University of South Carolina Scholar Commons

Theses and Dissertations

2017 Three Essays on Privatization, State Ownership, and Corporate Policies Ruiyuan Chen University of South Carolina

Follow this and additional works at: https://scholarcommons.sc.edu/etd Part of the Business Administration, Management, and Operations Commons

Recommended Citation Chen, R.(2017). Three Essays on Privatization, State Ownership, and Corporate Policies. (Doctoral dissertation). Retrieved from https://scholarcommons.sc.edu/etd/4178

This Open Access Dissertation is brought to you by Scholar Commons. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. THREE ESSAYS ON PRIVATIZATION , STATE OWNERSHIP , AND CORPORATE POLICIES

by

Ruiyuan Chen

Bachelor of Arts Wuhan University, 2008

Master of Science Bocconi University, 2011

Submitted in Partial Fulfillment of the Requirements

For the Degree of Doctor of Philosophy in

Business Administration

Darla Moore School of Business

University of South Carolina

2017

Accepted by:

Omrane Guedhami, Major Professor

Chuck Kwok, Major Professor

Allen Berger, Committee Member

Narjess Boubakri, Committee Member

Sadok El Ghoul, Committee Member

Cheryl L. Addy, Vice Provost and Dean of the Graduate School © Copyright by Ruiyuan Chen, 2017 All Rights Reserved.

ii DEDICATION

To my wife, Helen Wang, and my son, Zekun Chen.

iii ACKNOWLEDGEMENTS

I thank my family, advisors, friends, and colleagues who give me unconditional support along the journey. My deep gratitude first goes to my advisor, dissertation co-chair, and coauthor, Dr. Omrane Guedhami, for his constant support, encouragement, and guidance. I greatly appreciate everything he has done for me. His enthusiasm and thoroughness in research has always inspired to pursue greater accomplishment and never stop. I am also indebted to Dr. Chuck Kwok, my dissertation co-chair, for his generosity of providing help both in teaching and research. He is never tired of guiding me to be an excellent researcher, a competent teacher, and an upright citizen.

I would also like to thank my other dissertation committee members, Dr. Allen

Berger, Dr. Narjess Boubakri, and Dr. Sadok El Ghoul. I am extremely honored to know, learn, and work with these wonderful scholars, who never hesitate to offer help whenever

I need.

Let me also give my gratitude to my coauthors on various projects, Dr. Cui Lin, Dr.

Sali Li, Dr. Robert Nash, and Dr. Jeffery Pittman, and professors who taught me or guided me in Wuhan University, Bocconi University, and University of South Carolina.

I want to thank my friends, Zhongyu Cao, Li Huang, Xinming Li, Kun Liu, Yuqi

Peng, Dr. Guangzhi Shang, Dr. Yijia Zhao, Pengxiang Zhang, Xiaoying Zheng, Yaqin

Zheng, and Ying Zheng, for their support. Moreover, Beth Busby, our department secretary, and Scott Ranges, our business PhD program coordinator, have been extremely helpful, and I thank them.

iv Most importantly, I thank Dr. Helen Wang, who is my wife and best friend, for her endless love and support.

v ABSTRACT

This dissertation examines the impact of state ownership on corporate policies in the context of privatization. The first essay provides new evidence about the agency costs of state ownership on corporate cash holdings and new insight into the corporate governance role of country-level institutions. Consistent with agency theory, we find strong and robust evidence that state ownership is positively related to corporate cash holdings.

Moreover, we find that the strength of country-level institutions affects the relation between state ownership and the value of cash holdings. In particular, as state ownership increases, markets discount the value of cash holdings more in countries with weaker institutions.

The second essay examines the relation between state ownership, stock liquidity, and firm value. We first find that newly privatized firms, on average, are more liquid than other listed firms. However, partially privatized firms are more liquid than fully privatized counterparts. Both of these results suggest a non-linear relation between state ownership and stock liquidity. We find empirical support for this conjecture and the soft budget constraint associated with state ownership. We further find that the relation between state ownership and liquidity is stronger in countries with higher levels of state ownership of banks, and is weaker in countries with fewer limits on foreign banks and little or no political influence as measured by the independence of the supervisory authority. Finally, we show

vi that stock liquidity is related to firm value and cost of equity capital, suggesting that liquidity is a channel through which residual state ownership in NPFs affects valuation.

The third essay examines the impact of state ownership on trade credit provisions.

We find strong evidence that state ownership is positively associated with trade credit. This positive relation between state ownership and trade credit is stronger in countries with poorly developed financial markets, suggesting that implicit borrowing in forms of trade credit from state-owned enterprises provides an alternative source of funds for firms with little access to finance. Contrary to the price discrimination theory, which predicts that firms extend trade credit to extract marginal profit, we find that state ownership is more likely to provide trade credit in competitive industries. Moreover, we find that the market discounts the value of trade credit in the presence of state ownership.

vii TABLE OF CONTENTS

DEDICATION ...... iii

ACKNOWLEDGEMENTS ...... iv

ABSTRACT ...... vi

LIST OF TABLES ...... x

LIST OF FIGURES ...... xii

CHAPTER 1 INTRODUCTION ...... 1

CHAPTER 2 STATE OWNERSHIP AND CORPORATE CASH HOLDINGS : EVIDENCE FROM PRIVATIZATION ...... 3

2.1 INTRODUCTION ...... 3

2.2 LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT ...... 9

2.3 SAMPLE , VARIABLES , AND DESCRIPTIVE STATISTICS ...... 14

2.4 RESULTS ON THE RELATION BETWEEN STATE OWNERSHIP AND CASH HOLDINGS ...... 21

2.5 ADDITIONAL ANALYSIS : INSTITUTIONS , STATE OWNERSHIP , AND THE VALUE OF CASH ...... 34

2.6 CONCLUSION ...... 40

CHAPTER 3 STATE OWNERSHIP , LIQUIDITY AND VALUATION : EVIDENCE FROM PRIVATIZATION ...... 68

3.1 INTRODUCTION ...... 68

3.2 LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT ...... 78

3.3 SAMPLE AND VARIABLES ...... 83

3.4 RESULTS ...... 87

3.5 CONCLUSION ...... 99

viii CHAPTER 4 STATE OWNERSHIP AND TRADE CREDIT : EVIDENCE FROM PRIVATIZATION ...117

4.1 INTRODUCTION ...... 117

4.2 LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT ...... 123

4.3 SAMPLE , VARIABLES , AND DESCRIPTIVE STATISTICS ...... 127

4.4 RESULTS ...... 130

4.5 CONCLUSION ...... 141

CHAPTER 5 OVERALL CONCLUSION ...... 151

REFERENCES ...... 157

APPENDIX A – CHAPTER 2 VARIABLE DEFINITIONS AND DATA SOURCES ...... 170

APPENDIX B – CHAPTER 3 VARIABLE DEFINITIONS AND DATA SOURCES ...... 173

APPENDIX C – CHAPTER 4 VARIABLE DEFINITIONS AND DATA SOURCES ...... 175

ix LIST OF TABLES

Table 2.1 Distribution of the Sample of Newly Privatized Firms ...... 43

Table 2.2 Descriptive Statistics and Pearson Correlation Matrix ...... 45

Table 2.3 State Ownership and Cash Holdings ...... 49

Table 2.4 Endogeneity of State Ownership ...... 51

Table 2.5 State Ownership and Cash Holdings: Disgorging Cash via Investment and Payout, and Investment Efficiency ...... 53

Table 2.6 State Ownership and Cash Holdings: The Role of Market Power and the

Banking Sector ...... 54

Table 2.7 Comparison of Cash Holdings of U.S. Firms and NPFs ...... 55

Table 2.8 Time Trend of Cash Holdings of U.S. Firms and NPFs ...... 56

Table 2.9 State Ownership and Cash Holdings: Is China Different? ...... 58

Table 2.10 Investor Protection, State Ownership, and the Value of Cash ...... 60

Table 2.11 Press Freedom, State Ownership, and the Value of Cash ...... 62

Table 2.12 Political Accountability, State Ownership, and the Value of Cash ...... 64

Table 3.1 Sample Distribution ...... 100

Table 3.2 Descriptive Statistics and Correlation Matrix ...... 102

Table 3.3A Univariate Analysis ...... 103

Table 3.3B Univariate Analysis: The Role of Financial Crisis ...... 104

Table 3.4 Partial Privatization and Liquidity ...... 105

Table 3.5 State Ownership and Liquidity ...... 106

x Table 3.6 Endogeneity Tests ...... 108

Table 3.7 Robustness Checks ...... 110

Table 3.8 Soft Budget Constraint, State Ownership and Liquidity ...... 111

Table 3.9 State Ownership and Liquidity: The Role of Financial Crisis ...... 112

Table 3.10 Liquidity, Cost of Capital, and Valuation ...... 113

Table 4.1 Distribution of the Sample of Newly Privatized Firms ...... 142

Table 4.2 Descriptive Statistics and Correlation Matrix ...... 143

Table 4.3 State Ownership and Trade Credit ...... 144

Table 4.4 Robustness Checks ...... 145

Table 4.5 State Ownership, Financial Market Development, and Trade Credit ...... 147

Table 4.6 State Ownership, Product Market Competition, and Trade Credit ...... 148

Table 4.7 The Valuation Effect of State Ownership and Trade Credit ...... 149

Table 4.8 State Ownership, Product Quality, and Trade Credit ...... 150

xi LIST OF FIGURES

Figure 3.1 Scatter Plot of the Relation between State Ownership and Illiquidity ...... 114

Figure 3.2 Quadratic and Linear Relationship between State Ownership and Stock

Liquidity ...... 115

xii CHAPTER 1

INTRODUCTION

State ownership and privatization continue to be timely and relevant topics.

