each monitoring mechanism is an issue again other mechanisms of corporate which was beyond the scope of the paper. control must be relied upon. According to Moerland (1995), there exists It is usually believed that institutional a trade-off in the sense that the advantages investors find it in their best interest to of one mechanism will generally be the more effectively monitor company managers. disadvantages of the other. Questions also In the U.S.A., institutional shareholdings remain relating the effectiveness of moni- have increased over the last years, and a toring with corporate performance. Are few institutional shareholders have emerged firms with concentrated ownership more as very active monitors. The findings of profitable than firms in which ownership is Duggal and Millar (1994) suggest that not concentrated? If incumbent managers researchers should better split aggregate use defense tactics to deter takeovers, institutional ownership into different cate- are they completely insulated from the con- gories to obtain correct results.The results sequences of poor firm performance? The of this study show that in the Netherlands usual finding has been that takeover threat institutional shareholders like banks and improves performance of a firm. But, as has insurance companies do not have large been mentioned by Walsh and Seward holdings, and these have no relationship (1990), if managers in a poorly performing with the adoption of antitakeover devices. firm anticipate a takeover contest, they An implication of this finding is that active may adopt antitakeover devices to protect monitoring by institutional shareholders them. On the other hand, managers could may not take place in many countries. It is be more frequently dismissed when they highly unlikely that Dutch institutional perform poorly because another disciplinary shareholders lack the expertise and the mechanism—the managerial labor market— ability to serve as effective monitors. then starts functioning. A further analysis of Rather, the presence of small stakes may the role of different corporate governance explain why passivity remains the norm. It mechanisms represents an interesting area is also possible that active institutional of future research. monitoring may not be a representation of the general pattern in the U.S.A. There are many types of defense measures, and their effects also depend ACKNOWLEDGEMENTS on situation like the manner in which a particular device is introduced. Our analysis We would like to thank Tom Berglund, shows that it is difficult to say a priori Associate Editor Richard Bettis, Marc whether defense measures are good or bad Zenner and two anonymous referees for for shareholders. Even takeover defenses useful comments and suggestions. We have that are approved by shareholders have the also benefited from the comments of semi- potential to reduce shareholder wealth. nar participants at Tilburg University, the This lack of conclusive evidence regarding Workshop on Corporate Finance at the the stock price reaction could also be European Institute for Advanced Studies in attributed to the limitations of event- Management (Brussels), the International study methodology and the additional Meeting of the French Finance Association need of analysis using other characteris- (La Baule), the Symposium on Money, tics of firms that propose takeover defense Finance, Banking and Insurance (Karlsruhe) measures. and the Eastern Finance Association Although the results presented here show Meeting (Boston) where previous versions that one monitoring mechanism has been of this paper were presented. TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 659 NOTES 12 Because of space limitation and qualitatively similar findings, the regression results of only a 1 Although several studies have examined limited number of variables are presented. empirically the relationship between equity 13 Three measures could not be included in ownership and firm value (e.g., McConnell and the sample because the market model parame- Servaes, 1990; Slovin and Sushka, 1993), ters’ estimation period coincided with the event limited attention has been given to explore period of a previous measure. ownership concentration vis-à-vis multiple 14 The finding is robust as the two other defense measures. methodologies (the Market Adjusted Returns 2 Empirical evidence on the mixed role Model and the Market Model using postevent of institutional shareholders can be observed period data) show that stock prices decline by from different studies, such as Agrawal and 1.50 percent and 1.85 percent, respectively. Mandelker (1990), Bhagat and Jefferis (1991), 15 The Market Adjusted Returns Model and Brickley, Lease, and Smith (1988), Duggal and the Market Model using postevent period data Millar (1994), Pound (1988), Shivdasani show that stock prices increase by 1.33 percent (1993) and Van Nuys (1993). and 1.27 percent, respectively. 3 DeAngelo and Rice (1983) and Jarrell and 16 Once again, the results are materially Poulsen (1987) find an insignificant price indifferent to one particular methodology used effect; Linn and McConnell (1983) find a weak in calculating abnormal returns. The Market positive effect; and Jarrell and Poulsen (1988), Adjusted Returns Model and the Market Mahoney and Mahoney (1993), Malatesta and Model using postevent period data also show a Walkling (1988) and Ryngaert (1988) find a decline in stock returns (5.06% and 5.62%, significant negative share price effect. respectively). 4 Effect of ‘Bangeman Proposal’ on Barriers 17 The increase in stock price followed by a to Takeovers in the European Community, decline on the announcement of share issue Coopers & Lybrand Management Consultants, could also be seen as an indication of the break- Amsterdam, 1990. down of takeover negotiations.To verify this, we 5 Public corporations in the Netherlands searched the financial press throughout the issue predominantly bearer shares. event period, and found no report on any 6 We could not find a definitive announcement negotiation.This, of course, does not rule out the date for two companies; three companies were possibility of undisclosed information. closely involved in a merger or takeover. 18 We also searched the financial press to 7 Blockholders are owners of 5 percent or check if any specific event followed the more of the outstanding equity. announcement, but were unable to find any. 8 The logistic analysis is chosen here because 19 We also investigate whether the market the dependent variable is a binary, qualitative reaction varies with firms’ ownership concentra- variable. tion.The sample is divided into three portfolios: portfolio 1 contains firms with the lowest 9 As the variable Cblock combines both institutional and blockholders’ shares, we have concentration, portfolio 3 contains those with constructed another variable which estimates the highest concentration, and portfolio 2 is the share of all blockholders other than those between them. We do not find statistically held by institutional blockholders.The Herfindahl significant differences in cumulative abnormal measure was calculated by summing squared returns among these portfolios. Therefore, the percentage of shares owned by each blockholder. results are not reported here. A cross-sectional 10 The CBS-Total Return Index is a value- regression between ownership structure and weighted index representing all listed stocks. It announcement period abnormal returns also is the only market index available in the yields insignificant results. Our analysis, how- Netherlands which covers all listed companies. In ever, should be interpreted with caution because addition, the index is adjusted for cash dividends. the sample size is small and only one defense 11 Prowse (1995) reports average ownership mechanism is examined. concentration of the five largest shareholders to be 25 percent in the U.S.A., 21 percent in the U.K. and 33 percent in Japan. According REFERENCES to Bergström and Rydqvist (1990), the average ownership concentration of the largest Agrawal, A. and G. N. Mandelker (1990). shareholder in Sweden is 43 percent. ‘Large shareholders and the monitoring of 660 ON TAKEOVER AS DISCIPLINARY MECHANISM managers: The case of antitakeover charter “antitakeover” amendments on common amendments’, Journal of Financial and stock prices’, Journal of Financial Quantitative Analysis, 25, pp. 143–161. Economics, 11, pp. 361–399. Ambrose, B. W. and W. L. Megginson (1992). Mahoney, J. M. and J. T. Mahoney (1993). ‘The role of asset structure, ownership struc- ‘An empirical investigation of the effect of ture, and takeover defenses in determining corporate charter antitakeover amendments acquisition likelihood’, Journal of Financial on stockholder wealth’, Strategic Management and Quantitative Analysis, 27, pp. 575–589. Journal, 14(1), pp. 17–31. Bergström, C. and K. Rydgvist (1990). Malatesta, P. H. and R. A. Walkling (1988). ‘Ownership of equity in dual-class firms’, ‘Poison pill securities: Stockholder wealth, Journal of Banking and Finance, 14, profitability and ownership structure’, pp. 255–269. Journal of Financial Economics, 20, Bhagat, S. and R. H. Jefferis (1991). ‘Voting pp. 347–376. power in the proxy process’, Journal of McConnell, J. and H. Servaes (1990). Financial Economics, 30, pp. 193–225. ‘Additional evidente on equity ownership and Brickley, J. A., R. C. Lease and C. W. Smith corporate value’, Journal of Financial (1988). ‘Ownership structure and the voting Economics, 27, pp. 595–612. on antitakeover amendments’, Journal of McWilliams, V. B. (1990). ‘Managerial share Financial Economics, 20, pp. 276–291. ownership and the stock price effects of DeAngelo, H. and E. M. Rice (1983). antitakeover amendment proposals’, Journal ‘Antitakeover charter amendments and of Finance, 45, pp. 1627–1640. stockholder wealth’, Journal of Financial Moerland, P. W. (1995). ‘Alternative discipli- Economics, 11, pp. 329–359. nary mechanisms in different corporate Demsetz, H. (1983).‘The structure of ownership systems’, Journal of Economic Behaviour and and the theory of the firm’. Journal of Law Organization, 17, pp. 17–34. and Economics, 26, pp. 375–390. Pound, J. (1988). ‘Proxy contests and the Duggal, R. and J. A. Millar (1994).‘Institutional efficiency of shareholder oversight’, investors, antitakeover defenses and success of Journal of Financial Economics, 20, hostile takeover bids’, Quarterly Review of pp. 237–265. Economics and Finance, 34, pp. 387–402. Prowse, S. (1995). ‘Corporate governance in Franks, J. and C. Mayer (1990). ‘Capital an international perspective’, Financial markets and corporate control: A study of Markets, Institutions and Instruments, 4, France, Germany and the UK’, Economic pp. 1–63. Policy, 5, pp. 191–231. Ruback, R. S. (1988). ‘An overview of takeover Gordon, L. A. and J. Pound (1993). defenses’. In A. J. Auerbach (ed.), Mergers ‘Information, ownership structure, and and Acquisitions. University of Chicago shareholder voting: Evidence from share- Press, Chicago, IL, pp. 49–67. holder-sponsored corporate governance Ryngaert, M. (1988). ‘The effect of poison proposals’, Journal of Finance, 48, pill securities on shareholder wealth’, pp. 697–718. Journal of Financial Economics, 20, Jarrell, G. A. and A. B. Poulson (1987). ‘Shark pp. 377–417. repellents and stock prices: The effects of Shivdasani, A. (1993). ‘Board composition, antitakeover amendments since 1980’, ownership structure, and hostile takeovers’, Journal of Financial Economics, 19, Journal of Accounting and Economics, 16, pp. 127–168. pp. 167–198. Jarrell, G. A. and A. B. Poulsen (1988). Shieifer, A. and R. W. Vishny (1986). ‘Dual-class recapitalizations as antitakeover ‘Large share holders and corporate mechanisms, the recent evidence’, Journal of control’, Journal of Political Economy, 94, Financial Economics, 20, pp. 129–152. pp. 461–488. Linn, S. C. and L J. McConnell (1983). Slovin, M. B. and M. E. Sushka (1993). ‘An empirical investigation of the impact of ‘Ownership concentration, corporate control TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 661 activity, and firm value: Evidence from the 1989 Honey well proxy solicitation’, Journal death of inside blockholders’, Journal of of Financial Economics, 34, pp. 101–132. Finance, 48, pp. 1293–1321. Voogd, R. P. (1989). ‘Statutaire beschern- Song, M. H. and R. A. Walkling (1993). ‘The ingsmiddelen bij beursvennootschappen’, impact of managerial ownership on acquisition doctoral dissertation, University of Nijmegen, attempts and target shareholder wealth’, Kluwer. Journal of Financial and Quantitative Walsh, J. P. and J. K. Seward (1990). Analysis, 28, pp. 439–457. ‘On the efficiency of internal and external Van Nuys, K. (1993). ‘Corporate governance corporate control mechanisms’, Academy through the proxy process: Evidence from the of Management Review, 15, pp. 421–458. Chapter 26

Tim Jenkinson and Alexander Ljungqvist* THE ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE

Source: Journal of Corporate Finance, 7(4) (2001): 397–446.

ABSTRACT This article uses clinical evidence to show how the German system of corporate control and governance is both more active and more hostile than has previously been suggested. It provides a complete breakdown of ownership and takeover defence patterns in German listed companies and finds highly fragmented (but not dispersed) ownership in non-majority controlled firms. We document how the accumulation of hostile stakes can be used to gain control of target companies given these ownership patterns.The article also suggests an important role for banks in helping predators accumulate, and avoid the disclosure of, large stakes.

1. INTRODUCTION particularly important motive may be the expropriation of minority shareholders.The THERE IS A WIDESPREAD BELIEF THAT THE dynamics of hostile stakebuilding are German system of corporate governance complex and difficult to observe—in many exhibits a very low level of hostility. In the cases it is not possible simply to look at a stereotypical view of German finance, hos- share register and infer who is exerting tile tender offers are virtually unheard of, control over the company. This opaqueness with banks (rather than markets) assumed derives from the low level of transparency to play an important role in both the financ- of share stakes and weak regulation of paring and control of German corporations.1 ties acting in concert. Our article seeks to This article challenges some important overcome this opacity by taking a “clinical”, elements of this view. It is certainly true, or case study approach, looking in detail at notwithstanding the recent successful ten- the dynamics of stake accumulation, and der offer for Mannesmann by Vodafone– the control battles that ensued. Airtouch, that hostile tender offers have This approach is both a strength and played almost no role in disciplining incum- weakness of the paper.We identify 17 cases bent management.2 However, we suggest of hostile stakebuilding over an 8-year that there is a much greater incidence of period.This is clearly not a large number in outsiders accumulating hostile stakes or absolute terms, or relative to the total num- blocks in an attempt to gain control. ber of listed companies in Germany (fewer Hostile stakes are often built by coalitions than 600 at the time). However,one should of large investors who share dissatisfaction not jump to conclusions too quickly. In with the incumbent management, or have common with a number of other countries other motives for seeking control. One such in Europe (excluding the UK) ownership ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 663 concentration is very high in Germany.3 In Bulow et al., 1999) or a way to mitigate Section 2 we look in detail at the owner- the free-rider problem in monitoring man- ship structure, and takeover defences, of all agement (Butz, 1994; Mørck et al., 1989). German listed companies and find that as The second of these roles has recently few as 64 German companies may be attracted a lot of attention. One strand of vulnerable to hostile attack. The resulting the literature looks at acquisitions of 5% 3–4% per annum incidence of hostile or more in the US, that is, the emergence of stakebuilding is surprisingly similar to the new blocks. These ‘partial acquisitions’ are incidence of hostile tender offers in, for typically greeted with positive share price example, the UK.4 responses both for the target and the However, while we argue that this inci- buyer (see Mikkelson and Ruback, 1985, dence is economically significant, and has for public 13-D acquisitions, Wruck, 1989, not previously been identified, the low for private placements), indicating that absolute number of cases precludes econo- increases in ownership concentration are metric testing of formal hypotheses. On the value-increasing. Moreover, consistent with other hand, we are able, through the clini- the Berle–Means thesis, targets tend to cal approach, to analyse in considerable have performed poorly prior to the partial detail the behaviour of the various parties acquisition (Choi, 1991; Bethel et al., involved in the control contest. For exam- 1998) and be more diversified (Bethel et ple, we analyse the behaviour of banks in al., 1998), while subsequent target firm such battles and find their role to be much operating and financial behaviour is posi- more complex than has previously been tively affected (Spencer et al., 1998) and documented. Far from protecting incum- CEO turnover increases substantially bent management, on a number of occa- (Bethel et al., 1998). sions, German banks have been actively A second, and related, literature focuses involved in bringing about hostile changes on existing blocks. Perhaps surprisingly, of control by facilitating stakebuilding. We significant share blocks are common even show how banks can assist predator com- in the US, where corporate ownership is panies in the accumulation of hostile typically more dispersed than in Germany stakes, and how beneficial ownership can and other continental European countries be obscured. We also consider how the reg- (Barclay and Holderness, 1989).5 Trading ulatory environment allows such stake in such blocks is about twice as frequent as accumulation to occur and whether recent hostile tender offers (Barclay and Holder- important changes—such as the introduc- ness, 1991), and typically takes place at a tion of Germany’s first Takeover Code— premium to the post-trade market price will influence the way that corporate (Barclay and Holderness, 1989). Since control is exercised in the future. trading of existing blocks leaves ownership The contributions of this article are, concentration unchanged, Barclay and therefore, empirical. The findings are rele- Holderness (1989) suggest the block pre- vant to a number of different areas of mium is consistent with private benefits of research. First, there is a rich literature, control rather than value-increasing moni- mainly focused on US companies and mar- toring, though the two are not mutually kets, which investigates the links between exclusive: block trades are followed by an ownership structure and corporate per- increase in CEO turnover, of a magnitude formance. Most of this literature takes as usually only seen in hostile tender offers its starting point, the Berle and Means (Barclay and Holderness, 1991), and are (1932) thesis that dispersed ownership associated with subsequent business leads to an agency conflict between (weak) restructuring (Denis and Serrano, 1996). A owners and (strong) managers. Blocks in look at the treatment of minority share- this context are typically thought to per- holders when block trades are followed form one of two roles: a toehold prior to a by tender offers suggests that the block hostile takeover (Shleifer and Vishny, 1986; premium is not entirely due to private 664 ON TAKEOVER AS DISCIPLINARY MECHANISM benefits: minorities are typically bought out opportunity for hostile control changes to at a premium to the block price, despite the be effected via stakebuilding and for such absence of any legal compulsion (Barclay behaviour to be profitable. Such loose reg- and Holderness, 1991, 1992). In contrast, ulation exists in a number of continental in Germany, minorities are vulnerable to European countries including Germany. expropriation by majority owners: there is Hence, the behaviour that we focus on in no legal requirement to buy-out minority this article is, in a sense, quite predictable, stakes, and when offers to minority share- although no previous paper has, to our holders are made, they are typically at a knowledge, provided systematic evidence large discount to the price paid by the on the importance of hostile acquisitions in controlling shareholder. Germany via stakebuilding. Our article com- There are some important differences plements that of Franks and Mayer (2001), between the block trades discussed in the who consider the importance of share US literature and our cases. First, US Secu- stakes in Germany. They document a large rities and Exchange Commission (SEC) incidence of stake changes and suggest that regulations have tended to result in block such changes are the most significant influ- trades being negotiated and publicised ence on the turnover of supervisory board rather than covert. Second, partial acquisi- members. However, Franks and Mayer con- tions or trades of existing blocks take place sider the incidence and observed response against the background of an ownership to sales of share stakes in general, without structure that is significantly more dispersed distinguishing between the nature of these in the US than in Germany, particularly for sales. In contrast, we focus exclusively on large firms.6 Therefore, the economically hostile stakebuilding, where the objective is significant block size is likely to be differ- to influence control. ent in the US and Germany, and should The remainder of the article is organised depend on the concentration and composi- as follows. In Section 2, we start by dis- tion of the remaining ownership structure. cussing the rules governing corporate con- Third, three-quarters of the US cases trol in Germany, in particular, the control involve a simple sale of one existing block rights associated with share blocks of vari- (Barclay and Holderness, 1991), while our ous sizes.We also provide a snapshot of the article documents a much richer dynamic actual ownership patterns in Germany, which of blocks being built from open-market demonstrates the importance of large purchases acquired from existing block- share blocks. In Section 3, we outline our holders and combined with other blocks. In research methodology and data sources. that sense, we look at block-building, rather Section 4 presents evidence on hostile stake than block transfers. There has, to date, accumulation for a number of recent cases, been little analysis of block accumulation detailed chronologies of which are provided tactics, in the US or elsewhere.7 in Appendix A (in alphabetical order by Finally, our article is also related to the target company, for ease of reference). literature on optimal takeover regulation, Section 5 concludes. since we focus particularly on the ability to build hostile stakes without the knowledge of other market participants—something 2. OWNERSHIP AND CONTROL OF that the existing block literature assumes GERMAN COMPANIES away. For example, as Burkart (1996) notes, it is critical whether stakebuilding, Ownership of German companies is highly and hostile intentions, can be obscured concentrated, certainly in comparison with because as soon as such a strategy becomes US or UK companies. We document later transparent, free-riding (along the lines on in this section just how concentrated suggested by Grossman and Hart, 1980) is ownership structures are, but we begin by likely to result. Loose disclosure and regu- explaining the control rights associated lation of acting in concert will provide the with different-sized share blocks in Germany. ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 665 In many cases, the hostile stakebuilding contracts have no special legal standing, that we document later in the article does and are typically renewable every 4–5 years. not involve building a majority stake, but We start by considering the rights asso- rather takes the form of building a stake with ciated with dispersed shareholdings. In sufficient control rights to exert a major Germany, no special rights are attached to influence on the incumbent managers and share blocks of below 25%. Indeed, in other shareholders. The rules regarding common with many other EU countries, these control rights therefore play a critical small minority shareholders are afforded role in hostile stakebuilding strategies. little in the way of protection, either from expropriation by controlling shareholders, or equal treatment. For example, there is 2.1. Control rights no legal requirement to buy-out minority There are five economically significant stakes when a party acquires a stake in block sizes as defined by the rights of excess of 50%. While the recently intro- minorities and the discretionary powers of duced, and voluntary, Takeover Code does the dominant blockholders. These are sum- introduce such a requirement, it is notice- marised in Table 26.1. One important fea- able that the offer to minority shareholders ture of corporate control in Germany is the can be at a discount (of up to 25%) to the existence of pooling contracts. Pooling price paid in acquiring the controlling stake contracts are agreements between share- (see footnote 27 for more details). holders that oblige them to pool their votes. The first important threshold in terms of Hence, share blocks can be built directly control rights applies to blocks of 25% or via purchase and/or indirectly via a pooling more. Such a shareholding is referred to as contract. Either way, the control rights a blocking minority and gives veto powers summarised in Table 26.1 would apply to on corporate charter amendments, supervi- the single or pooled block. The pooling sory board changes, and profit transfer and

Table 26.1 Control rights in Germany The control rights associated with certain block sizes are related to the minimum required percentage of the votes cast at an annual or extraordinary general meeting. Corporate charters may in certain cases specify higher thresholds, for instance regarding changing the composition of the supervisory board.

Block size Control rights

25% No major formal rights other than voting pro rata. Shareholders who control at least 5% of the equity capital can demand that an annual general meeting be called. 25% Blocking minority control. Power of veto over corporate charter amendments, supervisory board changes, and profit transfer and control agreements. Stakes above 24.9% must be reported to the Cartel Office and to the target company, which in turn has a duty to publish a notification in certain newspapers. 50% Management control, although the existence of another block of 25% would limit discretion. 75% Super-majority control. Enables the blockholder to amend the corporate charter, change the composition of the supervisory board, and enter into profit transfer and control agreements.The lack of any blocking minority (25%) stake thus significantly increases effective control. 95% The blockholder can force a merger and minority shareholdings can be compulsorily purchased.The law requires that minority shareholders be paid ‘adequate’ compensation in the form of cash or shares in the parent company. The level of compensation is largely left unregulated. 666 ON TAKEOVER AS DISCIPLINARY MECHANISM control agreements. The latter agreements prices have nothing to do with fundamental are an important way by which the claims value—which can ‘only’ be found by of minority shareholders can be diluted. accountants acting as expert witnesses. Subject to ensuring the economic survival Even when the courts do find in the plain- of the dominated company, a controlling tiffs’ favour, the average imposed increase shareholder can dilute minorities in a vari- in bid prices or guaranteed dividends of ety of ways. First, group losses can be 25% (Wenger and Hecker, 1995) is not foisted disproportionately onto minorities, sufficient to close the discount to the mar- while profits can be transferred out of the ket price before the buy-out announcement. dominated company. Second, hidden reserves Finally, the law explicitly gives the majority (typically land and share stakes whose owner the right to cancel the control and/or book value is below market value) can profit transfer agreement if the court’s be sold and the proceeds transferred to the decision is unfavourable.11 majority owner.Third, group assets can be Hence, the ability to veto a profit trans- bought and sold at prices that are advanta- fer and control agreement represents an geous to the dominant shareholder.8 Fourth, important control threshold. It is also while minority shareholders must be offered worth noting that, in the absence of other the alternative of a guaranteed fixed divi- large shareholders, a 25% stake can in dend should they wish not to sell out to the practice provide more than mere veto influ- controlling shareholder, neither the bid ence, given an average presence of no more price nor the guaranteed dividend need in than 57% of votes at AGMs (Baums and any way be related to the share price, or the Fraune, 1994). price the bidder paid to acquire control.9 The next threshold in terms of control Both options are, in most cases, relatively rights occurs with blocks of 50% or more. unattractive: Wenger and Hecker (1995) Such a block gives management control of show that for 45 buy-out bids made to the company, but is subject to limits on the minority shareholders between 1983 and controlling party’s discretion due to the mid-1992, the price offered was 27.1% existence of a blocking 25% minority. In below the market price 2 days before the practice, 50% may not be enough to dis- announcement. miss incumbent management quickly if the Minority shareholders have no right of supervisory board is not on the controlling legal appeal against any such transfer of party’s side: it is the supervisory board which assets or profits per se, but can litigate appoints and dismisses the management, against the level of compensation (dividend and the supervisory board in turn can only or buy-out bid) offered within 2 months of be dismissed with a 75% super-majority. the date of the offer. Many such offers Given the various control rights that (such as the Boge case we examine below) accrue to blocking minority stakes of 25% do indeed result in lengthy litigation during or more, the next important control thresh- which minority shareholders attempt to old occurs with blocks of more than 75%. increase the attractiveness of the terms Such super-majority blocks give the they are offered. The average court case controlling party complete discretion in takes 5.3 years to conclude (Wenger and matters of supervisory board elections and Hecker,1995).There is, in general, no right profit transfer and/or control agreements. of final appeal to the Supreme Court, However, the incremental control rights which has led to different second-level associated with blocks of 75% accrue courts of appeal passing contradictory ver- entirely from the impossibility of a rival dicts, all of which have binding character. blocking minority stake, rather than Neither the law nor the courts have estab- additional legal rights. The final control lished any consistent framework for valuing threshold occurs with 95% ownership, a company or assessing its risk.10 As to when the remaining minority share- offers being made below trading prices, the holders can be compulsorily bought out on courts have adopted the line that market disadvantageous terms. ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 667 2.2. Ownership structure and multiple blocks, ownership in these firms is takeover defences not usually dispersed, but rather fragmented.This, we would argue, increases the Given the various control thresholds scope for hostile stakebuilding, especially identified, how frequently are blocks of each where blockholders are in disagreement size observed? Table 26.2 gives a complete over corporate policy or where blocks come breakdown of the ownership structure of up for sale. all listed German firms in September 1991 Table 26.3 lists the frequency of four by the size of the largest disclosed block, or types of takeover defences amongst the pooled block where a pooling contract population of all German listed companies exists. As can be seen from Panel A of in September 1991: instances where only Table 26.2, 72% of all listed companies non-voting preference shares are publicly have a majority owner: 17% have blocks in traded; voting right restrictions which cap excess of 95% of capital (or votes, if dif- the number of votes any individual share- ferent), 24.6% are super-majority con- holder can cast; limitations on the transfer- trolled, and 30.5% have a simple majority ability of shares (which give the target’s owner.Therefore, only 28% of companies— board the option not to register the shares 141 firms—are not at least majority con- in the new shareholder’s name, thus effec- trolled by one blockholder. Panels B and C tively disenfranchising the stake); and take a closer look at these 141 firms. 86 of departures from the principle of one-share- these have one or more blocking minority one-vote, as when certain classes of shares stakes (Panel B), where, of course, two or (usually held by friendly parties and not three such stakes would have amounted to traded) have greater voting power at all combined majority or super-majority control times or under certain circumstances (for had the various blockholders pooled their instance, for the purpose of supervisory board votes (which to our knowledge they had elections). Not surprisingly, few of the 402 12 not). The remaining 55 firms, in Panel C, majority controlled firms have takeover have no stakes in excess of 25%. Thirty- defences: 8.2% list only non-voting shares, seven of these have one or more disclosed 3.5% restrict the transferability of stock, non-blocking stakes, while 18 firms are and 4.7% have stock with differential classified as widely held because no stakes voting rights (Panel A). Limits on transfer- were disclosed at all (though Baums and ability and differential voting rights are the Fraune, 1994, claim that banks, despite main takeover defences amongst the 86 only owning an average of 13% of widely firms without majority owners but with one held firms’ equity, control more than 80% or more blocking-minorities (Panel B). of votes via proxies). Mannesmann, the However, 79% of these 86 firms have no target of Vodafone–Airtouch’s recent formal takeover defences. It is amongst the successful hostile tender offer, was one of firms without blocking-minority stakes that those rare widely held companies. takeover defences are most common. 42% From the point of view of corporate gov- of these 55 firms have defensive share ernance, the firms in Panels B and C are structures, with voting right restrictions very interesting. In the absence of a major- being the most popular (Panel C). ity owner, there are three alternatives for How many of the 141 non-majority con- how these companies are controlled. They trolled firms in Panels B and C could may either be (i) run by a coalition of non- potentially become targets of hostile stake- pooling blockholders, (ii) controlled by the building? Clearly not all of them: firms with dominant blockholder with the connivance three blocking-minority shareholders would of the remaining blockholders, or (iii) run be relatively hard to buy into, if only because by management without much shareholder each blockholder, being pivotal, would influence at all, in a way reminiscent of demand a control premium. To answer widely held firms in the US or the UK. the question, we propose four possible Given the presence of sizeable and often stakebuilding strategies (see Table 26.4). 668 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 26.2 Ownership of German stock exchange listed companies (1991)

Panel A: Majority control Number of firms (%) of sample

Single or formally pooled 95% block 95 17.0% Single or formally pooled super-majority 137 24.6% (75% ) owner of which have... One additional disclosed blockholder 11 Two or more additional blockholders 2 Single or formally pooled majority 170 30.5% (50% ) owner of which have... One additional disclosed blockholder 59 Two or more additional blockholders 22 Total with majority control 402 72.0%

Panel B: Blocking-minority control Number of firms (%) of sample

One blocking minority (25% ) block 47 8.4% of which have... One additional disclosed blockholder 8 Two or more additional blockholders 12 Two blocking minority (25% ) blocks 28 5.0% of which have... One additional disclosed blockholder 8 Two or more additional blockholders 3 Three blocking minority (25% ) blocks 11 2.0% of which have... One additional disclosed blockholder 3 Total with blocking-minority control 86 15.4%

Panel C: No blocking-minority control Number of firms (%) of sample

One or more non-blocking blocks ( 25%) 37 6.6% of which have... One block 12 Two blocks 9 Three blocks 5 Four or more blocks 11 Widely held (no blocks disclosed at all) 18 3.2% Total with no blocking minority control 55 9.8%