Following the recent financial crisis, government bailout programs aimed at rescuing financially distressed firms led to a significant increase in state ownership around the world, apparently reversing three decades of privatization that decreased the role of the state in the economy. This “reverse privatization” phenomenon has renewed the debate about the role of governments as shareholders, fostering research on the impact of state ownership on the valuation of corporate assets and equity, the cost of equity, the cost of debt, cash holdings, corporate risk-taking, governance quality, and corporate investment efficiency. Despite the raise of government ownership and “state capitalism”, privatization still remains on the reform agenda of several countries where governments continue to retain substantial stakes in fully and partially state-owned enterprises (SOEs).

There are two competing views of state ownership that result from governments imposing nonprofit-maximizing social and political objectives, while at the same time providing an implicit guarantee against failure. The political view of government ownership suggests that governments pursue political objectives (extract political rents) that rarely coincide with profit maximization. This is one of the main reasons behind SOEs’ inefficiency compared to private firms. As a result, it is only after firms’ control is transferred to private owners that former SOEs can break away from the “grabbing hand”

1 of the government, decrease minority shareholders’ expropriation, and improve their performance. However, state ownership can also have benefits as it carries an implicit guarantee on the firm’s debt because “it is less likely that a firm with state ownership would be allowed to fail. This unwillingness of governments to allow firm default is due to several reasons: pursuit of political and socially desirable goals, such as low unemployment and domestic investment; the desire to maintain key industries providing crucial services to the country; and the reluctance to be associated with a failed investment” (Borisova and

Megginson, 2011).

Against this backdrop, this dissertation examines the impact of state ownership on corporate policies. The first essay examines the relation between state ownership and corporate cash holdings and the role of the institutional environment in influencing the relation between state ownership, cash holdings, and firm value. The second essay examines the relation between state ownership, stock market liquidity, and firm value. The third essay examines the impact of state ownership on trade credit provisions.

2 CHAPTER 2

STATE OWNERSHIP AND CORPORATE CASH HOLDINGS : EVIDENCE FROM PRIVATIZATION

2.1 Introduction

In this essay, we investigate the impact of state ownership on corporations’ cash holdings. While a large body of research documents the effect of corporate governance on the costs and benefits of firms’ cash holdings (Dittmar and Mahrt-Smith, 2007; Dittmar,

Mahrt-Smith, and Servaes, 2003; Harford, Mansi, and Maxwell, 2008; Kalcheva and Lins,

2007), the literature has little to say about the role of state ownership on corporate cash holdings. Liquid assets help mitigate transaction costs (Meltzer, 1963; Miller and Orr, 1966;

Mulligan, 1997) but may also lead self-interested managers to consume private benefits at shareholders’ expense (Jensen, 1986). According to the former view, firms accumulate cash as a precaution against adverse economic shocks and the underinvestment problem

(Opler, Pinkowitz, Stulz, and Williamson, 1999). According to the latter view, firms hold more cash because it is easier for entrenched managers to expropriate value from cash than from other assets, particularly in countries with weak shareholder protection (Dittmar et al.,

2003; Kalcheva and Lins, 2007).

Using a large and unique sample of newly privatized firms (NPFs), we examine whether the level of residual state ownership affects how firms navigate this trade-off between the costs and benefits of cash holdings. We further investigate the role of the country-level institutional environment on the relation between state ownership, cash

3 holdings, and firm value. Kalcheva and Lins (2007) find that cash held by firms with entrenched management is valued at a discount when external country-level shareholder protection is weak. We extend this analysis by examining how shareholder protection, freedom of the press, and political accountability affect cash management.

Understanding the relation between state ownership and corporate cash holdings is important because of the dramatic increase in state ownership around the world that has primarily resulted from rescue efforts following the recent financial crisis (Borisova, Fotak,

Holland, and Megginson, 2015; Megginson, 2016). As of 2011, for instance, state ownership of equity accounts for nearly 20% of stock market capitalization worldwide

(Borisova, Brockman, Salas, and Zagorchev, 2012). Evidence on the role of the state as a large shareholder, however, is mixed. On the one hand, state ownership is associated with more agency problems, as managers of state-owned enterprises (SOEs) are typically entrenched bureaucrats who pursue political goals instead of the maximization of shareholder value. In line with this view, extant research shows that state ownership is associated with poor corporate governance, weak performance, and severe moral hazard problems (Boubakri, Cosset, and Saffar, 2013; Megginson and Netter, 2001; Shleifer and

Vishny, 1997). Contrastingly, state ownership is associated with implicit government guarantees, preferential access to credit, and soft budget constraints, particularly during times of financial distress (Borisova et al., 2015; Borisova and Megginson, 2011; Faccio,

Masulis, and McConnell, 2006). This evidence suggests that as government stakes increase, firms suffer less from financial constraints, adverse economic shocks, and underinvestment

4 problems. The impact of state ownership on corporate cash holdings is thus an open question. 1

To shed light on the potential relation between state ownership and corporate cash holdings, we compile a unique sample of 578 NPFs from 59 countries. Privatization, defined as the sale of SOEs or other government assets to the private sector (Megginson and Netter, 2001), provides an ideal setting for isolating the effect of state ownership on corporate cash holdings, for two reasons. First, privatization is associated with a dramatic change in ownership structure, as ownership is transferred from the government to private shareholders. 2 Second, the movement of SOEs to the private sector is associated with changes in agency problems, information asymmetry, and government implicit guarantees

(e.g., Borisova and Megginson, 2011; Boubakri, Cosset, and Guedhami, 2005; Denis and

McConnell, 2003; Guedhami, Pittman, and Saffar, 2009), and therefore is suitable for addressing questions on how state ownership affects cash holdings.

Consistent with the view that government ownership leads to serious information asymmetry and agency problems, we find that state ownership is positively associated with corporate cash holdings. Economically, an increase in government ownership from the 25 th to the 75 th percentile leads to a 30% increase in corporate cash holdings, while all other variables remain constant. This result continues to hold when we use alternative measures of state ownership. In particular, we find that firms hold more cash if the state maintains

1 The most prominent empirical studies of the impact of state ownership on cash holdings (Kusnadi et al., 2015; Megginson et al., 2014) only consider China. Our large, multinational sample (59 countries) allows us to examine the broader, cross-national implications of residual state ownership on corporate cash holdings. 2 We define a NPF as an entity in which state ownership has been recently reduced through privatization. A term commonly used in the privatization literature, NPFs may have a zero or a positive level of residual state ownership. Consistent with the findings of Megginson et al. (1994) and Boubakri and Cosset (1998), a majority of our sample of NPFs retains residual state ownership.

5 majority control (i.e., the government retains more than 50% of a firm’s shares) and if the

NPFs are politically connected as defined in Faccio (2006). Following prior studies of cash management (e.g., Gao, Harford, and Li, 2013; Harford et al., 2008), we also use alternative measures of cash holdings to test the sensitivity of our findings. For instance, to mitigate the concern that industry representation drives the relationship between state ownership and cash holdings, we use an industry-adjusted measure of cash holdings. We continue to find that firms hold more cash when state ownership is higher.

One may be concerned about the endogeneity of state ownership. Megginson and

Netter (2001, p. 346) argue that “sample selection bias can arise from several sources, including the desire of governments to make privatization look good by privatizing the healthiest firms first.” In addition, the relation between state ownership and cash holdings could be due to unobserved determinants of cash holdings that also explain residual state ownership following privatization. To address this concern, we first use firm fixed effects.

Consistent with our main results, we find that state ownership continues to load positively and is statistically significant at the 1% level. A disadvantage of a firm fixed effects model, however, is that it is not able to capture the effects of time-invariant factors (e.g., country- level variables such as the level of shareholder protection). As a result, we also employ instrumental variables, the Heckman selection model, and propensity score matching to address this issue. Our main results continue to hold after using each of these approaches.

To mitigate the concern that the relation between state ownership and cash holdings is driven by other firm- and country-level characteristics, we include several additional control variables. At the firm level, we include the ratio of R&D expenses to total sales, which proxies for information asymmetry, and the ratio of acquisition expenditures to total

6 sales, which captures managers’ attempts to increase the size of the firm. We also add proxies for profitability, asset sales, and new debt issues to account for the possibility that cash holdings in NPFs reflect increased profitability or deliberate corporate actions such as borrowing or selling assets. At the country level, we include the ratio of market capitalization to GDP to control for the development of a nation’s capital market. After taking these additional variables into account, we continue to find that state ownership is positively and significantly associated with corporate cash holdings.

Next, we examine why state ownership is associated with higher cash holdings, and whether this relation reflects higher agency problems. We address these questions in two ways. First, we examine how state ownership influences the use of cash holdings through firms’ investment and financial policies. Our results suggest that as residual state ownership increases, NPFs are more likely to use cash to increase capital expenditures and acquisitions, but not to increase research and development. We do not find evidence that cash is used to increase dividend payouts. In addition, consistent with Jaslowitzer,

Megginson, and Rapp (2016) and Chen, El Ghoul, Guedhami, and Wang (2017), we find that state ownership is associated with poor investment efficiency. Our initial findings are consistent with the contention that NPFs with higher residual state ownership are more likely to disgorge cash to achieve political goals rather than profit maximization. Second, we explore the role of market power and superior access to state-owned banks in influencing the relation between state ownership and cash holdings. Our findings suggest that state owners are likely to have superior access to financing and maintain high cash balances through financing from state-owned banks.

7 In additional analysis, we examine whether the quality of the institutional environment affects the value of cash holdings by our sample of NPFs. We find that as residual state ownership increases, capital markets more severely discount the value of cash holdings in countries where investor protection, freedom of the press, and political accountability are weaker. Overall, these results suggest that when country-level institutions are less effective in constraining consumption of political private benefits, the market value of cash holdings is lower for firms with a higher government stake.

Our study of the relation between state ownership and corporate cash holdings also makes an important contribution to the privatization literature. Extant research

[summarized in Megginson and Netter (2001)] generally identifies significant improvements in the financial and operating performance of NPFs. However, there is limited empirical evidence on the sources of these improvements. This essay extends this literature by providing preliminary insight into the sources of the performance improvements of NPFs. As we previously noted, the financial and operating performance of SOEs can be reduced by state misappropriation of corporate resources (i.e., the agency costs of state ownership). Privatization should mitigate such costs. One area where a reduction in the agency costs of state ownership could be directly measured is a firm’s cash management policies. Specifically, since cash is the most liquid asset, Caprio, Faccio, and

McConnell (2013), Kusnadi, Yang, and Zhou (2015), and Myers and Rajan (1998) argue that cash is most vulnerable to political extraction. Therefore, cash is an especially tempting target for state expropriation. Because privatization should dampen that temptation, one source of performance improvement following privatization is likely to be an increase in the value of cash holdings. By documenting that the value of cash holdings increases as

8 residual state ownership decreases, we provide evidence of a direct source of the value accretion associated with privatization.

Further, since we examine cash holdings across a large sample of institutionally diverse countries, our study provides interesting evidence regarding the effect of the institutional environment on contracting decisions. We know that institutions matter. In this essay, we attempt to shed light on which institutions matter more and why certain institutions may matter more under certain circumstances. Following La Porta, Lopez-de-

Silanes, Shleifer, and Vishny (1997), we consider the role of legal institutions. We also examine how additional institutional characteristics (i.e., effectiveness of the media, degree of political accountability) affect the private benefits of state control. By identifying the impact of multiple institutional factors, we broaden our understanding of how the institutional environment affects contracting decisions.

2.2 Literature Review and Hypothesis Development

2.2.1 Cash Holdings

Theoretically, a corporation should set its cash holdings at a level such that the marginal costs of cash holdings equal the marginal benefits of those holdings (Opler et al.,

1999). The costs of holding cash include the cost-of-carry, which is the opportunity cost of an additional dollar of cash, and the agency cost of cash holdings, which is the cost associated with self-interested managers engaging in wasteful spending or private consumption. On the benefits side, firms may hold cash for precautionary purposes. For instance, Myers and Majluf (1984) argue that firms hold more cash when information asymmetry is severe and the cost of raising external funds is high. Also, consistent with

9 the precautionary motive, Myers (1984) and Myers and Majluf (1984) contend that firms hold cash to prevent underinvestment in positive NPV projects.

Empirically, prior studies find evidence for the precautionary motive to hold cash.

Opler et al. (1999), for instance, show that publicly listed U.S. firms that face more difficulties in raising external capital (e.g., smaller firms and firms without a credit rating) hold more cash. Similarly, Duchin (2010) finds that public firms with higher cash flow uncertainty are likely to increase corporate cash holdings. McLean (2011) shows that firms with large R&D expenses, which proxy for firms’ precautionary motives, tend to issue shares to increase their cash positions. Other studies, however, find evidence supporting the agency-based explanation of cash holdings. In studies focusing on the U.S. context,

Harford (1999) finds that cash-rich firms are more likely to attempt to engage in acquisitions and that such acquisitions are associated with a negative stock market reaction and poor long-term operating performance. Additionally, Dittmar and Mahrt-Smith (2007) identify that the value of cash is lower in firms with severe agency problems (as measured by the intensity of antitakeover provisions and institutional ownership). In studies emphasizing the role of country-level agency problems, Dittmar et al. (2003) show that firms are likely to hold more cash in countries with weak investor protection. 3 Pinkowitz,

Stulz, and Williamson (2006) find that the market value of cash holdings by private sector firms is lower in countries with low investor protection. Kalcheva and Lins (2007) focus on the effects of firm-level agency problems and the institutional environment on cash holdings. They find that in countries with weak shareholder protection, private sector firms

3 Contrary to much of the cross-country literature, Harford et al. (2008) find that U.S. firms with weaker corporate governance structures tend to hold less cash. The authors argue that entrenched managers in the U.S. prefer to spend cash faster to avoid being scrutinized by external governance mechanisms (e.g., activists and raiders).

10 tend to hold more cash and firm value is lower when entrenched management holds more cash. Chen et al. (2012) consider how corporate governance affects the cash holdings of

Chinese firms. Using a sample of private and state-owned firms from China, the authors consider how an exogenous shock in the institutional environment affected the cash holdings and find that levels of cash holdings respond to changes in the strength of corporate governance. Given the mixed evidence (and the prior focus on private sector firms), the effect of state ownership on corporate cash holdings is ultimately an empirical question. We next discuss different theories and consider their testable predictions about how state ownership may affect cash holdings.

2.2.2 State Ownership

2.2.2.1 Soft Budget Constraints

According to the soft budget constraint theory developed by Kornai (1979, 1980), a government can soften or relax an SOE’s budget constraints by providing tax discounts, preferential access to credit, and other forms of support (Kornai, Maskin, and Roland,

2003). Anderson, Korsun, and Murrell (2000) find that privatization and decentralization are essential in reducing soft budget constraints because when the central government retains ownership in privatized firms, more than two-thirds of enterprises still have soft budget constraints. In a related study, Frydman, Gray, Hessel, and Rapaczynski (2000) identify performance differences between privatized firms controlled by outsiders and those controlled by governments, and argue that privatized firms controlled by governments are not as disciplined because state banks and tax authorities provide extra support to soften budget constraints. Faccio et al. (2006) describe yet another channel through which state ownership affects corporate decisions. In particular, they find that

11 politically connected firms are more likely to be bailed out by the government during times of financial distress. Similarly, Boubakri, Guedhami, Mishra, and Saffar (2012) show that firms increase leverage after a politician joins the board of directors. Chaney, Faccio, and

Parsley (2011) further document that politically connected firms have a lower cost of borrowing even though their reporting quality is poorer.

Taken together, extant studies suggest that state ownership is negatively associated with the transaction cost and precautionary motives, as state-controlled firms can raise external funds at lower cost. Consistent with this conjecture, Megginson, Ullah, and Wei

(2014) find for a sample of Chinese privatized firms that state ownership is negatively associated with corporate cash holdings. This discussion leads to our first hypothesis on the relation between state ownership and the cash holdings of NPFs.

Hypothesis 1a: NPFs hold less cash because of the soft budget constraints provided by residual state ownership.

2.2.2.2 Agency Problems

In an agency theory setting, SOE inefficiencies are a natural outcome of the separation of ownership and control. First, because SOEs are owned by the public but controlled by politicians, no individual owner has a strong incentive to engage in active monitoring (Laffont and Tirole, 1993; Vickers and Yarrow, 1988, 1991). In addition, managers of SOEs are evaluated according to achievement of political goals rather than profit maximization, and are less subject to pressures from the stock, product, or labor markets. Lacking both internal monitoring and external corporate governance mechanisms,

SOE managers have incentives to consume private benefits. In line with these arguments, empirical evidence shows that when the state retains a majority share of a privatized firm, the performance and governance improvements from privatization are less pronounced.

12 For instance, Boubakri and Cosset (1998), D’Souza and Megginson (1999), Guedhami et al. (2009), and Megginson, Nash, and Randenborgh (1994) document that state residual control undermines post-privatization restructuring and improvements in profitability, efficiency, and investment. Similarly, Borisova et al. (2012) identify that higher state ownership is associated with lower quality corporate governance. Jaslowitzer et al. (2016) and Chen, El Ghoul, Guedhami, and Wang (2017) find that state ownership is negatively associated with investment efficiency. Boubakri et al. (2013) show that state ownership is negatively related to corporate risk-taking.