Ownership information not available 15 2.7%

Grand total 558 100.0%

Source: Own calculations based on “Saling, 1992,” Hoppenstedt’s stock market yearbook. Notes: The law requires disclosure of blocks of more than 25% and more than 50%. Frequently, smaller blocks are also disclosed (as in Panel C). Non-disclosure need not imply non-existence. ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 669 Only non-voting Cap on Limited share Departures from 25%) 10 3 1 block 21 3 2 ) blocks 1 ) blocks 1 1 3 4 ) block 2 3 3 Defensive share structures of German stock exchange listed companies (1991) companies listed stock share structures exchange Defensive of German ) owner ) owner Grand totalGrand 37 21 25 32 Ownership information not available (15 firms) not available information Ownership 2 1 2 3 (% of all firms with no blocking minority) with no blocking (% of all firms 0% 30.9% 5.5% 5.5% Widely held (no blocks disclosed at all) disclosed (no blocks held Widely Total 7 0 17 2 3 3 Panel C: No blocking-minority control (55 firms) ( blocks One or more non-blocking (% of all firms with a blocking-minority)(% of all firms 2.3% 3.5% 7.0% 8.1% Three blocking minority (25% minority blocking Three Total 2 3 6 7 Two blocking minority (25% minority blocking Two Panel B: Blocking-minority control (86 firms) (25% minority One blocking Total owner) with majority (% of all firms 8.2% 33 3.5% 4.7% 14 19 Single or formally pooled majority or formally Single (50% 5 5 11 (75% Panel A: Majority control (402 firms) pooled 95% or formally Single Single or formally pooled super-majority or formally Single 7 6 6 Table 26.3 Number of firms with... shares listed voting rights transferability one-share-one-vote companies) listed (% of all German Source:“Saling, based on Own calculations 1992,” yearbook. Hoppenstedt’s stock market 6.6% 3.8% 4.5% 5.7% 670 ON TAKEOVER AS DISCIPLINARY MECHANISM mber of potential takeover targets of potential takeover mber ge number of votes to effectively rule out rule hostile to effectively of votes ge number Impediments to hostile stakebuildingOnly non- Cap on Limited Number Departures of firms Totalnumber voting shares voting rights share transferability one-share- takeover from without of firms listed one-vote defences 50%) Potential targets for hostile stakebuilders hostile targets for Potential less than 25% less than 25% • Firms with three blocking-minority stakes and a free float of and a free float stakes with three blocking-minority • Firms • Firms with two blocking-minority stakes and a free float of and a free float stakes with two blocking-minority • Firms • Firms with two blocking-minority stakes and free float of 25% and free float stakes with two blocking-minority • Firms • Firms with one blocking-minority stakes and a free float of 25% and a free float stakes with one blocking-minority • Firms • The remaining firms with one or more non-blocking stakes • with one or more non-blocking firms remaining The • Firms with one or more non-blocking stakes and free float of 50% and free float stakes with one or more non-blocking • Firms • Widely held firms of 50% and a free float stake with one blocking-minority • Firms Grand totalGrand 141 110 two existing blocking-minority stakes blocking-minority two existing Strategy 4:Strategy addition, in if bought out over take (potentially) could 25 1 1 2 0 21 one existing blocking-minority stake plus open-market purchases plus open-market stake blocking-minority one existing Strategy 3:Strategy addition, in if bought out over take (potentially) could 30 1 1 3 0 25 one or more or all existing non-blocking stakes non-blocking one or more all existing Strategy 2:Strategy addition, in if bought out over take (potentially) could 7 0 1 1 0 5 (free float stakebuilding. Table 26.4 Firms which could become the target of hostile stakebuilding (in order of increasing difficulty) companies) listed (% of all German B and C inNotes: Panels appear in which of the companies 26.1. is based on an analysis table This Table column, In the final lar to a sufficiently the applied rights or multiple voting to share transferability limits nu if the existing was reduced if and only 25.3% 19.7% Strategy 1:Strategy the open-market in over take (potentially) could 79 0 17 2 1 59 ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 671 First, a hostile bidder could attempt to build stock, one had sufficiently privileged stock stakes in firms that have a free float of to block hostile approaches. Depending on 50% or more, without buying any existing the effectiveness of each of these takeover blocks.The companies that potentially fall into barriers, the number of firms at risk this category are widely held firms; compa- from hostile stakebuilding could fall to 110 nies with one blocking-minority stake; and (if all takeover strategies are considered) firms with one or more non-blocking- or as low as 64—just 11% of all German minority stakes. Both Mannesmann and listed firms—if only the first two strategies Thyssen belong in this category. Checking are considered realistic. for the presence of additional disclosed stakes and computing free float, we find 79 companies with a free float of 50% or more.13 Second, a bidder could take over a 3. DATA AND METHODOLOGY company by buying one or more or all existing non-blocking stakes; there were The cases for our clinical analysis were seven firms potentially at risk from this identified using the Financial Times strategy. A third takeover strategy is to Mergers and Acquisitions (FTMA) data- buy-out one existing blocking-minority stake base. FTMA contains structured templated and to make up the difference to 50% via information on changes in the ownership open-market purchases. Thirty firms had structure of European firms, covering one or two blocking minority stakes and (i) listed and unlisted firms; (ii) takeover enough free float for this strategy poten- bids and stake purchases; and (iii) transac- tially to allow a majority takeover.The final tions which are completed as well as those strategy involves a bidder buying out two merely rumoured or still under negotiation. existing blocking minority stakes, for All ‘forms’ compiled by FTMA between instance, in companies which have two or January 1988 and December 1996 were three such stakes and too little free float to obtained for transactions involving German make any of the other strategies viable. companies as targets. This yielded 2511 Therefore, of the 141 non-majority conforms, some of which pertain to the same trolled, some are much more vulnerable to firm at different points in time, or to differ- hostile stakebuilding than others. In prac- ent bidders for the same firm at the same tice, the first two strategies look much more point in time.Two types of filters were used feasible than the last two, suggesting the to manage this very large amount of raw potential takeover targets might number data. ‘Negative’ filters were used to around 86, which represents just 15.4% of eliminate, without further investigation, the total number of listed German firms. FTMA-reported transactions with a low If the defensive features summarised in likelihood of being motivated by ‘hostility’. Table 26.3 are effective, some of these 86 These negative filters were: firms are unlikely to be taken over.Table 26.4 takes this into account by looking in detail 1. cases of apparent initiation of cross- at the nature of any takeover defences.Two shareholdings (which require both companies, for instance, listed only non- parties’ agreement) voting preference shares, making them vir- 2. “participation in capital increases tually immune to hostile stakebuilding.14 A via subscription” (which presumably further 20 had voting right restrictions, are agreed between buyer and tar- though we will argue later that the empiri- get, and thus not hostile) cal effectiveness crucially depends on the 3. unlisted companies other than those level of the cap. Of the 11 companies with captured using a positive filter (see limited share transferability, eight below) restricted transferability sufficiently to 4. complete or partial disposals of make a bidder’s life difficult. Finally, of the divisions or other operating units 21 companies that had multiple classes of 5. privatisations by the Treuhandanstalt 672 ON TAKEOVER AS DISCIPLINARY MECHANISM Second, we used positive filters designed deals conventionally classified as friendly to identify cases with a high probability of and hostile, and suggest that both being motivated by hostility, and devoted management resistance and publication more time to these cases: thereof are strategic bargaining ploys to improve the bid terms for target share- 1. firms known to be potential takeover holders. While this point is pertinent to targets using our knowledge of their situations where stakebuilding is followed ownership structure discussed above by tender offers (which affect target share- (no majority owner, no pooled holders), very few of our cases involve majority etc.) subsequent tender offers (and those that 2. multiple filings over time for the do offer minority shareholders very poor same firm terms). 3. cases where FTMA reported On the basis of the news stories, 17 transactions by ‘undisclosed bidders’ prima facie hostile cases were identified. We make no claims as to the comprehen- For firms not eliminated by the negative siveness of our search strategy: our filters filters, all electronically available news story may well have filtered out hostile cases, headlines were read, and where appropri- particularly amongst unlisted firms. The ate, the story itself,15 around the year indi- filters were necessary, however, given the cated in FTMA to establish the nature of otherwise unmanageably large number of transaction: friendly, negotiated, don’t know, FTMA filings, each of which would have openly hostile, etc. At this stage, a 6th neg- required a news search to establish the ative filter was used to exclude companies background. in financial distress as the change in owner- All news stories before and after the ship structure presumably reflects rescue FTMA date were then read line by line for operations. Most effort was spent reading the 17 cases. In some instances, this about transactions that passed the positive involved reading more than 1000 articles filters. (e.g. 1283 articles on the Continental In common with existing research in the case). From these readings, we established area of hostile takeovers, we do not have a the players, the sequence of events, precise definition of hostility. Instead, we information on ownership changes and look for such ‘telltale signs’ as public prices paid (where available), the bidder’s resistance by target management to block- (apparent) motivation, the target’s building, for instance taking the form of response, the outcome, etc. We augmented verbal exchanges or actions designed to the news information with security price ward off further block-building (e.g. dis- data (from Datastream) where we had couraging further stake sales) or reduce information on dates (e.g. date of block the new blockholder’s influence (e.g. impo- purchase) and with ownership data from sition of voting restrictions). Clearly, this the standard German sources: Hoppenstedt’s unavoidably biases our sample identifica- Saling stock market yearbook; tion since we must rely on reportable and Com-merzbank’s Wer gehört zu wem? reported reactions by target management. Tri-annual register of corporate ownership; Where disputes take place, and stay, behind and the electronic Amadeus database, closed doors, we are unable to establish the which uses ownership files compiled by blockbuilder’s intent and the target’s Creditreform, a credit reference agency. reaction. While this is unavoidable, a We refrain from performing an event potentially more serious concern is study on our cases for three main reasons. Comment and Schwert’s (1997) argument First, in contrast to the US or the UK, that using press coverage to distinguish where event dates are easy to identify from between hostile and friendly tender offers SEC or stock exchange filings, it is may be misleading. Comment and Schwert extremely difficult to identify the dates of find little economic difference between German block trades: the notification ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 673 requirement is to the target company in the industry cases. In only two instances are first instance. Second, the way blocks are the control contests of a conglomerate acquired is often, and deliberately, covert, nature.This suggests that the bidders’ moti- even where formal notification would in vation may often be industrial or strategic principle be required. Finally, block trades rather than disciplinary: stakebuilders may often receive press coverage only once seek to reduce overcapacity in their indus- further buying activity or target responses try or increase market power rather than add up to a pattern, resulting in substantial ‘punish’ wayward managers. Since target reporting delays. managers would expect to lose their jobs whatever the stakebuilders’ motivation, their resistance to the control changes is 4. THE CASE STUDIES not surprising. The strong horizontal (and vertical) bias results in the Cartel Office The 17 cases of hostile stakebuilding pro- being involved in a relatively high propor- vide an insight into the way corporate gov- tion of the cases. Indeed, the Cartel Office’s ernance is exercised in Germany. Brief frequent involvement supports the view summaries of each individual case are that market concentration or consolidation provided in Appendix A. In this section, we motives play an important role in our cases. first describe the sample companies in The final column of Table 26.5 attempts terms of firm characteristics, shareholder to categorise the stakebuilders’ apparent structure, and defences in place. We then motivations. In their study of the US block turn to the various strategies employed by market, Bethel et al. (1998) distinguish hostile stakebuilders, including the forma- between ‘activist’ block-buyers (usually tion of coalitions, the role of banks and individuals whose announced intention is to ways to avoid disclosure. We next consider influence firm policy) and ‘strategic’ block- the defensive actions taken by targets, buyers (other companies whose purchases including voting restrictions, denial of are unopposed by target management). Our board representation and the use of defen- stakebuilders do not easily fit this distinction: sive coalitions and white knights. Finally, most of the stakebuilders are companies— we discuss the outcomes of the different not individuals as documented in the US by types of control battle that are observed. Holderness and Sheehan (1985) and Bethel et al. (1998)—but we cannot call them ‘strategic’ since they were opposed 4.1. Company characteristics by target management and announced ‘activist’ style intentions. The classifica- Table 26.5 lists the 17 targets and their tions we offer are instead based on the respective suitors.The control contests were extent to which the stakebuilders sought to initiated between 1987 (Axel Springer become involved—ranging from gaining Verlag) and 1994 (Kolbenschmidt) though, influence over corporate decision-making due to the often covert nature of stake- via blocking minority stakes (as in the AMB, building, precise dating is difficult.The tar- Buderus, Th. Goldschmidt and Holzmann gets are invariably from ‘traditional’ cases) to gaining majority control (as in industries, ranging from construction the Axel Springer Verlag case) or effecting (Philipp Holzmann and Dywidag) to insur- a takeover (as in the Asko, BIFAB, Boge, ance (AMB and DBV). One of the most Bopp and Reuther, Conti, DBV, Dywidag, noticeable characteristics of the case stud- Hoesch, and Kolbenschmidt cases). We ies is that the stakebuilders mostly operate label six of the nine intended takeovers as in the same industry as their targets. ‘consolidating takeovers’ because the Indeed, many cases are reminiscent of stakebuilders—all of whom operated in the Jensen’s (1993) over-capacity argument, for same industry—made public statements instance, the two construction cases, the concerning the need to remove overcapacity Krupp–Hoesch steel case, or any of the car (steel being an obvious example).The large 674 ON TAKEOVER AS DISCIPLINARY MECHANISM target ate decision making ate decision investment” 25% via block corporate decision-making corporate ontal takeover consolidating Industrial Stakebuilder’s lomerate/ – disciplining ia/publishing horizontal over control gaining ividual management ocess engineering horiz pulp and paper many/UK tools machine vertical greenmail possibly Sweden/Sweden pulp and paper/ management all from Germanyall from med Germany ind Country relationship apparent erke vs. APV Ger brothers vs.brothers heirs Springer corpor chemicals Germany conglomerate takeover possibly metal products Stakebuilders and targets Stakebuilders ühle Nobel conglomerate vs. Veba vs. Stora SCA Germany/ cong Noll (SEN) Widmann (Dywidag) Widmann Feldm Versicherung (DBV)Versicherung Switzerland and F.A.Brockhaus (BIFAB) Communication Münchener Beteiligungen Münchener Target company Industry Stakebuilder of origin Industry with target motive a ase 1988 Seitz–Enzinger– engineering process Klöckner W 1992 Wünsche conglomerate WCM and Diekell Germany/ conglomerate/ – target disciplining 1994 Kolbenschmidt car parts T&N UK car parts horizontal takeover 1989 Holzmann Philipp construction Hochtief Germany construction horizontal influence gaining 1991 Hoesch steel Krupp Germany steel horizontal takeover consolidating 1988 1991 Th. Goldschmidt metallurgy/industrial VIAG vs. Veba/Rütgers Germany/ conglomerate/ – influence, gaining 1988 und Dyckerhoff construction Group Walter Germany construction horizontal takeover consolidating 1993 Deutsche Beamten– insurance Commerzbank/Winterthur Germany/ bank/insurance horizontal takeover 1989 Continental tires Pirelli Italy tires horizontal takeover consolidating 1994 Buderus systems/ heating and Berger Bilfinger Germany construction vertical strategic “long-term 1989 Bopp and Reuther engineering process IWKA Germany pr 1988 Boge car parts Sogefi Italy car parts horizontal takeover consolidating 1988 Institut Bibliographisches publishing Maxwell UK media/publishing horizontal takeover 1987 Springer Axel Verlag media/publishing vs. group Kirch Burda 1992 Asko retailing Metro Switzerland retailing horizontal takeover consolidating B Table 26.5 year 1991 und AMB Aachener insurance AGF France insurance horizontal over influence gaining a contest began. the control which in as the year is defined Base year ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 675 number of such ‘consolidating takeover’ bids Feldmühle Nobel and Hoesch). This is not lends further support to the importance of surprising as such caps clearly make it ‘industrial’ rather than disciplinary motiva- more difficult (although as we shall see tions as the primary driver of hostile stake- later, not impossible) for a stakebuilder to building. Only the two cases involving establish a controlling interest. targets that were conglomerates appear to The final block of Table 26.6 provides have been motivated by a desire to remove some performance indicators prior to the ‘underperforming’ management. Finally, control contests. The targets are mostly APV’s stakebuilding in the SEN case is large companies, with nominal annual sales perhaps best understood as an instance of averaging DM 3.6 billion in the 3 years intended greenmail, as we will show later on. prior to the control contests. The smallest A final striking feature of the stake- target (BIFAB) had annual sales of DM 44 builders is their nationality: nine cases million, and the largest target (Asko) had involved non-German stakebuilders, includ- DM 13.2 billion. All but four of the targets ing most of the earliest cases. This is con- had experienced positive sales growth in sistent with ‘outsiders’ undermining the the 3 years leading up to the control prevailing ‘governance consensus’. contests, with double-digit growth rates in In line with our discussion in Section 2.1, five cases. The earnings dynamic is more in Table 26.6, we classify the target mixed, with three firms moving into losses companies according to their shareholding over the 3 years pre-contest, three firms structure in the run-up to their control con- experiencing negative earnings growth, and tests.Three of the companies were majority the remaining 11 seeing sometimes quite controlled prior to the control contest. substantial increases in earnings. In the SEN had a 50.01% owner but still became year their control contests began, four the target of a control contest. Wünsche firms were loss-making and six firms was majority controlled by two brothers earned less than an 8% return on equity. who later fell out. Bopp and Reuther, the The remaining seven, however,would appear only non-listed target in our sample, was to be performing healthily, with return on controlled by a large number of family equity ranging from 10% to 22%. shareholders and one outside 25% blockholder, who originally pooled their 4.2. Stake accumulation tactics votes. Of the remaining 14 companies, there was a blocking minority shareholder in six Hostile stakes can be formed either via cases, and in the remaining eight companies, open-market purchases or by purchasing ownership was initially dispersed. blocks from existing shareholders. As The middle block of Table 26.6 docu- Table 26.7 shows, stakes are often built ments the takeover defences in place in the partially through open-market purchases, run-up to the control contests. Eight of the but limited secondary market liquidity typ- 17 targets had some form of takeover ically precludes a controlling stake being defence. Not surprisingly, the three majority accumulated without also buying (or form- controlled firms in Panel A had no formal ing coalitions with) existing blocks. In con- takeover defences. Amongst the six trast to the predictions of the theoretical companies in Panel B, which initially had literature (such as Zwiebel, 1995) there blocking minority shareholders, two are usually multiple large blockholders in (Axel Springer Verlag and DBV) limited our cases. This greatly increases the com- the transferability of shares, and one plexity of the game for each individual (Th. Goldschmidt) had differential voting blockholder.In particular,each blockholder rights. The most popular defence for will fear not being part of a controlling companies with fragmented ownership coalition when a hostile stakebuilder structures appears to be caps on voting emerges, given the high risk of being diluted rights, which were in place in four of the eight in Germany. It is not altogether surprising, companies in Panel C (Asko, Continental, therefore, that in a large number of cases 676 ON TAKEOVER AS DISCIPLINARY MECHANISM b 0.7 0.3 5.1 sales b on on c growth losses c 7.3 into move n.a. 10.8 into move n.a. Sales Earnings Return Return b Performance prior to stakebuilding a Cap on Limited Departure Sales Defenses voting share from one- DM growth several smaller stakes smaller several one smaller stake one smaller with pooling contract with pooling blocking-minority stake blocking-minority had majority control,had majority one minority outside blocking Shareholder structure prior to rights transferability share-one- million (%) (%) equity Pre-contest target company ownership structure,Pre-contest target company ownership defenses, takeover and performance Buderus stake, one blocking-minority 2696 3.9 11.4 16.2 2.5 BIFAB stakes, two blocking-minority 44 8.1 51.8 n.a. Verlag stakes smaller several Panel B. Blocking minority stake(s), prior to stakebuilding SpringerAxel stake, one blocking-minority yes 2220 28.4 218.8 19.1 2.6 Wünsche stakes two blocking-minority 1833 6.6 29.5 21.2 1.5 SEN one 50.01% stake, one 378 Reuther contract, bound by a pooling losses Table 26.6 Target company stakebuildingPanel A. Majority controlled, prior to stakebuilding Bopp and shareholders, family several vote 185 (%) (%) ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 677 3.8 4.7 10.2 2.3 25.9 2.7 0.4 29.6 3.3 0.1 losses 3.9 into move n.a. 3.0 1197 7.2 16.0 22.0 4.1 8846 d e stakes several smaller stakes smaller several several smaller stakes smaller several hle Nobel small stakes several 5% Feldmü Dywidag small stakes several 1459 9.2 Continental at least two non-blocking 5% 6588 21.2 26.2 16.0 2.6 Boge stakes two non-blocking 537 8.8 25.4 7.3 1.0 Asko small stakes several 5% 13,188 20.1 Panel C. Fragmented ownership, prior to stakebuilding AMB small stakes several yes n.a n.a. 4.8 5.4 n.a. Th. Goldschmidt stake, one blocking-minority yes Hoesch stake at least one non-blocking 15% 10,426 8.7 29.7 5.3 0.9 Kolbenschmidt stake, one blocking-minority 1449 e rumours. response to stakebuilding Adopted in c 2 years. and the previous the base year in rate growth Measured as the average d 1.3% of shares have 11.6% votes. DBV stakes two blocking-minority yes HolzmannPhilipp stake at least one non-blocking yearbook. stockSource: market issues of the Saling and various Hoppenstedt Bilanzdatenbank data comes from accounting a shares listed. non-voting have only we analyse None of the target firms b 3 years. and the previous the base year in Measured as the average 2199 12.6 35.8 4.7 1.7 3937 32.3 142.7 12.9 1.2 678 ON TAKEOVER AS DISCIPLINARY MECHANISM Bank assists ✓✓ ✓✓ ✓ ✓ ✓ ✓✓ Acting in Bank stake- Disclosure ✓✓✓ ✓✓✓ 1 share stake from family shareholders;VIAG’s family of from origin ategy 25% further stake later to gain majority control majority to gain further later stake (the deceased founder) targeted bid response to Maxwell’s in family threats; takeover DBV from to protect bought a have option on 16%,have option unclear; source Burda Springer Axel from bought 24.9% stake brothers by brother Kai Wünsche, Kai by brother Wünsche. the CEO of SEN with KW’s subsidiary (KW had bid for the for (KW had bid SEN with KW’s subsidiary WCM with a bank) to parked 6% were (of which and Dieckell, who pooled; sale was welcomed this to at a 25% discount sold was allegedly stake The price market brother,W-J. 25% Wünsche sold 40% block); at DM175 remainder for APV tender Kommunale Landesbank) to foil forced merger of merger forced Landesbank) to foil Kommunale shareholders, agreement despite pooling minority shareholders (stake held by Badische held (stake shareholders minority lding str Stakebui Th. Goldschmidt Veba: Metallgesellschaft, 20% from to increased DBV at the IPO, took a stake Commerzbank ostensibly BIFAB Meyer from stake bought initial Langenscheidt Buderus unknown, of IPO at time but possibly Panel B: Blocking minority stake(s), prior to stakebuilding SpringerAxel Verlag bought 10% at the IPO; Kirch also to claimed Wünsche feud with his to end a long-standing possibly SEN by SEN on 40% holding options APV offered Table 26.7 Target Source of stake concert stake builder problems Panel A: Majority controlled, prior to stakebuilding Bopp and Reuther of family a group from IWKA bought stake ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 679 ✓✓ ✓ ✓✓✓ ✓ ✓✓✓ ✓ ✓✓✓ ✓ ✓✓✓ ✓ ✓✓✓ ✓✓ incumbent management incumbent knight—who sold stake on to Walter on to stake sold knight—who and their coalition of five others (40%, others of five and their coalition originally to win failed once the coalition assembled) secretly contest due to house bank supporting the proxy Walter from gaining control by selling stake to white stake by selling control gaining from Walter 24.9% stake; to prevent were Holzmann’s intentions “white knight” bought Holzmann’s pivotal Advanta from the founding family, Siemens and its house the founding bank; from each from Asko (which Asko owned via a subsidiary) via owned Asko (which Asko each from to cement a joint venture;to cement a joint however, then Metro the open-market bought further shares in emerged when there were rifts in the family in rifts were when there emerged (with whom concert) they acted and boosted in purchases open-market via this shareholder coalition and rumours of a rift and rumours coalition shareholder rival 20% unknown;rival the two new blockholders (3%) investors institutional (4.9%),(10%) and Commerzbank its house bank between management and family shareholders between management and family (47%) who were in liquidity crisis; liquidity in (47%) who were Magna Canada (2.5%);International and other hle Nobel owner bought out the two nephews of former Stora Feldmü Hoesch purchases open-market Holzmann Advanta from on) stakes Hochtief bought (options BogeContinental open-market open-market AskoDywidag and its house bank, Metro WestLB, bought 10% acquired a total of 38% stakebuilder,Walter, hostile Panel C: Fragmented ownership, prior to stakebuilding AMB AB Skandia from AGF bought 1.8% block Kolbenschmidt Metallgesellschaft from T&N bought stakes 680 ON TAKEOVER AS DISCIPLINARY MECHANISM existing pooling agreements break down in given the horizontal nature of many of the face of a hostile stakebuilder.For exam- these takeovers, the threshold beyond ple, in the Th. Goldschmidt case, the family which Cartel Office approval is required. shareholders had a pooling agreement with Disclosure problems are of particular Allianz that gave the coalition majority concern since most German companies— control. When VIAG emerged as a stake- unlike British or American ones—issue builder,Allianz let its pooling contract expire bearer shares, making it hard to identify and sold out to VIAG, who thereby gained one’s shareholders. During the period of our control. In the Bopp and Reuther case, the investigation, company law required only pooling agreement that established major- holdings in excess of 25% and 50% to be ity control was actually violated by certain revealed. This contrasts with the 5% shareholders, which precipitated the control threshold in the US (Regulation 13-D) and battle in the first place. the 3% threshold in the UK. The ability of It is also interesting to observe the role a predator to build a stake without other of banks during the control contests. As market participants realising is important Table 26.7 shows, banks held significant as it reduces the likelihood of other share- stakes in 12 of the 17 cases, and often holders eliminating potential takeover sided with the predator (either through sale gains by free-riding,17 and it prevents the or pooling). In some cases, the bank stakes potential target from initiating some form appear to have been held for a long time, of anti-takeover protection such as limiting although there are also examples where the votes that can be cast by any individual investment banks essentially built stakes in shareholder. Germany’s disclosure rules partnership with a predator, and appear to have been progressively tightened in recent play a pivotal role in brokering a control- years,18 but it is unclear whether this in ling coalition. Perhaps the most interesting itself limits the ability to build a secret example is the Hoesch case, where three stake: as our cases demonstrate, banks or banks held significant stakes: Deutsche Bank other friendly parties can act in concert to (Hoesch’s house bank);WestLB (Krupp’s— help companies build large stakes by com- the stakebuilder’s—house bank) and Credit bining a number of smaller stakes that indi- Suisse (who was acting for Krupp). Credit vidually do not have to be disclosed. Note Suisse, on behalf of Krupp, secretly accu- also that the law requires no notification mulated an initial 24.9% stake. Later in for block increases between 25% and the battle, Credit Suisse accumulated a fur- 49.9%, so a stakebuilder could secretly ther 20% stake, pledging it in support of build a near-controlling block. Krupp. While WestLB did not declare its Our case studies show how weak, in prac- 12% stake in support of Krupp during the tice, disclosure rules are in Germany. For battle (to avoid voting restrictions), it later example, by splitting up his 38% stake worked with Krupp to have voting restric- between himself and some of his companies tions removed and seal the takeover. and two children, Walter avoided revealing Interestingly, Krupp also enjoyed the support his ownership interest in Dywidag. In the of Deutsche Bank—Hoesch’s house bank— Feldmühle Nobel case, the Flick brothers which was estimated to control around 12% were able to assemble a secret 40% stake of the votes (partly via proxies).16 via open-market purchases and undisclosed The secretive nature of stakebuilding in agreements with institutional blockholders. the Krupp case is a common feature of our Indeed, having sold this stake to Veba, the case studies. There is clearly much to be Flick brothers secretively assembled a second gained from not alerting other sharehold- stake, which they used to extract further ers that a significant stake, or coalition of surplus from Veba. Effective stakes can also stakes, is being assembled.Two particularly be hidden by the use of option contracts, such relevant considerations in this respect are, as in the Axel Springer Verlag case, where first, the rules governing disclosure of the predator had a disclosed stake of 10% stakes and acting in concert, and second, but an undisclosed option on a further 16%. ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 681 Stakes in excess of 24.9% must be response to the emergence of an unwanted reported to the Cartel Office, whose per- stakebuilder. The main defensive actions mission is required before they are increased. are to forge an alliance with friendly block- However, even the Cartel Office rules can holders, seek the assistance of the house be stretched. For example, in the Hochtief bank, line up a white knight, and reduce the case, Commerzbank enabled its client to stakebuilder’s influence by denying him a acquire an effective stake in excess of the supervisory board seat. As Table 26.8 24.9% threshold by purchasing an addi- shows, almost all the target companies in tional block and granting a call option on it our sample attempted to forge (or main- to Hochtief. Commerzbank was paid an tain) a defensive alliance of friendly block- undisclosed fee to finance the cost of carry; holders. In many cases, white knights were hence, all the economic risks were borne by sought by the incumbent management, in Hochtief who, in all but title, thus ‘owned’ some cases apparently motivated, at least the stake.19 Monopolies Commission reports in part, by a desire to thwart a foreign have highlighted a number of other such suitor. This was certainly the case in the cases. A particularly interesting—albeit battle for BIFAB, where the board “resis- extreme-case is the 1985 takeover of ted a takeover by a foreign firm” and found Deutsche SB-Kauf by Asko, a rival retailer. a white knight in Langenscheidt, who ulti- Acting in concert with three banks, Asko mately took the firm private. Similarly, the avoided a Cartel Office investigation by attempted takeover of Boge by Italian rival arranging for all parties to hold no more Sogefi was resisted in favour of a German than 24.9% each, thus accumulating a white knight (Mannesmann). However, stake of 99.6%! white knight defences do not always work The critical issue is clearly the regulation as expected. For example, Advanta adopted of parties acting in concert, which histori- the role of white knight in defence of cally has been very weak in Germany. Such Dywidag against the unwelcome attention matters are beginning to come under of Walter,but then sold the critical controlling review by the German courts in the wake of stake to Walter! the case brought by the Cartel Office in the The other main defence that we observe Hochtief–Holzmann bid. However, while is denial of representation on the supervi- there may be some tightening of the rule sory board. As noted in Section 2, the regarding breaching the Cartel Office’s supervisory board appoints, or removes, the 24.9% threshold, it seems less likely that management board, and a blocking minor- regulations regarding the disclosure of ity can prevent changes in the supervisory beneficial ownership or, more crucially, board. Hence, we tend to see such defences effective control, will be changed sufficiently employed where initial ownership includes to remove the ability to build secret stakes a blocking majority. The Dywidag case in excess of the formal disclosure limits. illustrates how powerful a blocking minority can be in this respect: even a stake as 4.3. Defensive actions large as 40% need not guarantee even a single seat on the supervisory board in the During our sample period, German firms face of target opposition. Denial of super- were prevented by law from using such visory board representation can certainly popular US defence tactics as share buy- thwart a blockholder’s attempts to change backs, issuing shares with multiple voting management or corporate strategy. However, rights, poison pills or recapitalisations tar- in some cases where the initial goal of the geted at white knights. However, they did stakebuilder was “co-operation and influ- have a number of other defences against ence” such denial by the target can provoke hostile stakebuilders.We distinguish between a full takeover. For example, the board of pre-existing takeover defences, such as the Th. Goldschmidt denied major blockholders voting restrictions that we documented in Veba and VIAG (who between them, held Table 26.6, and defensive actions taken in nearly 46% of the shares) representation 682 ON TAKEOVER AS DISCIPLINARY MECHANISM ✓ ✓ ✓ ✓ ✓ ✓ Bank Deny Caps on Limited Defensive assists White supervisory total votes transferability ✓✓ ✓✓ ✓✓ ✓ ✓✓✓ ✓✓ ✓✓✓✓ ✓ ✓✓ ✓✓ Defensive actions Voting restrictions ✓✓ ✓✓✓✓ Defensive actions Defensive hle Nobel Asko Boge Dywidag Hoesch Kolbenschmidt Panel C: Fragmented ownership, prior to stakebuilding AMB Continental Feldmü BIFAB Buderus DBV Th. Goldschmidt Table 26.8 TargetPanel A: Majority controlled, prior to stakebuilding Bopp and Reuther SEN Wünsche Panel B: Blocking minority stake(s), prior to stakebuilding Axel Springer Verlag alliances target KnightHolzmann board seat per block of stock ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 683 on the supervisory board. Initially Veba and significantly, at no stage declared its support VIAG were demanding a “change of direc- for Krupp—thus avoiding accusations of tion” but their frustration ultimately acting in concert. Surprisingly, Deutsche resulted in VIAG assembling a controlling Bank, who controlled about 12%, also stake, without offering to buy-out the fam- sided with Krupp despite being Hoesch’s ily blockholders who had denied it a seat on house bank. Hence, this case illustrates the the supervisory board. significant difference between voting A more potent form of defence, fre- restrictions of 5%, which can be a major quently employed by companies with frag- impediment to hostile stakebuilders, and mented ownership, can be the use of voting restrictions of 15%, which are typically restrictions. While it is not unheard of that much less effective in defending a target a company institutes voting restrictions company.22 It also illustrates that a target after a hostile stakebuilder has emerged, as cannot, as a matter of course, rely on its in the Feldmühle Nobel case, more usually house bank for its defence. In general, proxy restrictions are already in place.20 Such votes give banks both substantial influence restrictions typically limit the proportion of (because voting restrictions do not apply to votes that can be cast by any individual proxy votes which are ultimately not owned shareholder or in some cases, groups of by the banks themselves) and valuable shareholders operating under a pooling information about changes in ownership, agreement. However,voting restrictions can which could be used to obstruct, or accel- also take the form of limited transferability erate, a bid’s progress. However,proxies are of voting rights. For example, Continental not always effective. While Deutsche Bank had a restriction that no shareholder could successfully used its proxy votes to defend exercise more than 5% of the votes. Not Continental against Pirelli, its proxies surprisingly, this resulted in numerous proved ineffective in defending Feldmühle blockholders holding precisely 5% blocks. Nobel once shareholders themselves had When the Pirelli bid emerged, a group of decided to accept Stora’s bid. minority shareholders demanded an extraordinary general meeting to remove the 4.4. Outcomes voting right restriction.This required a simple majority and was duly carried. However, Classifying the outcomes of the control this decision was later overturned in the contests requires some care, as the ‘success’ courts when Continental alleged that Pirelli of a stakebuilder’s control challenge must had acted in concert with its allies.21 The be measured against his initial intentions. predator group should, therefore, have been For instance, where stakebuilders intended subject collectively to a 5% voting limit. to gain ‘influence’ (or,perhaps, co-operation) Continental continued to deny Pirelli and rather than majority control, a measure of its allies votes in excess of 5%, and this their success is whether they gain seats on restriction was clearly critical in thwarting the supervisory board, or whether they can the attempted takeover. effect changes in management or corporate Voting restrictions appear,however,to be strategy. In this spirit, Table 26.9 sum- much less potent as a defence when they marises the outcomes of the 17 control are set at slightly higher levels. For exam- contests. Judged relative to the initial ple, the 15% voting right restriction in intentions, the original hostile stakebuilder Hoesch’s corporate charter was not an was successful in 10 of the 17 cases. It is effective defence against Krupp. The 15% noticeable that there is no apparent restriction meant that it was only neces- relationship between initial ownership sary for Krupp (with its 24.9% stake) to structure and the success of the stake- convince two other significant blockholders building: the two main reasons for failure in to vote in favour of removal of the voting our cases are either the intervention of the restriction. One of these was WestLB, its Cartel Office or the appearance of a second house bank, who had a 12% stake but, ‘white knight’ stakebuilder. 684 ON TAKEOVER AS DISCIPLINARY MECHANISM 4.1 0.3 0.0 5.1 21.2 24.0 1.5 2.1 n.a. n.a. controlling Outcomerelative Return on equity Average Return on sales Average Average Average to of 3 years of 3 years of 3 years of 3 years Original original Management Supervisory pre- post- pre- post- apparent apparent Control board board Tender to contest contest contest contest erke, merger Outcomes stake to two stake brother beingbrother from removed the management managementboard. succeeded He then 26% his sold resulted in one resulted in target builders change resigned the latter sold its stake to its stake sold but ultimately shareholder controllingshareholder W Klöckner succeeded opposed seats for forced discount opposed greenmail builder change board vetoed at 18% sided with IWKA sided who then bought stakes remaining pooling broke agreement when down Finanz Hannover Wünsche feud a family disciplining stake- control CEO SEN stakebuilder rival possibly stake- control management bidder rival yes, n.a. 1.6 Reuther succeeded Defensive and 43% stake.and 43% takeover builder change dismissed Table 26.9 Target SummaryboughtPanel A: Majority controlled, prior to stakebuilding Bopp IWKA consolidating stake- motive control management motive outcome changes changes minorities (%) (%) (%) (%) ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 685 ) 0.7 4.4 Table 26.9 continued ( breakdown of abreakdown un- premium pooling successful: pooling agreement,with white Langenscheidt control gaining taking and later to sold knight prompted the prompted builder change at 10% who ultimately with Kirch when part of the Burda stake to the was sold family, Springer co-operated This broke down broke This making seat Springer family.Springer decision- operation given against theagainst corporate co- then CEO was removed sold out once the sold pressured the CEO to resign. Rumours about outsiders, who the CEO change led to 28% rise. share price stakebuilders The private BIFAB BIFAB the stakebuilding takeover stake- control Yes, n.a. 17.8 Springer initially family Verlag pooled stakes control over builders operation CEO succeeded: initially resignations denied Panel B: Blocking minority stake(s), prior to stakebuilding Axel and Burda Kirch gaining stake- co- multiple bidder 19.1 25.0 2.6 2.9 686 ON TAKEOVER AS DISCIPLINARY MECHANISM 22.0 15.2 4.1 3.2 4.7 8.0 1.7 2.5 Office stakebuilder succeeded Outcomerelative Return on equity Average Return on sales Average Average Average to of 3 years of 3 years of 3 years of 3 years involvement” un- seat Originalapparent original apparent Control board Management Supervisory board Tender to contest pre- contest contest post- contest pre- post- Continued This was denied. This VIAG then sought board representation. 45.8% stake) pooled and takeover (despite stakes, which they possibly succeeded seat denied gradually to 50% increased and then sold on to a Swiss insurer to increase.The was stake stake, which itundertook not succeeded helda 25% helda builder change blocked Cartel 25% but was block by blocked increase it to increase 25% via successful: stake, to tried bought a 15% strategic builder change given schmidt both bought influence, builder change initially Th. Gold-Veba VIAG and gaining stake- control bidders DBV Commerzbank takeover stake- control Table 26.9 Target SummaryBuderus stakebuilder motive “long-term stake- motive no bidder outcome changes changes minorities 16.2 (%) 21.4 (%) 2.5 (%) 3.4 (%) ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 687 ) 3.8 2.1 Table 26.9 continued ( n.a. 11.9 board turned after AMB’safter co- rights on rights stakeblocking operation supervisory influence builder operation resigned succeeded: placed 25% with Rheinmetall, also later acquired the 25% remaining granted options granted to T&N. Ultimately, Commerzbank acquired the and block to 2nd bidder appeal, blockCommerzbank sold Controlling the Cartel Office. an Pending by Cartel Office. prevented from by control taking successful: blocked block, but were un- by Allianz to by Allianz majority gain control stake owned stake bought Veba’s and a stake CEO against Panel C: Fragmented ownership, prior to stakebuilding AMB AGF won voting gaining stake- co- CEO 5.4 6.2 n.a. n.a. Kolben- a T&N built takeover stake- control schmidt builder controlling change 688 ON TAKEOVER AS DISCIPLINARY MECHANISM 3.3 5.5 0.1 0.5 Outcomerelative Return on equity Average Return on sales Average Average Average to of 3 years of 3 years of 3 years of 3 years Original original Management Supervisory pre- post- pre- post- apparent apparent Control board board Tender to contest contest contest contest Continued were ultimatelywere defeated by the court upholds bought 5% un- management stakes, but successful: board split also sold out to also sold Mannesmann a few months later Mannesmann. stake-initial Sogefibuilder to sold “grey knight” sold out to sold coalition successfully un-successfully formed.The groupcontrolling defensive successful: 64% discount coalition takeovercoalition builder change offer at then removed and takeover welcomed Voting succeeded Voting restrictions chair ousted. takeover builder change removed tinental each allies takeover builder change removed; Con- and Pirelli consolodating stake- no CEO 16.0 10.9 2.6 1.0 Panel C: Fragmented ownership, prior to stakebuilding Boge a defensive consolodating stake- control yes, initial 7.3 5.7 1.0 0.5 Table 26.9 Target SummaryAsko board supervisory consolidating motive stake- control motive outcome changes changes chairman minorities (%) (%) (%) (%) ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 689 ) 0.6 Table 26.9 continued ( 2.7 5.0 0.4 0.4 itial stake- disciplining stake- control yes, 10.2 13.4 2.3 2.1 24.9% stake takeover builder change board offer stakes buying both buying Veba. took Stora by control stake in to opposition control gains Flick brothersFlick a then built stakebuilder restrictions. 2nd blocked byblocked successful: voting management un- discount “white knight” Advanta, who on a then sold 25% critical Walter to stake 40% stake in40% stake with Dywidag to a view takeover. Dywidag responded by welcoming succeeded (despite 40% stake) constructed a takeover builder change seat denied Continental’s rightContinental’s cap and share price the failure voting to remove restrictions fall in voting and its advisor, succeeded members Hoesch a built Krupp consolidating stake- control management yes, share 5.3 n.a. 0.9 NobelVeba by building target builder change at 15% Feldmühle In Dywidag consolidating Walter stake- control bidder 690 ON TAKEOVER AS DISCIPLINARY MECHANISM Outcomerelative Return on equity Average Return on sales Average Average Average to of 3 years of 3 years of 3 years of 3 years Originalapparent original apparent Control board Management Supervisory board Tender to contest pre- contest contest post- contest pre- post- Continued exercising on options by held stakes Cartel Office increasing itsincreasing 20% long-held viastake unsuccessful: by blocked prevented from influence builder change removed. Among allies Krupp’s its house were bank and the target’s house bank restriction was restriction voting rights voting Credit Suisse,Credit also took a 20% stake. With other helpfrom blockholders, a suspended banks Holzmann Hochtief gaining stake- no 12.9 15.1 1.2 1.0 Table 26.9 Target Summary motive motive outcome changes changes minorities (%) (%) (%) (%) Source: data: accounting yearbook. for stock market issues of the Saling and various Hoppenstedt Bilanzdatenbank ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 691 An alternative way to assess the outcome changes resulting directly from the control of the control contests is to look at actual contest.We find senior management changes changes in control. Consider first the three in seven23 of the 17 cases.This rate of post- companies that were initially majority con- contest management turnover is roughly trolled (see Panel A). It is noticeable that comparable to the finding of Bethel et al. control changed hands in each case, which (1998) that about 22% of US CEOs are demonstrates that pooling agreements can replaced following ‘activist’ or ‘strategic’ break down when rival stakebuilders appear, block purchases. German CEOs appear to and that blocking minority stakes can have be at risk even when the stakebuilders considerable influence even when there is a ultimately do not gain control, as in the controlling stake or coalition. The stake- Continental case. builders’ apparent motivations in this group How were minority shareholders treated were very heterogeneous, ranging from in those cases where majority control was IWKA’s consolidating takeover bid to APV’s gained? As noted in Section 2, German greenmail and the disciplinary actions corporate law neither requires that minor- taken in the Wünsche case. Four of the six ity shareholders be made a buy-out offer targets in Panel B, which initially had one nor that they receive ‘equal treatment’ in or more blocking minority stakes (but no the terms offered. As can be seen from controlling stake), saw control change, Table 26.9, in the few cases where tender though in two cases it was parties other offers were made to minority shareholders, than the original stakebuilders who gained such offers have typically been at a signifi- control. In one case (Axel Springer Verlag) cant discount to the price offered to block the target finally agreed to ‘co-operate’ sellers in the course of gaining control. with the stakebuilder. In the final case Specifically, in three of the five cases where (Buderus) the Cartel Office effectively offers were made to minorities, the discounts blocked a change in control (the Cartel ranged from 15% to 64%. Only in one Office also blocked T & N’s stakebuilding in case (BIFAB) were minorities offered a Kolbenschmidt, although this simply led to premium to the market price, of 10%. The control passing to a third party). implications of the rules regarding minori- Of the eight cases in Panel C, where the ties can be seen in the takeover of SEN by initial ownership structure was fragmented, KW. Rival stakeholder APV had no realistic the initial stakebuilder took control in three chance of taking majority control of SEN cases, a rival stakebuilder took control in since KW owned 50.01%. However,its 40% two cases, and in one case (AMB) the tar- stake had considerable value as it could be get company agreed to cooperate with the used to block KW’s merger with SEN. stakebuilder. In the remaining two cases Blocking minority stakes can, therefore, (Holzmann and Continental), control did potentially be used for greenmail: in the not change hands. In the Holzmann case, event, KW bought APV’s stake at a the attempt by Hochtief to gain control was premium to its purchase price and then effectively blocked by the Cartel Office. In proceeded to offer a coercive dilution deal the remaining case, involving Continental, to the remaining (non-blocking) minorities. court decisions on voting restrictions The Feldmühle Nobel case similarly involved were the decisive factor in thwarting the a substantially lower bid to minorities once predator—though Continental’s supervi- Stora had gained control. sory board nevertheless dismissed the CEO This evidence contrasts with the US, for his alleged intransigence in dealing with where most of the blockholders investigated Pirelli’s bid. by Barclay and Holderness (1992) volun- Overall, therefore, we observe control tarily buy out the minorities at a premium changes in 12 cases, co-operation with the to the block price. The ability to dilute the stakebuilder in two cases, and no change value of minority shareholdings in Germany in control in the remaining three cases. does, of course, provide a powerful incentive Table 26.9 also documents management for takeover and is likely to reduce Grossman 692 ON TAKEOVER AS DISCIPLINARY MECHANISM and Hart (1980) free-rider problems. a pattern of ownership is the exception However, the sense of inequity resulting rather than the rule. In many other coun- from such dilution frequently results in pro- tries, ownership is concentrated in large tracted lawsuits, and is likely to have been blockholders who, either individually or in a reason why the Ministry of Finance coalition, exercise control. If the US/UK recently introduced its voluntary Takeover corporate governance problem is one of Code, which limits (but specifically does “strong managers, weak owners”, as sug- not remove) the ability to dilute minority gested by Roe (1994), in continental shareholdings.24 Europe, and many other countries, the cor- Finally, we consider how the control con- porate governance problem is rather one of tests affected the performance of the target “strong block owners, weak minorities.” companies. Given our small sample size and While a theoretical literature on blocks has the uncertainty about precise event dates, recently developed, there is little systematic we cannot offer formal statistical tests. empirical evidence on their importance or Instead, the final columns of Table 26.9 their impact on corporate control. This offer an impressionistic glance at perform- article provides such evidence for Germany. ance changes by comparing return on sales The article challenges a number of con- (RoS) and return on equity (RoE) in the ventional views of the way corporate 3 years before and the 3 years after the control is exercised in Germany. First, there control contests began. On average across is a widespread belief that there is a very the sample, RoE hardly changed (11.3% low incidence of hostile acquisition. While before, 12.5% after)—mirroring Bethel there have, to date, been only two cases of et al.’s US findings that ‘activist’ and hostile tender offers for German companies, ‘strategic’ block purchases lead at best to we have documented a more important and modest performance gains—but RoS nearly common means of gaining control: through doubled (from 0.7% to 1.3%). At the indi- the building of hostile stakes. Given the pat- vidual company level, nine firms increased tern of ownership of German listed compa- their RoE and eight firms increased their nies (summarised in Section 2), with RoS. Although this suggests that stake- around 87% of firms having at least one building had some positive effect on target blockholder owning 25% or more, it is not financial performance, the predominantly surprising that a relatively active market horizontal nature of the control contests exists in stakes, both for liquidity reasons cautions against attributing this necessarily and for friendly as well as hostile changes to enhanced efficiency. Also, the signs of in control. We identify 17 cases over an the performance changes do not appear to 8-year period that can be classified as hos- correlate with the control contest outcomes: tile stakebuilding. At first sight, this may half of the cases in which control changed look like an insignificant threat. However, hands experienced a decline in RoS, while our analysis of the ownership structure of the other half saw an increase. So, to the German companies suggests that the num- extent that hostile stakebuilding has an ber of firms facing a realistic risk of hostile effect on financial performance, it appears acquisition may be as low as 64. This sug- unrelated to the final outcome and it is gests a much higher incidence (at around uncertain whether it derives from increases 3–4% per year) of hostility in Germany in market power or in efficiency. than has previously been suggested, and is of a similar magnitude to the incidence of hostile takeovers in the UK. 5. CONCLUSIONS Second, these cases reveal the considerable power that can be associated with In some countries, notably the US and UK, blocking minority stakes, even in cases where ownership of companies is dispersed and a controlling shareholder, or coalition of control is exercised, at least in part, through shareholders, exists.The incremental control tender offers to shareholders. However,such rights associated with 25% stakes ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 693 derive, in the main, from the ability to block corporate control we observe? A full answer transactions such as profit transfer and is beyond the scope of the present article, control agreements that can be used to but there are certainly a number of areas reduce the value of minority shareholdings. of concern. First, the ability of a controlling Establishing a blocking minority stake is, blockholder to expropriate minorities could therefore, frequently the initial stage in a significantly raise the cost of capital. It is control contest, and is often followed by an interesting to observe that the recently attempt to build a coalition with other introduced Takeover Code in Germany—the large stakeholders with a view to gaining adoption of which is voluntary26 for firms— control. As we see from the cases, the makes some attempt to protect minorities. emergence of a rival coalition can even Companies acquiring stakes in excess of cause existing pooling agreements between 50% are now required (provided they have stakeholders to collapse. adopted the Code) to make an offer for the Third, we have illustrated the complex outstanding shares, but the regulations role that banks can play in corporate gov- regarding the terms of such offers provide ernance. The dominant role accorded to very little, if any, additional protection for German banks in much of the academic lit- minorities and are far weaker than those erature is as major providers of finance and operating in other countries.27 also—via their representation on supervi- Second, corporate governance in Germany sory boards, their direct equity ownership is both unpredictable and lacking in trans- and their control of proxy votes—as impor- parency. Battles often involve a protracted, tant monitors of corporate performance. and clandestine, shuffling of stakes between We believe that this stereotype is, at best, rival coalitions and the revising of pooling only partially accurate. There is growing agreements. Even large blockholders can evidence that banks do not provide a higher find themselves, apparently without warning, proportion of finance for investment in as members of the suppressed minorities. Germany than elsewhere, and that the Furthermore, once a hostile stakebuilder effectiveness of the monitoring role has appears there is frequent recourse to the often been overstated.25 In this paper, we courts, whose decisions are, on occasion, identify another important role for banks, unpredictable and lack consistency. As a namely their role in assisting companies result, some of the bids considered in this pursuing a strategy of hostile stakebuild- paper took over 5 years to reach their con- ing. We document many cases where banks clusion. If a guiding principle for the design play a pivotal role in building, brokering of corporate governance systems is and concealing stakes. In contrast, it is reasonable speed and certainty, the German striking how few examples we find of banks system frequently fails to achieve either. actively defending target companies from a Germany’s market for stakes is becom- hostile stakebuilder.Such behaviour may, of ing more liquid. Banks are, in general, course, be compatible with the view that reducing their stakes in companies, in banks actively monitor German companies response to tax changes and a general shift and help to effect changes in corporate in opinion regarding their role in corporate governance in the case of failing firms. governance. The latter has many roots: the However, it is important to recognise that recent string of embarrassing failures of this role is performed not by the compa- control and monitoring at Schneider, nies’ house banks (who are often believed Balsam, Metallgesellschaft and Klöckner– to be acting as monitoring intermediaries Humboldt–Deutz; the increasingly global drawing on their privileged information), financial outlook of many large German but by the banks assisting the predator.This companies; and the new focus on share- role has not previously been recognised. holder value permeating many German An important question remains: given companies and banks. As banks reduce their the ownership structure that exists in stakes, and their influence over corporate Germany, how efficient is the system of governance, cases of hostile stakebuilding 694 ON TAKEOVER AS DISCIPLINARY MECHANISM will surely increase. An important gap in to the original cross-shareholding structure, the existing theoretical literature is the exercising its right of refusal, bought part analysis of control battles where multiple of this stake (at a 5.5% premium over the large blocks exist, and this remains an area market price), increasing its 6% holding to for future research. 