The above-mentioned literature suggests that state ownership is associated with poor corporate governance and more severe agency problems. Since managers of SOEs may hold more cash in an effort to expropriate minority shareholders, this theory predicts a positive relation between state ownership and the cash holdings of NPFs. Therefore, our alternative hypothesis on the association between state ownership and cash holdings is as follows.

Hypothesis 1b: NPFs hold more cash because of the agency costs of residual state

ownership.

2.2.2.3 Institutional Environment and the Agency Costs of State Ownership

Our second hypothesis relates to our analysis of whether the strength of a country’s institutional environment affects the agency costs of residual state ownership and thus the value of cash holdings of NPFs. If stronger institutions help mitigate the agency problems of state ownership, we should expect a positive relation between the strength of a country’s institutions and the value of cash holdings. This leads to our second hypothesis.

13 Hypothesis 2: The value of cash holdings of NPFs is higher in countries with a

stronger institutional environment.

2.3 Sample, Variables, and Descriptive Statistics

2.3.1 Sample Selection

To empirically test the relationship between state ownership and cash holdings, we construct a sample of 578 NPFs from 59 countries over the period 1981–2014. In our multivariate analysis, we also construct various samples by matching NPFs to privately owned U.S. firms. Our sample of NPFs comes mainly from Boubakri et al. (2013), which we update using Privatization Barometer, Thomson Reuters SDC Platinum Global New

Issues database, and SDC Platinum Mergers & Acquisitions database. This sample provides an ideal setting in which to investigate the role of state ownership on cash holdings.

First, we track the change in ownership after the first privatization, which allows us to analyze the time-varying effect of state ownership on cash holdings. Second, these data cover firms in countries with diverse institutional environments, which allows us to investigate the relation among cash holdings, state ownership, and institutional quality. We obtain cash holdings and other financial statement information from Compustat Global and ownership data from Boubakri et al. (2013), firms’ annual reports, Bureau van Dijk’s Osiris database, and Bloomberg. Since the behavior of financial firms (SIC codes between 6000 and 6999) is heavily influenced by a country’s regulatory environment, we exclude these firms from our sample.

Table 2.1 summarizes the sample distribution by year, industry, and region. Panel

A reveals significant growth in privatizations over time. In particular, more than 90% of our sample firms were privatized in the 1990s and 2000s. Panel B shows that the sample

14 firms are widely distributed across industries [defined according to Campbell’s (1996) classification], with 31% in utilities, 16% in basic industry, and 14% in transportation.

Similarly, Panel C shows that the sample firms are widely distributed across regions, with

46% located in East and South Asia and the Pacific, 41% in Europe and Central Asia, 9% in Latin America and the Caribbean, and 4% in Africa and the Middle East.

2.3.2 Variables

2.3.2.1 Cash Holdings

Following prior literature (e.g., Harford et al., 2008), we calculate Cash Holdings as the ratio of cash and marketable securities to total sales. For robustness, we also use the ratio of cash and marketable securities to total assets (Gao et al., 2013) and the ratio of cash and marketable securities to net assets, where net assets is the difference between total assets and cash and marketable securities (Opler et al., 1999). To mitigate concerns that the relationship between state ownership and cash holdings is driven by industry-specific characteristics, we additionally use an industry-adjusted measure of cash holdings. This metric (Industry-Adjusted Cash) is equal to the difference between a firm’s Cash Holdings and its industry median Cash Holdings (Gao et al., 2013; Harford et al., 2008). 4 The

Appendix summarizes the definitions for these and all other variables that we use in our analyses.

2.3.2.2 State Ownership

We capture state ownership using three indicators. First, State Ownership is the percentage of shares held by a government. Second, Control is a dummy variable equal to

4 We also follow Dittmar and Mahrt-Smith (2007) to estimate excess cash. Our measure of excess cash is the residual from a regression of cash on its determinants. Our main results remain statistically unchanged when we use excess cash as our dependent variable.

15 one for firms in which the state maintains control (i.e., the government holds more than 50% of the firm’s shares) following partial privatization. Our third measure of state ownership is Political, which is a dummy variable equal to one for firms that are politically connected

(Faccio, 2006). 5

2.3.2.3 Control Variables

Our regressions include several firm- and country-level control variables to ensure that the relation between state ownership and cash holdings is not driven by confounding factors. In particular, we follow Gao et al. (2013), Harford et al. (2008), and Opler et al.

(1999) and include firm size (Size), leverage (Leverage), net working capital to total sales

(Net Working Capital), cash flow to total sales (Cash Flow), capital expenditures to total sales (Capital Expenditures), sales growth (Sales Growth), a dividend dummy (Dividend), and the standard deviation of cash flow (Cash Flow Volatility) as controls. Additionally, following Dittmar et al. (2003) and Kalcheva and Lins (2007), we include private credit to

GDP (Private Credit) to control for a country’s banking sector development.

2.3.2.4 Institutional Variables

We also examine the effect of institutional development on corporate cash holdings and consider the role of institutions on the relation among state ownership, cash holdings, and firm value. Specifically, we include measures of institutional strength that relate to investor protection, freedom of the press, and political accountability. We discuss each of these in turn.

Investor protection

5 Faccio (2006, p. 369) identifies a firm as politically connected “…if at least one of its large shareholders (anyone controlling at least 10 percent of voting shares) or one of its top officers (CEO, president, vice- president, chairman, or secretary) is a member of parliament, a minister, or is closely related to a top politician or party.”

16 Protection of the rights of minority shareholders varies significantly across countries (La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997, 1998). Dyck and

Zingales (2004) and Johnson, La Porta, Lopez-de-Silanes, and Shleifer (2000) show that the private benefits of control are greater in countries with weaker legal protection of minority shareholders. Therefore, the state may extract larger private benefits of control from NPFs when legal protection of minority shareholders is weak. Since this expropriation of minority shareholders could include extracting value from the firm’s assets, we expect a positive relation between the strength of shareholder protection and the value of cash holdings. To measure the strength of shareholder protection, we follow Kalcheva and Lins

(2007) and use the revised anti-director rights index of Djankov, La Porta, Lopez-de-

Silanes, and Shleifer (2008) to construct Shareholder Rights .6

Freedom of the press

Bushman, Piotroski, and Smith (2004) and La Porta, Lopez-de-Silanes, Shleifer, and Vishny (2002) show that expropriating states frequently attempt to suppress information to conceal the predatory behavior of politicians. However, active and independent media are likely to expose diversion of SOE resources for political or personal advantage, decreasing the value of expropriation opportunities and increasing politicians’ accountability (Dinc and Gupta, 2011; Haw, Hu, Hwang, and Wu, 2004; Hellman, Jones, and Kaufman, 2003). 7 Thus, by raising the costs of state diversion, active and independent

6 In robustness tests, we also use the anti-self-dealing index from Djankov et al. (2008), the rule of law metric from the World Governance Indicators (WGI), and legal origin from La Porta et al. (1998). Our main results remain unchanged. 7 The monitoring role of the media is consistent with the conjecture of Glaeser, La Porta, Lopez-de-Silanes, and Shleifer (2004) and Haw et al. (2004) that an educated public is less tolerant of government malfeasance. The assumption that more effective monitoring by the media will reduce the state’s propensity to consume private benefits of control is similar to the argument of Becker (1968) that the crime rate is lower when potential perpetrators face a higher likelihood of being caught.

17 media are expected to positively affect the value of cash holdings by NPFs. To capture the role of the media, we follow Dyck and Zingales (2002) and Brunetti and Weder (2003) and use Freedom House’s freedom of the press index. Freedom House assesses each country’s media independence by assigning numerical scores ranging from 0 (the most free media) to 100 (the least free media). We construct Press Freedom as 100 minus the original

Freedom House index. Higher values of Press Freedom indicate that a country’s media are more independent and hence more likely to expose state corruption or misappropriation.

Political accountability

A politician’s cost of extracting private benefits from SOEs is a function of not only the probability of being exposed, but also the severity of the penalties imposed upon detection. One such penalty for a politician is to be voted out of office. Prior work suggests that politicians are less likely to engage in expropriation in a more politically competitive environment (Beck, Clarke, Groff, Keefer, and Walsh, 2001; Bortolotti and Pinotti, 2003;

Dyck, Volchkova, and Zingales, 2008; Johnston, 1997). Building on these studies, we argue that politicians who are more accountable to voters should be less willing to expropriate wealth from firms with residual state ownership. Thus, by limiting the state’s ability to extract private benefits, greater accountability to voters should be positively associated with the value of cash holdings.

We employ several proxies for the constraints imposed by a competitive political environment (i.e., political accountability). First, we use the democracy index, Democracy , from the International Country Risk Guide (ICRG). This index ranks how democratic vs. autocratic a country’s political system is. Politicians from countries with a more democratic political system are likely to face greater electoral competitiveness and hence are more

18 likely to be constrained from exploiting private benefits of control. Since higher index values indicate a more democratic government (i.e., more competitive elections), we expect a positive relation between Democracy and the value of cash holdings by NPFs.