21%. Meanwhile, another 10% block came on the market which AMB was eager for another German insurer,Volksfürsorge, ACKNOWLEDGEMENTS to buy (AMB had recently acquired a majority interest in Volksfürsorge thanks to The paper has benefited from helpful dis- Fondiaria). Hostilities ended when the cussions at Birkbeck College, Humboldt chairman of AMB’s supervisory board bro- University Berlin, London Business School, kered a deal with AGF, against the opposi- Lund University, Oxford’s Saïd Business tion of his own CEO, who subsequently School, Stockholm School of Economics, resigned. The peace agreement entailed a the 1997 European Corporate Governance partial registration of AGF’s votes, opera- Network conference (Milan), the 1997 tional co-operation, and a commitment by European Finance Association meetings AMB to take a 5% noyeau dur stake in (Vienna), and the 1997 TMR Network AGF in preparation for AGF’s planned pri- Conference (Florence). We are grateful to vatisation. Fondiaria, opposed to AGF’s Patrick Bolton, Ernst-Ludwig von Thadden, accommodation, sold its 21% holding to a Ekkehart Boehmer, Colin Mayer, Mike consortium of German banks and insurers Burkart, Luc Reeneboog, and Dr. Ruppelt (in preference to AGF’s counter-offer). (head of the policy unit of Germany’s Federal These German institutions then controlled Cartel Office) for useful comments and to 38%, in response to which AGF increased Dirk Schiereck for generously making the its stake to 33.55% via purchases from accounting data available. Any views UK institutional investors (at the time, a expressed herein are ours alone, as are any draft EC Directive suggested harmonising remaining errors. Finally we thank the the blocking minority threshold to 33 1/3%). Editor and an anonymous referee for par- Following subsequent skirmishes between ticularly useful comments and suggestions. AGF and the German institutions over the chairmanship of the supervisory board, AGF finally entered into a standstill APPENDIX A agreement not to increase (or decrease) its block (with a registered 27.49% of the A.1. Aachener und Münchener votes) until December 1999, forestalling— Beteiligungen (AMB) at least for some time—a majority takeover of AMB. The control contest began when French insurer AGF built a 25% 1 share stake in A.2. Asko a bid to become “actively involved in insurer AMB’s management”, but was rebuffed by The control contest began when rival AMB’s board who refused to register AGF’s retailer Metro increased its declared (and votes (though the board could not prevent indeed so far friendly) 10% stake in open- AGF from voting Skandia’s 1.8% stake with market purchases, while two close allies of whom AGF was acting in concert). AGF’s Metro (including its house bank) also held legal challenge against AMB’s refusal to 10% each. Asko’s supervisory board chair register failed in a lower court on the (who was CEO for 18 years and founded grounds that a “board has a legitimate duty the firm) opposed Metro’s bid for control, to defend itself against a hostile takeover.” despite Asko’s financial difficulties. In an Part of AMB’s cross-shareholdings unrav- attempt to fight off the predator, Asko elled when Royal Insurance decided to sell revealed that 50% of the votes in its its 18.8% stake. Fondiaria, the third party main trading subsidiaries were held by a ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 695 foundation close to members of its supervi- and the family came to an arrangement, sory board, implying that a new owner electing first one of his associates and later could not control Asko’s operations (this Kirch himself onto the supervisory board. was part of a defence structure erected in Throughout the 10 years of control battles, response to two hostile takeover attempts there was extremely high turnover amongst by rival retailer co-op AG in 1978 and top executives, including four fired CEOs in 1981). Asko itself had a 5% voting restric- one 12-month period. tion. The battle ended once the supervisory board removed its chairman in a vote of confidence, unravelled the special rights of A.4. Bibliographisches Institut and the foundation, and removed the voting F.A. Brockhaus (BIFAB) right restriction. Metro then increased its stake to 55%, in the open-market and from The control contest began when Maxwell other blockholders (believed to be its asso- Communications privately approached ciates, though the sellers’ identity was BIFAB’s board with a takeover offer at a never confirmed). 53% premium to the share price. At the time, there were three blockholder groups: the Brockhaus family with around 15%, its A.3. Axel Springer Verlag pooling partner the Meyer family with around 38%, and Rheinpfalz Verlag with The control contest began when, following 27.34%. While it is not known what the flotation of the company and the death prompted Maxwell’s bid, the fact that the of its founder, three rival blockholders Meyer family did eventually sell out hints at emerged at this publishing house: the family either of: (i) its desire to divest its stake for (which, though in a minority, controlled the some unknown reason, which put BIFAB in boards), the Burda brothers Franz, Frieder play and prompted Maxwell’s bid; (ii) the and Hubert (to whom the late founder had imminent breakdown of the controlling tried to give majority control, which was coalition of the two families; or (iii) the blocked by the Cartel Office), and an out- possibility that the large premium Maxwell sider,Leo Kirch. Initially, Franz and Frieder was willing to offer tempted the Meyer Burda co-operated with the family to con- family into considering a sell out. BIFAB’s tain Kirch’s influence, denying him a seat board responded to Maxwell’s approach by on the supervisory board and not removing saying it “resisted a takeover by a foreign the CEO, as he demanded. Then they sur- firm” and was reported to be looking for a prisingly agreed to pool their 25.9% stake (German) white knight to stave off with Kirch’s declared 10% and his unde- Maxwell’s “hostile takeover bid.” The clared option on a further 16%, giving the Brockhaus family added to this opposition new coalition majority control. Only a claiming it would not sell under any cir- month later, however, Franz and Frieder cumstances, nor would it tolerate a sale to Burda sold their block to the family. The Maxwell by the Meyer family. A white family was not yet safe, though: claiming he knight was quickly found in rival publishing had first refusal on his brothers’ stake, house Langenscheidt, which bought the Hubert Burda challenged the sale in court. Meyer family’s 38%, giving it majority As defensive measures, the Springer family control in coalition with the Brockhaus refused to register Kirch’s additional 16% family (whose patriarch joined Langen- of votes, signalled they would not register scheidt’s board). Langenscheidt shortly Hubert’s block should he win in court afterwards increased its stake to 65.34% (which he finally did not), and entered into by buying out Rheinpfalz. One year later, a cross-shareholding agreement with Monti Langenscheidt and Brockhaus jointly took of Italy (in a further twist, the family and BIFAB private; the buy-out offered minor- Monti later fell out, with Monti threatening ity shareholders a 10% premium to the to sell its 10% to Kirch). Eventually, Kirch trading price. 696 ON TAKEOVER AS DISCIPLINARY MECHANISM A.5. Boge management favoured Arvin as its new majority owner,its current majority owners, The control contest began when rival car led by Commerzbank, instead sold their parts company Sogefi, controlled by Carlo stakes to Mannesmann (to whom Com- de Benedetti, emerged as the holder of a merzbank shortly afterwards also brokered 24.9% block in Boge, assembled in the the sale of VDO) at an undisclosed price in open-market. In response to Sogefi’s acqui- a deal described as “a defensive measure to sitions and newspaper speculation that de prevent de Benedetti from taking over.” Benedetti sought a majority stake, A few months later,Sogefi also sold its stake Commerzbank and Boge’s industrial part- to Mannesmann. Mannesmann proceeded ner VDO Adolf Schindling, who at Boge’s to offer to buy-out the remaining 2% flotation 18 months earlier, had each taken minority shareholders, at a 64% discount a 10% stake, increased their stakes to to the price it paid Sogefi, offering one 15% and 17.5%, respectively, and claimed Mannesmann share for every two Boge that they, together with an unnamed third shares (1:2).Though a shareholder lawsuit blockholder, controlled a majority of aimed at annulling the forced integration of Boge’s votes. Both Boge and VDO rejected Boge was unsuccessful, Mannesmann nev- Sogefi’s proposal for a three-way merger, ertheless increased its offer to 1:1 plus a stating they wished to remain independent. cash payment of DM 80/Boge share. The At Boge’s subsequent AGM, Sogefi dis- control contest lasted 2.5 years in total. closed an increased stake of 28.3%, which they increased to over 45% over the following year. Meanwhile, doubts had A.6. Bopp and Reuther (B & R) emerged over the actual existence of that The control contest began when IWKA third blockholder and thus over Boge’s acquired a 42.9% block from a group of ability to muster a friendly majority coali- family shareholders following years of poor tion against Sogefi. Boge’s management performance. Two years earlier, the family eventually placed a 6% stake with its US owners of B & R had sold a 25.1% stake joint venture partner,TRW, openly express- to financial investor Hannover Finanz (HF) ing its preference for closer co-operation in preparation for a possible subsequent with TRW over Sogefi.When Sogefi further stock market listing.The twenty-odd family boosted its stake, to 47.88%, Com- shareholders and HF had signed a pooling merzbank revealed that a friendly pool now agreement which secured pre-emptive rights controlled a slim majority: Commerzbank over share stakes and “a say in any impor- (24%),VDO (17.6% plus 2.5% held by its tant decisions affecting the company’s Swiss subsidiary) and TRW (6%). Free future.” This pooling agreement was vio- float at this point was a mere 2%, down lated when some family members sold their from 80% when the company went public 42.9% block without notifying either the 3 years earlier. When over the following company’s board or their pooling partners. year, the co-operation with TRW went sour Hinting at resistance against IWKA’s and TRW put its stake up for sale, intrusion from the remaining 32% family Commerzbank sought a friendly buyer of blockholders (led by B & R managing the 50.1% pooled block (TRW’s 6% was director Carl-Friedrich Reuther), IWKA clearly pivotal. However, it is likely that affirmed its wish to majority-control B & R TRW was prevented by the pooling con- and to “exercise its influence on the man- tract from selling it to the highest bidder if agement or supervisory board with a view that bidder was unacceptable to its pooling to improving the company’s poor operating partners). One bidder, US car parts group performance,” and consequently offered to Arvin Industries, was publicly rebuffed, buy-out the remaining shareholders. B & prompting Arvin to consider a public R’s board countered by pointing out that a counter-offer or to buy Sogefi’s near- 54% majority was still bound via a pooling majority block. Interestingly, while Boge’s contract between some of the remaining ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 697 family shareholders and HF (though that this option had been rejected to ensure contract was due to expire within 11 months Buderus’ independence. However, shortly and even the board had to admit that the afterwards, Commerzbank and Dresdner pool did not agree on the desirability of revealed they had each taken a 10% stake IWKA’s new stake; the 54% figure also “as a long-term financial investment and suggests that 3.1% of the family holdings not for resale to potential takeover bid- did not rally around Reuther’s defence). ders.” A month later, Bilfinger and Berger Reuther himself was engaged in negotiating (B & B) revealed at Buderus’ AGM that it the sale of the combined 54% block held had assembled a 15% block and planned by the remaining family shareholders and to discuss its long-term strategic vertical HF to Britain’s Siebe (it later emerged involvement with Buderus’ management. that the family had offered to buy-out HF While Buderus did not openly condemn the at 170% of book value, and that Siebe’s emergence of this block, its management 195% bid for the 54% majority stake val- did declare their intention to keep the com- ued the company more highly than IWKA’s pany independent of outside influence 180%). However, within 2 weeks, IWKA (curiously motivated as being in the interest had secured HF’s support (in spite of the of Buderus’ core customers!). Significantly, latter’s pooling commitment) and intended Dresdner Bank acted as B & B’s house to dismiss B & R’s board at the forthcom- bank, was its only declared blockholder ing extraordinary annual meeting.The deal (25.1%), and chaired its supervisory with Siebe fell through once HF switched board. This constellation led to speculation its support to IWKA, and HF was later that Dresdner Bank had facilitated the sued by family members for violation of the assembly of B & B’s 15% stake (possibly pooling contract (the outcome of this suit in connection with the IPO) and might have is unknown). Following the EGM, the taken its own 10% stake for the benefit— remaining family shareholders gave up and if not on behalf of—B & B. Not surpris- sold their 32% stake to IWKA at the lower ingly, therefore, B & B applied to the price of 180% of book value. IWKA Federal Cartel Office to increase its exchanged the management, began to Buderus stake to 25%, prompting Buderus restructure the company, and a few years to declare that it could not see any synergy later also bought out HF. gains and did not need B & B as a strategic partner. Only once the Cartel Office ruled against the stake increase did Buderus A.7. Buderus soften its tone and entered into a dialogue with its 15% shareholder and supervisory The control contest began shortly after board member B & B. To date, B & B has Metallgesellschaft floated its 79.9% stake not divested its Buderus stake. in Buderus in a public offering lead-managed by Deutsche Bank and co-managed by Dresdner Bank and Commerzbank with a A.8. Continental (Conti) mandate to spread the shares widely. At the time of the book-building, Buderus The control contest followed a year of announced that a number of institutional heavy trading (7.6 times total Conti share investors had taken stakes of between 1% capital) and persistent stock market and 3% and jointly (though not in coali- rumours—repeatedly denied by Conti and tion) controlled a majority of votes. the subsequent bidder, Pirelli—that a Buderus’ management welcomed the fact takeover bid was imminent. In the run-up to that the company no longer had a majority Pirelli’s intentions being confirmed, owner; it is known that Buderus’ manage- the Italian tyre manufacturer—faced ment and Metallgesellschaft had consid- with worldwide over-capacity in the tyre ered selling to a single investor instead of industry—privately approached Conti with placing the shares in the market, but that a merger proposal, and reportedly received 698 ON TAKEOVER AS DISCIPLINARY MECHANISM encouragement from two members of the launched a public tender offer. Events took supervisory board: Ulrich Weiss, of Deutsche an unexpected turn when a group of minor- Bank (Deutsche Bank was Conti’s house ity shareholders demanded an extraordi- bank, chaired the supervisory board, and nary general meeting to repeal the voting held 5% of its shares) and Friedrich right restriction (which required a simple Schiefer, a management board member of majority) and force a decision on the merger 5% shareholder Allianz, the insurance proposal (which required the approval of company. Having assembled a 5% stake in 75% of votes at the EGM). Deutsche Bank open-market purchases, Pirelli eventually and Morgan Grenfell put together a defen- announced its bid for control in the form of sive coalition of banks, proxy votes and car a reverse takeover by Conti of Pirelli’s tyre manufacturers large enough to block the division, adding that it had already secured merger proposal: Daimler Benz (in which the support of an unnamed majority of Deutsche Bank held a 28.3% stake and Conti’s shareholders; Conti’s share price whose supervisory board it chaired), fell by 7%. Pirelli was at pains to stress Volkswagen (whose CEO was Urban’s that its proposal was friendly. Behind the predecessor at Conti), and BMW; Conti’s scenes, Conti CEO Horst Urban enlisted share price fell by 5.5%. Nevertheless, at Morgan Grenfell (Deutsche’s investment the EGM, the voting restriction was over- banking division) to advise on the takeover turned with a 65.97% majority (though defence and lobbied his board to reject the this was later to be opposed in the courts proposed merger. Urban also promised his by a minority shareholder as well as Conti unions that there would be no redundancies itself); Conti’s share price rose 5.2%. if Conti stayed independent—which is a Conti still vowed to defend its independ- significant form of defence as employee ence. Interestingly, the arithmetic of the representatives controlled half the supervi- EGM indicates that Pirelli did not in fact sory board. Though allegedly originally in control a majority of votes. Shortly after- favour of Pirelli’s approach, supervisory wards, Conti’s supervisory board relieved board chairman Weiss helped defeat (with CEO Urban of his duties, reportedly for his the votes of the employee representatives) continued opposition to talks. Uncondi- the other board members’ suggestion that tional talks were resumed and continued management be instructed to negotiate over the next 8 months, though the man- with Pirelli. Urban then publicly declared agement board was still publicly divided on the proposal “a hostile takeover attempt, the merits of a merger. Conti began to despite all assurances to the contrary.”The restructure by closing overseas factories share price continued to fall. Over the fol- and selling off non-core divisions. Just lowing few months, stakes of up to 5% before a cross-shareholding deal between each were disclosed by some of Pirelli’s Conti and Pirelli was to be announced, it backers, including two of Pirelli’s own emerged that Pirelli had given its backers shareholders (Italian merchant bank indemnity guarantees to reimburse any Mediobanca and Sopaf), Fiat, Allianz’ losses on Conti shares. As Conti shares had Italian subsidiary and Merrill Lynch, lost roughly 45% of their value since Pirelli’s advisors. After an unusually public Pirelli and its partners bought their stakes, war of words, Conti demanded a standstill Pirelli came under pressure from its banks agreement as a pre-condition for talks, to find funds to cover its position and was including a commitment that Pirelli abstain eventually forced to call off the deal and from attempting to remove Conti’s 5% vot- restructure and refinance its balance sheet. ing right restriction. Pirelli rejected this However, Pirelli also bought options on its demand. Morgan Grenfell was actively allies’ combined 32.4% in Conti. At that engaged in finding a white knight or at least point, Conti appealed to the courts alleging a 25% blocking coalition, a solution publicly Pirelli had broken securities laws by acting favoured by Urban. Pirelli had still neither in concert and not disclosing the contracts named its alleged majority supporters nor with its allies.This challenge, later accepted ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 699 by the court, invalidated the EGM’s decision the overall stake would be reduced back to to remove the voting right restriction. Over 25.1% in due course, and, it appeared the next 4 months, new Pirelli allies bought later, its agreement that DBV could veto stakes in Conti and Pirelli tried again: Commerzbank’s choice of buyer. When at the next AGM, it moved to have the Zurich Insurance emerged as a possible restriction lifted, a proposal deemed buyer, DBV’s management expressly “hostile” by Conti’s new CEO who called declared they would block any attempt at a Pirelli’s intentions “sinister.” Conti refused hostile takeover and that they would not to let Pirelli and its allies vote all their allow any shareholder to build up a major- shares at the AGM, on the grounds that ity or dominant stake —which would seem they constituted a concert party and as a credible threat given (i) the restrictions such were limited to 5%; Pirelli’s motion on votes, (ii) the free float of below 50%, was defeated, though Pirelli managed to and (iii) Commerzbank’s earlier commit- obstruct Conti’s proposed capital increase. ment to ensure DBV’s independence. The When finally a superior court resurrected a talks with Zurich Insurance collapsed, decision of a pre-bid AGM to raise the partly because DBV opposed Zurich majority required to remove the voting Insurance’s insistence on majority control. right restriction from 50% to 75%, Pirelli Four months later, in spite of its long- gave up and let its 5% stake and the standing commitment not to hold more options it held be placed by Deutsche Bank than a blocking minority stake, Com- with German companies friendly to Conti. merzbank increased its stake to 50% 1 These placements were supported by a share via open-market purchases. Once it financial guarantee from the state of Lower controlled DBV, the bank swiftly proceeded Saxony where Conti is headquartered. The to sell majority control to Winterthur, a control contest lasted 2.5 years in total. Swiss insurer.As DBV’s CEO was chosen to head all of Winterthur’s German sub- A.9. Deutsche Beamten–Versicherung sidiaries, it seems unlikely that DBV objected to the sale; however, since (DBV) Commerzbank had acquired majority con- The control contest began soon after public trol opposition on DBV’s part was, one pre- sector insurer DBV was privatised via a sumes, no longer an attractive proposition. public offering of 50% minus 2 shares.The IPO was lead managed by Commerzbank, A.10. Dyckerhoff und Widmann which not only took a 25% 1 share (Dywidag) stake (another 25% 1 share block still being in the public sector) but also bound The control contest began at a time when itself not to increase the size of its block Dywidag’s ownership structure was highly and agreed that DBV was to remain inde- fragmented, with a maximum free float of pendent. One way to ensure independence only 21% and the two largest blocks was the choice of restricted transferability owned by Holzmann, a rival construction of shares, which requires the board’s regis- company whose shareholding was viewed tration before votes can be exercised. as friendly, and industrialist Max Aicher. However,when DBV’s share price soon began During the late 1980s, Ignaz Walter, to fall below the offer price, Commerzbank, acquisitive owner of several regional con- perhaps in an effort to support the share struction companies, embarked on what he price, acquired a further 23.3% over the later called a ‘strategy of slow takeover’ of following few months from an unnamed Dywidag by clandestinely buying up minor- investor.This stake increase ostensibly had ity blocks from various sources, including a the approval of DBV’s board, and anyway stake held by his house bank, Bayern LB, was accompanied by Commerzbank’s and 2% bought on the open-market. While pledge not to have the votes of the addi- there was speculation that Walter was tional shares registered, its assurance that behind the stake purchases, he repeatedly 700 ON TAKEOVER AS DISCIPLINARY MECHANISM denied being a blockholder in Dywidag. By the sale price of various assets. When a the end of 1991, Walter controlled 40% of hostile tender offer was rumoured, FeNo’s Dywidag’s votes, which he had previously management restricted voting rights to a failed to declare, and explicitly denied hav- maximum of 5% per shareholder, aided by ing; it appears that technically, he never Deutsche Bank which controlled about crossed the 24.9% disclosure rule simply 55%, mostly via proxies but also via its by spreading the 40% stake over various 8% stake in FeNo. A year later, the Flicks associated parties, including his children. and five associated parties sold a previ- Once Walter declared his stake as well as ously undisclosed 40% block (assembled his intention to take over the company, with the help of Merrill Lynch via open- Dywidag’s management strongly and pub- market purchases and secret direct agree- licly resisted his hostile endeavours. ments with institutional blockholders) to Despite being the largest shareholder, Veba, which Veba then boosted to 51%. Walter was denied a seat on the supervi- However, when Veba failed to remove the sory board. In December 1991, a financier 5% restriction and take control of FeNo, called Dieter Bock proposed a “friendly” the Flicks assembled another stake of takeover of Dywidag, a move that was wel- between 10% and 20% and began to comed by the board. Rather than inviting oppose Veba for not launching a full bid. all Dywidag’s shareholders to tender their The Flicks’ actions are consistent with a shares, Bock proposed to consummate the strategy of trying to maximise the bid value takeover via the negotiated friendly acqui- in the ensuing auction, which they were sition of two key shareholders’ large stakes: well-placed to do given that their 10–20% the 24.9% stake held by Holzmann, and stake, spread over several parties, gave the Aicher block, which by now had been them more clout than Veba derived from its increased to 24.7%, along with the small 51% stake.Two further suitors emerged in stake owned by Dumez of France via a the form of Sweden’s Stora Kopparberg cross-shareholding arrangement.This 51% and SCA, both of which bid for the company. stake would block the 40% stake owned by Veba eventually sold its block to Stora, who Walter. However, events did not turn out as also bought out the block jointly held by the expected. Bock duly acquired the Flicks and Merrill Lynch, as well as SCA’s Holzmann stake, but then announced, in 5% toehold stake. Once Stora owned 85%, May 1992, that Advanta had failed to com- minority shareholders were offered a buy- plete the assembly of a controlling stake in out price 15% below the bid price paid to Dywidag, blaming Max Aicher for reneging Veba. The firm was split up and restruc- on the sale contract (a view disputed by tured, in spite of the management’s opposi- Aicher). Bock also disclosed that Advanta tion.The control contest lasted 15 months. had sold the 24.9% stake in Dywidag it had acquired from Holzmann to Walter, A.12. Th. Goldschmidt finally giving Dywidag’s hostile suitor majority control! Dywidag’s CEO resigned. The control contest began when two con- No buy-out offer was made to minority glomerates, VIAG and Veba, independently shareholders.The Cartel Office subsequently and potentially in rivalry, bought stakes fined Walter DM 500,000 for failure to from an existing corporate shareholder and register changes in its ownership interest in part of the family. When management and Dywidag. the remaining family shareholders affirmed their desire to remain independent, the new A.11. Feldmühle Nobel (FeNo) blockholders pooled their 45.8% stakes and demanded “a change of direction,” but The control contest began when the Flick were kept at bay by the insiders who brothers, Friedrich Christian and Gert- refused to grant representation on the Rudolf (the former owner’s grand-nephews) supervisory board.The contest was resumed accused the management of not maximising when (i) Veba came under pressure from ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 701 its shareholders to sell its 27.95% stake Krupp made its public announcement. (25.02% of votes) and (ii) the family’s Initially, it was only Hoesch’s unions which coalition with insurer Allianz broke down, called Krupp’s bid a hostile takeover. when Allianz chose to let its pooling con- Hoesch’s CEO, Kajo Neukirchen, subse- tract (via which the family controlled the quently developed a more confrontational firm) expire in 1996 and put its 10.38% tone in public, and eventually Hoesch’s stake (9.29% of votes) on the market. supervisory board declared its opposition VIAG beat the family to both stakes, and in to the clandestine nature of Krupp’s stake- April 1997, controlled 50.34% of the building, though it was in principle willing votes to the family’s 39.94%. No offer to to consider merger plans on the basis of a the minority shareholders was made. The voluntary discussion amongst equals. At control battle lasted 5 years in total. the same time, it took the unusual step of suspending two members of the manage- A.13. Hoesch ment board who were believed to favour Krupp’s bid, widely seen as a hardening of The control contest began when high trad- positions. For a brief moment, it looked as ing volumes in Hoesch shares throughout if British Steel might step in as a white 1991 prompted speculation of a possible knight to rescue Hoesch. Krupp’s Cromme (foreign) takeover bid. In October 1991, reacted by noting that he would take over Krupp, a rival steel maker, revealed it had Hoesch whether or not Hoesch co-operated. bought a 24.9% stake secretly accumu- Significantly, he also claimed Deutsche lated on its behalf by Credit Suisse. Hoesch’s Bank had been notified of the impending share price fell by 9.7% in response, while bid some 2 weeks in advance, and had Krupp’s rose by 8.6%. Krupp made clear welcomed it, which contradicted Deutsche its intention to acquire a majority block, a Bank’s public insistence on its uninformed plan for which it claimed to have received and neutral role. When Hoesch demanded the prior support of various banks and evidence of Krupp’s alleged majority coali- financial institutions with holdings in Hoesch. tion, Cromme provided notary evidence of One of these was WestLB, Krupp’s house the support of a further 30.4% of the bank and chair of its supervisory board, votes, including a 20% block held at Credit which declared it had a 12% Hoesch stake Suisse, but excluding WestLB’s 12%. on its trading books, though it denied to Shortly afterwards, Krupp announced it have pledged the shares or the votes to had bought a further 26% (likely to have Krupp. This is not altogether surprising, included at least part of Credit Suisse’s since a formal agreement would have 20% stake) at an undisclosed price, reduced Krupp’s influence given Hoesch’s thus increasing its 24.9% stake to a corporate charter, which capped the votes majority block; Hoesch’s share price fell by of any stake or formal pooling of votes at 4%. Krupp had still not made a formal 15%. Deutsche Bank was Hoesch’s house tender offer. Over the next few months, bank and chaired its supervisory board. Krup and its allies removed the voting Nevertheless, there were persistent right restriction, sealed a merger agree- rumours that Deutsche Bank controlled a ment and saw off legal challenges from block of perhaps 10% which was friendly three minority shareholders. Amongst its to Krupp! Krupp’s CEO, Gerhard Cromme, allies were not only WestLB, its own was at pains to stress this was no hostile house bank, but also Deutsche Bank, bid, but a defensive move as Krupp would Hoesch’s house bank, which controlled an have suffered had Hoesch been taken over estimated 12% of Hoesch via proxies. by a (foreign) rival. The initial reaction Hoesch’s CEO Neukirchen resigned. from Hoesch’s management was muted Krupp’s CEO Cromme was elected with no particular indication of opposition, “Manager of the Year” by TopBusiness and except perhaps grumblings about not hav- Manager Magazin. The contest only lasted ing been informed until a few days before a few months. 702 ON TAKEOVER AS DISCIPLINARY MECHANISM A.14. Philipp Holzmann threatening the three companies’ managers with fines of up to DM 1 million each. The control contest began when Advanta, a Simultaneously, Hochtief also purchased company controlled by financier Dieter the 4.9% stake in Holzmann, which Bock (see above: Dywidag), announced it Commerzbank had acquired 13 years ear- had acquired, from an unnamed source, a lier. In a filing with the Cartel Office, 10.25% stake in Holzmann, Germany’s Hochtief later disclosed it had held a call largest construction company by turnover. option on the 4.9% stake all along, though At the time, a key 20% Holzmann share- neither Hochtief nor Commerzbank had holder was Hochtief, Germany’s second- previously disclosed this. Indeed, Com- largest construction company. Bock’s merzbank had declared the disputed stake ultimate intentions were unclear until as its own until the very onset of hostilities, Advanta sold its stake to BfG, one of its thus helping Hochtief obscure its true own- house banks. That BfG simultaneously ership interest in Holzmann. These revela- granted Hochtief a call option on the stake tions followed assurances by Commerzbank only became apparent when Hochtief noti- and Hochtief that there were no undis- fied the Federal Cartel Office of its inten- closed stakes or contracts pertaining to tion to raise its 20% stake via exercising such stakes; Commerzbank and Hochtief the BfG option, thus triggering a manda- only eventually disclosed their contractual tory anti-trust review. Furthermore, arrangements due to requirements under Hochtief declined to rule out increasing its the new securities trading law. stake further, raising the prospeck of a takeover. While analysts welcomed the potential bid, Holzmann’s board issued a A.15. Kolbenschmidt statement reaffirming its commitment to remaining independent. At the same time, The control contest began when T & N there was speculation—first denied and acquired options on a combined 52.5% later confirmed—that Deutsche Bank, block after financially troubled Metallge- Holzmann’s house bank, dominant share- sellschaft put its 47% block on the market holder and chair of supervisory board, was (the remainder were options on 2.5% from willing to contemplate reducing its 25.9% Magna International of Canada, and 3% to 10%, thus deserting its client (Deutsche from institutions which held a combined Bank was doing brisk trade with Hochtief’s stake of 10%). As a horizontal merger, the majority owner, RWE). Holzmann’s main deal was subject to Cartel Office approval. line of defence, therefore, was the anti-trust Within 3 months, target management card which duly paid off: the Cartel Office asserted their desire to remain independent ruled against Hochtief on competition of any majority shareholder, sought a dif- grounds, blocking any future increase in ferent, friendly but minority buyer (Dana) Hochtief’s Holzmann stake from the pre- for a 25% block, and organised a work- bid level of 20%. Hochtief arranged for force petition in protest against T & N’s Commerzbank, its house bank and minority planned takeover. Due to Cartel Office shareholder,to purchase the 10.25% block opposition and problems over its UK from BfG, paid Commerzbank an undis- asbestos liabilities,T & N failed to exercise closed fee to finance the cost of carry, and its options by their expiry, prompting its eventually signed a purchase agreement for bank, Commerzbank, to acquire a total of the stake with Commerzbank contingent on 49.99% on T & N’s behalf and grant Hochtief winning its appeal against the T & N a new option (Magna’s 2.5% option Cartel Office’s ruling before the superior was extended). Shortly afterwards, the court in Berlin. When these transactions Cartel Office blocked the deal and T & N came to light, the Cartel Office started an appealed. After a further options extension, investigation into the legality of Hochtief’s T & N appealed—unsuccessfully—to the dealings with Commerzbank and BfG, EU competition authorities to overrule the ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 703 German decision. Finally, Commerzbank critic of the management and supervisory placed half of its stake with Rheinmetall, boards, lobbying to oust the supervisory another car parts maker, which subse- board for failing to carry out its control quently received Cartel Office clearance to duties. Wünsche eventually sold his 26% take management control of Kolbenschmidt. blocking minority stake to two outsiders at Within a few months, Rheinmetall took a 24% discount to the market price. The management control by purchasing T & N’s new blockholders, one of whom wanted the option on the remaining 24.99% stake still company broken up, pooled their votes held by Commerzbank and buying a further and—even though they were still bound by 3% in the open-market. Wünsche’s original pooling agreement with his brother—put pressure on the CEO to eventually step down. Once Kai Wunsche A.16. Seitz–Enzinger–Noll (SEN) had agreed to resign as CEO to move onto The control contest began when SEN’s the supervisory board, the two outside minority shareholders, and then its own blockholders sold their stakes to a new board, objected to the forced merger with a financial investor group, who also acquired division of its 50.01% parent, Klöckner- a 4.9% stake from WestLB to form a Werke (KW). In that climate, APV man- blocking minority stake of 25.1%.The new aged to acquire a 40% block from a local CEO acquired a 10% stake on assuming savings bank (BaKoLa) and the Seitz fam- office, while Kai Wünsche reduced his then ily, and offered to buy-out both KW’s con- 42% stake to 30%. When the CEO change trolling block and outside shareholders was first rumoured, the share price rose (prompting KW to offer to match APV’s 28% in 1 day. The new CEO proceeded to bid. APV won support for this deal from sell off a string of peripheral businesses the SEN board and the trade unions, argu- with a view to refocusing the company. ing that unlike KW, APV would not rationalise the firm. In coalition with the remaining minority shareholders, APV then NOTES obstructed KW’s attempts at controlling SEN, for instance by voting against super- * Corresponding author. Finance Department, visory board appointees; while KW sought Stern School of Business, New York University, an EGM to sack the 10 out of 12 supervi- 44 West Fourth Street, Suite 9–190, New York, sory board members hostile to it. In the NY 10012-1126, USA. Tel.: 1-212-998- end, APV abandoned its control bid and 0304; fax: 1-212-995-4233. sold out to KW, making a 14.6% return E-mail address: [email protected] over 1 year on the stake sale. Controlling (A. Ljungqvist). 90% of votes, KW then forced a profit 1 The following illustrate this conventional transfer and control agreement on SEN, wisdom: Carney (1997, p. 78), “...no market and offered to buy-out minorities at a for corporate control exists in Germany to cure 12.9% discount to the price it paid APV, even the most extreme monitoring problems.” Grundfest (1990, p. 105), “...in both Germany and an 18% discount to the market price. and Japan, corporate investors and intermediaries are able to reach deep into the inner workings of portfolio companies to effect fun- A.17. Wünsche damental management change.They do so with- The control contest began when the family out the need for a hostile takeover or proxy coalition, which in total controlled over contest.” Allen and Gale (1994, p. 9): “Banks are heavily involved in the control of industry two-thirds of votes, broke down. After being and form long-term relationships with firms.There dismissed from the management board by is little publicly available information about his brother Kai Wünsche (the CEO and firms and there is no active market for corpo- 44% blockholder) for alleged insider deal- rate control.” Franks and Mayer (1990, p. 208) ing, W-J. Wünsche became a vociferous “...banks protect firms from interference from 704 ON TAKEOVER AS DISCIPLINARY MECHANISM external parties, in particular from hostile 9 There is a third option: minorities can be takeovers.” offered a certain fraction of the controlling 2 Krupp’s recent (ultimately unsuccessful) parent’s dividend, which, since the law does not bid for rival steel producer Thyssen was possibly specify what dividend policy the parent has to Germany’s first-ever truly Anglo–U.S. tender follow, allows for minorities to be diluted with- offer: being open to all shareholders and offer- out any compensation whatsoever. ing a 25.5% premium. Vodafone’s tender offer 10 One appeals court decided in 1990 that the for Mannesmann was, in our view, the first compensation should be based solely on the successful hostile tender offer for a German basis of future cash flows, without regard to hid- company. den reserves or other peripheral assets, both of 3 See Becht and Roell (1999). which could be realised solely for the benefit of 4 The incidence of hostile takeovers is quite the majority shareholder. Recently, another variable over time, but Jenkinson and Mayer appeals court chose a discount factor signifi- (1994) report an average of 40 hostile bids per cantly above the risk-free rate to calculate the annum in the U.K. over the period 1984–1989. required minority compensation, after having Taking the number of potential targets as used a discount factor below the risk-free rate around 1500 (excluding investment trusts), this in a previous decision. would result in an incidence of under 3% per 11 This option will be optimally exercised once annum in the U.K. the dependent company has insufficient capacity 5 They quote figures from a 1984 survey by to generate enough cash to pay the guaranteed the SEC that around 20% of NYSE and AMEX dividend post-dilution. companies have at least one outside shareholder 12 Thyssen, the target of Krupp’s unsuccessful owning more than 10% of the common stock. 1997 hostile bid, would feature in our Panel B, 6 It should be noted that many small to given its two non-majority blockholders: an medium sized firms in the U.S. have concen- investment company (controlled by Allianz, trated shareholding structures. Germany’s largest insurer, and Commerzbank) 7 There has, however, been some interesting with a 25% stake and the Thyssen family analysis of optimal ownership structure in the trust holding under 9%. presence of blockholders (e.g. Bolton and von 13 This is the maximum number of firms with Thadden, 1998; Zwiebel, 1995). Zwiebel free float of 50% . Where there are undis- assumes that there are private benefits of con- closed stakes, true free float may be less than trol that can be divided and shared amongst the level we calculated. blockholders who form controlling coalitions. 14 Non-voting shares need not always be an Zwiebel suggests that the existence of a large effective takeover barrier, as the experience of blockholder will tend to discourage small block- computer manufacturer Nixdorf illustrates. holders, who would then find it more difficult to Nixdorf became the target of takeover rumours form controlling coalitions. Hence, large block- when one family (voting) shareholder was holders tend to ‘create their own space’. In con- alleged to be looking to sell out.The reason why trast, our article documents the existence of Nixdorf’s CEO, Klaus Luft, took these rumours multiple competing large blockholders who are seriously enough to declare his opposition is a engaged in trades (often of a clandestine German corporate law stipulation that prefer- nature). The apparent motivation for block ence shares become enfranchised if a company building is not Zwiebel’s pursuit of a share in passes the preferred dividend in two consecutive the (divisible) private benefits of control, but to years—a condition which in the Nixdorf case effect a control change via the accumulation of was met. As soon as Luft resigned from the a majority or controlling stake. board (without giving any reasons), Siemens 8 For instance, minority shareholders in took over control by buying 51% of the voting Volksfürsorge, an insurance company, recently shares from Deutsche Bank, the family and a suffered dilution twice within a short space charitable trust in a move apparently master- of time. Volksfürsorge was first made to buy minded by Deutsche Bank. (As we cannot attrib- a stake in its parent’s health insurance ute hostile intentions to Siemens, we do not subsidiary, which KPMG had valued using a include this case in our clinical analysis below.) particularly low discount rate of 2.9%. 15 Electronically available news sources are: Subsequently, Volksfürsorge was ordered to sell Reuters Textline, German and international news- its legal insurance subsidiary to its parent who papers, and newswires such as Press Association. had commissioned a valuation on the basis of a 16 This case demonstrates that the conven- much higher discount rate of 12%! tional view that a German house bank will ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 705 defend its client from takeover is not generally 21 This information came to light when it accurate. In an interview with the Frankfurter emerged that Pirelli had entered into contracts Allgemeine Zeitung, Hilmar Kopper, then chief with some of its allies guaranteeing to compen- executive of Deutsche Bank, said he had known sate them for any fall in Continental’s share of Krupp’s planned takeover of rival steel group price. Fall it duly did (by around 45%) and Hoesch beforehand. “It has my full support Pirelli’s exposure became known when its banks because it makes good industrial sense,” Kopper exerted pressure to cover its position. said. In his view, Deutsche Bank (whose man- 22 Limiting the transferability of shares can agement board member, Herbert Zapp, headed also be a major impediment to a predator, as in Hoesch’s supervisory board) had no obligation the case of AGF’s stake in AMB. In this case, to defend Hoesch against the bid: “How [should the court upheld the refusal of AMB’s board to Deutsche Bank have defended Hoesch]? register the shares on the grounds that a board Everyone knew someone was buying shares, but has a right to defend itself against a hostile no one knew who. Secondly, why should stakebuilder. Deutsche Bank defend Hoesch? Does Hoesch 23 Counting Asko’s supervisory board chair- have a right to a defence? Or is Deutsche Bank man as ‘management’, which he de facto obliged to maintain [industrial] structures?” appeared to be. 17 As suggested, for example, by Shleifer and 24 The Takeover Code was introduced in Vishny (1986). However, incentives for minori- October 1995, and includes a requirement that ties to free-ride in Germany are likely to be minorities should receive an offer when major- small relative to the fears of expropriation. ity control changes hands. However,its adoption 18 The new securities trading law has intro- is voluntary, and the rules provide little protec- duced two lower reporting thresholds of 5% tion for minorities. See footnote 27 for more and 10%. In addition, the Berlin Supreme Court details. ruled in 1991 that if asked at their AGMs, com- 25 See, for example, Rajan and Zingales panies have a duty to disclose shareholdings of (1995), Corbett and Jenkinson (1996) and (i) 10% or more or (ii) a minimum market Edwards and Fischer (1994). value of DM 100 million. The market value 26 By September 1996, 1 year after it came threshold means that share stakes of less than into force, only around one-third of listed com- 1% in DAX-30 companies are declarable. panies in Germany had actually adopted the Code. 19 Hochtief later went a step further and The Takeover Commission reviewed 12 cases in signed a purchase agreement for the stake with its first year, only one of which led to a public Commerzbank, effectively raising its holding in censoring. Holzmann to 35.15%. The agreement and 27 A party that acquires a stake in excess of exchange of title and consideration were contin- 50% (via open-market purchases or a private gent on Hochtief winning its appeal against the deal) must offer to buy out minority sharehold- Cartel Office’s ruling that it should not be ers unless the acquiring company has merged allowed to increase its stake above 20%. This with the target company or entered into a profit ruling prevented Hochtief voting the additional transfer and/or management control agreement stake pending regulatory approval, but the within the first 18 months after gaining control. purchase agreement bound Commerzbank ‘not In the absence of a merger or control agree- to act against Hochtief’s interests’, a clause ment, the controlling company is required to which would almost certainly tie Com- make a public offer for the remaining shares merzbank’s hands at Holzmann’s AGM. When within the next 3 months. If, during the initial 18 these transactions came to light, the Cartel month period after having gained control, the Office started an investigation into the legality acquiring company has not bought additional of Hochtief’s dealings with Commerzbank and shares, the price offered to minorities must not another bank (BfG), threatening the three be less than the price paid on purchases during companies’ managers with fines of up to DM 1 the 6 months before gaining control minus 25% million each. (denote this price P1). If the acquiring company 20 A Frankfurt court recently ruled a charter has made additional purchases since gaining amendment inadmissible, that would have control, the price offered to minorities must be allowed the Dresdner Bank management board the maximum of P1 and the weighted average to institute a contingent 10% voting right of the prices paid on such additional purchases. restriction in the event of a hostile bidder To those who have grown accustomed to observing emerging, subject only to the supervisory bid premia being offered to shareholders in the board’s approval. event of a takeover, such arrangements will 706 ON TAKEOVER AS DISCIPLINARY MECHANISM hardly appear too onerous for the bidder! This control. Journal of Financial and Quantitative contrasts, for example, with the U.K. City Code Analysis 26, 391–407. on Takeovers, which demands a full bid once a Comment, R., Schwert, G.W., 1997. Hostility in shareholder obtains a stake in excess of 30%, at takeovers: In the eyes of the beholder? a price no worse than the highest price that the mimeo, University of Rochester. bidder paid during the previous 12 months. Corbett, J., Jenkinson, T.J., 1996. The financing of industry, 1970–1989: an international REFERENCES comparison. Journal of the Japanese and International Economies 10 (1), 71–96. Allen, F., Gale, D., 1994. A welfare comparison Denis, D.J., Serrano, J.M., 1996. Active of the German and US financial systems, LSE investors and management turnover following Financial Markets Group Discussion Paper unsuccessful control contests. Journal of no. 191. Financial Economics 40 (2), 239–266. Barclay, M.J., Holderness, C.G., 1989. Private Edwards, J.S.S., Fischer, K., 1994. Banks, benefits from control of public corporations. Finance and Investment in West Germany Journal of Financial Economics 25 (2), Since 1970. Cambridge Univ. Press, 371–395. Cambridge. Barclay, M., Holderness, C.G., 1991. Negotiated Franks, J., Mayer, C.P., 1990. Capital markets block trades and corporate control. Journal and corporate control: a study of France, of Finance 46 (3), 861–878. Germany and the U.K. Economic Policy 10, Barclay, M., Holderness, C.G., 1992. The law 191–231. and large-block trades. Journal of Law and Franks, J., Mayer, C.P., 2001. Ownership and Economics 35, 265–294. control of German corporations. Review of Baums, T., Fraune, C., 1994. Institutionelle Financial Studies, forthcoming. Anleger and Publikumsgesellschaft. Die Grossman, S., Hart, O., 1980. Take-over bids, Aktiengesellschaft 3, 106–124. the free-rider problem and the theory of the Becht, M., Roell, A., 1999. Blockholdings in corporation. Bell Journal of Economics 11, Europe: an international comparison. European 42–64. Economic Review 43, 1049–1056. Grundfest, J.A., 1990. Subordination of Berle, A., Means, G., 1932. The Modern American capital. Journal of Financial Corporation and Private Property. Macmillan, Economics 27 (1), 89–114. New York. Holderness, C.G., Sheehan, D.P., 1985. Raiders Bethel, J.E., Liebeskind, J.P., Opler, T., 1998. or saviors? The evidence on six controversial Block share purchases and corporate perform- investors. Journal of Financial Economics 14 ance. Journal of Finance 53 (2), 605–634. (4), 555–579. Bolton, P., von Thadden, E.-L., 1998. Blocks, Jenkinson, T.J., Mayer, C.P., 1994. Hostile liquidity and corporate control. Journal of Takeovers—Defence, Attack and Corporate Finance 53 (1), 1–26. Governance. McGraw-Hill, London. Bulow, J., Huang, M., Klemperer, P., 1999. Jensen, M., 1993.The modern industrial revolu- Toeholds and takeovers. Journal of Political tion, exit, and the failure of internal control Economy 107 (3), 427–454. systems. Journal of Finance 48 (3), Burkart, M., 1996. Economics of takeover reg- 831–880. ulation, mimeo. Stockholm School of Mikkelson, W., Ruback, R., 1985. An empirical Economics. analysis of the interfirm equity investment Butz, D.A., 1994. How do large minority share- process. Journal of Financial Economics 14, holders wield control? Managerial and 523–553. Decision Economics 15 (4), 291–298. Mørck, R., Shleifer, A., Vishny, R.W., 1989. Carney, W.J., 1997. Large bank stockholders in Alternative mechanisms for corporate con- Germany: saviours or substitutes? Journal of trol. American Economic Review 79 (4), Applied Corporate Finance 9 (4), 74–81. 842–852. Choi, D., 1991. Toehold acquisitions, share- Rajan, R.G., Zingales, L., 1995. What do we holder wealth and the market for corporate know about capital structure? Some evidence ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 707

from international data. Journal of Finance Wenger, E., Hecker, R., 1995. Übernahme-und 50 (5), 1421–1460. Abfindungsregeln am deutschen Aktienmarkt— Roe, M.J., 1994. Strong Managers, Weak Eine kritische Bestandsaufnahme im Owners: The Political Roots of American internationalen Vergleich. Ifo Studien Corporate Finance. Princeton Univ. Press, 41, 51–87. Princeton. Wruck, K.H., 1989. Equity ownership concen- Shleifer, A., Vishny, R., 1986. Large sharehold- tration and firm value: evidence from private ers and corporate control. Journal of Political equity financings. Journal of Financial Economy 94 (3), 461–488. Economics 23, 3–28. Spencer, C., Akhigbe, A., Madura, J., 1998. Zwiebel, J., 1995. Block investment and Impact of partial control on policies enacted partial benefits of corporate control. by partial targets. Journal of Banking and Review of Economic Studies 62 (2), Finance 22, 425–445. 161–185. Chapter 27

Jens Köke THE MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY: A GOVERNANCE DEVICE?

Source: Journal of Corporate Finance, 10(1) (2004): 53–80.

ABSTRACT This study examines whether changes in ultimate firm ownership (control) play a disciplinary role in a bank-based economy.We focus on Germany as the prototype of a bank-based system. We find that poor performance makes a change in control more likely; this suggests a disciplinary role. Tight shareholder control acts as a substitute for control changes, strong creditor control as a complement. Following a change in control, management turnover increases, but not as a consequence of poor performance, and performance does not improve significantly.These findings are inconsistent with a disciplinary role of the market for corporate control in bank-based Germany.

1. INTRODUCTION This study aims to fill this gap. The question it tries to answer is: ‘Does the THE MARKET FOR CORPORATE CONTROL market for corporate control fulfill a represents an important governance disciplinary function in a bank-based device. For market-based economies, the economy?’ To address this question, we disciplinary role of control changes is well examine the frequency, causes, and documented (see Jensen and Ruback, consequences of changes in ultimate firm 1983; Jarrell et al., 1988 for the US, and ownership for Germany. We choose Franks and Mayer, 1995 for the UK).To a Germany because it is often regarded as large extent, this market is based on the prototype of a bank-based country hostile takeovers in both countries. For (OECD, 1995). We investigate changes in bank-based economies, the typically low ownership because this should capture all frequency of hostile takeovers suggests forms of control changes. And we define a that there is no market for corporate change in control as a change in ultimate control (OECD, 1998). But other forms of ownership because ownership patterns in control changes may also play a discipli- bank-based countries are often complex nary role (see Schwert, 2000 who shows (La Porta et al., 1999), making changes in that hostile and friendly takeovers in the direct ownership less meaningful. US are not distinguishable in economic An important characteristic of bank- terms). Overall, systematic evidence on the based economies is that there is a number nature of control changes in bank-based of alternative governance mechanisms to countries is sparse. changes in control. For Germany, the most MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 709 relevant are tight shareholder control, changes in block ownership and their continuous creditor control, and control by relation to management turnover for a the supervisory board. These nonmarket small sample of listed firms. This study is mechanisms might act as a substitute for the first to provide systematic evidence on disciplinary control changes. Therefore, we the causes and consequences of control examine whether their presence makes changes, based on a large sample of listed control changes less likely to occur. and nonlisted firms.Third, this study identi- Another characteristic of bank-based fies changes in ultimate firm ownership countries is the large number of nonlisted based on individual shareholder data. corporations. We take advantage of this Therefore, it allows for changes within characteristic. If a stock market listing shareholder categories and in this sense provides for an extra monitoring tool (e.g. extends the work of Denis and Sarin via regular analysts’ reports), we should (1999) who examine changes between expect nonmarket governance devices to be shareholder categories. It improves on stronger for nonlisted firms. Therefore, we Franks and Mayer (2001) whose work is investigate whether the causes and conse- based on changes in direct ownership. quences of control changes differ between The study proceeds as follows: Section 2 listed and nonlisted firms. reviews the literature on control changes The empirical analysis is based on a and other governance devices to formulate sample of almost 1000 listed and non- testable hypotheses. Section 3 describes listed German corporations for the years the data used for this study and defines the 1987–1994. Our main findings are: concept of control that is applied to iden- (1) poor performance makes a change in tify the ultimate owner.Section 4 describes control more likely, being a necessary the frequency of control changes in condition for control changes to be discipli- Germany as well as the type of buyers and nary; (2) tight shareholder control (creditor sellers of control blocks. Section 5 analyzes control) has a negative (positive) impact the causes and consequences of control on the probability of control change, indi- changes and compares the results with cating that shareholder control (creditor evidence from market-based countries. control) acts as a substitute (complement) Section 6 concludes. for control changes; (3) following a change in control management turnover increases, but irrespective whether performance was 2. HYPOTHESES good or poor before the transaction; and (4) performance does not improve signifi- As a starting point of our analysis, we cantly after a change in control. We con- hypothesize that changes in control have a clude from these results that the market for disciplining function in bank-based as well corporate control in Germany does not as market-based economies. To make this work as a governance device. working hypothesis testable, we now This study contributes to the literature in develop more specific hypotheses. Basically several ways. First, previous evidence on they say when firms are likely to experience ownership changes almost exclusively a change in control and what the conse- refers to market-based economies; closely quences of a control change are likely to related studies are Bethel et al. (1998) be. We pay particular attention to key and Denis and Sarin (1999). This study characteristics of the German system of examines a typical bank-based economy. corporate governance. Second, previous studies on Germany provide valuable insights into the nature of Hypothesis 1. Poorly performing firms are ownership changes; Jenkinson and more likely to experience a change in control. Ljungqvist (2001) assemble case study It is widely recognized that firm perform- evidence on hostile stake-building, and ance may suffer if management does not Franks and Mayer (2001) examine act in the interest of shareholders. The 710 ON TAKEOVER AS DISCIPLINARY MECHANISM market for corporate control moderates concentrated ownership also has costs: this divergence of interest because First, shareholders owning large blocks are managers who fail to create shareholder poorly diversified. This may render their value can be disciplined by shareholders monitoring too risk-averse (Fama and acquiring control. This discipline can take Jensen, 1983). Second, the interests of the the form of a takeover, closer shareholder concentrated shareholder do not need to monitoring, or dismissing management coincide with those of other investors. This (Shleifer and Vishny, 1986; Jensen, 1988; may allow the concentrated owner to Scharfstein, 1988). Hence, we expect redistribute wealth from others (Shleifer control changes to affect particularly and Vishny, 1997). Third, disciplinary poorly performing firms. action via takeovers becomes more difficult because incumbent large shareholders Hypothesis 2. The availability of nonmarket must be willing to sell their blocks. governance devices makes firms less likely Regardless of whether concentrated to experience a change in control. ownership is good or bad, we expect changes in control to be less likely if For market-based economies, the market ownership is concentrated. For Germany, for corporate control is often described as this negative relation between shareholder a critical governance mechanism (Jensen, concentration and the probability of a 1988; Shleifer and Vishny, 1997). For a change in control should be particularly bank-based economy such as Germany, non- pronounced because share concentration is market monitoring devices can be expected comparatively high (see, for example, Becht to play a larger role because hostile control and Böhmer, in press). transactions are rare (Franks and Mayer, The second characteristic concerns 1998) and because other constituencies lending relationships. Shleifer and Vishny such as large shareholders or large credi- (1997) argue that large creditors are tors typically have considerable power similar to large shareholders because they (Mayer, 1988; OECD, 1995). Key charac- have large investments in the firm and teristics of the German system of corporate therefore a strong incentive to monitor. governance are: (a) highly concentrated Additionally, large creditors typically have share ownership, allowing for tight share- a variety of control rights and therefore holder control, (b) close relationships to sufficient power to monitor. Hence, agency banks, allowing for tight creditor control problems should be smaller for firms under and long-term lending, and (c) a two-tier strong bank influence, making a discipli- board structure with representatives of nary change in control less profitable. For banks and shareholders on the supervisory Germany, we might expect that bank board, supporting tight control by creditors influence is particularly strong. One reason and shareholders. We expect that a firm is that historically German banks have will be less likely to experience a change in acted as so-called house banks, providing control if these alternative governance long-term loans to long-term clients mechanisms are present. (Baums, 1994). Another reason is that Concentrated share ownership is an German banks traditionally have owned important vehicle to overcome the tradi- large equity positions in other German tional free-rider problem (Shleifer and corporations and therefore have been Vishny, 1986). A large shareholder has the represented on many supervisory boards. incentive to collect information and to Mayer (1988) characterizes these lending monitor management because the expected relationships as an integral part of the private benefits from monitoring exceed the German system of corporate finance. expected costs. Hence, agency problems The third characteristic concerns the should be smaller for firms with concen- two-tier board structure of German trated ownership, making a disciplinary corporations (see Hopt, 1997 for a detailed change in control less profitable. But description). The management board MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 711 (Vorstand) manages the firm according to on the probability of a control change is its own business judgement, and the more pronounced for nonlisted firms. supervisory board (Aufsichtsrat) oversees management. The supervisory board has Hypothesis 4. Changes in control are the legal duty to supervise management, followed by corporate restructuring and and its directors are not allowed to improvements in performance. simultaneously serve on the management board.This strict separation of management If changes in control have a monitoring and control makes the supervisory board a function, we will expect operational potentially strong monitoring institution, changes to take place after a control and its presence should reduce the scope change. Jensen and Ruback (1983) argue for performance improvements following a that the market for corporate control is change in control. However, German super- part of the managerial labor market, in visory boards are subject to codetermina- which alternative management teams com- tion. This might weaken the board’s pete for the right to manage corporate monitoring function because it tends to resources. Hence, we expect increasing slow decision-making processes and endan- turnover rates for the management board. gers confidentiality. Therefore, we expect Turnover rates for the supervisory board the supervisory board’s impact on the should also increase because it comprises probability of a control change to be weak. shareholders’ representatives. But bringing an organization back on track requires Hypothesis 3. The availability of nonmarket further organizational changes. Jensen governance devices has a stronger negative (1986) argues that management is likely to impact on the probability of a control change invest free cash flow in inefficient projects for nonlisted firms than for listed firms. if monitoring is not strict enough. Reverting these investments should be A fourth characteristic of bank-based accompanied by asset sales or layoffs. governance systems is that their stock Ultimately, we expect performance to markets are smaller and less liquid than the improve after a change in control. stock markets in market-based economies Irrespective whether control changes are (OECD, 1995). In turn, the number of disciplinary or not, there are reasons to nonlisted corporations is typically large. expect other firm characteristics to be Whereas listed and nonlisted corporations related to the probability of a change in are similar regarding regulatory character- control. Among these characteristics are istics (see below), we expect their gover- insider ownership, ownership complexity, nance structures to differ for two reasons. and firm size. Insider ownership can First, nonlisted firms lack the monitoring weaken the ability of investors to take over function of the stock market, for example the firm, for example because the insiders in form of regular analysts’ reports. This belong to the founder’s family and want to makes it more difficult for investors to rec- keep the business ‘within the family’. Based ognize the necessity of a change in control on disciplinary arguments, the role of (Holmström and Tirole, 1993). Second, insider ownership is not clear. Insider own- nonlisted firms are likely to have higher ership aligns managers’ and shareholders’ shareholder concentration and higher interests (Jensen and Meckling, 1976), but leverage because they did not sell their it also restricts optimal governance due to shares to the public and thus need to rely risk-averse monitoring (Fama and Jensen, on debt. This implies that both shareholder 1983); the first scenario decreases the control and creditor control are likely to be need for disciplinary action, the second stronger, and that the scope for profitable scenario increases it. Regarding ownership control changes is accordingly smaller. In complexity, Bebchuk et al. (2000) predict summary, we expect that the negative that cross ownership and pyramid struc- impact of nonmarket governance devices tures may deter control changes; both types 712 ON TAKEOVER AS DISCIPLINARY MECHANISM of complexity are well documented for 2 years after change in ownership. In Section Germany (La Porta et al., 1999). Finally, 3.2, we investigate whether dropping firms large firms may be less likely to experience with fewer continuous years of data intro- a change in control due to wealth duces a selection bias, particularly which constraints or diversification strategies of variables are most likely to be affected. investors. Shleifer and Vishny (1992) show The sample is fairly representative for that the market for corporate control is the universe of large German corporations. less liquid as firm size increases. Taking the number of all incorporated German firms in the year 1992 as a benchmark, coverage is high for listed firms 3. DATA AND MEASUREMENT (66.6%), all of which are public corpora- 1 ISSUES tions. For nonlisted firms, coverage is small for public corporations (13.9%) and weak for private corporations (0.03%). We examine the frequency, causes, and Controlling for firm size, the sample consequences of control changes using a includes 48% of all public corporations sample of listed and nonlisted German with total sales exceeding 100 million DM, firms for the years 1987–1994. Firms from and more than 3% of large private former Eastern Germany are included after corporations. Industry coverage is suffi- 1990. We use only incorporated firms ciently representative for manufacturing because disclosure requirements are the industries. There is also a significant strictest for them.The inclusion of nonlisted number of firms from other industries such firms is essential because the determinants as wholesale trade and construction. of control changes might differ between Details on sample selection and data listed and nonlisted firms, particularly for a sources are provided in Appendix A. bank-based economy such as Germany. To This study does not distinguish between identify changes in control, we apply a private and public corporations because concept of control that is based on ultimate they are similar in many regulatory charac- share ownership (see Section 3.3). This is teristics (e.g., liability status, disclosure significant because German corporations requirements, and taxation).2 But it does are often owned through pyramids or cross- distinguish between listed and nonlisted ownership structures (La Porta et al., corporations because their governance 1999), making changes in direct share structures are likely to differ.3 As argued ownership less meaningful. above, we expect tighter shareholder control and creditor control for nonlisted 3.1. Sample firms, reflected in differences of ownership and capital structure. For the present data In total, the analysis of the frequency of set, we find that shareholder concentration control changes (Section 4) is based on is significantly higher for nonlisted firms data from 946 firms (4882 firm years). (Table 27.1). But we do not find that This sample (in the following, Sample I) creditor control is tighter, at least when includes firms with at least two continuous measured by leverage or the ratio of bank years of data because changes in ownership debt to total debt. However, we find major must be calculated. The analysis of causes differences between listed and nonlisted and consequences of control changes firms regarding ownership complexity and (Section 5) is based on a subsample and types of owners. At this point, we take these encompasses data from 664 firms (4433 observed differences as indicative for firm years). This sample (in the following, differences in governance structures. The Sample II) includes firms with at least four observed differences are generally robust continuous years of data because analyzing to excluding nonlisted subsidiaries, which the consequences of control changes were increasingly sampled (by the data requires data on the year prior and the provider) during the early 1990s. Whether MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 713 Table 27.1 Characteristics of listed and nonlisted firms