Second, we use Checks, which reflects the degree to which political checks and balances limit the ability of government officials to engage in unilateral decision-making and in turn rent-seeking behavior (see, e.g., Djankov, La Porta, Lopez-de-Silanes, and

Shleifer, 2002, 2010; Keefer, 1999; Keefer and Knack, 1997). This measure is from the

Database of Political Institutions and is based on the number of political actors with veto power in a country’s government (e.g., whether the executive and legislative branches are controlled by different parties, the number of parties in coalition governments). Henisz

(2002) shows that governments that do not have political checks and balances in place are more likely to divert resources from firms with residual state ownership. We therefore expect a positive association between Checks and the value of cash holdings by NPFs.

As a measure of another “check” on governmental power, we use La Porta, Lopez- de-Silanes, Pop-Eleches, and Shleifer’s (2004) Judicial Independence index, which reflects the degree to which a country’s judicial system is autonomous from its other branches of government. This measure is based on factors such as the tenure of judges and the prevalence of case law. By limiting unilateral decision-making within a government, greater judicial independence should reduce the expropriation of SOE resources by politicians. Accordingly, we expect a positive relation between Judicial Independence and the value of cash holdings by NPFs.

Our last political accountability variables capture the opacity of a state’s operations.

Specifically, we use the disclosure variable from Djankov et al. (2010): Disclosure by

19 Politicians, which indicates the extent to which politicians are required to disclose their finances and business activities. Higher disclosure score suggests more transparency. To the extent that expropriating governments prefer to operate in the shadows (Bushman et al.,

2004), a more transparent government should be associated with lower private benefits of control. We therefore expect the value of cash holdings by NPFs to be higher in countries with more stringent disclosure requirements.

2.3.3. Descriptive Statistics

Panel A of Table 2.2 provides summary statistics. To avoid the influence of outliers, we winsorize all financial variables at the 1% level on both sides of the sample distribution.

Our main dependent variable, Cash Holdings , has a mean (median) of 0.12 (0.09). Residual state ownership (State Ownership) has a mean (median) of 0.25 (0.12), in line with a sharp decline in state ownership after privatization (Boubakri et al., 2005), although governments maintain control (Control) in 29% of the firms in our sample. Of the privatized firms, 33% are politically connected (Political). This is consistent with the finding by Bortolotti and

Faccio (2009) that governments frequently transfer ownership rights without relinquishing control rights after privatization.

Panel B of Table 2.2 presents Pearson correlation coefficients between the firm- level variables. Consistent with the agency hypothesis, we find that State Ownership,

Control , and Political are all positively and significantly related to cash holdings. As predicted by the precautionary motive, we find that Cash Flow Volatility is positively related to cash holdings. A proxy for external monitoring, Leverage , is negatively related to cash holdings.

20 Panel C reports the means of our institutional indices by country. With respect to shareholder protection, Western European countries exhibit substantial variation in the revised anti-director index, while the Asian countries (except China), Brazil, and Ghana have high values on this index. Turning to freedom of the press, countries with low press freedom are generally developing or transitioning economies. The sample countries display wide variation on the political accountability indices. For example, Western European countries have high scores on judicial independence and democracy, while India has the highest average number of checks and balances. Most countries mandate disclosure by politicians (notable exceptions include China, Denmark, and Finland).

Overall, the above statistics show that our sample countries vary in the extent of their institutional development. This cross-sectional variation mitigates concerns of selection bias in privatization data (e.g., Megginson and Netter, 2001). 8 Most importantly, this diversity in terms of institutional development allows us to empirically examine how cross-country variation in institutional factors affects the market value of cash holdings.

2.4 Results on the Relation between State Ownership and Cash Holdings

2.4.1 Main Evidence

Table 2.3 presents results for regressions of the different measures of cash holdings on our proxies for state control as well as firm characteristics, country-level institutional variables, and industry and year fixed effects. In our baseline specification (Model 1), we use the natural logarithm of Cash Holdings as the dependent variable and test for the effect

8 These data also emphasize the large institutional differences between China and the other countries in our sample. Megginson et al. (2014), in one of the most prominent studies of the relation between state ownership and cash holdings, only consider Chinese firms. Our large, institutionally diverse sample allows us to more broadly consider how institutional heterogeneity can impact the relation between state ownership and cash holdings. Jaslowitzer et al. (2016) also recognize the importance of a sample’s institutional diversity when considering the relation between state ownership and financial decision-making.

21 of State Ownership . Specifically, we estimate the following regression (superscripts omitted for simplicity):

ln ℎ = + ℎ + +

+ + ℎ

+ + ℎ +

+ ℎ + ℎℎ ℎ

+ + + + . 1

To control for within-firm correlation, we present significance levels based on robust standard errors adjusted for clustering at the firm level.

We find that State Ownership is positively associated with corporate cash holdings.

This relation is statistically significant at the 1% level. It is also economically significant, with the coefficient on State Ownership suggesting that moving state ownership from the

25 th percentile (0.00) to the 75 th percentile (0.51) results in a 30% (=0.51×0.587) increase in corporate cash holdings (with all other variables constant). In line with the univariate results, these findings support the agency view of state ownership (Hypothesis 1b).

Models 2 and 3 include alternative proxies for state control. In Model 2, we replace

State Ownership with the dummy variable Control , which is equal to one if the government retains control over the partially privatized firm (i.e., owns more than 50% of the firm’s shares). We find that Control loads with a positive and statistically significant coefficient

(at the 1% level). This result is also economically significant, with corporate cash holdings increasing 31% as Control changes from zero to one (while holding all other variables constant). In Model 3, we replace State Ownership with the dummy variable Political, which is equal to one if a firm is politically connected, as defined by Faccio (2006).

Providing additional evidence that state influence is positively associated with corporate

22 cash holdings, we find that Political loads positively and is statistically significant at the

1% level. This finding is also economically significant, with cash holdings increasing 23% as Political changes from zero to one (while holding all other variables constant). These results are consistent with those based on State Ownership and support the agency view of state ownership (Hypothesis 1b). 9

In Models 4 through 6, we employ alternative measures of cash holdings. In Model

4, we use Cash/Assets . We find that the coefficient on State Ownership continues to be positive and statistically significant at the 1% level, with the magnitude of the economic impact of state ownership on corporate cash holdings comparable to that in Model 1. In

Model 5, we use Cash/Net Assets as our dependent variable. Our results remain unchanged:

State ownership loads positively and is statistically significant at the 1% level. In Model 6, we use an industry-adjusted measure of cash holdings, which is equal to the difference between a firm’s actual cash holdings and its industry-median cash holdings [based on

Campbell’s (1996) 12 industry categories]. We continue to find that State Ownership is significantly positively associated with cash holdings. Finally, in Model 7, we omit the country-level institutional variables and include country fixed effects. We find that State

Ownership continues to load positively and is statistically significant at the 5% level. These results confirm our findings that state ownership is a key determinant of corporate cash holdings.

In summary, our results in Table 2.3 are consistent with the agency view of state ownership, indicating that firms are likely to hold more liquid assets as the government’s

9 To help verify robustness, we follow Boubakri et al. (2013) and also capture state influence with a dummy variable Golden share , which indicates whether the government holds a golden share in the NPF. The unreported results indicate that Golden share is positively and significantly (at the 1% level) related to corporate cash holding, consistent with Hypothesis 1b.

23 ownership stake increases. We further note that our multinational results are very different from those of Megginson et al. (2014). Using a sample of Chinese firms, Megginson et al.

(2014) identify a negative relation between state ownership and cash holdings (consistent with the soft-budget hypothesis). However, Megginson et al. (2014) only consider firms in

China. As we document in Table 2.2 (Panel C), the institutional environment in China is very different from that of the rest of the world. 10 We contend that the unique political, social, economic, and cultural environment of China contributes to the differences between our findings and those of Megginson et al. (2014). We further explore the effect of those institutional differences in our subsequent empirical analysis (in Section V).

2.4.2 Endogeneity

In the context of privatization, a major econometric concern is selection bias. As

Megginson and Netter (2001, p. 346) point out, “sample selection bias can arise from several sources, including the desire of governments to make privatization look good by privatizing the healthiest firms first”. Further, governments may retain higher stakes in firms with more cash holdings to extract greater private benefits or because few shareholders may be interested in investing in privatized firms. In addition, the relation between state ownership and cash holdings could be driven by unobserved determinants of cash holdings that also explain residual state ownership. We address these issues using several approaches in Table 2.4.

First, we use a firm fixed effects model to mitigate concerns that corporate cash holdings are driven by time-invariant unobservable variables. Model 1 presents the results.

10 In unreported robustness tests, we re-do the analysis of Table 2.3 using data only for our Chinese firms and then using data for the total sample (excluding China). Our findings for the China-only sample confirm those of Megginson et al. (2014). Our findings for the total sample (after excluding China) are virtually the same as those we present in Table 2.3.