Listed firms Nonlisted firms

All All Without subsidiaries

Ownership structure Largest block 60.6% 85.9%*** 70.0%*** Ownership concentration 47.3% 80.3%*** 58.1%*** Cross ownership 6.6% 9.5%*** 6.8% Pyramid 35.6% 57.0%*** 50.9%*** Type of owner Individual 33.5% 28.1%*** 31.5% thereof: insider 22.5% 12.0%*** 12.9%*** Financial firm 9.2% 9.2% 7.1%*** Nonfinancial firm 24.0% 48.6%*** 38.2%*** Government 1.9% 4.7%*** 3.4%*** Dispersed shares 31.5% 9.3%*** 19.8%*** Capital structure Leverage 41.3% 39.0%*** 39.0%*** Bank debt 35.8% 20.7%*** 27.4%*** Number of observations 2460 2422 1141

Comparison of listed and nonlisted firms, based on the mean of various firm-specific characteristics. All characteristics are defined in Appendix A.The types of owners are: individual (family, partnership, or foundation), financial firm (bank, insurance company, or investment fund), nonfinancial firm, government authorities, and dispersed shares (shares not held in a block). Insiders are executive and nonexecutive directors (or members of their families). The test statistics are heteroscedastic t-tests of equal means; the tests compare listed firms (column 2) with nonlisted firms (columns 3 and 4, respectively). The sample comprises 946 firms (Sample I). *** Significant at the 0.01 level. this sampling procedure induces some type we find some differences that are of sampling bias is examined next. statistically as well as economically significant: For nonlisted firms, firms in Sample II have lower leverage (measured by 3.2. CHARACTERISTICS OF ENTERING debt to assets), better performance (measured AND EXITING FIRMS by industry-adjusted return on assets, return on equity, or earnings loss), and larger firm Next we check for a sampling bias. For size (measured by total assets). For listed example, if poor performance increased the firms, we find no systematic difference likelihood of firm failure or takeover, the between the two samples except for firm sample could contain systematically fewer size, which is larger in Sample II.4 poorly performing firms. This would bias Second, we examine what type of firms any observed correlation between perform- enter and exit the full sample (Sample I) ance and the likelihood of a control change during the years 1987–1994 (Table 27.2). towards zero. For firms entering in year t, the comparison First, we examine whether Sample II, group is firms in year t that are in the which is used for the analysis of causes and sample at least since year t 1. For firms consequences of control changes (Section exiting at the end of year t, the comparison 5), systematically excludes some type of group is firms in year t that are still in the firms. Comparing firms with at least four sample in year t 1. We find that entering continuous years of data (Sample II) with firms tend to be smaller and have higher firms that have fewer than four observations ownership concentration. Listed firms 714 ON TAKEOVER AS DISCIPLINARY MECHANISM

Table 27.2 Characteristics of entering and exiting firms

Entry analysis Exit analysis

Firms in sample Firms entering at least since Firms exiting Firms still in in year t year t 1 at end of year t sample in year t 1

Listed firms Largest block 67.3% 60.8%*** 84.0% 60.3%* Ownership concentration 54.5% 47.5%*** 73.8% 47.2% Debt-to-assets ratio 41.0% 41.6% 41.1% 41.2% Return on assets (ROA) 4.6% 1.2%*** 0.1% 1.5% Return on equity (ROE) 6.9% 1.3%*** 0.2% 1.8% Change in ROA 0.4% 0.3% 4.6% 0.6% Change in ROE 0.4% 0.9% 3.3% 1.6% Earnings loss 3.5% 10.1%*** 20.0% 8.9% Total assets (mn DM) 527.5 2714.2*** 27.0 2518.4*** Number of observations 171 2094 5 2094 Nonlisted firms Largest block 89.1% 85.7%*** 82.1% 84.8% Ownership concentration 85.1% 79.8%*** 75.5% 78.6% Debt-to-assets ratio 43.8% 37.9%*** 47.2% 37.8%** Return on assets (ROA) 0.1% 0.8% 0.4% 1.0% Return on equity (ROE) 1.7% 1.3% 3.9% 1.6% Change in ROA 0.9% 0.2%* 2.3% 0.4% Change in ROE 1.7% 0.2% 1.3% 0.6% Earnings loss 20.0% 13.0%*** 19.3% 12.3% Total assets (mn DM) 692.4 1453.6*** 461.8 1379.3*** Number of observations 448 1842 60 1842

Characteristics of firms entering the sample after 1987, the first year of the sample, or exiting the sample before 1994, the last year of the sample, compared with characteristics of firms that do not enter or exit the sample, separately for listed and nonlisted firms. For firms entering in year t, the comparison group is firms in year t that are in the sample at least since year t 1. For firms exiting the sample at the end of year t, the comparison group is firms in year t that are still in the sample in year t 1. Firms are compared at the mean of each firm characteristic. All characteristics are defined in Appendix A. All performance measures except the loss indicator are calculated as the difference to median industry performance (two-digit industry level).The test statistics are heteroscedastic t-tests of equal means. The sample comprises 946 firms (Sample I). * Significant at the 0.10 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level. typically show better performance when reduce the explanatory power of poor entering, whereas nonlisted firms show performance in a regression explaining the poorer performance. Regarding exiting probability of control change. Second, a firms, the main difference to firms remain- systematic bias due to sample attrition is ing in the sample is firm size, which tends unlikely because, with the exception of to be smaller for listed and nonlisted firms firm size, exiting firms are not signifi- leaving the sample. cantly different from other sample firms. There are two lessons from this sample This result could be expected because composition analysis. First, it is possible 60% of sample exits are cases of nonre- that Sample II is somewhat biased towards porting due to name changes, and not due well-performing firms, but only for non- to bankruptcy or ownership change (see listed firms. As discussed above, this can Appendix A). MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 715 3.3. CONCEPT OF CONTROL firm in the aeronautics industry (Figure 27.1). It is 100% owned by The identification of the ultimate owner for Dornier, the direct shareholder (level 1). In each firm is based upon German corporate turn, Dornier is owned by the Dornier fam- law and involves two steps. First, we ily (42.4%) and by Deutsche Aerospace identify the ultimate owner for every direct (57.6%) (level 2). This latter firm is part shareholder of a firm using the following of the Daimler-Benz conglomerate, being three rules: owned with 83.0% by the automobile giant (level 3). Daimler-Benz itself is in the ● Rule 1 (strong ownership rule): A hands of Deutsche Bank (28.2%), chain of control is pursued to the Mercedes Automobil-Holding (25.2%), next level if the shareholder being and the Emirate of Kuwait (14%).The rest analyzed is owned to 50% or more of the shares is dispersed. Applying our by a shareholder on the next level, concept of control, we find a continuous while all other shareholders on the chain of majority stakes between Dornier next level own less than 50%. Aeronautics and Daimler-Benz, but several ● Rule 2 (weak ownership rule): If shareholders owning minority stakes larger rule 1 does not apply, a chain of than 25% in Daimler-Benz. Hence, we control is pursued to the next level if identify Daimler-Benz as the ultimate the shareholder being analyzed is owner of Dornier Aeronautics. owned to 25% or more by a share- The strength of our concept of control is holder on the next level, while all that it clearly identifies one single ultimate other shareholders on the next level owner and at the same time accounts for own less than 25%. complex ownership structures. Identifying ● Rule 3 (stop rule): If neither rule 1 one controlling shareholder is important nor rule 2 applies, a chain of control for this study because it allows to examine is not pursued further. changes in control. If instead multiple controlling shareholders were defined, for These rules guarantee that no more than example by pursuing each ownership chain one ultimate owner is identified for every to its end, it would remain unclear which of direct shareholder. Note that if a share- the so-defined controlling shareholders holder has split his ownership stake in a actually controls the respective firm. particular company into several smaller Taking into account complex ownership stakes, for example into two blocks of 50% structures is important because pyramid held by two subsidiary firms, we combine structures are widespread in bank-based these smaller stakes into one single block. economies. If instead control were strictly We set the first cutoff point at 50% defined at the direct level of ownership, this because German law allows an investor would assume that shareholders on higher owning 50% of all shares to appoint man- levels cannot influence decisions at the agement.5 The second cutoff point is set at base level. Whereas control might be 25% because an investor owning 25% of partially diluted in pyramids, for example the shares has the right to veto decisions. due to information asymmetries or transaction After having identified the ultimate owner costs, full dilution is unlikely since pyramids of every direct shareholder,the second step are attractive forms of organization. is to apply the three rules to all direct One weakness of our concept of control shareholders.6 This allows to identify one could be that it uses fixed thresholds. single shareholder that is in ultimate control. We recognize that the selection of these When no single shareholder fulfills the cri- thresholds to some extent is arbitrary (see teria, the respective firm is seen to have no also Short, 1994). But as argued above, ultimate owner. these thresholds are based on German cor- For illustration, consider the example of porate law and therefore associated with Dornier Aeronautics, a nonlisted mediumsized explicit control rights. In addition, the size 716 ON TAKEOVER AS DISCIPLINARY MECHANISM

Dornier Aeronautics GmbH

100.0% Dornier GmbH

57.6% 42.4% Deutsche Aerospace AG Dornier family

17.0% 83.0% Dispersed shares Daimler-Benz AG

28.2% 25.2% 14.0% 32.6% Deutsche Bank AG Mercedes Automobil-Holding AG Emirate of Kuwait Dispersed shares

Figure 27.1 Ownership structure of Dornier in 1992.

distribution of share blocks for German the precondition for bank voting power corporations shows large peaks at exactly based on proxy rights. Hence, if there is a these thresholds, indicating that sharehold- discrepancy between German banks’ ers hold these blocks in awareness of the ownership rights and voting rights, it is associated control rights (Köke, 2001; Becht probably small. In summary, the concept of and Böhmer, in press). Therefore, varying control has potential weaknesses. But these these thresholds should not significantly weaknesses do not appear to induce a change the main findings of this study. systematic bias in the identification of the Another weakness could be that the ultimate owner for each sample firm. concept of control is based on officially reported ownership rights instead of voting rights.7 Ownership rights do not need to coincide perfectly with voting rights for two 4. FREQUENCY OF CONTROL reasons: First, shares might have voting CHANGES caps or carry multiple voting rights. Second, shares that are officially reported In this section, we determine the frequency as dispersed might be under de facto of changes in ownership, the size of traded control of banks because the German blocks, and the types of buyers and sellers system of proxy voting gives banks the right of control blocks. We distinguish between to execute voting rights of deposited shares. changes in the ultimate owner of each Unless shareholders have given instruc- direct share block (block trades) and tions, banks can multiply their influence changes in the ultimate owner of each sam- this way. We believe that these weaknesses ple firm (control changes). For illustration, do not induce a systematic bias. Although consider the following example (Figure 27.2). legal during 1987–1994, limitations on In the year 1990, Boge, a large nonlisted voting rights do not seem to be widespread. firm in the rubber industry, is owned by four Gorton and Schmid (2000) report that large shareholders. Applying the concept of only 14 of the largest German firms had control, we identify Carlo de Benedetti as voting rights restrictions. Hence, for the the ultimate owner of Boge. In the year vast majority of sample firms, voting rights 1991, Mannesmann becomes the new do coincide with ownership rights. As for ultimate owner of VDO Adolf Schindling, a proxy voting, recent evidence suggests that direct shareholder of Boge. This is a block proxy voting is extremely unlikely to trade from the Schindling-Rheinberger enhance German banks’ voting power family to Mannesmann. Note that de considerably (Edwards and Nibler, 2000). Benedetti remains the ultimate owner of In addition, only few of our sample firms Boge. In the year 1992, Mannesmann have a dispersed shareholder base, which is acquires almost all shares of Boge, becoming MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 717

1990: Boge AG

44.3% 20.1% 24% 6% SOGEFI S.p.A./ Italy VDO AdoIf Schindling AG Commerzbank AG TRW Inc.

100.0% 100.0% Carlo de Benedetti Schindling-Rheinberger family

1991: Boge AG

44.3% 20.1% 24% 6% SOGEFI S.p.A./ Italy VDO Adolf Schindling AG Commerzbank AG TRW Inc.

100% 51% 49% Carlo de Benedetti Mannesmann AG Schindling-Rheinberger family

1992: Boge AG

98% 2% Mannesmann AG Dispersed Shares

Figure 27.2 Ownership structure of Boge AG in 1990, 1991, and 1992. its new ultimate owner.These block trades (Renneboog, 2000) and 9.0% for the UK lead to a change in control. (Franks et al., 2001). For the US, Bethel et In total, the sample encompasses 2460 al. (1998) report a rate of 6.6% for blocks firm years for listed firms and 2422 firm larger than 5%. Hence, the frequency of years for nonlisted firms. This is the total block trades in Germany is similar to other number of cases in which block trading countries. For nonlisted firms, the respec- could occur. For listed firms, in 258 cases tive annual rate is 7.0%; no comparable (10.5% on average in any given year), we figures from other countries are available. observe that a block is purchased by a new As exemplified by the case of Boge (Figure shareholder,who did not own a block before 27.2), the trade of a large block does not the purchase. For nonlisted firms, the need to be associated with a change in respective rate of block purchases is 7.3%. control over the entire firm. An example of Table 27.3 shows the size distribution of this is a new shareholder purchasing a large blocks purchased by new shareholders. For block, while the majority of shares remains listed firms, the purchased block is a in the hands of the ultimate owner.To iden- majority block in 107 cases (41.5%), and tify changes in control, we check for name a minority block in 62 cases (24.0%).This changes of the ultimate owner.We find that strongly suggests that new shareholders a change in control takes place in 205 cases come in because they want to take over in listed firms (8.3% on average in any control of the respective firm. given year), and in 146 cases in nonlisted To put these results into perspective, we firms (6.0%).8 Hence, the frequency of con- compare them with evidence from other trol changes is somewhat lower than sug- countries. For Germany, we find that a new gested by the number of blocks traded. We shareholder purchases a block larger than conclude that not all block trades lead to 10% in 9.9% of listed firms, on average in change in control but that a large part does. any given year. This rate of block pur- To see whether a particular shareholder chases compares to 12.4% for Belgium type dominates control changes, we explore 718 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 27.3 Size distribution of blocks purchased by new shareholders

Size of purchased block

[5%; 10%] [10%; 25%] [25%; 50%] [50%; 100%] Total number Total Percentage Total Percentage Total Percentage Total Percentage of purchases (%) (%) (%) (%)

Listed 258 11 4.3 74 28.7 62 24.0 107 41.5 firms Nonlisted 178 6 3.4 24 13.5 21 11.8 125 70.2 firms

Size distribution of blocks that are purchased by a new shareholder (i.e. a shareholder that did not own a block before the transaction) for the period 1987–1994, separately for listed and nonlisted firms. Block size is observed at the direct level of ownership, block ownership is determined at the ultimate level, applying the concept of control. Therefore, purchases by new shareholders refer to changes in ultimate ownership of direct share blocks. Column 2 shows the total number of block purchases by new shareholders. Columns 3, 5, 7, and 9 disaggregate these purchases into four size classes. Columns 4, 6, 8, and 10 relate the disaggregate number of blocks to the number of all block purchases by new shareholders (column 2). The sample comprises 946 firms (Sample 1). the transaction partners more closely. For 5. CAUSES AND CONSEQUENCES OF listed firms, most control blocks are sold by CONTROL CHANGES nonfinancial firms (33.2%) and individuals (29.3%). In turn, the largest number of In the following, we analyze the causes and control blocks is purchased by nonfinancial consequences of control changes (Sections firms (41.5%) and individuals (25.9%). 5.1 and 5.2, respectively).The aim is to test Compared to the fraction of listed firms the hypotheses lined out in Section 2. The under ultimate ownership of nonfinancial analysis is based on Sample II, which firms (28.6%) and individuals (42.4%), it contains firms with at least four continuous becomes clear that nonfinancial firms trade years of data. Throughout, we distinguish control blocks more often than would between listed and nonlisted firms. In be expected by their frequency as ultimate Section 5.3, we compare our findings from owners. In comparison with all other share- the bank-based German economy to previ- holder types, nonfinancial firms are the ous findings from market-based economies. most active traders of control blocks. For nonlisted firms, nonfinancial firms buy an even larger fraction of control blocks 5.1. Causes of control changes (56.8%), more than three times the frac- To examine the antecedents of control tion individuals purchase. These findings changes, we apply a univariate and a stand in sharp contrast to evidence from multivariate analysis. The univariate analy- market-based governance systems. For the sis provides first evidence on the relation US, Bethel et al. (1998) report that between control changes and key firm 13.1% of share blocks are purchased by characteristics such as performance and strategic investors such as Gulf and governance structure. The multivariate Western or IBM. For the UK, Franks et al. analysis investigates the causes of control (2001) report a comparable ratio of changes in a more systematic fashion using 14.5%. In summary, although the fre- a logistic regression model. quency of control changes is similar for bank-based and market-based economies, 5.1.1. Univariate analysis the players differ fundamentally. Whether this affects the mechanics of control Table 27.4 compares firms that experience changes is examined next. a change in ultimate ownership in year 0 MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 719 Table 27.4 Causes of control changes: univariate results

Listed firms Nonlisted firms No change Change No change Change in control in control in control in control

Return on assets (ROA) year – 2 1.3% 1.3% year – 1 0.8% 0.3% year 0 1.8% 0.9% 0.9% –1.0% Earnings loss year – 2 9.9% 6.2%** year – 1 13.7%** 12.4% year 0 8.0% 11.8%* 13.8% 15.5% Ownership concentration year – 2 36.1%*** 60.2%*** year – 1 36.3%*** 61.4%*** year 0 49.8% 40.2%*** 82.1% 68.3%*** Bank debt year – 2 39.3%*** 29.8%*** year – 1 40.6%*** 31.2%*** year 0 33.7% 40.3%*** 18.8% 30.8%*** Supervisory board year – 2 100.0% 67.0%* year – 1 100.0% 67.0%* year 0 100.0% 100.0% 56.8% 67.0%* Insider ownership year – 2 14.6%*** 12.6% year – 1 12.2%*** 10.3% year 0 27.3% 9.7%*** 13.5% 5.4%*** Cross ownership year – 2 6.8% 7.2% year – 1 6.2% 7.2% year 0 5.8% 8.1% 10.2% 7.2% Total assets (mn DM) year – 2 1327.7*** 960.2*** year – 1 1426.8*** 1030.0*** year 0 3089.1 1630.9*** 1598.0 1012.8*** Number of observations 1443 978 1391 621

Characteristics of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6), compared with characteristics of firms that do not experience such a change in any year (columns 3 and 5), separately for listed and nonlisted firms. Firms are compared at the mean of each firm characteristic. For firms with a change in control, statistics are calculated for the 2 years preceding the change (year – 1 and year – 2) and the year of change (year 0). For other firms, statistics refer to the whole period of observation. All characteristics are defined in Appendix A.The test statistics are heteroscedastic t-tests of equal means. The sample comprises 664 firms (Sample II). * Significant at the 0.10 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level.

(change in control) with firms that do not change (8.0%). These purely descriptive experience such a change in any of the results suggest that poor profitability gen- sample years (no change). We find that erally does not induce a change in ultimate profitability (measured by industry- ownership. Only in case of an earnings loss adjusted return on assets) is weaker during a change in control becomes more likely. the 2 years preceding a change in control We take these results as indicative that the (year –1, year –2). But the difference to German market for corporate control firms not experiencing a change in control (measured by changes in ultimate owner- is statistically insignificant. The proportion ship) addresses governance problems of loss-making listed firms is significantly (reflected in poor performance) and thus higher among firms in which a control potentially works as a disciplining device. change is going to occur within 1 year Next, we consider three alternative (13.7%) than in firms without a control governance devices: (1) shareholder control, 720 ON TAKEOVER AS DISCIPLINARY MECHANISM (2) creditor control, and (3) supervisory board (or their families), we find that firms board control. Concerning shareholder con- not experiencing a change in control tend trol, we find that ownership is significantly to show higher insider ownership; but this less concentrated in firms experiencing a holds only for listed firms. This result sug- change in control.This finding is consistent gests that insiders try to hinder changes in with the view that concentrated ownership control. Finally, measuring ownership ensures good governance, making a discipli- complexity by a dummy that indicates nary change in control less profitable and whether the ultimate owner belongs to the therefore less likely. It is also consistent well-known web of German financial and with the view that concentrated ownership nonfinancial firms (Wenger and Kaserer, makes it more difficult for outside 1998)9, we find no evidence that complex investors to assemble a controlling block. ownership structures deter control purchases. Concerning creditor control, we find that the ratio of bank debt to total debt is sig- 5.1.2. Multivariate analysis nificantly higher among firms experiencing a change in control. Taking bank debt as a The above univariate analysis provides a proxy for bank influence, this implies that first indication of the determinants of control changes are more likely for firms control changes. To take into account that under bank influence. This result is incon- different variables can simultaneously sistent with the view that strong creditors affect the probability of a control change, continuously monitor borrowers and we examine the determinants of control thereby reduce the need for disciplinary changes using a logistic regression model control changes. Rather, banks appear to (Table 27.5). This model predicts assist control changes. control changes using 1-year-lagged meas- The supervisory board is mandatory for ures of performance, alternative governance listed firms. Thus, we cannot determine devices, and other deterrents of control whether its presence is related to the changes. To investigate the difference probability of a control change for listed between listed and nonlisted firms, we run firms. For nonlisted firms, there is weak the regression on the pooled sample and evidence that a larger proportion of firms include interaction terms for nonlisted experiencing a change in control has a firms, i.e. a dummy that indicates a supervisory board than firms not experi- nonlisted firm multiplied by the respective encing a change in control. Hence, the firm characteristic. Hence, the columns for presence of a supervisory board does not nonlisted firms in Table 27.5 show to which appear to make control changes less likely. extent their slope parameters differ from This finding is inconsistent with the view those for listed firms. that monitoring by a board modifies the Model (1) confirms that firms making need for disciplinary action. As argued in earnings losses are more likely to experi- Section 2, some firm characteristics can be ence a change in control.Tight shareholder expected to reduce the probability of a control, measured by ownership concentration, change in control, irrespective whether these serves as an alternative governance device, changes are disciplinary or not: insider reducing the probability of a control ownership, ownership complexity, and firm change. Tight creditor control does not size. Measuring firm size by total assets, we reduce the probability of a control change. find that firm size tends to be larger for To the contrary, firms with strong creditors firms that do not experience a change in are more likely to experience a change in control. This supports the view that firm control. We find no evidence that the size deters control purchases, for example supervisory board acts as a substitutive due to wealth constraints of investors. governance device to changes in control. As Furthermore, measuring insider ownership argued in Section 2, this lack of monitoring by the fraction of shares owned by mem- power may be due to co-determination of bers of the management or supervisory supervisory boards. Among the other MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 721 deterrents of control changes, only insider creditor control for addressing governance ownership is significantly negatively related problems in large German firms. Our finding to the probability of a control change. The that strong creditor control enhances the determinants of control changes do not probability of a control change is consis- differ between listed and nonlisted firms as tent with recent case-study evidence: the column for nonlisted firms shows. The Jenkinson and Ljungqvist (2001) docu- only exception is the dummy indicating ment many cases where German banks play earnings losses, which is significantly smaller an assisting role in building hostile stakes. for nonlisted firms. In fact, taking the sum Finally, we do not find evidence that the of both earnings loss coefficients, we obtain determinants of control changes differ the respective slope parameter for non- between listed and nonlisted firms.Thus, we listed firms; this parameter is not statisti- do not find support for H3. cally different from zero at the 0.10 level. Hence, for nonlisted firms, poor performance 5.1.3. Sensitivity of results does not affect the probability of a change in control. As argued in Section 3.2, this To test the sensitivity of our results, we result might be due to a selection problem conduct a number of robustness checks because our sample contains systematically (Table 27.6). Our benchmark is Model (2) fewer poorly performing nonlisted firms. from Table 27.5. First, we examine whether Model (2) investigates the positive our results depend on the choice of the impact of bank debt on the probability of a performance measure. Choosing a dummy control change more closely. The aim is to that indicates low interest coverage instead see whether this effect is due to the credi- of an earnings loss leaves all results quali- tor position of German banks or to their tatively unaffected (Model 2a).When using shareholder position. As argued in Section industry-adjusted return on assets, we find 2, German banks are often large creditors that the performance variable looses its as well as large shareholders. To examine significance, but the other results remain this issue, we interact bank debt with the qualitatively unaffected (Model 2b).10 total burden of debt (leverage) and with These findings confirm that poor perform- the fraction of shares ultimately owned by ance makes a control change more likely, banks (bank ownership). Table 27.5 shows but only if performance is very poor.Model that bank debt looses its explanatory (2c) examines whether our results depend power. Instead, the interaction of leverage on the choice of the measure for share- and bank debt is significantly positive.This holder control. We find that measuring indicates that the banks’ role in promoting shareholder concentration by the size of the control changes comes from their strong largest share block instead of the creditor position. In comparison, banks do Herfindahl concentration index does not not appear to enhance their monitoring affect our findings qualitatively. This result role based on their shareholder position. In could be expected because shareholder summary, we find support for H1 that poor concentration in Germany is typically very performance makes firms more likely to high (see also Table 27.1). experience a change in control, but only if Next, we examine the impact of owner- the firms’ financial difficulties are severe. ship complexity more closely. Replacing We find partial support for H2 because the dummy that indicates cross ownership tight shareholder control functions as a with a dummy that indicates pyramid substitutive governance device to control structures, we find essentially the same changes. But we find no support for the results (not tabled). In particular, the view that creditors or the supervisory impact of ownership complexity remains board act as alternative governance insignificant. These findings stand in con- devices. These findings are consistent with trast to the popular view that complex Edwards and Nibler (2000) who show that German ownership structures hinder shareholder control is more important than control changes. 722 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 27.5 Causes of control changes: multivariate results Model (1) Model (2) Listed firms Nonlisted firms Listed firms Nonlisted firms

Earnings loss 0.863** (0.034) 1.300* (0.052) 0.791** (0.050) 1.261* (0.062) Ownership 0.992*** (0.000) 0.353 (0.384) 0.981*** (0.000) 0.352 (0.388) concentration Bank debt 1.072*** (0.001) 0.191 (0.701) 0.160 (0.815) 1.040 (0.346) Bank debt 2.114** (0.031) 1.310 (0.460) leverage Bank debt 0.378 (0.695) 0.176 (0.932) bank ownership Supervisory 0.013 (0.952) 0.018 (0.934) board Insider 0.797*** (0.006) 0.845 (0.165) 0.821*** (0.005) 0.836 (0.172) ownership Cross ownership 0.053 (0.874) 0.342 (0.459) 0.014 (0.968) 0.327 (0.486) Log (total 0.050 (0.130) 0.017 (0.293) 0.040 (0.234) 0.013 (0.435) assets) Number of 4181 4181 observations Log 1030.9 1028.4 likelihood

Results of logistic regression models predicting changes in control by various firm characteristics. The models estimate the slope parameters for listed firms (column listed firms) and the difference in slope parameters compared to nonlisted firms (column nonlisted firms). All characteristics are defined in Appendix A. Both models include an intercept and time dummies (not reported). The p-values (reported in parentheses) are based on robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance– covariance matrix. * Significant at the 0.10 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level.

Finally, we run Model (2) on two 5.2. Consequences of control subsamples. First, we exclude all nonlisted changes subsidiary companies because their disclosed annual reports could to some extent be To analyze the consequences of control influenced by biased transfer prices within changes, we compare firms that experience conglomerates. Then, these reports would a change in ultimate ownership with firms not reflect the subsidiaries’ true financial that do not experience such a change in any status. Second, we exclude all firms from of the sample years. Table 27.7 examines Eastern Germany for which data are the effects of control changes on corporate included since 1990. The reason is that restructuring; Table 27.8 the effects on their governance structures might be performance. All figures are calculated for fundamentally different from those in the year of ultimate ownership change Western Germany. Both corrections do not (year 0), the two following years (year 1, change the results of Model (2) estimated year 2), and the year prior to the change on the full sample (not tabled). In (year – 1). summary, we are confident that our main Regarding corporate restructuring, we findings from Section 5.1.2 are robust to use the following measures: management alternative definitions of explanatory vari- turnover, asset divestitures, employment ables as well as the inclusion of potentially growth, and cutting of labor costs. questionable types of firms. Concerning management turnover, we MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 723 ) (0.076) (0.091) * * listed firms 0.006 (0.977) 1029.1 0.267 (0.897) 1.268 (0.475) 0.760 0.770 (0.205) 1.199 (0.001) (0.004) (0.030) (0.079) * ** *** *** 0.050 (0.135) 0.027 (0.158) 0.393 (0.683) 0.125 (0.854) 1.019 (0.356) 2.117 0.717 0.923 0.838 -values (reported parentheses) are based in -values p pe parameters for listed firms (column firms listed for pe parameters 0.005 (0.980) 1029.6 0.283 (0.891) 0.823 (0.179) 1.404 (0.425) 0.443 (0.276) (0.001) (0.005) (0.032) ** *** *** 0.044 (0.185) 0.011 (0.523) 0.279 (0.771) 0.405 (0.443) 0.873 (0.212) 0.246 (0.717) 1.198 (0.273) 0.927 0.826 ). the robust- for A.The model Appendix benchmark in are defined All characteristics (0.021) ** 1015.3 1.318 (0.458) 2.113 0.843 (0.171) 0.033 (0.987) 1.383 0.408 (0.321) nonlisted firms (0.005) (0.006) (0.000) (0.029) *** ** *** *** 0.033 (0.329) 0.015 (0.364) 0.031 (0.927) 0.370 (0.432) 0.013 (0.969) 0.296 (0.527) 0.031 (0.928) 0.280 (0.551) 0.367 (0.703) 0.829 0.190 (0.779) 1.084 (0.326) 0.979 (0.062) * 0.018 (0.934) 0.012 (0.957) 1028.4 0.176 (0.932) 1.261 1.310 (0.460) 2.139 0.836 (0.172) 0.352 (0.388) (0.005) (0.000) (0.050) (0.031) *** ** ** *** 0.821 0.040 (0.234) 0.013 (0.435) 0.014 (0.968) 0.327 (0.486) 0.378 (0.695) 0.160 (0.815) 1.040 (0.346) 0.981 0.791 2.114 in slope parameters compared to nonlisted firms (column firms compared to nonlisted slope parameters in Listed Nonlisted Listed Nonlisted Listed Nonlisted Listed Nonlisted Model (2) Model (2a) Model (2b) Model (2c) Causes of control changes:Causes of control tests robustness difference Significant at the 0.01 level. Significant Significant at the 0.05 level. Significant Significant at the 0.10 level. Significant * ** *** board observations assets) bank ownership adjusted ROA leverage coverage concentration ownership on robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance–covariance matrix. standardon robust errors, the variance–covariance for sandwich estimator White/Huber the which are calculated using ness checks is Model (2),ness checks is Model from is taken which 27.5. (not reported). Table dummies and time an intercept include All models The and the Insider Number of Number 4181 4145 4181 4181 Results of logistic regression models predicting changes in control by various firm characteristics.The models estimate the slo estimate characteristics.The models firm by various control in changes predicting models regression Results of logistic Log (total Supervisory Ownership Bank debt Cross ownership Cross Bank debt Table 27.6 loss Earnings Log likelihood Industry Largest block Low interestBank debt 0.952 724 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 27.7 Corporate restructuring following control changes

Listed firms Nonlisted firms No change Change No change Change in control in control in control in control

Turnover of CEO year 1 10.2% 12.8% year 0 14.0% 21.1%** 14.2% 17.9% year 1 19.5%* 25.6%** year 2 21.1%** 20.5% Turnover of year 1 10.9% 11.2% executive directors year 0 11.9% 19.6%*** 11.5% 20.6%*** year 1 15.8%* 23.4%*** year 2 16.5%** 16.2% year 1 6.9% 6.0% Sales of fixed assets year 0 6.2% 8.8%** 6.6% 5.2%* year 1 7.4% 8.5% year 2 7.3% 5.8% year 1 16.5% 12.1% Sales of financial assets year 0 16.2% 20.0% 14.8% 16.7% year 1 18.6% 13.8% year 2 17.4% 17.7% year 1 4.6% 0.1% Growth rate of year 0 4.6% 5.5% 1.3% 3.2% employment year 1 1.1% 1.1% year 2 2.0%*** 3.2% year 1 4.6% 3.3%** year 0 4.6% 4.3% 6.4% 4.9% Growth rate of year 1 10.9%** 13.6%** labor costs year 2 4.3% 6.8%

Number of observations 1443 978 1391 621

Characteristics of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6), compared with characteristics of firms that do not experience such a change in any year (columns 3 and 5), separately for listed and nonlisted firms. Firms are compared at the mean of each firm characteristic. For firms with a change in control, statistics are calculated for the year preceding the change (year –1), the year of change (year 0), and the 2 years following the change (year 1 and year 2). For other firms, statistics refer to the whole period of observation. All characteristics are defined in Appendix A. The test statistics are heteroscedastic t-tests of equal means. The sample comprises 664 firms (Sample II). * Significant at the 0.10 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level.

observe that the replacement rates of the For the supervisory board, we also find CEO and other executive directors strongly strongly increasing turnover rates, both for increase following control changes. For the chairman and the other board members example, for listed firms that experience a (not tabled). change in control CEO turnover rates more Concerning asset divestitures, we find than double from 10.2% in the year pre- that control changes are accompanied by ceding the change to 21.1% in the year of increasing sales of fixed assets, but only change. This turnover rate is significantly for listed firms. Specifically, sales rates of above average CEO turnover in listed firms fixed assets increase from 6.9% in year – 1 that do not experience a change in control. to 8.8% in year 0. The difference in the MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 725 rate of asset sales to firms without a contrary, labor costs per employee change in control is significant at the 0.05 significantly increase after a change in level. Regarding employment, we find that ultimate ownership.This might reflect that in the second year after the control the new owner is working to improve change, the average growth rate of employ- incentives for employees. ment is negative for listed and nonlisted Finally, the evidence does not support firms, but the difference in growth rates the view that control changes improve between firms with and without a change performance (Table 27.8). Neither for in control is significant only for listed listed firms nor for nonlisted firms we firms. Hence, employee layoffs increase, observe significant improvements in but not before 2 years after the change in operating performance following control control.This time lag suggests that adjust- changes. Only the fraction of loss-making ment costs play a role in reorganizing a listed firms declines to a level that is firm. Concerning cost cutting, we find no comparable to the industry benchmark. In evidence of decreasing labor costs. On the summary, we find no support for H4.

Table 27.8 Performance following control changes

Listed firms Nonlisted firms No change Change No change Change in control in control in control in control

Return on assets (ROA) year 1 0.8% 0.3% year 0 1.8% 0.9% 0.9% 1.0% year 1 1.0% 1.9% year 2 1.1% 0.4% Return on equity (ROE) year 1 2.7% 0.3% year 0 2.1% 0.1% 1.4% 4.6%** year 1 2.4% 2.1% year 2 2.4% 2.6% Change in ROA year 1 0.6% 1.1% year 0 0.4% 0.3% 0.1% 1.6% year 1 0.4% 2.3%** year 2 0.2% 1.5% Change in ROE year 1 1.0% 2.0% year 0 0.4% 5.0% 0.2% 8.6%* year 1 0.2% 6.0% year 2 0.7% 7.2% Earnings loss year 1 13.7%** 12.4% year 0 8.0% 11.8%* 13.8% 15.5% year 1 11.2% 11.0% year 2 11.8% 11.9% Number of observations 1443 978 1391 621

Performance of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6), compared with performance of firms that do not experience such a change in any year (columns 3 and 5), separately for listed and nonlisted firms. Firms are compared at the mean of each performance measure. For firms with a change in control, statistics are calculated for the year preceding the change (year 1), the year of change (year 0), and the 2 years following the change (year 1 and year 2). For other firms, statistics refer to the whole period of observation. All performance measures are defined in Appendix A; all except for the loss indicator are calculated as the difference to median industry performance (two-digit industry level). The test statistics are heteroscedastic t-tests of equal means. The sample comprises 664 firms (Sample II). * Significant at the 0.10 level. ** Significant at the 0.05 level. 726 ON TAKEOVER AS DISCIPLINARY MECHANISM 5.3. Comparison with market-based between ownership and disciplinary action economies against management for Germany. In turn, our finding that changes in ownership are To put our findings into perspective, we not related to the dismissal of poorly compare them to evidence from market- performing management is consistent with based economies. Regarding the causes of Franks and Mayer (2001). This questions control changes, Bethel et al. (1998) show the disciplinary nature of control changes for the US that poor performance makes in a bank-based economy. block purchases more likely (see also Denis What is similar to market-based and Sarin, 1999; Maksimovic and Phillips, economies is the role of insider ownership. 2001). Unlike in the US, ownership For the US, Barber et al. (1995) and changes in Germany become more likely Bethel et al. (1998) show that insiders can only if performance is extremely poor.This deter changes in control when owning large suggests that control changes in a bank- blocks. This is consistent with our evidence based economy are not well suited to from Germany. Likewise, consistent with address regular performance declines. But Bethel et al. (1998) and Mulherin and they might come into play when all other Boone (2000), we document that larger governance mechanisms clearly have failed. firms are less likely to experience a change An alternative monitoring device to in control, although the impact of firm size control changes is tight shareholder disappears in the multivariate analysis. control.We are not aware of any study that Regarding the consequences of control tests for a substitutive relationship between changes, recent studies on market-based these two governance mechanisms for a economies document various types of market-based economy. For Germany, we restructuring. CEO turnover strongly find that tight shareholder control makes increases following block purchases of control changes less likely, and we take this activist investors (Bethel et al., 1998) and as evidence for this substitutive relation- following large increases or decreases in ship. To explicitly test for the disciplinary insider ownership (Denis and Sarin, 1999). role of both shareholder control and Similarly, asset divestitures and employee control changes, we examine their joint layoffs increase, and the frequency of impact on management turnover in a mul- mergers and acquisitions decreases after tivariate framework. We look at manage- changes in ownership (Bethel et al., 1998, ment turnover because it is often regarded for the US; Franks and Mayer, 1995, for as a key indicator of disciplinary action the UK). For Germany, we also observe (Bethel et al., 1998; Denis and Sarin, higher rates of management turnover.Sales 1999).Table 27.9 confirms that changes in of fixed assets increase as well, but only control are associated with higher rates weakly, and employee layoffs increase, but of CEO turnover.The impact of ownership not before 2 years after the control change. concentration is also positive but insignifi- In contrast to the US (Bethel et al., 1998), cant. When we interact both governance the impact on performance is weak. Only devices with a dummy that indicates an the fraction of loss-making firms declines earnings loss in the previous year, we find after control changes. that the interaction term is insignificant for In summary, some of the causes and control changes but significant for owner- consequences of control changes are ship concentration. These findings suggest similar for bank-based and market-based that control changes as well as tight share- economies. But since we cannot find holder control make CEO turnover more evidence that control changes help disci- likely, but only tight shareholder control pline poorly performing management, we disciplines poor performance. The benefi- cannot support the working hypothesis of cial role of tight shareholder control stands this study. Control changes in bank-based in contrast to Kaplan (1995) and Franks Germany do not play the disciplinary role and Mayer (2001) who find no relation that they play in market-based economies. MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 727 Table 27.9 Logistic regressions predicting CEO turnover Listed firms Nonlisted firms

Change in control 0.612*** (0.001) 0.049 (0.888) Change in control earnings loss in t 1 0.563 (0.536) 1.558 (0.370) Ownership concentration 0.247 (0.153) 0.044 (0.792) Ownership concentration earnings loss in t 1 0.958** (0.026) 0.058 (0.937) Number of observations 3187 Log likelihood 1335.9

Results of logistic regression model predicting CEO turnover by two governance devices (change in control, tight shareholder control) and their interaction with poor performance (earnings loss in previous period). The model estimates the slope parameters for listed firms (column listed firms) and the difference in slope parameters compared to nonlisted firms (column nonlisted firms). All characteristics are defined in Appendix A.The model includes an intercept and time dummies (not reported). The p-values (reported in parentheses) are based on robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance– covariance matrix. ** Significant at the 0.05 level. *** Significant at the 0.01 level.

6. SUMMARY AND CONCLUSIONS not make control changes less likely. This suggests that the board cannot substitute This paper sheds light on the market for for control changes; this stands in contrast corporate control in a bank-based country. to evidence from the US (Kini et al., We focus on Germany because high owner- 1995). Codetermination may explain the ship concentration and tight bank relation- weak monitoring power of German supervi- ships make it the prototype of a bank-based sory boards. Contrary to our expectation, system. We examine the frequency, causes, we find that control changes are more and consequences of changes in corporate likely for firms under strong creditor influ- control and compare our findings to those ence. This suggests that creditor control from market-based countries. and control changes are complementary; In contrast to the common perception, this assertion is supported by recent Germany has an active market for large case-study evidence on hostile stake-building share blocks. We find that the frequency of in Germany (Jenkinson and Ljungqvist, block trades is similar to the US and the 2001). UK. Since the traded blocks are typically Whereas the observed causes of control very large, block trades usually lead to a changes are consistent with a disciplinary change in control. The most active traders role of control changes, the observed of control blocks are nonfinancial firms. consequences are not. We find that follow- This stands in contrast to market-based ing a change in control management, economies where nonfinancial firms play turnover increases, but irrespective only a minor role in control transactions. whether performance was good or poor Regarding the causes of control changes, before the control transaction. For listed we find that poor performance makes a firms, also sales of fixed assets and layoffs change in control more likely.This indicates increase. In contrast to evidence from the that control changes in Germany may play US (Bethel et al., 1998; Denis and Sarin, a disciplinary role as in market-based 1999), we cannot confirm that sales of economies. We also find that high owner- financial assets increase or labor costs ship concentration makes control changes decrease. Similarly, we cannot confirm that less likely. This is consistent with the view performance significantly improves after a that tight shareholder control acts as a change in control. In summary, although substitute for disciplinary control changes. there is an active market for corporate The presence of a supervisory board does control in Germany and although there are 728 ON TAKEOVER AS DISCIPLINARY MECHANISM some indications that control changes ACKNOWLEDGEMENTS address poor performance, the evidence on the consequences is inconsistent with a Financial support from the German disciplinary role of control changes. Science Foundation (DFG) is gratefully Finally, we find that the causes and acknowledged (grant no. BO 934, 7-2). consequences of control changes do not Comments from Axel Börsch-Supan, Dirk differ between listed and nonlisted firms. Guennemann, Silke Januszewski, Melanie Even more, the impact of nonmarket gover- Lührmann, Mattias Nilsson, Joachim nance devices (such as shareholder control) Winter, an anonymous referee, seminar on the probability of control change or participants at the University of Mannheim managerial turnover is quantitatively and the Centre for European Economic similar for listed and nonlisted firms.These Research in Mannheim, and participants of results suggest that a stock market listing the Financial Management Association does not provide for an extra monitoring meeting in Paris and the European Finance tool, and nonmarket governance devices Association meeting in Barcelona are need to be similarly strong for listed and highly appreciated. Matthias Braun, Gregor nonlisted firms to have a disciplinary Führich, and Heiko Truppel provided impact. We take this as additional evidence excellent research assistance. for the weakness of market-based governance devices in Germany. Thus, can we answer the question posed APPENDIX A in this paper: Does the market for corporate control fulfill a disciplinary function in The first main pillar of data comes from a bank-based economy? The answer seems Hoppenstedt’s Balance Sheet Database to be no. Nonetheless, our results should be (BSD). An important feature of this data interpreted with some caution. First, it is source is that it contains information on possible that the results of a disciplinary listed and nonlisted corporations, both change in control are not observable within public (AG) and private (GmbH). We take our 2-year time horizon; major perform- 1986 as the starting year because a change ance improvements may need several years in disclosure rules hinders comparability of to materialize. In Germany, it is difficult to the annual reports before and after the conduct extensive restructuring rapidly, for year 1986.11 The last year of the sample is example due to legal protection against 1994 because publication of our main dismissal. Hence, turning around an ailing source of ownership data has been discon- firm may indeed take longer than in tinued in this year. For the period market-based economies (see Denis and 1986–1994, BSD contains 5222 firms Kruse, 2000 for evidence from the US). (22, 732 firm years) for which consoli- Second, we do not know whether our dated balance sheet data are available results are unique to the period of (Table 27.10). We eliminate firms from the 1987–1994 examined in this study. It may utility, traffic, and telecommunications be that recent changes in capital market industries because they were still predomi- regulation and the emergence of the nantly government-owned during the period European Monetary Union have changed of observation. As a matter of comparabil- the nature of the German market for ity, we also eliminate firms that primarily corporate control. For example, increasing operate in the banking and insurance competition may have strengthened man- business, while operating little in nonfinancial agers’ focus on shareholder value and activities. Selection by industry causes therefore the disciplinary characteristic of 1752 firm deletions. control changes. Therefore, it would be The second main pillar—data on interesting to investigate the causes and ownership structures—is obtained from consequences of control changes during annual reports published by Bayerische other time periods and for other countries. Hypotheken-und Wechsel-Bank (in short, MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 729 Table 27.10 Data selection procedure

Selection criterion Firms Firm years

Consolidated balance sheet data for the years 1986–1994 5222 22,732 Mining, manufacturing, construction, and trade 3470 15,148 Ownership data 1485 6367 No missing values 1455 6205 Sample I (at least two continuous years of data) 946 4882 Sample II (at least four continuous years of data) 664 4433

Hypobank).These reports contain informa- nonlisted firms, mainly for private tion on direct ownership of common stock corporations (GmbH). Third, some firms (Stammaktien) for all listed German changed their names during the period of corporations. In addition, Hypobank observation, for example following takeovers provides information on direct ownership of or restructuring of conglomerates. Since common stock for large nonlisted corpora- changes in ownership are crucial to the data tions (Stammaktien for the nonlisted AG collection procedure, we adjust for name and Gesellschafteranteile for the GmbH changes and obtain a panel on ownership which cannot be listed). Hypobank reports structures through the years 1986–1994. the size and the name of a direct owner In consequence, we are left with owner- when the size of the ownership block ship data for 1485 firms (6367 firm exceeds 5%. In general, ownership rights years). Because of missing values for as reported by Hypobank correspond to important balance sheet items, another 30 voting rights (see Section 3.3). firms must be eliminated. This selection Ownership information from Hypobank procedure generates a sample of 1455 cannot readily be used in our analysis for firms (6205 firm years) with at least three reasons. First, ownership information 1 year of balance sheet and ownership from Hypobank only refers to the direct level data during the years 1986–1994. As this of owners. But this analysis requires to study examines changes in ownership, we identify the ultimate owner of each sample further eliminate 509 firms for which we firm. In contrast to Becht and Böhmer (in have only 1 year of data. For almost all of press), who rely on voting rights information these 509 firms, the year 1994 is the provided by the German Securities Exchange single year. This is due to the fact that Commission (Bundesaufsichtsamt für den Hoppenstedt substantially increased firm Wertpapierhandel), this study reconstructs coverage in 1994. The resulting sample voting rights information in a bottom-up that contains 946 firms (4882 firm years) approach from information on direct is labeled ‘Sample I’ in this study. For the ownership rights. Our concept of control, as analysis of causes and consequences of outlined in Section 3.3, represents the control changes (Section 5), we require methodological tool to achieve this aim. A firms to display at least four continuous second drawback is that Hypobank does not years of data. This generates ‘Sample II’ directly reveal ownership information on which contains 664 firms (4443 firm medium-sized nonlisted firms. Therefore, we years). Table 27.10 summarizes the construct the relevant ownership structures sample selection procedure. by searching the information on investments To test for a potential sample selection in subsidiaries and affiliated companies, bias, we collect data on the firms’ survival which is given in the appendix to each status. For firms leaving the sample before company in Hypobank. To further enlarge 1994, information is obtained from the our sample, we search the Mannheim BSD and MUP databases and from tele- Company Database (MUP) located at the phone interviews.We find that 39 out of 65 ZEW in Mannheim. We thereby obtain firms that exit the sample before 1994 still ownership structures on many medium-sized existed in 1994 without a change in ultimate 730 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 27.11 Definition of variables

Performance Return on assets (ROA) Earnings before interest, taxes, depreciation and amortization (EBITDA) divided by total assets Return on equity (ROE) EBITDA/total equity Change in ROA ROAt ROAt – ROAt-1 Change in ROE ROEt ROEt – ROEt-1 Earnings loss EBITDA 0 Ownership structure Largest block Size of largest share block n Ownership concentration Herfindahl index H P2, j1 j

with Pj proportion of a firms’ shares owned by shareholder j Insider ownership Fraction of shares owned by executive and nonexecutive directors or their families Cross ownership Cross 1 if ultimate owner is part of the web of firms identified by Wenger and Kaserer (1998) and if ultimate owner indirectly owns a share block in itself, 0 otherwise Pyramid Pyramid 1 if ultimate owner is located on the second or higher level in the ownership structure Capital structure Bank debt Bank debt to total debt Leverage Total debt to total assets Interest coverage EBITDA/interest payments; interest coverage is low if EBITDA/interest payments 1 Board structure Turnover of CEO CEOt 1 if chairman of management board (Vorstand) changes from year t – 1 to t, 0 otherwise. Turnover of executive directors (Jt Lt)/(Et Et–1) with Jt (Lt) number of joining (leaving) directors executive directors and Et number of executive directors Supervisory board Board 1 if firm has a supervisory board, 0 otherwise Restructuring Sales of assets sjt Sjt/Tjt with Sjt sales of asset class j in year t and Tjt stock of asset class j at the beginning of year t Growth rate of employment nt (Nt – Nt–1)/(Nt–1) with Nt total number of employees in year t Growth rate of labor costs ct (Ct – Ct–1)/(Ct–1) with Ct total labor expenditures in year t

ownership, but simply changed their name not generate a random sample of nonlisted or quit reporting due to reasons determined medium-sized firms. Given that we include within the firm. In 11 cases, operation was only medium-sized firms that are sub- shut down due to liquidation or bankruptcy. sidiaries of conglomerates, our sample In 14 cases, the respective firm had been might be biased. For example, we lack taken over by another entity. And in one medium-sized firms that are directly owned case, operation was shut down voluntarily. by families. But given the advantage of a Hence, 60% of firm exits during the sample smaller average firm size in the total sam- period are not related to bankruptcy or ple, we regard this as a necessary cost to be ownership change. incurred. We recognize that the selection proce- All variables used in this study are dure of nonlisted medium-sized firms does defined in Table 27.11. MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 731 NOTES 10 When we do not correct for industry performance, we also obtain an insignificant 1 The database includes all firms listed on coefficient for performance. The same holds any German stock exchange that do not belong when we use other measures of performance to the financial industry and industries under such as return on equity, return on sales, or strict regulation such as utility, traffic, and 1-year changes in each of these measures of telecommunications (see Appendix A). performance. 2 One major difference between private 11 In 1985, several changes were introduced (GmbH) and public corporations (AG) is that in in German corporate law (§289 HGB), most of private corporations, the general meeting of them triggered by the European Community’s shareholders can give instructions to Fourth Company Law Directive on the harmo- management, but not in public corporations. nization of national requirements pertaining to Therefore, management could in principle be financial statements. under tighter control in GmbH firms. This difference is unlikely to be of major relevance in our sample because ownership concentration is REFERENCES high for both types of firms. 3 Regulatory characteristics of listed and nonlisted corporations are also very similar.One Barber, B.M., Palmer, D., Wallace, J., 1995. difference is that disclosure rules are more Determinants of conglomerate and strictly enforced for listed corporations. predatory acquisitions: evidence from the However, after inspecting the annual reports, 1960s. Journal of Corporate Finance 1, we cannot find evidence that the nonlisted 283–318. sample firms do not meet the exchanges’ listing Baums, T., 1994. The German banking system requirements. and its impact on corporate finance and 4 Detailed results are available upon request. governance. In: Aoki, M., Patrick, H. (Eds.), 5 A 50% majority is sufficient to dismiss management after its regular period of office, The Japanese Main Bank. Oxford Univ. Press, and a majority of 75% to dismiss management Oxford, pp. 409–449. during its period of office (§103 (1) AktG). Bebchuk, L.A., Kraakman, R., Triantis, G., 6 Of course, this step is required only if there 2000. Stock pyramids, cross ownership, and is more than one direct shareholder. dual class equity. In: Morck, R.K. (Ed.), 7 Information on voting rights is provided by Concentrated Corporate Ownership. the German Securities Exchange Commission University of Chicago Press, Chicago, pp. (BAWe) and is used by Becht and Böhmer (in 295–315. press). Unfortunately, this information is avail- Becht, M., Böhmer, E., 2002. Voting control in able only from 1996 onwards, and the method- German corporations. International Review of ology applied by the BAWe to attribute voting rights in complex ownership structures has Law and Economics (in press). serious drawbacks, as shown by Becht and Bethel, J.E., Liebeskind, J.P., Opler, T., 1998. Böhmer. Block share purchases and corporate 8 Overall, 73.2% of sample firms do not performance. Journal of Finance 53 (2), experience any change in control. For about 605–634. 20% of firms, the ultimate owner changes once, Denis, D.J., Kruse,T.A., 2000. Managerial disci- for about 6% twice. Hence, several consecutive pline and corporate restructuring following changes in control are not very likely. performance declines. Journal of Financial 9 Originally, the web described by Wenger Economics 55, 391–424. and Kaserer (1998) consists of 39 firms. From Denis, D.J., Sarin, A., 1999. Ownership and these firms, we select those that own shares in each other because only those are truly ‘cross- board structures in publicly traded corpora- held’. We identify the following web members: tions. Journal of Financial Economics 52, (a) financial corporations: Allianz, 187–223. Commerzbank, Deutsche Bank, Dresdner Bank, Edwards, J., Nibler, M., 2000. Corporate gover- Münchner Rück, and the predecessors of nance: banks versus ownership concentration Hypovereinsbank; (b) nonfinancial corpora- in Germany. Economic Policy 15 (2), tions: Linde, Siemens, and VIAG. 237–267. 732 ON TAKEOVER AS DISCIPLINARY MECHANISM Fama, E.F., Jensen, M.C., 1983. Separation of Kaplan, S.N., 1995. Corporate governance and ownership and control. Journal of Law and incentives in German companies: evidence from Economics 26, 301–325. top executive turnover and firm performance. Franks, J., Mayer, C., 1995. Ownership and European Financial Management 1, 23–36. control. In: Siebert, H. (Ed.), Trends in Kini, O., Kracaw, W., Mian, S., 1995. Corporate Business Organization. Mohr, Tübingen, takeovers, firm performance, and board pp. 171–200. composition. Journal of Corporate Finance 1, Franks, J., Mayer, C., 1998. Bank control, 383–412. takeovers, and corporate governance in Köke, J., 2001. New evidence on ownership Germany. Journal of Banking and Finance structures in Germany. Kredit und Kapital 34 22, 1385–1403. (2), 257–292. Franks, J., Mayer, C., 2001. Ownership and La Porta, R., Lopez-de-Silanes, F., Shleifer, A., control of German corporations. Review of 1999. Corporate ownership around the world. Financial Studies 14, 943–977. Journal of Finance 54 (2), 471–517. Franks, J., Mayer,C., Renneboog, L., 2001.Who Maksimovic, V., Phillips, G., 2001. The market disciplines the management of poorly for corporate assets: who engages in mergers performing companies? Journal of Financial and asset sales and are there efficiency gains? Intermediation 10, 209–248. Journal of Finance 56 (6), 2019–2065. Gorton, G., Schmid, F.A.,2000. Universal bank- Mayer, C., 1988. New issues in corporate ing and the performance of German firms. finance. European Economic Review 32, Journal of Financial Economics 58, 29–80. 1167–1189. Holmström, B.,Tirole, J., 1993. Market liquidity Mulherin, J.H., Boone, A.L., 2000. Comparing and performance monitoring. Journal of acquisitions and divestitures. Journal of Political Economy 101 (4), 679–709. Corporate Finance 6, 117–139. Hopt, K.J., 1997.The German two-tier board: a OECD, 1995. Financial markets and corporate German view on corporate governance. In: governance. Financial Market Trends 62, Hopt, K.J., Wymeersch, E. (Eds.), 13–35. Comparative Corporate Governance: Essays OECD, 1998. Shareholder value and the market and Materials. Walter de Gruyter, Berlin, in corporate control in OECD countries. pp. 3–20. Financial Market Trends 69, 15–37. Jarrell, G.A., Brickley, J., Netter, J., 1988. The Renneboog, L., 2000. Ownership, managerial market for corporate control: the empirical control, and the governance of companies evidence since 1980. Journal of Economic listed on the Brussels stock exchange. Journal Perspectives 2, 49–68. of Banking and Finance 24 (12), Jenkinson, T., Ljungqvist, A., 2001. Hostile 1959–1995. stakes and the role of banks in German cor- Scharfstein, D., 1988. The disciplinary role of porate governance. Journal of Corporate takeovers. Review of Economic Studies 55, Finance 7 (4), 397–446. 185–199. Jensen, M.C., 1986. Agency costs of free cash Schwert, W.G., 2000. Hostility in takeovers: in flow, corporate finance, and takeovers. the eyes of the beholder? Journal of Finance American Economic Review 76, 323–329. 55 (6), 2599–2640. Jensen, M.C., 1988.Takeovers: their causes and Shleifer, A., Vishny, R.W., 1986. Large share- consequences. Journal of Economic holders and corporate control. Journal of Perspectives 2, 21–48. Political Economy 94 (3), 461–488. Jensen, M.C., Meckling, W.H., 1976. Theory of Shleifer, A., Vishny, R.W., 1992. Liquidation the firm: managerial behavior, agency costs, values and debt capacity: a market equilibrium and ownership structure. Journal of Financial approach. Journal of Finance 47 (4), Economics 3, 305–360. 1343–1366. Jensen, M.C., Ruback, R., 1983.The market for Shleifer, A., Vishny, R.W., 1997. A survey of corporate control. Journal of Financial corporate governance. Journal of Finance 52 Economics 11, 5–50. (2), 737–783. MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 733 Short, H., 1994. Ownership, control, financial model which should not be imitated. structure, and the performance of firms. In: Black, S.W., Moersch, M. (Eds.), Journal of Economic Surveys 8 (3), 203–247. Competition and Convergence in Wenger, E., Kaserer, C., 1998. The German Financial Markets. Elsevier, Amsterdam, system of corporate governance: a pp. 41–78. Chapter 28

Julian Franks* and Colin Mayer HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE

Source: Journal of Financial Economics, 40(1) (1996): 163–181.

ABSTRACT This paper examines the disciplining function of hostile takeovers in the U.K. in 1985 and 1986. We report evidence of high board turnover and significant levels of post-takeover restructuring. Large gains are anticipated in hostile bids as reflected in high bid premiums. However, there is little evidence of poor performance prior to bids, suggesting that the high board turnover does not derive from past managerial failure. Hostile takeovers do not therefore perform a disciplining function. Instead, rejection of bids appears to derive from opposition to post-takeover redeployment of assets and renegotiation over the terms of bids.