24 We find that State Ownership loads positively and is statistically significant at the 10% level. This result suggests that the positive relation between state ownership and corporate cash holdings is not driven by time-invariant firm effects.

Second, we estimate instrumental variable regressions. In particular, we use the variable Collectivism , which is equal to 100 minus the value of Hofstede’s (2001) individualism index. We use Collectivism as an instrument since Boubakri, Guedhami,

Kwok, and Saffar (2016) show that residual state ownership is higher in privatized firms located in more collectivistic societies. In the first-stage regression (unreported), we regress

State Ownership on Collectivism along with the full set of control variables. Consistent with Boubakri et al. (2016), we find that Collectivism is positively and significantly associated with State Ownership . We use two tests to check the validity of our instrument.

We conduct an F-test of the excluded exogenous variable. The results reject the null hypothesis of the F-test that the instrument does not explain state ownership. Also, we conduct a Kleibergen-Paap rk LM test and reject the null hypothesis that the model is under-identified at the 1% level. Model 2 shows the results of the second-stage regression.

We again find that State Ownership is significantly positively associated with corporate cash holdings.

Third, we perform a Heckman two-stage analysis to address sample selection concerns. In the first stage, we use a Probit model to predict whether governments retain control over the privatized firms. In particular, we regress Control on Collectivism , the full set of control variables, and industry and year fixed effects (as in Model 1 of Table 2.3).

This allows us to estimate the inverse Mills ratio ( Lambda ). In the second stage, we include

Lambda as an additional independent variable in the cash holdings regression. The results

25 in Model 3 show that state ownership is significantly positively associated with corporate cash holdings. In addition, Lambda loads negatively and is statistically significant at the 5% level.

Fourth, we employ propensity score matching to match firms under government control with firms not under government control based on observable firm characteristics.

In the first stage, we use the same Probit model as in the Heckman first-stage analysis. We then match state-controlled firms with the closest propensity score. In the second stage

(Model 4), we estimate the regression using the matched sample. Consistent with our main analysis, the instrumental variable analysis, and the Heckman analysis, we continue to find that the coefficient on State Ownership loads positively and significantly at the 1% level.

2.4.3 Alternative Explanations

An additional concern is that the relation between state ownership and corporate cash holdings has alternative explanations. We investigate several possibilities, which we summarize in the Internet Appendix. First, when information asymmetry is severe, outside investors may be less likely to invest in privatized firms, in which case residual state ownership is likely to be high. At the same time, it may be difficult for such firms to raise external financing, in which case they are likely to hold more cash. We test this explanation by including the ratio of research and development expenses to sales ( R&D ) as an additional control variable. This approach follows prior studies that use R&D expenditures to proxy for information asymmetry (e.g., Dittmar et al., 2003). We find that R&D is not statistically significant. More importantly, the coefficient on State Ownership remains positive and statistically significant at the 1% level.

26 Second, since Boubakri et al. (2013) find that state ownership is associated with less risk-taking, there may be a negative relation between acquisition activity and state ownership, which could result in higher cash holdings in firms with greater state ownership.

To test this possibility, we control for the ratio of acquisitions to sales. The results show that the relation between State Ownership and cash holdings is robust to adding acquisition activity as a control variable. We continue to find that State Ownership loads positively and is statistically significant at the 1% level.

Third, we include additional country-level variables to address the concern that the relationship between state ownership and cash holdings is driven by the development of the country’s stock market. Specifically, we include the ratio of market capitalization to

GDP ( Market Capitalization ). Consistent with the agency view, we continue to find that

State Ownership is significantly positively associated with corporate cash holdings. This suggests that firms with a high government stake tend to hold more cash.

Fourth, it is possible that the observed cash holdings in NPFs reflect increased profitability or a deliberate corporate decision such as borrowing or selling assets. We test these explanations by controlling for the lagged return of assets ( Lagged ROA ) to account for profitability, the change of the ratio of plant, property, and equipment to total assets

(Asset Sales ) to account for asset sales, and new debt issuance ( Debt Issuance ) to account for borrowing. Although we find that profitability is related to cash holdings, we continue to find that State Ownership is significantly positively associated with cash holdings.

Finally, we exclude firms from the utility industry because they are highly regulated and account for roughly 31% of our sample, which could be driving our results. Our main findings remain statistically unchanged when we drop the utility industry from our analyses.

27 2.4.4 Why Is State Ownership Related to More Cash Holdings? An Investigation of the

Dissipation of Cash

Our main evidence suggests a positive relation between state ownership and cash holdings. This relation, which is robust to endogeneity and alternative explanations, is consistent with the agency view of state ownership. The more interesting question is why.

To further investigate, we consider how state owners disgorge cash through their influence on firms’ investment and financial policies. More specifically, following Gao et al. (2013), we examine whether state owners use cash to increase investment (i.e., capital expenditures, acquisitions, and R&D) or increase payouts (the sum of dividends and repurchases). Table

2.5 reports the results.

Our focus is on the impact of the interaction between state ownership and cash holdings on investment and financial policies in the next year, including capital expenditures (Model 1), acquisitions (Model 2), R&D (Model 3), and payout policy (Model

4). In Model 1, the coefficient of State Ownership×Cash Holdings is positive and statistically significant at the 5% level, suggesting that higher state ownership increases the likelihood that NPFs spend cash on capital expenditures in the next period. Similarly, in

Model 2 we estimate positive and statistically significant coefficients on State

Ownership×Cash Holdings , suggesting that NPFs are more likely to spend cash on acquisition over the next period when residual state ownership is high. In Model 3, we find that the coefficient on State Ownership×Cash Holdings is negative, but statistically insignificant. This suggests that NPFs are not more likely to spend cash holdings on R&D over the next period as state ownership increases. Additionally, in Model 4, we examine whether NPFs with higher state ownership are more likely to use cash to increase their

28 payouts. The results show that the coefficient on State Ownership×Cash Holdings is negative, but statistically insignificant.

In summary, our results suggest that as state ownership increases, NPFs are more likely to use cash to increase capital expenditures or acquisitions, but not to increase R&D expenses. According to Harford et al. (2008), evidence on whether firms spend cash on capital expenditures, acquisitions, or R&D may not necessarily indicate whether these investments are optimal or not. However, prior studies (e.g., Eberhart, Maxwell, and

Siddique, 2004; Harford, 1999; Masulis, Wang, and Xie, 2007) show that acquisitions by firms with cash tend to be value-destroying, while R&D investment is value-increasing.

We therefore interpret our findings to be more consistent with the agency theory explanation.

To further confirm this interpretation, in Model 5 we examine the impact of state ownership on investment efficiency. If state ownership is associated with sub-optimal investment decisions, as suggested by agency theory and empirical evidence (e.g., Liu and

Siu, 2011; Firth, Malatesta, Xin, and Xu, 2012; Jaslowitzer et al., 2016), then we should observe lower investment efficiency associated with state ownership in NPFs. To test this prediction, we follow Jaslowitzer et al. (2016) and examine the relation between state ownership and investment sensitivity to Q. The coefficient of State Ownership×Q is negative and statistically significant at the 5% level, suggesting that higher residual state ownership in NPFs reduces investment efficiency. This is consistent with our expectation that agency costs of state ownership will be manifested by sub-optimal, politically motivated investment by SOEs.

29 Taken together, our results regarding the impact of state ownership on the dissipation of cash in NPFs suggest that NPFs with higher residual state ownership are more likely to disgorge cash in a manner consistent with efforts to achieve political goals.

Furthermore, as would be expected due to the agency costs associated with state ownership, these investment decisions are not economically efficient.

2.4.5 State Ownership and Cash Holdings: The Role of Market Power and the Banking

Sector

In this section, we explore two channels through which state-owned companies may be able to maintain high cash balances when they are less profitable and less efficient than private firms. More specifically, we examine whether market power and superior access to state-owned banks drive the positive relation between state ownership and cash holdings.

In the following empirical analysis, we interact state ownership with proxies for market power and access to government-owned banks at the country level. Table 2.6 reports the results. In Model 1, we explore the role of market power, which we measure with the ratio of sales to industry sales ( Market Share ). Model 1 indicates that the coefficient on the interaction term State Ownership×Market Share is negative but statistically insignificant.

In this regression, the coefficient on State Ownership continues to be positive and statistically significant at the 1% level. We interpret this evidence as inconsistent with the explanation that SOEs with greater market power hold more cash.

In Model 2, we explore whether superior access to state-owned banks explains the high cash balances documented in NPFs with higher residual state ownership. We measure government ownership of banks ( Government Ownership of Banks ) at the country level obtained from Barth, Caprio, and Levine (2013). We expect state-owned firms to benefit

30 from better access to external financing in countries with higher government ownership of banks. Therefore, our focus is on the interaction term between State Ownership and

Government Ownership of Banks . We find that the coefficient of State

Ownership×Government Ownership of Banks is positive and statistically significant at the

1% level. This suggests higher cash holdings of stated-owned firms in countries with greater government ownership of banks. Overall, these results are consistent with

Jaslowitzer et al. (2016), who find that state ownership is associated with external finance frictions in countries with no or limited government ownership of banks. These results suggest that state-owned companies are likely to have superior access to financing and may maintain higher cash balances through financing from state-owned banks.

2.4.6 Do NPFs and U.S. Companies Differ in their Cash Holdings?

Prior literature (e.g., Bates, Kahle, and Stulz, 2009; Graham and Leary, 2016) reports that U.S. companies doubled their cash holdings during the last three decades, and there is no consistent evidence that agency conflicts contributed to this increase. Our earlier results could reflect this temporal pattern. In this section, we investigate how NPFs differ from U.S. companies and whether there is a time trend that explains corporate cash holdings.

In Table 2.7 we re-run our main regressions using samples of both NPFs and U.S. firms and include a dummy variable, U.S. , to identify U.S. firms. In Models 1 and 2, we use the full sample of U.S. firms and NPFs. We find that the U.S. dummy loads negatively at the 1% level in both models. In Models 3 to 6, we use samples matched by (1) size and industry and (2) all explanatory variables (by applying propensity score matching). We obtain the propensity scores from a first-stage Probit model, where the dependent variable

31 is a dummy variable equal to one if a firm belongs to the NPFs sample, and zero otherwise

(i.e., a U.S. firm). This model incorporates all the explanatory variables used in the main cash holdings regression model, including industry and year effects. We then match, without replacement, each NPF with the U.S. firm with the closest score, and we re- estimate our regressions using the matched samples. Using these matched samples, in

Models 5 and 6 we continue to find negative and statistically significant coefficients on the

U.S. dummy. This again suggests that U.S. firms hold less cash than NPFs.

In Table 2.8 we explore whether the time trend, instead of state ownership, explains cash holdings. Table 2.8 reports the results relating time trend to cash holdings of U.S. firms and NPFs using the full and matched samples. Using the full sample in Models 1 and

2, the coefficients Time Trend and U.S.×Time Trend are statistically insignificant at conventional levels. Models 3 and 4 replicate the regressions using the sample matched by size and industry. We find that the interaction terms between U.S. and Time Trend are positive and significant, suggesting that the increase in cash holdings over time is more pronounced for U.S. firms. This is consistent with Figure 2. We obtain consistent evidence in Models 5 and 6 using the propensity score matched sample based on all explanatory variables.

2.4.7 Is China Different?

As discussed above, the unique political, social, economic, and cultural environment of China may contribute to differences in cash policies of Chinese NPFs compared to those in other countries. To test this possibility, we replicate Models 1-3 of

Table 2.3 after including a dummy variable, China , to designate NPFs from China. We also consider its interaction terms with State Ownership , Control , and Political to

32 investigate how China is different from other countries in terms of the effect of state ownership on cash holdings. The results, which are reported in the Internet Appendix, indicate that the coefficient on China is positive and statistically significant across the three models, suggesting that NPFs tend to hold more cash in China. More importantly, the interaction terms between China and three indicators for state ownership ( State Ownership ,

Control , and Political) are negative and statistically significant. This implies that Chinese

NPFs hold less cash as government stake increases, which is consistent with the findings of Megginson et al. (2014). These findings persist when we apply propensity score matching to match Chinese NPFs with NPFs from the other countries according to all explanatory variables.

2.4.8 The Role of Financial Crises11

To better understand the relation between state ownership and cash holdings, we investigate the role of the recent global financial crisis (2008 to 2010 period) and the ongoing Euro crisis (the period after 2010) on this relation. These crises are characterized by substantial credit contraction, which may have changed the relationship between state ownership and cash holdings. The results, which are reported in the Internet Appendix, we find that firms tend to have significantly less cash holdings during the global financial crisis period. More importantly, we find that as state ownership increases, firms tend to hold more cash during financial crises. 12 We do not document similar results when we examine

11 In additional analysis, we examine the cash holdings of rescue firms, which refer to publicly listed firms that were acquired by governments or government-controlled entities during the recent global financial crisis. In unreported results (for the sake of brevity), we find that rescue firms tend to hold less cash compared to their NPFs peers, but increase their cash holdings following acquisitions by governments or government- controlled entities. These findings reinforce the positive relation between state ownership and cash holdings. 12 This is consistent with Jaslowitzer et al. (2016) who present evidence that the financial policies of state- owned firms appear to be formulated to promote stability (especially during periods of financial crisis).

33 the effect of the Euro crisis on the relation between state ownership and cash holdings for a sample of European countries.

2.5 Additional Analysis: Institutions, State Ownership, and the Value of Cash

Since the U.S. scores highly on shareholder rights and law enforcement, Harford et al. (2008) argue that entrenched managers in the U.S. are not as entrenched as their counterparts in countries with weak shareholder protection. More specific to our study, prior literature examines how country-level shareholder protection (Pinkowitz et al., 2006) interacts with firm-level agency problems, and thus affects the value of cash (Kalcheva and

Lins, 2007). Because existing research documents that the strength of a country’s institutions affects the value of firms’ cash holdings, we expect to observe cross-country differences in the value of cash holdings. In this section, we examine the effect of institutions on the relation between state ownership and the value of cash holdings. In particular, we regress firm value, as captured by Tobin’s Q ( Tobin’s Q ), on institutional factors ( Institution ), interactions between the institutional factors, cash holdings, and state ownership, and the full set of controls. Specifically, we follow Kalcheva and Lins (2007) and estimate variants of the following regression (superscripts omitted for simplicity):

34 = + ℎ + ℎ ×

+ ℎ + ℎ × ℎ

+ ℎ × ℎ × +

+ + + ℎ

+ + ℎ +

+ ℎ + +

+ +

+ . 2

As in the cash holding regression, we report significance levels based on robust standard errors clustered at the firm level. The institutional characteristics we consider are

(1) shareholder protection ( Shareholder Rights ), (2) freedom of the press ( Press Freedom ), and (3) political accountability ( Democracy , Judicial Independence, Checks, and

Disclosure by Politicians ). We discuss each of these factors in turn.

2.5.1 The Role of Shareholder Protection

Table 2.9 presents results of the valuation model using Shareholder Rights, which is the revised anti-director rights index of Djankov et al. (2008). A higher anti-director rights score indicates stronger investor protection. To facilitate comparison with Kalcheva and Lins (2007), in Model 1 we regress Tobin’s Q only on Cash Holdings , its interaction with Shareholder Rights , and the controls. We find that the coefficient on the interaction between Shareholder Rights and Cash Holdings is positive but not statistically significant. 13

13 Note that we base our measure of shareholder protection on Djankov et al.’s (2008) updated index rather than the index used in Dittmar et al. (2003) and Kalcheva and Lins (2007). Thus, our results may differ from these papers because of differences in the underlying data.

35 In Model 2, we assess the interaction between firm- and country-level governance on the value of cash holdings. Specifically, we regress Tobin’s Q on State Ownership , Cash

Holdings×State Ownership , Cash Holdings×State Ownership×Shareholder Rights , and the controls . Focusing on the interactions, we find that the coefficient on Cash

Holdings×State Ownership is negative and statistically significant, suggesting that outside investors discount the value of cash holdings in firms with high government stakes. The coefficient on the three-way interaction between Cash Holdings, State Ownership, and

Shareholder Rights is positive and significant. This indicates that when shareholder protection is poor, investors further discount the value of cash holdings in firms with more state ownership. These results are economically significant. For a firm with mean cash holdings of 0.12, an increase in state ownership from the 25 th percentile (0.00) to the 75 th percentile (0.51) and a decrease in shareholder protection from 6 to 1 are associated with a

0.05 (0.164×0.12×(0.51-0)×(1-6)) reduction in Tobin’s Q . Since the mean Tobin’s Q is

1.34, this corresponds to a 4% (0.05/1.34) decline in Tobin’s Q .

In Model 3, we re-run the specification in Model 2 using Control rather than State

Ownership . Consistent with the results in Model 2, we find that the coefficient on Cash

Holdings×Control is negative and statistically significant while the coefficient on the three-way interaction between Cash Holdings, Control, and Shareholder Rights is positive and significant. These results are also economically significant. Compared to non–state- controlled firms, for a state-controlled firm with mean cash holdings of 0.12, a decrease in shareholder protection from 6 to 1 is associated with a 7% (0.155×0.12×(1-6)/1.34) decline in Tobin’s Q .

36 In Model 4, we replace State Ownership in Model 2 with Political and continue to find that the two-way interaction between Cash Holdings and Political is statistically negatively related to firm value. The three-way interaction has a positive and significant coefficient. Economically, the results suggest that compared with non–politically connected firms, for a politically connected firm with mean cash holdings of 0.12, a decrease in shareholder protection from 6 to 1 will lead to an 8% (0.186×0.12×(1-6)/1.34) decrease in Tobin’s Q .

Taken together, the results in Table 2.9 show that while a NPF’s value is lower when the state holds more cash, consistent with the agency perspective, this negative effect is less pronounced in countries with greater shareholder protection. More specifically, our results suggest that the strength of shareholder protection affects the extent of agency conflicts between politicians and minority investors: in countries with weak shareholder protection, the value of cash holdings is discounted more by outside investors. These findings highlight the importance of considering a country’s institutional environment when examining agency conflicts between the state and minority investors.