1. INTRODUCTION Shepro (1990, p. 3) believe that they are ‘a tremendously expensive and imprecise solu- HOSTILE TAKEOVERS ARE CONTROVERSIAL.To tion.’ Shleifer and Summers (1988, p. 37) some, takeovers are an important method take a wider social perspective and ‘see of correcting managerial failure. hostile takeovers as destructions of valu- Rappaport (1590), for example, believes able corporate cultures ...which have that the recent wave of takeover activity extremely serious allocative consequences’. ‘has changed the attitudes and practices of This paper addresses the disciplining U.S. managers’. He argues (p. 100) that ‘it function of takeovers, specifically examining represents the most effective check on whether hostile takeovers are associated management autonomy ever devised. And it with the dismissal of the management and is breathing new life into the public corpo- prior poor performance. Evidence on poor ration’. Grossman and Hart (1980, p. 329) performance of target firms comes from associate this discipline with hostile pre-bid measures of financial performance takeovers: ‘...since the threat of raids as well as bid premiums and post-takeover encourages good management and raids restructurings. We find clear evidence of only occur in events where the company high board turnover and significant levels is worth more to the raider than it is cur- of restructuring in hostile takeovers. Large rently worth, there is no reason on effi- gains are anticipated, as reflected in high ciency grounds for society to restrict raids’. bid premiums paid to target shareholders. To others, takeovers are a poor form of However, using a number of different corporate governance. Peter Drucker has benchmarks, we find little evidence that declared that ‘there can be absolutely no hostile takeovers are motivated by doubt that hostile takeovers are exceed- poor performance prior to bids. We ingly bad for the economy.’ Herzel and therefore reject the view that hostile HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 735 takeovers perform a disciplinary role. subsequently acquired by another bidder in Instead, we argue that opposition to bids by a later bid battle.There were a total of 325 incumbent management reflects disagree- bids for listed companies, including unsuc- ment over the price the bidder is willing to cessful bids. Of the 80 hostile bids, 35 were pay and its intentions to restructure the successful, 23 remained independent (to company. July 1990), and 22 were subsequently Section 2 describes the sample and the acquired. In five cases there were multiple methodology. Section 3 reports results on bids involving more thann one bidder. If hostile takeovers and managerial control two companies bid for one target and the changes. Section 4 provides evidence on the target management rejected one bid but bidding process in hostile bids. Section 5 accepted the other,then the former bid was concludes the paper. classified as ‘rejected’ and the latter as ‘accepted’. This occurred only once in our sample.The level of hostile takeover activity 2. DATA, HYPOTHESES, AND is similar to prior years: according to the METHODOLOGY annual report of the U.K. Takeover Panel, 23% of bids for publicly listed targets in The U.K. and U.S., unlike many other coun- 1969–1971 were classified as hostile. The tries, have active markets for corporate level of hostile takeover activity is also not control. In 1985 and 1986 there were 80 dissimilar to that recorded in the U.S. by hostile bids in the U.K. Jensen (1988) Bhagat, Shleifer, and Vishny (1990), who reports that there were 40 hostile tender find 63 successful hostile takeover bids and offers in the U.S. in 1986. In contrast, 22 unsuccessful contests during the three Germany has had just three cases of hostile years 1984–1986.1 acquisitions of nonfinancial corporations in The sample data is drawn from AMDATA, the whole of the post-Second World War a database of takeover bids, buyouts, and period. The absence of a market for corpo- divestments. The sample includes all 35 rate control in most of continental Europe successful hostile take overs, all 23 unsuc- is attributable to the structure of capital cessful hostile takeovers, a set of 35 markets, with small numbers of quoted randomly selected accepted bids from companies and ownership of quoted com- AMDATA, and a group of nonmerging panies concentrated in the hands of a small firms over the performance period 1980 to number of investors (see Franks and 1986.The nonmerging sample was matched Mayer, 1994). by industry (London Stock Exchange The U.K. is thus one of the few countries Industry Classification) and size (equity other than the U.S. whose market for cor- market capitalization in 1990) with the porate control can be studied. In fact, the targets of hostile bids (and of uncontested U.K. has certain advantages over the U.S. bids) using the London Share Price Data in that there are fewer antitakeover provi- Base (LSPD). sions enshrined in either corporate charters The unsuccessful hostile takeovers, the or state legislation. For example, the randomly selected accepted bids, and the Takeover Code in the U.K. explicitly pre- nonmerging firms provide benchmarks of vents the application of poison pills once a comparative performance for the sample of takeover bid has been launched. successful hostile takeovers. We compare This paper examines hostile takeovers in the performance of firms involved in suc- the U.K. in 1985 and 1986. A hostile cessful hostile bids with the performance of takeover is defined as one in which the first firms involved in other takeovers (accepted offer is opposed by the incumbent manage- bids). We also compare targets of hostile ment. A bid is successful if the bidder takeovers and nonmerging firms to see how acquires the target on the basis of its first firms that become hostile takeover targets or a revised offer. A bid is unsuccessful if perform relative to other firms of similar the target firm remains independent or is size in the same industry. Finally, we compare 736 ON TAKEOVER AS DISCIPLINARY MECHANISM successful hostile takeovers and failed basis. In contrast, the post-bid asset sales hostile takeovers. Failed takeovers provide of targets of unsuccessful bids are readily evidence of target management’s response observable. Consequently, we also analyze to the threat of a hostile bid and can be the asset sales of targets of unsuccessful bids compared with the bidder’s response in a over a period of five years around the bid. successful takeover. 2.2. Anticipated bid benefits and 2.1. Board turnover and prebid performance measures post-takeover restructuring If hostile bids serve a disciplinary function If hostile takeovers perform a disciplinary there must also be evidence of anticipated role, we should see a high level of manage- gains associated with the restructurings rial turnover followed by large-scale and poor performance prior to the takeover restructuring. These actions are necessary Anticipated merger benefits are measured but not sufficient conditions for takeovers using bid premiums based on abnormal to be disciplinary because they could reflect share price returns in the month of the first disagreement over a strategic redeploy- bid. If bids are revised. we examine the ment of assets. To evaluate management month of the first bid to the month of the turnover,we collect data on all members of last bid. the board of directors of the target firm, We use four measures of pre-bid finan- including the name of the director, the date cial performance. The first is share price of resignation or retirement, and the even- returns over five years prior to the month of tual position of the director in the merged the announcement of the initial bid.The bid company. announcement date is from Financial Times Our information comes from company Mergers & Acquisitions (FTM&A). If bids accounts of the target and bidder firms and are revised, the date of the first bid is from company announcements issued by obtained from FTM&A, London Stock the Stock Exchange. The number of target Exchange Microfiches, and Textline.Textline company board members promoted to provides extracts from newspapers such as the acquiring company’s board comes from the Financial Times. Returns are based accounts of the target company and the upon share price and dividend data collected acquiring firm. Telephone interviews with from LSPD and Datastream. Abnormal company representatives corroborated and returns are calculated using a market supplemented the data. The directors’ his- model ( 0, 1).The second measure tories were followed for two years after the of financial performance is based on takeover. changes in dividends per share and dividend We use two measures of post-takeover omissions in each of the two years prior to restructuring.The first measure is significant the financial year in which the takeover sales of assets during the two years after occurred. The third measure of pre-bid takeover. Asset sales for successful hostile financial performance is the cash flow rate and friendly bids are classified as signifi- of return on assets employed in 1980 and cant if more than 10% of the total fixed in the last accounting year before the bid, assets of the merged firm are sold in the calculated from the cash flow gross of corresponding year. Data on asset sales are interest and depreciation divided by the based on annual accounts and telephone sum of the market value of equity and the interviews with company secretaries to the book value of debt.The last measure is based board of directors or equivalent persons. on Tobin’s Q ratio, the ratio of the market Post-bid studies of successful takeovers value to replacement cost of a firm’s suffer from the difficulty of separately assets.The ratio is calculated for 1980 (or identifying asset sales in the target firms the first year of the listing) and the last from those of the bidder,since the financial accounting year before the bid.The market accounts are reported on a consolidated value of equity and book value of debt proxy HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 737 for the numerator.The replacement cost of significant at the 1% level for both total each company’s assets was taken from its and inside board resignations. annual accounts. Until 1985, companies in The levels of inside director resignations the U.K. were required to report the cur- in both successful and unsuccessful hostile rent cost of their assets. Current cost is the bids are higher than in nonmerging firms. replacement cost based on an up-to-date For example, in a sample of ten nonmerg- cost of the assets, after providing for depre- ing firms drawn from the lowest decile of ciation to reflect the age and usage of the share price performance in 1985, only assets. Replacement costs are based on an 19% of the board resigned during the year index of prices of new assets. of bad performance and the subsequent year, compared with 90% in successful hostile bids and 39% in unsuccessful bids. 3. TEST RESULTS Clearly, hostile takeovers are associated with a high level of board turnover. These Sections 3.1 and 3.2 examine the extent to results are consistent with Klein and which hostile takeovers give rise to man- Rosenfeld (1988) who find that firms suc- agement board turnover and post-takeover cessfully defending themselves through restructuring. Sections 3.3 and 3.4 evalu- greenmail experience above-average man- ate the extent to which hostility and mana- agement turnover.The high turnover of the gerial control changes are associated executive members of the boards in unsuc- with higher bid premiums and poor pre-bid cessful hostile bids is also consistent with performance. Hirshleifer and Thakor’s (1994) prediction that hostile bids reveal information about the quality of management to the monitors of the firm. 3.1. Board turnover

Table 28.1 reports that 50% of the direc- 3.2. Post-takeover asset sales tors resign after accepted bids, compared with 90% after successful hostile bids. If Hostile takeovers are also associated with outside directors are excluded, the differ- restructurings. Based on accounts of merged ence barely changes. According to panel 2, companies and telephone interviews, asset the differences in proportions between sales exceed 10% of the value of total successful hostile and accepted bids are fixed assets of the merged firms in 53% of

Table 28.1 Proportion of directors who resigned in the targets of accepted and successful and unsuccessful hostile bids (size of sample, n. in parentheses)

Panel 1. Resignations Accepted Successful hostile Unsuccessful hostile

All directors 50% (34) 90% (31) 39% (23) Inside directors 50% (26) 88% (26) 36% (22)

Panel 2: t-statistic: difference in proportions All directors Inside directors

Successful hostile vs. accepted 3.52a 3.00a Unsuccessful hostile vs. successful hostile 4.01a 3.76a

Source: Annual reports. Stock Exchange microfiches. Part of the source data for 1986 is contained in appendices A and B of Franks and Mayer (1990). It includes resignations in the two years subsequent to takeover. a Significant at the 1% level. 738 ON TAKEOVER AS DISCIPLINARY MECHANISM successful hostile bids. In contrast, only create greater shareholder, value than in 26% of accepted bids had significant asset other takeovers. Furthermore, if takeovers sales, a difference that is statistically sig- are disciplinary, then takeovers which give nificant at greater than the 5% level. The rise to managerial control changes should high level of asset disposals is consistent have higher bid premiums than those which with the results of Dann and DeAngelo do not. Table 28.3 finds that anticipated (1988, p. 88), who find that ‘the median benefits are higher in successful hostile bids restructuring involves assets constituting than in other bids: bid premiums average 23% of the target’s pre-offer equity value’. 30% in the month of a bid compared with It is also consistent with Bhide (1989) who only 18% for accepted bids. [By way of finds more evidence of sales of assets comparison, Franks and Harris (1989) arising from past acquisitions of targets in report premiums of 42% for hostile bids hostile than in friendly bids. and 28% for uncontested and unrevised Table 28.2 provides a detailed examina- bids in the U.S.] The difference of 12 tion of the proportion of asset disposals for percentage points is statistically significant unsuccessful hostile bids. The table sug- at the 5% level. When bid premiums are gests that prior to the bid and in the year of computed for a period which includes any the bid, annual sales of assets total 8–9% bid revisions, the difference is 15 percentage of total fixed assets. Post-bid asset sales points and is significant at the 1% level. are larger, with the proportion increasing There is therefore some support for the from 13% in the year after the acquisition proposition that bid premiums are higher in to 19% three years after takeover. In the hostile bids. Bradley, Desai, and Kim third year after the acquisition, asset sales (1988) suggest that competition in the are significantly greater than 10% (at the bidding market may produce benefits to the 5% significance level), with 60% of com- target shareholders at the expense of the panies having sold more than 10% of their shareholders of the bidder. assets. This finding is consistent with the Unsuccessful hostile bids have bid pre- observation of Dann and DeAngelo (1988, miums that are considerably lower than p. 88) that ‘hostile bids rarely prevail in those in successful hostile bids. In fact, they cases in which the planned restructurings are much closer to the bid premiums in were implemented by target managers’. accepted bids, suggesting that the market is anticipating some restructuring after 3.3. Anticipated benefits and bid unsuccessful bids but not at as high a level premiums as in successful hostile bids. We also examine whether the high level It might be expected that the higher levels of board turnover in hostile bids (reported of restructurings in hostile takeovers would earlier) is associated with high bid premiums.

Table 28.2 Asset disposals for a five-year period around the bid for unsuccessful bids occurring in 1985 and 1986 (size of sample. t-statistic, and proportion 10% in parentheses) B 1a BB 1B 2B 3 Averageb Asset 8.5% 9.2% 13.3% 17.2% 19.4% 11.0% disposals (32, 1.0,0) (24, 0.4,25) (19,1.0,42) (17,1.6,53) (15,2.2.60) (33,0.7.42) as a % of tangible fixed assets Source: Annual reports. t-statistic indicates whether the figure is different from 10%. a B is the year of the bid. b Firms for which data do not exist for some of the five years have been included for the calculation of the average. HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 739 Table 28.3 Bid premiums for three samples: targets of accepted bids, successful hostile bids, and unsuccessful hostile bids (sample size, t-statistic, and proportion positive in parentheses)

Panel 1. Eid premiums Accepted Successful hostile Unsuccessful hostile

Bid month 18.44% 29.76% 21.50% (34,6.09,88) (32, 8.30, 100) (43, 6.27, 87) Initial bid to final bid month 18.44% 33.88% 22.40% (34, 6.09, 88) (33, 7.39,97) (46, 5.98, 89)

Panel 2: t-statistic: difference in bid premiums Initial bid to final Bid month bid month

Successful hostile vs. accepted 2.29a 2.80b Unsuccessful hostile vs. successful hostile 1.41 1.94c

Source: LSPD. Bid premiums in panel 1 are calculated using a market model with parameter values of 1 and 0. a Significant at the 5% level. b Significant at the 1% level. c Significant at the 10% level.

We define a change in managerial control level of managerial control changes. This as occuring when either the chief executive evidence is not consistent with a disciplinary or the chairman of the target firm resigns role for hostile takeovers. Section 4 pro- or when there are no promotions from the vides an alternative explanation for the high target to the main board of the acquiring bid premiums in hostile takeovers based on firm. This happens in 88% of successful bargaining over the terms of the offer. hostile bids and 60% of accepted bids.The difference is statistically significant at the 2% level.We also repeat the analysis using 3.4. Pre-bid performance an alternative definition: a control change Finally, if hostile bids are disciplinary, we occurs when either the chief executive or would expect high board turnover to be the chairman resigns and there are no pro- associated with poor performance of the motions to the main board of the acquiring target prior to the bid. We use four meas- firm. According to this definition, there are ures of performance: share prices, dividend control changes in 24% of accepted bids changes, cash flow rates of return, and and 66% of successful hostile bids. Since Tobin’s Q. the subsequent results are similar we report the evidence from the first test only. Bid premiums for targets with control 3.4.1. Pre-bid share price performance changes are almost identical to those without control changes: 25.23% versus The first panel of Table 28.4 reports abnor- 25.25%. This result is similar to Martin mal share price performance for successful and McConnell (1991) who report abnor- hostile bids, unsuccessful hostile bids, and mal returns of 31.33% for the sample in accepted bids, as well as for two samples of which there is turnover of the top manager nonmerging firms paired by industry and and 33.77% when there is no change; they size with the samples of accepted bids and find that the difference is not statistically successful hostile bids. Abnormal returns significant. Thus, while hostile takeovers are shown for five years, two years, and one give rise to higher bid premiums, the higher year prior to the date when the market first premiums are not attributable to the high began to anticipate the bid which is 740 ON TAKEOVER AS DISCIPLINARY MECHANISM Table 28.4 Abnormal share price returns for the period up to five years before the bid for various samples of merging and nonmerging firms (size of sample, t-statistic, and proportion positive in parentheses)

Panel 1. Abnormal returns Matched Matched Unsuccessful Accepted nonmerging Successful nonmerging hostile bids samplea hostile bids samplea bids

Average 4.53% 0.83%b 0.14% 0.14% 2.71% over (28, 0.99, 64) (19, 0.23, 53) (32, 0.04, 50) (26, 0.03, 38) (34, 0.72, 50) 5 yearsb 2 years 7.75% 7.84% 6.09% 2.26% 5.07% prior (32, 1.10, 53) (18, 0.67, 50) (34, 0.84, 38) (25, 0.84, 38) (44, 0.78, 41) 1 year prior 7.77% 1.99% 7.68% 7.90% 2.84% (35, 1.11, 51) (19, 0.22, 58) (34, 1.17, 41) (26, 1.14, 54) (44, 0.57, 48)

Panel 2. t-statistic: difference in abnormal returns Average over 5 years 2 years prior 1 year prior

Successful hostile less accepted 0.79 1.37 1.61 Accepted less nonmerging 0.46 1.14 0.51 Successful hostile less nonmerging 0.05 0.83 1.63 Successful hostile less unsuccessful hostile 0.53 1.15 0.60

Source: LSPD and own calculations. Abnormal returns in panel 1 are calculated using a market model with parameter values of 0 and 1. a Based upon the same industry using the Stock Exchange Industrial Classification and the company with the closest market equity capitalisation. b Some companies had incomplete data for the five years because they were not listed at the beginning of the period. c One company has been excluded because it was an outlier showing an abnormal return of 161% over five years.

assumed to be three months prior to described as one and two years prior to the the announcement date as reported by the bid, the difference is around 14 percentage London Stock Exchange. Results are points. Although economically meaningful, reported on an equally weighted basis. The panel 2 records that this difference is not t-statistics in the second panel test the significant at the 5% level. The difference hypothesis that the average abnormal in returns is reduced to less than 5 percentage returns in the two relevant samples are points when abnormal returns are esti- equal. The t-statistics are calculated as mated over the five years prior to the bid. t (M1 M2)/SD, where M stands for When matching the accepted and hostile 2 2 the cell mean SD =/(S1 /N1 + S2/N2, with bids with nonmerging firms of equivalent N the number of observations averaged in a size and industry, nonmerging firms record cell mean and S the cross-sectional deviation almost identical performance to successful about the cell mean. hostile bids over the five years prior to a A comparison of accepted bids with bid and somewhat worse (but not statisti- successful hostile bids in panel 1 suggests cally significantly so) performance than that the pre-bid performance of targets of accepted bids. Abnormal returns are higher successful hostile bids is worse than that of for unsuccessful bids than for successful accepted bids: in each of the periods bids, although the difference over five years HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 741 is small and none of the differences are prior to a bid. The dividend behavior of statistically significant. An analysis of medi- targets of unsuccessful hostile bids is very ans discloses similar differences between similar to targets of successful bids. accepted and successful hostile bids on an Differences in dividend policy may equally weighted basis two years prior to a reflect changes in payout policy rather than bid and a smaller difference one year differences in underlying performance. before a bid. Using the Wilcoxon test, the However, the payout ratios of the three differences in median are significant at samples calculated for the six-year period the 13% level two years before a bid and 1980–85 (or 1981–86 for 1986 bids) are at the 14% level one year before a bid. similar: 38.3% for accepted bids, 40.4% for successful hostile bids, and 43.3% for 3.4.2. Dividend changes unsuccessful hostile bids.The differences in the payout ratios between the three sam- The second measure of financial perform- ples are not statistically significant and the ance reports firms’ dividend decisions in t-statistics are all less than one.The results each of the two years prior to a bid using suggest that there is little difference in per- four classifications: omission, increase, formance between the samples. decrease, or no change. Poor performance This result stands in marked contrast to should be associated with reductions and the dividend behavior of a sample of 80 omissions in dividends. Panel 1 of Table 28.5 firms in the lowest quintile of shareprice shows that in the two years prior to a bid a performance reported by LSPD in 1985. majority of companies in all three samples The sample had an average abnormal of bids (successful hostile, unsuccessful return of 63%. Dividends were omitted hostile, and accepted bids) increased their or reduced in 41% of cases compared with dividends; very few reduced them. The divi- 12% for successful hostile bids. In the dend was omitted entirely in only six cases. remaining cases, dividends were held con- A similar pattern emerges in the second stant (29%) or increased (30%) For 80 panel for dividends declared over the one firms with zero abnormal returns, dividends

Table 28.5 Proportion of companies omitting or changing dividends in accepted and hostile bids (size of sample, n, in parentheses)

Panel 1. 2 years prior Successful Unsuccessful Accepted hostile hostile

Omitted 5.6% (3) 5.9% (2) 2.8% (1) Decrease 14.3% (5) 2.9% (1) 5.7% (2) No change 22.9% (8) 14.7% (5) 17.1% (6) Increase 51.4% (18) 73.5% (25) 71.4% (25) N/A 2.9% (1) 2.9% (1) 2.8% (1)

Panel 2. 1 year prior Successful Unsuccessful Accepted hostile hostile

Omitted 2.9% (1) 5.9% (2) 2.9% (1) Decrease 8.6% (3) 5.9% (2) 5.7% (2) No change 11.4% (4) 11.8% (4) 11.4% (4) Increase 77.1% (27) 76.5% (26) 74.3% (26) N/A 0.0% (0) 0.0% (0) 5.7% (2)

Source: Annual reports. 742 ON TAKEOVER AS DISCIPLINARY MECHANISM were omitted and reduced in 9% of cases successful hostile bids are not statistically and increased in 77% of the cases. These different from those of unsuccessful bids. results are remarkably similar to the dividend performance of the sample of hostile takeover targets reported in Table 28.5. 3.4.4. Tobin’s Q This evidence suggests that when there is The failure to uncover differences in poor performance as measured by share pre-bid rates of return between targets of prices, dividends are more likely to be omit- hostile bids and other firms may be due to ted or reduced.The results strongly suggest the stock market capitalizing poor per- that the sample of targets of hostile bids formance at the start of the period. If this with its very low level of dividend reduc- is the case, then there will be differences in tions or omissions is not representative of the Tobin’s Q ratios, the ratio of market to firms with poor share price performance. replacement cost valuations of firms. Table 28.7 shows that the Q ratios of tar- 3.4.3. Pre-bid cash flow rates of return gets of successful and unsuccessful hostile bids in 1980 are similar to those of Table 28.6 reports cash flow rates of return accepted bids.Thus the difference in abnor- on capital for 1980 and 1985/86. Rates of mal returns between hostile and accepted return for targets for successful hostile bids bids revealed in Table 28.4 is not a conse- (17.82%) are similar to those of accepted quence of differences in initial market val- bids (20.25%) as well as to their matched ues. However,the nonmerging sample has a sample of nonmerging companies (17.15%). significantly higher Tobin’s Q than the suc- None of the differences is statistically cessful hostile bids. There is, therefore, significant. Rates of return for targets for some evidence that poor performance is

Table 28.6 Cash flow to asset ratios in 1980 and 1985/1986 for various samples of merging and nonmerging firms (size of sample, t-statistic, and proportion positive in parentheses)

Panel 1. Cash flow/(market value of equity book value of debt) Matched Matched nonmerging Successful nonmerging Unsuccessful Accepted bids samplea hostile bids samplea hostile bids

1980 28.20% 21.74% 22.28% 18.24% 24.49%b (23, 6.04, 96) (18, 8.81, 100) (27, 6.69, 100) (23, 6.42, 100) (26, 8.66, 100) 1985, 20.25% 21.61% 17.82% 17.15% 17.50% 1986 (30, 5.73, 97) (18, 5.27, 94) (29, 8.72, 100) (24, 10.3, 100) (30, 13.67, 100)

Panel 2. t-statistic: difference in cash flow ratio of two samples 1980 1985, 1986

Successful hostile less accepted 1.03 0.60 Accepted less nonmerging 1.28 0.22 Successful hostile less nonmerging 0.92 0.25 Successful hostile less unsuccessful hostile 0.51 0.13

Source: Datastream. LSPD, own calculations. Definitions: Cash flow Profit after tax Depreciation Interest payments, Book value of debt ST loans LT loans Cash balances Deferred tax. a Based upon the same industry, using the Stock Exchange Industrial Classification and the company with the closest market equity capitalization. b One company has been excluded from the sample because it showed a cash flow ratio of 306.64%. HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 743 Table 28.7 Tobin’s Q for 1980 (or first year of listing) and 1985/1986 (or last year of listing/annual report) for various samples of merging and nonmerging firms (size of sample, t-statistic, and proportion with Tobin’s Q 1 in parentheses)

Panel 1 Tobin’s Q Matched Successful Matched Unsuccessful Accepted nonmerging hostile nonmerging hostile bids samplea bids samplea bids

1980 0.72% 0.92% 0.82% 1.12% 0.80 (23, 3.8, 26) (17, 1.0, 47) (27, 2.7, 19) (24, 1.0, 38) (24, 3.1, 25) 1985, 1986 1.35 1.25 1.04 1.11 1.22 (29, 3.7, 76) (18, 2.1, 73) (29, 1.0, 55) (25, 2.0, 60) (27, 3.0, 74)

Panel 2. t-statistic: difference in Tobin’s Q of two samples 1980 1985, 1986

Successful hostile less accepted 0.92 2.95c Accepted less nonmerging 1.73b 0.67 Successful hostile less nonmerging 2.12d 1.06 Successful hostile less unsuccessful hostile 0.21 2.20d

Source: Datastream, LSPD, own calculations. t-statistic indicates whether Tobin’s Q is different from one. Definition:Tobin’s Q (Market value of equity Book value of debt)/Replacement cost of assets. a Based upon the same industry using the Stock Exchange Industrial Classification and the company with the closest market equity capitalization. b Significant at the 10% level. c Significant at the 1% level. d Significant at the 5% level. being capitalized in the targets of hostile of targets of hostile bids is not easily dis- bids when the comparison is made with tinguished from that of targets of accepted firms in the same industry. bids. The performance of targets of hostile For 1985 and 1986 the Tobin’s Q ratios bids is not inferior to that of a nonmerging of targets of hostile bids are lower than sample of firms matched on the basis of those of accepted bids in the year of the size and industry; the only exception is for merger (1.04 versus 1.35); the t-statistic Q ratios. This evidence is consistent with for the difference is 2.95. In addition, Martin and McConnell (1991) who find no the Q ratio of successful hostile bids is difference in pre-bid performance of lower than that for the targets of unsuc- targets of accepted and rejected bids. cessful hostile bids (1.04 versus 1.22, These findings raise the question of t-statistic 2.2). Thus, the performance whether poor pre-bid performance is more of targets of hostile takeover bids is signif- closely associated with managerial control icantly worse than that for the other two changes than with the hostility of bids. We samples. However, the average Q ratios of examine this proposition by repeating the the targets of accepted and hostile bids are above pre-performance tests on the sample both greater than one, providing evidence of of all takeovers partitioned by control poor relative, rather than poor absolute, changes. A very similar picture to that performance. described for the hestile and accepted partition emerges. (Tables relating to this 3.5. Summary of findings section are available from the authors on request.) To summarize, three different measures indi- Over five and two years prior to the bid, cate that the pre-bid financial performance takeovers with control changes display 744 ON TAKEOVER AS DISCIPLINARY MECHANISM superior abnormal share price performance Martin and McConnell’s (1991) finding of to those without control changes, the oppo- worse pre-bid performance in takeovers site of the result predicted by evidence of where there are high executive changes managerial failure. Only in the year prior to stands in contrast to our results for the U.K. the bid is there worse performance although the difference is not statistically significant. [Kini, Kracaw, and Mian (1993) record an inverse relation between performance and 4. REVISED BIDS CEO turnover for insider-dominated boards and no relation for outsider-dominated The previous section reports a high level of boards.] The proportion of firms cutting or board turnover in hostile take-overs, associ- omitting dividends is very small in relation ated with high levels of sales of assets and to the proportion raising dividends for both high bid premiums. However, there is no samples. There is slightly more evidence evidence of high bid premiums when that firms omit dividends prior to control takeovers are partitioned on the basis of change bids but no evidence that they cut control changes. In addition, neither hostile them. Cash flow rates of return are lower in take-overs nor control changes are associ- the control change sample both five years ated with poor pre-bid performance. The before the bid and in the year of the bid. evidence therefore strongly suggests that However, the differences are not statisti- hostile takeovers are not associated with a cally significant. Finally, five years prior to disciplinary role in this sample of U.K. the bid the Tobin’s Q ratio of control takeovers. change targets is slightly higher (0.79) The evidence of a high level of post- than that of noncontrol change targets takeover restructuring in conjunction with (0.70), although in the year of the bid it is high board turnover is more consistent with lower (1.17 versus 1.32); the differences board dismissal being a result of disagree- are not statistically significant. In both ment with the bidder over the future cases, the Q ratios are greater than one and restructuring of the company. Another therefore suggest profitable investment explanation, consistent with the high bid opportunities rather than poor performance. premiums in hostile bids, is that rejection In summary, there is little evidence that of bids is part of the process of negotiating takeovers with control changes are the the terms of bids. To evaluate this possibil- result of poor past performance. Several ity, we collected data on bid revisions for studies examine the relation between board our sample of successful and unsuccessful turnover and performance.Weisbach (1988) bids. In Table 28.8, we show that hostile finds a strong association between prior bids were revised in 25 out of the 35 suc- performance and the probability of director cessful hostile bids.When the hostile bidder resignations for companies with outsider- did not revise the terms after rejection by dominated boards. Warmer, Watts, and the target, there was a high rate of bid Wruck (1988) only find an inverse relation failure (20 out of 30 cases). between the probability of a management If rejection of bids were a matter of change and a firm’s share price performance negotiation over price. we would expect to when the latter is extremely good or bad. observe bids being rejected, revised, and Gilson (1989) finds a significantly higher then accepted (if the bid succeeded); alter- level of executive turnover in financially natively, the bid would fail because other distressed than nondistressed firms. bidders would emerge in the same or in a Whereas the last two studies find a subsequent bid. The 13 successful hostile relation between very poorly performing bids that were finally accepted (and there- companies and high board turnover. we fore recommended) by target management show that the targets of hostile take-overs can be interpreted in this way. However, we are performing neither poorly nor worse also observe 12 successful hostile bids than the targets of accepted bids. However, being rejected to the end, even after the HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 745 Table 28.8 Number of bids subject to revision partitioned by hostility of bid

Revised Unrevised

Accepted 0 35 Successful hostile: finally accepted 13b 0 Successful hostile: contested to the end 12c 10 Unsuccessful hostile bidsa 15d 20

Source: Financial Times Mergers & Acquisitions, Extel and Textline. a Some data on revisions of unsuccessful bids missing. b Average length of the window: approximately 45 days. c Average length of the window: approximately 75 days. In three cases, the length was greater than 100 days. One of them was referred to the Monopolies and Mergers Commission, another was referred to the Takeover Panel because of alleged irregularities with the offer, and the other was subject to multiple bidders. d Average length of the window: approximately 78 days. In two of them the length was greater than 100 days. One of them was referred to the Monopolies and Mergers Commission and the other was subject to multiple bidders. bids were revised. For this sample, rejection return are higher in ultimately accepted may reflect the breakdown of the bargain- than in rejected bids in 1980 but lower in ing process or outright hostility towards 1985/86. the particular bidder. The latter is consis- In sum, there is no evidence of poor pre- tent with a disciplinary role for takeovers bid performance in ultimately rejected bids. or disagreement over the redeployment of Continuing opposition to the bid appears to assets after the takeovers. reflect a breakdown over negotiations or A further test of the disciplining function disagreement over the redeployment of involves examining whether a sample of assets after the takeover.The fact that post- hostile bids that were ultimately rejected takeover restructuring is significantly had worse performance than a sample of higher in hostile than accepted takeovers revised bids that were ultimately accepted. points to the latter as being an important Table 28.9 reports three measures of per- consideration. formance Tobin’s Q, abnormal share price returns, and cash flow rates of return. If initial rejection but ultimate acceptance of 5. CONCLUSION bids has to do with bargaining over price, then we would predict (i) higher Q ratios This paper examines whether hostile for bids that are ultimately accepted (S.2 takeovers act as a discipline on manage- in table 28.9), (ii) larger abnormal returns ment. Evidence comes from the perform- and (iii) higher cash flow rates of return. ance of targets of hostile takeovers before, The sample sizes are small, between 10 during, and after acquisition. We report and 15 for each partition of hostile bids. that hostile takeovers are associated with Table 28.9 reports that the pattern of high levels of boardroom changes, higher in results is not consistent for any of the three hostile than accepted bids and higher in measures, except that the Q ratios for con- targets of successful than unsuccessful tested bids that are ultimately accepted are hostile bids. There are also larger asset greater (0.87) than for the sample of bids disposals and bid premiums in hostile that are ultimately rejected (0.73) in takeovers. However, targets of hostile 1980. In 1985/86, the Q ratios are almost takeovers do not perform worse prior to identical. Abnormal returns are smaller for bids than either targets of other acquisi- ultimately accepted bids over five years, tions or nonmerging firms. By only one though the difference in averages is not measure (Tobin’s Q) is there any evidence statistically significant. Cash flow rates of of statistically significant worse performance 746 ON TAKEOVER AS DISCIPLINARY MECHANISM cumbent of sample, Interest payments.Interest Book Book value of debt). Book value 0.91 1.59 1. Depreciation 0 in parentheses) 0 in 0 and 0.87, 40) (10, 3.06, 100) (11, 7.82. 100) Profit after tax after Profit 1 year 1980 1985/86 8.33% 27.12% 13.92% 1 year 1980 1985/86 Cash flow (Market value of equity value (Market Cash flow 0.58, 46) (12, 0.09, 31) (10, 7.07, 100) (11, 4.79, 100) 0.28, 47) (13, 2 years 3.00% 5.31% 0.85% 18.72% 21.60% 2 years 0.14 0.69 0.28,47) (12, 1.40, 47) (13, 5 years 4.20% 5 years 1 or abnormal returns and cash flow rates of return rates and cash flow returns 1 or abnormal Deferred tax.Deferred of return rates Cash flow Q Book value of debt).Book value Replacement cost of assets. Cash flow 0.64 0.98 Cash balances 3.1, 10) (11,0.2,45) (12, 1.0, 30) (11, 1.2, 64) (13, LT loans LT (Market value of equiry value (Market 0.87 , rejected by the in are ultimately which returns, bids abnormal successful hostile (i) revised for of return rates and cash flow Q (10, Tobin’s Q1980 1985/86 (10, Abnormal returnsTobin’s Q1980 Cash flow rates of return 1985/86 Abnormal returns Cash flow rates of return Q ST loans ST -statistic,Tobin’s with and proportion t Tobin’s management (S.1), management (S.2) (size accepted by the incumbent are ultimately which bids successful hostile and (ii) revised :Tobin’s : Datastream, LSPD, calculations. own values with parameter model are based on a market returns Abnormal accepted bids 0.87% 1.08% rejected bids 0.73% 1.02% 0.50% S.2: Ultimately Panel 2 t-statistic value of debt value S.1: Ultimately Panel 1. Tobin’s Q, abnormal returns, and cash flow rates of return Table 28.9 S.1 less S.2 Source Definitions HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 747 of targets of hostile bids. There is no evi- The definition of hostility is similar to that in dence of either high bid premiums or poor this paper,where the target management initially pre-bid performance when takeovers expressed opposition to the bid. involve managerial control changes. The market for corporate control does not REFERENCES therefore function as a disciplinary device for poorly performing companies. Bhagat, Sanjay, Andrei Shleifer, and Robert W. Rejection of bids is more consistent with Vishny, 1990. Hostile takeovers in the 80s: opposition to the anticipated redeployment The return to corporate specialization, of assets by the bidder and negotiation over Brookings Papers on Economic Activity, the terms of bids. We find that there is a 1–72. high level of bid revisions and large bid pre- Bhide, Amar, 1989, The causes and conse- miums in hostile takeovers out no evidence quences of hostile takeovers, Journal of of poor performance where negotiations Applied Corporate Finance 2, 36–59. over the terms of the bid break down.Thus, Bradley, Michael, Anand Desai, and E. Han Kim, our results provide a picture of a market 1988, Synergistic gains from corporate for corporate control in the U.K. that is acquisitions and their division between the more consistent with dissatisfaction over stockholders of target and acquiring firms, the bid terms and the redeployment of Journal of Financial Economics 21, assets after the takeover than with a disci- 183–206. plinary role. Dann, Larry and Harry DeAngelo, 1988, Corporate financial policy and corporate control: A study of defensive adjustments in asset NOTES and ownership structure, Journal of Financial Economics 20, 87–127. * Corresponding author. Franks, Julian R. and Robert Harris, 1989. This paper is part of an ESRC funded project, Shareholder wealth effects of corporate number W102251003, on ‘Capital Markets, Corporate Governance, and the Market for takeovers: The UK experience 1955–1985, Corporate Control’. We are grateful to Nick Journal of Financial Economics 23, Carrick, Luis Corrcia da Silva, and Marc 225–249. Goergen for valuable research assistance. We Franks, Julian R. and Colin P. Mayer, 1990, wish to thank John Chown for financial assis- Capital markets and corporate control: A tance on this project and for helpful discussions. study of France, Germany and the UK, The paper has been presented at the 1993 Economic Policy 10, 191–230. Annual Meeting of the American Finance Franks, Julian P. and Colin P. Mayer, Association in Anaheim, at the Annual Meeting 1994, Ownership and control in Germany, of the French Finance Association in Paris, at Working paper (London Business School, the Annual Meeting of the European Finance Association in Athens, at a conference on London). European Restructuring at INSEAD and at Gilson, Stuart C., 1989, Management turnover seminars in Essex, Keele, Oxford, Paris, and and financial distress. Journal of Financial Warwick.We are particularly grateful to discus- Economics 25, 241–267. sants Patrick Bolton, Stuart Gilson, Kristian Grossman, Sanford J. and Oliver D. Hart, 1980, Rydgvist, and Paul Seabright for valuable sug- Takeover bids, the free rider problem, and the gestions. We are also grateful for comments theory of the corporation, Bell Journal of from Swami Bhaskaran, Michael Brennen, Economics 11, 42–64. Michel Habib, David Hirsbleifer, Ernst Maug, Herzel, Leo and Richard W. Shepro. 1990. Narayan Naik, Kjell Nyborg, and Walter Torous. Bidders and targets: Mergers and acquisi- We also wish to thank Richard Ruback (the editor) and John Pound (the referee) for tions in the U.S. (Blackwell, Oxford). many helpful comments in the revision of the Hirshleifer, David and Anjan Thakor, 1994, manuscript. Managerial performance, boards of directors 1 Their sample includes hostile takeovers for and takeover bidding, Journal of Corporate targets with a value greater than $50 million. Finance 1, 63–90. 748 ON TAKEOVER AS DISCIPLINARY MECHANISM Jensen, Michael, 1988, Takeovers: Their causes Rappaport, Alfred, 1990, The staying power of and consequences, Journal of Economic the public corporation. Harvard Business Perspectives 2, 21–48. Review 1. 96–104. Klein, April and James Rosenfeld, 1988, Shleifer, Andrei and Larry H. Summers, Targeted share repurchases and top manage- 1988, Breach of trust in hostile takeovers, in: ment changes. Journal of Financial A.J. Auerbach ed., Corporate takeovers: Economics 20. 493–506. Causes and consequences (National Bureau Kini, Omesh,William Kracaw, and Shehzad Mian, of Economic Research, Chicago, IL). 1993. Corporate takeovers, firm performance and Warner, Jerold B., Ross L. Watts, and Karen H. board composition.Working paper (Pennsylvania Wruck. 1988, Stock prices and top manage- State University. University Park, PA). ment changes, Journal of Financial Martin. Kenneth J. and John J. McConnell, Economics 20, 461–492. 1991, Corporate performance, corporate Weisbach, Michael S. 1988, Outside directors takeovers and management turnover. Journal and CEO turnover, Journal of Financial of Finance 46. 671–687. Economies 20, 431–460. Name index

Abowd, J. 390, 394n, 437 Basu, S. 553n, 641n Acharya, V. 420, 441 Baumol, W. 43n, 56, 502, 539 Adedeji, A. 607 Baums,T. 154, 196, 198, 216n, 498, Adler, H. 79 666–667, 710 Admati, A. 138, 147, 149, 154n, 223, 228, Baysinger, B.D. 303, 552n, 596 229, 247, 493 Bebchuk, L. 3, 64, 99, 216n, 217n, 257, 275, Aghion, P. 64, 72, 149, 290, 315 278, 367–368, 711 Agrawal, A. 221, 224, 230, 243, 307n, 364, Becht, M. 2, 133, 157–162, 166, 188n, 191, 400, 572, 620, 649, 659 216n, 307n, 496, 517n, 518n, 704n, 710, Alchian, A. 53, 129n, 154n 716, 729, 731n Alford, A. 625, 640 Beck, P. 602 Ali, A. 625 Beck,T. 101, 636 Allen, F. 54, 101, 136, 138–139, 148, 155, Belsley, D. 619 495, 517n, 605, 607, 703n Bennedsen, M. 99, 182 Ambrose, B. 648 Benston, G. 58 Anderson, D. 42n Berg, S. 564 Ang, J. 8, 111–130 Berger, A. 115–116, 129n Aoki, M. 80, 102, 154n Berger, P. 44n, 553 Argyris, C. 23 Berglöf, E. 66, 72, 107n, 138, 149, 154n, Asquith, P. 43n, 64, 70, 572, 574 307n, 517n Atiyah, P. 344n Bergström, C. 69, 182, 654, 659 Audretsch, D. 502, 517 Berkovitch, E. 210 Ayres, I. 223 Berkovitch, W. 290 Berle, A. 58, 100, 163, 450, 452, 491, 663 Baber,W. 407, 416n Bernard, V. 553n Bacon, J. 307n Bertrand, M. 539 Baird, D. 64 Bethel, J. 221–222, 235, 314, 323, 340, 552, Baker, G. 42n, 44n, 45n, 474 572, 663, 673, 691–692, 709, 717, 718, Baker, H. 605, 620n 726–727 Balakrishnan, S. 559–560, 562 Bhagat, S. 59, 70, 75, 364, 552n, 659n, 735 Baldwin, C. 42n, 45n Bhargava, A. 507, 511 Banerjee, S. 289, 304, 307 Bhattacharya, U. 641n Banz, R. 553n Bhide, A. 44n, 46n, 79, 493, 517, 738 Barber, B. 245–247, 426, 448n, Bianchi, M. 159, 170, 188n, 280n, 307n 479–481, 726 Bianco, M. 160, 307n, 321 Barca, F. 52, 60, 66, 68, 78, 99, 159, 307n, Bittlingmayer, G. 552 342n, 517n Black, B. 36, 45n, 46n, 65, 79, 225, 229, Barclay, M. 60, 69, 175, 261, 288–289, 248, 364, 552n 572, 663–664, 691 Black, F. 374–376, 383–384, 395, 399, Barney, L. 601 402–404, 409, 411–412, 414–416, 417n, Barry, B. 42n, 45n 419, 422, 428, 432, 447 Barth, J. 175 Blackman, S. 43n Bartlett, R. 452 Blair, M. 517n 750 CORPORATE GOVERNANCE AND CORPORATE FINANCE Blake, D. 224–225, 228 Carhart, M. 532–534, 553n Blasi, J. 80 Carleton, W. 222 Bleeke, J. 560, 562, 565 Carlin, W. 101 Bloch, L. 159, 165, 170, 188n, 307n Carney, W. 703n Boehmer, E. 166, 188n, 194, 198, 216n, 498, Carpenter, J. 394n, 435 508, 517n, 518 Carroll,T. 452 Boeker,W. 573 Carter, M. 420, 440 Bognanno, M. 390, 394n Casavola, P. 160 Böhmer, E. 159, 161, 307n, 710, 716, Chan, S. 560 729, 731n Chance, D. 420–423, 434, 440–441, 447 Bolton, P. 64, 72, 73, 137, 149, 154n, 199, Chandler, A. 13–14 210, 289–290, 315, 494, 704n Chang, S. 601, 615 Bond, H. 562 Chapelle, A. 307n Bond, S. 464n Charkham, J. 52, 63, 71, 80, 225, 248n Boone, A. 726 Chatfield, R. 605 Boot, A. 138, 155n Chew, D. 248n Borde, S. 560 Chidambaran, N. 368, 419–448 Borokhovich, K. 344n, 539 Chirelstein, M. 64 Bosch, F.van den 498 Chirinko, R. 198, 453, 495–496, 518n Bower, J. 42, 45n Choi, D. 572, 663 Boycko, M. 45n, 53, 60, 76–77 Chordia,T. 544 Bradley, M. 738 Christensen, C. 17, 43n Brancato, C. 225 Chumpeter, J. 12 Brandt, R. 43n Chung, K. 210, 605 Bratton, W. 207, 217n, 292 Church, J. 519n Brealey, R. 341 Ciscel, D. 452 Brennan, M. 197, 204, 217n, 544 Claessens, S. 100, 163–164, 169, 175, 624 Breusch,T. 507, 511 Clark, K. 42n Brickley, J. 44n, 45n, 224, 249n, 481, Clark, L. 45n 572, 659n Clark, R. 57, 344 Brill, J. 470 Coase, R. 8, 55, 57, 94 Brioschi, F. 188 Coates, J. 525, 546, 550, 552n Bristow, D. 343n Coffee, J. 65, 79, 95, 103, 105, 223, Brown, D. 571 258–259 Brown, L. 641 Cole, R. 111–130, 601, 615 Brown, S. 559 Comment, R. 552n, 672 Brudney, V. 64 Conrad, J. 245 Brunarski, K. 539 Conyon, M. 367, 371–396, 453 Buckingham, L. 371 Cook, P. 392 Bulow, J. 61, 663 Copeland,T. 343n, 559–560, 565 Burgstahler, D. 627 Corbett, J. 705n Burkart, M. 70, 103, 211, 218, 289, 315, Core, J. 368, 417, 431, 467–485, 550 494, 664, 694 Correia da Silva, L. 216n Burkhardt, W. 45n Cosh, A. 394n Burnham, J. 19, 23, 67 Cotter, J. 223, 364 Burrough, B. 56 Coughlan, A. 58, 280n, 296, 339, 364, 389 Butler, H. 303, 552n Cowe, R. 371 Butz, D. 663 Cramton, P. 223 Byrd, J. 364 Crawford, R. 154n Cremer, J. 70 Cable, J. 495–496, 518n Crespi-Cladera, R. 188n Cadbury, A. 222–223, 227, 229, 237, 248n, Crutchley, C. 600 314, 347, 371, 572, 602, 605, 607, 609, Crystal, G. 371 613, 617, 619 Cubbin, J. 494 Cameron, R. 107n Cuny, C. 403, 416n, 435 Campbell, J. 154n Cannella, A. 565n Daems, H. 293, 304–306, 309n Cantrijn, D. 645, 647–659 Dahya, J. 256, 278, 314, 347–365, 488 NAME INDEX 751 Daines, R. 525, 534–535, 546, Eckbo, B. 548 549–550, 553n Edwards, J. 71, 99, 102, 464n, 517n, Dalton, D. 573 518n, 705n Danielson, M. 553n Eisenberg, M. 65, 344n Dann, L. 44n, 69, 738 Eisenberg,T. 229, 243 Dark, F. 45n Elliott, J. 290 Das, S. 559–560, 564, 566n Elston, J. 198, 368, 450–464, 495–496, 518n Datta, D. 560, 562 Emmons, W. 453 David, R. 94 Enriques, L. 639 Davis, E. 225 Ernst, D. 560, 562, 565 De Bondt, W. 596n Erramilli, M. 562 De Wulf, H. 305 Esty, B. 44n, 45n DeAngelo, H. 23, 37, 42n, 44n, 45n, 60, 69, Ezzamel, M. 394n 75, 200, 477, 572, 605, 620n, 649, 652, 659, 738 Faccio, M. 133–134, 163–188, 221–250 DeAngelo, L. 23, 37, 42n, 44n, 45n, 60, 75, Fader, B. 43n 200, 477, 572, 605, 620n Fama, E. 55, 58–59, 113, 130n, 288, 315, Dechop, J.R. 452 518n, 529, 533–537, 541–542, 544–545, Dechow, P. 411, 626–627 553n, 557, 572, 600, 604, 710–711 DeFusco, R. 400 Fan, J. 625, 636, 640 Degeorge, F. 627 Farber, H. 437 Del Guercio, D. 222, 224, 228–229, 233, 245 Farinha, J. 488, 599–621 Delde, H. van 44n Finer, S. 98 Demange, G. 138, 155n Fischel, D. 52, 63–64, 67, 79, 94 DeMarzo, P. 136, 138–139, 145, 149, 155n Fischer, K. 71, 99, 102, 464n, 517n, Demirguc-Kunt, A. 101 518n, 705n Demsetz, H. 65, 68–69, 129n, 130n, 154, Fisher, L. 59 198–199, 209, 222, 224, 288–289, 321, FitzRoy, F. 454, 456, 463n 358, 453, 467, 470, 474, 484, 493, 495, Fluck, Z. 74, 493, 600 503–504, 510, 535, 571, 600, 604 Fowler, K. 573, 596n Denis, D.J. 44n, 45n, 67, 164, 249n, Franke, R. 562 259–260, 286, 321, 339, 364, 444, 663, Frankel, A. 518n 709, 726, 728 Franks, J. 65–67, 70, 154n, 194, 196–197, Denis, D.K. 259–260, 286, 321, 364, 444 200, 204, 217n, 218n, 221–222, 239, 248n, Dennis, D. 200, 290, 297, 308n 256, 261, 286, 313–344, 364, 453, Desai, A. 738 495–496, 518n, 646n, 649, 664, 703n, Dewatripont, M. 54, 71–72, 74, 149 708–710, 717–718, 726, 734–747 Dial, J. 26, 45n Fraune, C. 198, 498, 666–667 Diamond, D. 61, 68, 72, 74, 116, 138, 223, Frederikslust, R. van 4, 488, 557–566 228–229, 494, 571 Freeman, H. 620 Dichev, I. 627 Friedland, C. 453 Dobrzynski, J. 37 Friedman, P. 557, 564 Dobson, J. 130n Froot, K. 410 Dodd, P. 73 Doidge, C. 636 Galbraith, J. 62 Donaldson, G. 26, 44n Gale, D. 54, 71, 101, 136, 138–139, 148, Dong, X. 130n 155n, 495, 517n, 703n Douglas, W. 79 Galgano, F. 260 Dow, G.K. 130n Garcia-Cestona, M. 159, 188n, 307n Drucker, P. 248n, 734 Gardiol, L. 182 Duffie, D. 136, 138–139, 149, 156n Garvey, G. 539 Duggal, R. 658, 659n Gaver, J. 407–408, 416n, 485n Dyck, A. 192, 217, 624, 636–638 Gaver, K. 407–408, 416n, 485n Dyl, E.A. 453 Gedajlovic, E. 495–496, 518n, 519n Gerschenkron, A. 79, 103 Easterbrook, F. 52, 63–64, 67, 74, 79, 94, Gersick, C. 45n 599 Gersovitz, M. 61 Eaton, J. 61 Gertner, R. 64, 71, 73, 571 752 CORPORATE GOVERNANCE AND CORPORATE FINANCE Gibbons, R. 45n, 395n Hausman, J. 507, 511, 519n Gibson, M. 28n Hawkins, J. 222, 224, 228, 233, 245 Gillan, S. 552n Hay, D. 193 Gilson, R. 344n Healey, P. 288, 297 Gilson, S. 45n, 46n, 64, 68, 103, 261, 265, Hecker, R. 194, 666 281, 288, 340, 364, 420, 569, 570–571, Heckman, J. 444 574, 596n, 744 Heiss, F. 209 Glaeser, E. 105 Hellwig, M. 54, 71, 102, 107, 138, Goergen, M. 2, 133–134, 159, 161, 170, 494, 518n 188n, 191–218, 342n, 495–496, 518n Helyar, J. 56 Goldberg, L. 368, 450–464 Hemmer,T. 404 Gomes, A. 61, 74, 182 Henry, P. 12, 106 Gompers, P. 487, 489, 523–553 Hermalin, B. 302, 308n, 314, 339, 343n, Goodstein, J. 573 364, 552n, 573 Gordon, L. 224, 601, 615, 648 Hickman, K. 364 Gorton, G. 66n, 68, 70, 99–100, 154n, Higgins, H. 641n 495–496, 518n, 716 Hill, C. 571 Goshen, Z. 196, 340 Himmelberg, C. 338, 428, 470 Granick, D. 390 Hirschey, M. 571 Grasshoff, U. 454, 463n Hirschman, A. 42n, 104 Green, O. 348 Hirshleifer, D. 737 Greenbury, R. 222–223, 237, 248n, 371, Hoffman, L. 560 373–374, 391, 395n Hofstede, G. 562, 566n Greene, W. 444 Holder, M. 604 Gregg, P. 395n Holderness, C. 44n, 45n, 60, 64–65, 69, Griffith, R. 620n 154n, 163, 175, 261, 288–289, 321, Gromb, D. 211, 218n, 494 339–340, 572, 663–664, 673, 691 Gross, N. 43 Holmstrom, B. 54, 58, 223, 395n, 553n Grossman, S. 55–57, 67, 69, 71, 74, 76, 92, Holthausen, R. 552 99, 286, 288, 493, 664, 691, 734 Holyoke, L. 43n Grout, P. 154n Hope, O. 641n Grundfest, J. 44n, 46n, 63, 79, 703n Hopt, K. 63, 517n, 710 Guay, W. 367, 398–417, 431, 444, 447, Horner, M. 60, 177, 182 479, 484n Hoshi,T. 70, 79, 102, 154n Gugler, K. 159, 162n, 307n Hoskisson, R. 596n Guy, F. 395n Hsiao, C. 501, 507–508, 511–513 Hubbard, G. 428, 518n Hackman, J. 45n Huddart, S. 403, 493 Hall, B.H. 44n, 45 Hughes, A. 394n Hall, B.J. 374, 378, 391, 395n, 420, Hung, M. 425, 639–640 443, 474 Hunt, B. 56 Hall, M. 519n Huson, M. 364 Hall, R.E. 45 Hutton, A. 222, 248n Hamao, Y. 154n Hwang, L. 625 Hambrick, D. 565n Hampel, R. 371, 373 Idson,T. 453 Hanka, G. 539 Ikenberry, D. 553n Hansen, R. 600 Innes, R. 139, 155n Hansmann, H. 54, 76 Ishii, J. 487, 523–553 Harrigan, K. 42n, 565 Harris, M. 45n, 69, 71–72, 74, 99 Jackson,T. 64 Harris, R. 738 Jahera, J. 553n Hart, O. 54–57, 62, 64, 67, 69, 71–72, 74, Jain, B. 600 76, 92, 99, 137–139, 149, 223, 286, 288, James, C. 571 493, 495, 601, 664, 692, 734 Jarrell, G. 44n, 59–60, 224, 249n, 648–649, Haubrich, J. 58 659n, 708 Haugen, R. 398, 400 Jefferis, R. 659n Hauser, S. 130 Jegadeesh, N. 553n NAME INDEX 753 Jenkinson,T. 200, 645, 662–706, 709, Kroll, L. 451 721, 727 Kruse,T. 249n, 264, 339, 728 Jensen, M. 7–9, 11–46, 52, 55–59, 63, 67, Kumar, R. 420 70–71, 75, 79, 92, 111–113, 115, 119, 128–129, 130n, 175, 223, 288, 315, 324, La Porta, R. 8, 91–107, 154n, 163–165, 171, 372, 384, 393, 395n, 398, 400, 417n, 450, 175, 187n, 191–192, 194, 255, 257, 261, 452, 469, 493–494, 515, 517, 518n, 538, 278, 280, 314–315, 341, 344n, 534, 539, 570–572, 574, 599–600, 603–605, 623–625, 634–638 673, 708, 710–711, 735 Lai, J. 488, 568–596 Jepson, J. 249n Lang, L. 133–134, 163–188 Jeunink, A. 642–659, 645 Larcker, F. 467–485 Jo, H. 605 Lasfer, M. 133–134, 221–250 John, K. 45n, 64, 249n, 420, 570 Lehmann, 3, 487, 491–519 Johnson, S. 95, 100–101, 105, 175, 199, 257 Leibenstein, H. 502 Johnson, W. 60 Leuz, C. 489, 623–641 Jong, A. de 159, 188n, 307n Levine, R. 101, 107n, 637 Joos, P. 625 Lewellen, W. 44n, 59, 541 Jorion, P. 403, 416n Lichtenberg, F. 45 Joskow, P.L. 288 Liebeskind, J. 572 Lin, J.T. 111–130, 221 Kabir, R. 307n, 645, 647–659 Linn, S. 652, 659n Kahn, C. 154n, 224 Lins, K. 164, 230, 243 Kang, J. 66, 68, 102, 221, 280n, 364 Lintner, J. 603 Kaplan, S.N. 37, 43n, 44n, 45n, 58–59, Lipton, M. 15, 45n 62–63, 66, 68, 75, 221, 258, 270–272, Ljungqvist, A. 200, 646, 662–706, 709, 721, 280n, 286, 288, 340, 364, 389, 518n, 535, 727 553n, 726 Loderer, C. 44n, 59, 341, 541 Karpoff, J. 222, 224, 243, 248n, 249n, 552n, Long, J. de 57–60 553n Loomis, C. 26 Kaserer, C. 517n, 720, 730, 731n Lopez-de-Silanes, F. 8, 76, 91–107, 154n Kashyap, A. 70, 79, 154n Lorsch, J. 45n Kaul, G. 245 Loughran,T. 62, 541 Keasey, K. 228, 248n Lynch, L. 420, 440 Keene, P. 43n Lyon, J. 245, 247, 426, 448n, 479–481 Keim, D. 605, 607 Kensinger, J. 560 MacAvoy, P. 341 Keown, A. 560 MacBeth, J. 535 Kester, C. 45n McCahery, J. 207, 216n, 217n, 292 Keynes, J. 103 McConnell, J. 59–60, 67, 69, 223, 237, 243, Kim, E.H. 738 248n, 249n, 258, 260, 261, 269, 339, 343n, Kim, H.J. 453 344n, 347–365, 470, 571, 652, 659n, 739, King, R. 101 743–744 Kini, O. 341, 364, 600, 727, 744 McCraw,T.K. 13, 105 Klausner, M. 525, 546, 549, 550, 553 McGuire, J. 452 Klein. A 364, 737 McGurn, P. 552n Klein, B. 154n McMillan, J. 470, 472 Knoeber, C. 221, 224, 230, 243, 307n, 364, McMurray, D. 12, 21 620 McWilliams, V. 649 Knoesel, J. 194 Magee, R. 60 Kogut, B. 560, 562 Magnet, M. 34 Koh, J. 559–560, 564 Mahoney, J. 649, 652, 659n Köke, J. 209, 646, 708–731 Mahoney, J.T. 649, 652, 659n Kole, S. 468 Main, B.G. 378, 391, 394n, 395n, 396n Kothari, S. 416n Maksimovic, V. 101, 726 Koza, M.P. 559–560, 562 Malabre, A. 42n Kraakman, R. 45n Malatesta, P. 60, 552, 659n Kracaw, W. 364, 744 Maldmann, R. 62 Kremp, E. 159, 165, 170, 188n, 307n Mallette, P. 573, 596n 754 CORPORATE GOVERNANCE AND CORPORATE FINANCE Mallin, C.A. 227 Mullins, D. 572, 574 Mandelker, G. 221, 400, 572, 649, 659n Munro, J. 16 Mann, S. 44 Murphy, K. 26, 35, 45n, 58–59, 224, 248n, Manne, H. 4, 63, 67, 129n, 493n, 645 287, 290, 305, 364, 367, 371–396, 400, Marris, R. 56, 539 419–420, 443, 452, 469, 474, 574 Marsh, P. 297, 316 Muus, C.K. 166, 177, 188n Martin, J. 15n, 261, 560 Myers, L. 627 Martin, K. 44n, 67, 364, 739, 743–744 Myers, S. 70, 72, 74, 103, 341, 410, 416n, Maskin, E. 71 570, 600 Masson, R. 452 Maug, E. 154n, 224 Nachman, D. 138–139, 155n Mayer, C. 65–67, 70, 73–74, 78, 101, 104, Nagarajan, N. 60 154n, 191, 200, 217n, 218n, 221–222, Nakamura, M.N. 102 248n, 256, 261, 280n, 286, 313–344, 364, Nalebuff, B. 42n 453, 491, 494–496, 517n, 518n, 646, 649, Nanda, D.A. 557, 559, 560, 623–641 664, 703n, 704n, 708–710, 726, 734–737 Nantell,T.J. 560 Mayers, D. 410, 601, 615 Narayanan, M. 210 Means, G. 58, 100, 163, 450, 452, 663 Neff, R. 43n Meckling, W. 8, 45n, 55, 57–58, 70, 92–94, Nenova,T. 107n, 177, 624, 636 99, 111–112, 115, 119, 128–129, 175, Netter, J. 44n 288, 398, 400, 417n, 450, 571, 600, Neun, S. 453 603–604, 711 Newbitt, S. 224 Meeus, D. 310n Newman, H. 60 Megginson, W. 76, 177, 648 Nibler, M. 716, 721 Mehran, H. 601, 615 Nicodano, G. 274 Mello, A. 129, 199, 210 Noe,T.H. 138–139, 155n, 448n Merton, R. 297, 403, 414 Norton, L. 469 Metrick, A. 487, 523–553 Novaes, W. 182 Meulbroek, L. 553n Nuys, K. van 224, 248n, 659n Meyer, J. 308n, 678, 695 Mian, S. 364, 744 Ofer, A. 297n Michaely, R. 288, 553n, 605, 607 Ohashi, K. 138n Michel, A. 606 Organisation for Economic Co-operation and Mikkelson, W. 60, 221, 364, 572, 663 Development (OECD) 63–68, 73–74, 154, Milgrom, P. 395n, 401, 407 225, 517n, 708, 710–711 Millar, J. 658, 659n Osano, H. 2, 133, 155 Miller, G. 339–340 Miller, K. 43n Pagan, A. 507, 511 Miller, M. 37, 39, 71, 92, 390, 605, 620n Pagano, M. 52, 62, 78, 260, 494 Miller,T. 452 Page, D. 553n Millstein, I. 341 Palepu, K. 38, 43n, 44n, 45n, 67, 78, Minns, R. 228 288, 297 Minton, B. 66, 68, 221, 280n, 286n Palia, D. 479 Mitchell, M. 553n Palmer, D. 518 Modigliani, F. 37, 39, 71, 92, 107n, 620n Panunzi, F. 70, 218n, 494 Moerland, P. 491, 649, 658 Parkinson, J. 344n Mohanram, P. 557, 559–560 Parrino, R. 364, 539 Moore, J. 55, 64, 72, 74, 76, 92, Parsons, J. 129n, 199, 210 137–139, 149 Partch, M. 364 Moore, M. 308 Patell, J. 481 Morck, R. 59, 67, 69, 101–102, 199, 223, Patrick, H. 102 249n, 258, 260, 269, 297, 344n, 453, 467, Patton, A. 452 470, 484, 535, 541, 552n, 553n, 601, 603 Peasnell, K. 611 Morris, D. 193 Peck, S. 394n, 395n Moses, D. 288 Perlitz, M. 517n Moyer, R. 605 Perotti, E. 66, 107n, 154n Mulherin, J. 726 Perry,T. 396n Mullainathan, S. 539 Petersen, M. 115–116 NAME INDEX 755 Pfeffer, J. 470 Ruback, R. 43n, 45n, 56, 67, 221, 493, 539, Pfleiderer, P. 138, 147, 149, 154n 572, 650, 663, 708, 711 Phillips, G. 726 Rydqvist, K. 60, 62, 79, 177, 182, 654, 659n Pinnell, M. 526, 548–549 Ryngaert, M. 60, 659n Pistor, K. 105 Pontiff, J. 43n, 70 Sabow, S. 470, 472 Porter, M. 22, 45n, 80, 102, 248n, 517n Sadler, G. 395n, 396n Posner, R. 56, 344n Sahlman, W. 17, 29, 45n Poulsen, A. 44n, 45n, 60, 224, 249n, 572, Saly, P. 420 648, 649, 659n Santerre, R. 453 Pound, J. 24, 36, 44n, 45n, 46n, 63, 126, Sapienza, P. 102 224, 493, 517n, 596n, 601, 615, Sarin, A. 200, 249n, 364, 709, 726–727 648, 659n Scharfstein, D. 64, 67, 70–73, 79, 137, 154n, Pourciau, S. 308n 315, 710 Prabhala, N. 368, 419–448 Scharfstein, D.S. 137, 139 Pratt, J. 400 Schipper, K. 572, 574 Prowse, S. 66, 154n, 225, 653, 657, 659n Schmid, F. 66, 68, 70, 99–100, 453–454, Pruitt, S. 210 456, 462, 463n, 495–496, 518n, 716 Pugh, W. 553n Schmidt, K. 70, 495 Puia, G. 560, 562 Schmidt, R. 58, 280n, 296n, 339, 364, 389 Scholes, M. 374–376, 381, 383–384, 390, Rahi, R. 138 395n, 399, 402–404, 406, 409–410, 412, Rajan, R. 69–70, 74, 101, 107n, 115–116, 414–416, 417n, 419, 422, 428, 432, 447 129n, 571, 705n Schooley, D. 601 Ramseyer, M. 104 Schrand, C. 416, 417n Rao, R.P. 553n Schumpeter, J. 12 Rappaport, A. 734 Schut, G. 4, 488, 557–566 Rau, P.R. 541 Schwalbach, J. 394n, 453–454, 456, 463n Raviv, A. 45n, 69, 71, 72, 74, 99 Schwert, R. 25, 60, 552n, 672, 708 Rechner, P. 573 Sebora,T. 453 Rediker, K. 601, 605 Segerx, F. 463n Rees, W. 620n Sen, P.K. 559–560, 564 Reese, J. 470 Senbet, L. 398, 400 Reishus, D. 288, 340, 364 Senbet, L.W. 249n Renneboog, L. 2, 133–134, 159, 161, 170, Sengupta, S. 559–560, 564 188n, 191–218, 256, 280n, 285–310, Serrano, J. 66, 663 313–344, 364, 717 Servaes, H. 69, 164, 223, 230, 237, 243, Rice, E. 37, 44n, 60, 75, 649, 652, 659n 248n, 249n, 258, 260, 269, 339, 344n, 470, Richman, L. 42 571, 600, 603, 659n Roberts, D. 452 Seth, A. 601, 605 Roberts, J. 395n, 401, 407 Seward, J. 648, 657–658 Rock, K. 605 Seyhun, H. 553n Roe, M. 24, 36, 44n, 45n, 46n, 52, 65, 78–80, Shane, S. 562 216n, 517n, 518n, 692 Shapiro, D. 76, 495–496, 518n, 519n Roëll, A. 133, 157–162, 260, 307n, 704n Shaw, W. 290 Rogoff, K. 61 Sheehan, D. 44n, 45n, 64–65, 154n, 289, Roll, R. 59 321, 339–340, 572, 673 Romano, R. 52, 64, 79, 223–224, 314 Sheikh, S. 620n Rose, M. 288 Shepro, R. 734 Rose, N. 396n Shin, H.-H. 494, 535, 553n Rosen, S. 379, 386, 392, 395n Shivdasani, A. 66, 68, 221, 280n, 364, Rosenbaum, V. 525–526, 548 572–573, 659n Rosenfeld, A. 59, 541 Shleifer, A. 2, 7–9, 15, 43n, 44n, 45n, 52–82, Rosenfeld, J. 364, 737 91–107, 126, 133, 154n, 163, 199, 207, Rosenstein, S. 364, 573 221, 224, 285–286, 289, 315, 492–494, Rosett, J. 43n, 70 517n, 518n, 535, 539, 541, 548, 552n, Rozeff, M. 600–601, 603–604 553n, 571, 576, 623–625, 648, 663, 705n, Rubach, M. 453 710, 712, 734–735, 737 756 CORPORATE GOVERNANCE AND CORPORATE FINANCE Short, H. 228, 248n, 495, 517n, 518n, 715 Tomlinson, L. 229 Sicherman, N. 44n Torregrosa, P. 600 Siegel, D. 44n, 45n, 297 Towers, P. 472 Siegfried, J. 452 Townsend, R. 71, 138 Singh, H. 74, 221, 560, 562 Travlos, N. 256, 347–365 Sisneros, P. 605 Treisman, D. 104 Skinner, D. 626–627 Trigeorgis, L. 45n Slatter, S. 571 Tufano, P. 400 Slovin, M. 659n Tyrell, M. 177 Smith, A. 37, 44n, 45n Smith, C. 68, 71, 398, 399, 400, 401, 404, Udell, G. 115–116, 129n, 130n 408, 410, 474, 572, 574, 600, 605, 659 Smith, M. 224 Venkatraman, N. 559–560, 564 Snell, S. 571 Vermaelen,T. 307, 541, 553n Song, M. 649 Vetsuypens, M. 420 Spencer, C. 663 Viénot, M. 285, 308n Stafford, E. 553n Vietor, R. 43n Stapledon, G. 198, 216n, 225, 228, 248n, Vijh, A. 480, 482, 541 249n, 339 Vishny, R. 2, 7–9, 45n, 52–107, 126, 133, Starks, L. 364 154n, 163, 199, 207, 221, 224, 285–286, Stein, J. 45n, 62, 71, 75, 80, 553n 289, 315, 492–494, 517n, 535, 539, 541, Steiner, I. 45n 548, 552n, 553n, 571, 576, 604, 623–625, Stephens, S. 44n, 45n 648, 663, 705n, 710, 712, 735 Stevenson, H. 17 Vives, X. 517n Stewart, G. 35, 44n Volpin, P. 255, 257–283 Stigler, G. 53, 94, 453 Voogd, R. 651 Stiglin, L. 43n Stiglitz, J. 58, 76, 492 Wahal, S. 223–224, 245, 247 Stomper, A. 307n Wahlen, J. 623, 626 Strangeland, D. 199 Walkling, R. 44n, 57, 59–60, 541, 649, 659n Strong, J. 308n Walsh, J. 648, 657–658 Strong, N. 566n Ware, R. 519n Stulz, R. 44n, 45n, 59, 69, 72, 102, 106, Warner, J. 45n, 63, 71, 73, 264, 286, 248n, 398, 400–401, 410, 493–494, 535, 296–297, 308n, 339, 364, 389, 493 541, 553n, 600, 604 Watson, B. 21 Subrahmanyam, A. 544 Watson, R. 394n Subramanian, G. 539 Watts, R. 45n, 63, 364, 399, 408, 474, 605, Sudarsanam, P.S. 4, 221, 343n, 488, 568–596 627, 744 Summers, L. 15, 44n, 62, 70, 80, 734 Wei, K.C.J. 553n Sundaram, R. 420 Weigandz, J. 3, 453, 487, 491–519 Sushka, M. 659n Weinstein, D. 69, 102 Swartz, L. 25 Weisbach, M. 43n, 45n, 59, 63, 264, 286, 297, 302–303, 308n, 314, 339–340, 342n, Tedlow, R. 44n 343n, 364, 389, 552n, 572–573, 604, 744 Teoh, S. 62, 288 Weiss, L. 64, 519n Teubner, G. 63 Welch, I. 62 Thadden, E.L. von 72, 107n, 138, 145, 149, Wenger, E. 194, 517n, 666, 720, 730, 731n 154n, 199, 210, 494, 704n Weston, J. 600, 611 Thakor, A. 138, 155n, 737 White, H. 64, 292, 409, 412, 459, 507, Thaler, R. 553, 596n 512–514, 619 Thomas, J.K. 553n Whitehouse, C. 620n Thompson, J. 493 Whyte, A. 560 Thonet, P.J. 495 Wiant, K. 560 Tiemann, J. 44n Williams, J. 113, 128, 244n Tirole, J. 54, 72, 74, 76, 138, 149, 154n, Williamson, O. 54, 56, 72, 154n, 288, 539 155n, 711 Willig, R.D. 76 Titman, S. 74, 553n Winter, R. 604 Todd, R.B. 420 Winton, A. 154n, 224 NAME INDEX 757 Wizman,T. 43n, 70 Yafeh,Y. 66, 69, 102 Wolfenzon, D. 99, 182, 257, 275 Yang, F. 535, 553n Wolff, E. 43n Yermack, D. 58, 229–230, 243, 261, 395n, Wolfram, C. 77, 396n 410, 411, 413, 416n, 417n, 431, 436–437, Wolken, J. 114–115 441, 552n Womack, K. 553n Yeung, B. 199 Wong,T. 62, 625, 636, 640 Young, L. 470 Woodburn, L. 115 Wruck, K. 16, 26, 38, 43n, 44n, 45n, 63, 69, Zechner, J. 138, 147, 149, 154n, 307n 364, 663, 744 Zeckhauser, R. 126 Wu, D. 507, 519n Zellner,W. 43n Wuh Lin, J. 111–130 Zenner, M. 364, 396n Wulf, H. de 305 Zervos, S. 101 Wurgler, J. 101 Zhou, X. 394 Wyatt, J. 364, 573 Zimmerman, J. 287, 290, 305, 364, 574, 627 Wyatt, S. 560 Zingales, L. 52, 56, 60, 62, 69, 74, 78, 99, Wymeersch, E. 289, 310n 101, 107n, 175, 182, 192, 199, 210, 217n, Wysocki, P. 489, 623–641 258, 260, 535, 624, 636–638, 705n Zorn,T. 602 Xie, F. 553n Zwiebel, J. 675, 704n