2.5.2 The Role of Freedom of the Press

We assess the role of freedom of the press in Table 2.10 by replicating the regressions in Table 2.9 (after replacing Shareholder Rights with Press Freedom ). In

Model 1, we find that the coefficient on Cash Holdings×Press Freedom is positive and statistically significant at the 10% level, suggesting that cash holdings are more valuable in countries with independent media.

In Model 2, we regress Tobin’s Q on interactions among Cash Holdings, State

Ownership , and Press Freedom . As before, the coefficient on Cash Holdings is positive

37 and statistically significant, while the coefficient on Cash Holdings×State Ownership is negative and insignificant. Moreover, the coefficient on the three-way interaction between

Cash Holdings, State Ownership, and Press Freedom is positive and significant. This result shows that active and independent media increase the value of cash holdings for firms with more residual state ownership. Indeed, for a firm with mean cash holdings of 0.12, an increase in state ownership from the 25th percentile (0.00) to the 75 th percentile (0.51) and a decrease in press freedom from 75 (free) to 25 (less free) will lead to a 12%

(0.051×0.12×(0.51-0)×(25-75)/1.34) reduction in Tobin’s Q .

In Model 3 (Model 4), we regress Tobin’s Q on interactions between Cash Holdings,

Control (Political ), and Press Freedom . The results are consistent with those in Model 2.

In particular, compared to non–state-controlled firms, for a state-controlled firm with mean cash holdings of 0.12, a decrease in press freedom from 75 (free) to 25 (less free) will lead to a 17% (0.038×0.12×(25-75)/1.34) reduction in Tobin’s Q . Similarly, compared with non–politically connected firms, for a politically connected firm with mean cash holdings of 0.12, a decrease in press freedom from 75 (free) to 25 (less free) will lead to a 13%

(0.030×0.12×(25-75)/1.34) reduction in Tobin’s Q .

Overall, the results in Table 2.10 indicate that in countries with independent media, and hence more transparent environments, cash is valued more by outside shareholders of

NPFs with residual state ownership. These results are consistent with independent media reducing the scope for private benefits extraction in NPFs.

2.5.3 The Role of Political Accountability

In Table 2.11, we replicate the analysis in Table 2.9 using several measures of political accountability: Checks , Democracy , Judicial Independence , and Disclosure by

38 Politicians . Political accountability represents a more direct channel through which managers of NPFs are likely to be subject to external discipline. Panel A reports results using the extent of checks and balances ( Checks ) in a country. Model 1 shows that the coefficient on Cash Holdings×Checks is positive and statistically significant at the 1% level, which suggests that cash holdings are more valuable in countries with more checks and balances. In Model 2, we regress Tobin’s Q on interactions between Cash Holdings,

State Ownership , and Checks . We find that the coefficient on State Ownership×Cash

Holdings is negative but not statistically significant. The three-way interaction between

Cash Holdings, State Ownership, and Checks has a positive and significant coefficient.

When we replace State Ownership with Control and Political in Models 3 and 4, respectively, we continue to find that politicians may be able to extract private benefits more easily in countries with fewer checks and balances. This leads outside investors to more substantially discount the value of cash holdings.

Panel B of Table 2.11 presents the regression results using Democracy . Consistent with our findings in Panel A, the results for Democracy suggest a positive relation between political accountability and the market value of cash holdings by NPFs.

Panel C of Table 2.11 further assesses the role of political accountability by using

Judicial Independence. In Model 1, we find that the coefficient on Cash Holdings×Judicial

Independence is positive and statistically significant. This suggests that the value of cash holdings is higher in countries with more judicial independence. In Model 2, we regress

Tobin’s Q on interactions between Cash Holdings, State Ownership , and Judicial

Independence . We find that the coefficient on State Ownership×Cash Holdings is negative and significant. The three-way interaction between Cash Holdings, State Ownership, and

39 Judicial Independence has a positive and significant coefficient. In Models 3 and 4, we replicate the regression in Model 2 after replacing State Ownership with Control and

Political, respectively. The results for Models 3 and 4 are identical to those of Model 2.

Panel D of Table 2.11 presents findings using Disclosure by Politicians . We find that the coefficient on State Ownership×Cash Holdings is negative and significant. The interaction between Cash Holdings, State Ownership, and Disclosure by Politicians has a positive and significant coefficient 14 .

In summary, Table 2.11 provides consistent evidence that while the cash holdings of NPFs are discounted by external investors, an institutional environment of greater political accountability helps alleviate concerns that managers of firms with residual state ownership will expropriate minority shareholders.

2.6 Conclusion

In this study, we investigate the impact of state ownership on corporate cash policy.

Using a unique sample of 578 NPFs from 59 countries over the period 1981–2014, we find, consistent with agency theory, that state ownership is positively related to corporate cash holdings. This result is robust to endogeneity tests, alternative measures of cash holdings, various measures of state ownership, and additional control variables. We further find that country-level institutions affect the agency conflicts between politicians and minority shareholders. Building on Kalcheva and Lins (2007) and Pinkowitz et al. (2006), we show that stronger institutional environments moderate the negative relation between residual state ownership and the value of cash holdings by NPFs. In particular, when the government has a higher stake in partially privatized firms, the value of cash holdings by

14 We further use two enforcement indices that measure registrar strength and checking unit strength, respectively, and find the results are in line with those in Panel D of Table 2.11.

40 these firms is lower in countries with weaker investor protection, lower press freedom, and reduced political accountability. Overall, this essay contributes to the corporate finance literature by providing evidence that cash holdings are affected by state ownership and by broadening our understanding of how the institutional environment affects contracting decisions.

We provide evidence consistent with the premise that oversight resulting from a more effective institutional environment (e.g., independent media, stronger legal protection, greater political accountability) may curb the state’s propensity to expropriate wealth from SOEs and therefore may contribute to higher market values of cash holdings by NPFs. Future research should focus on whether investment banks and other financial intermediaries also serve an important monitoring function. For example, Leland and Pyle

(1977), Booth and Smith (1986), Titman and Trueman (1986), Megginson and Weiss

(1991), Chemmanur and Fulghieri (1994), and Fang (2005), among others, identify the significant contribution of financial intermediaries as information producers. This monitoring by underwriters, coupled with the increasingly important impact of coverage by top analysts affiliated with the most prestigious banks (Mola and Loughran, 2004;

Huang and Zhang, 2011; Liu and Ritter, 2011), can be instrumental in reducing information asymmetry and managing stockholder/manager conflicts.

Moreover, focusing on the value of the “certification” provided by prestigious underwriters, authors such as Logue (1973), Johnson and Miller (1988), Carter and

Manaster (1990), Carter, Dark, and Singh (1998), Chemmanur and Fulghieri (1994), Chen and Ritter (2000), Gao and Ritter (2010), and Golubov, Petmezas, and Travlos (2012) recognize how an investment bank’s desire to protect its reputational capital instills

41 incentives for it to diligently scrutinize the financial and managerial activities of the issuing firm. Golubov et al. (2012) further contend that the motivations of the intermediary to more meticulously evaluate the issuer are intensified in transactions involving larger reputational exposure. Since Megginson and Netter (2001) and Boutchkova and Megginson (2000) report that privatizations are frequently the largest and most highly visible equity offerings in a nation’s capital market history, the reputation stakes during share-issue privatizations should be high and the investment bank should thus be compelled to rigorously evaluate the issuing NPF. Accordingly, future research should consider the additional oversight brought on by the capital-raising process as another factor in the institutional environment that affects the NPF’s cash management policies.

42

Table 2. 1 Distribution of the Sample of Newly Privatized Firms

This table presents the distribution of the sample of 578 privatized firms. The full sample contains 5,254 firm-year observations. Panel A reports sample distribution by privatization year. Panel B reports sample distribution by industry based on industry groupings in Campbell (1996). Panel C reports sample distribution by region based on World Bank group classification.

Panel A: By privatization year Panel B: By industry Year Obs. % Firms % Industry Obs. % Firms % 1981 15 0.29 1 0.17 Basic industry 837 15.93 96 16.61 1983 7 0.13 1 0.17 Capital goods 260 4.95 26 4.50 1984 4 0.08 1 0.17 Construction 380 7.23 44 7.61 1985 20 0.38 2 0.35 Consumer durable 420 7.99 45 7.79 1986 44 0.84 4 0.69 Food and tobacco 255 4.85 35 6.06 1987 121 2.30 8 1.38 Leisure 143 2.72 13 2.25 1988 96 1.83 8 1.38 Petroleum 360 6.85 39 6.75 1989 166 3.16 12 2.08 Services 84 1.60 10 1.73

43 1990 150 2.85 15 2.60 Textiles and trade 139 2.65 18 3.11

1991 369 7.02 33 5.71 Transportation 713 13.57 74 12.80 1992 427 8.13 45 7.79 Utilities 1604 30.53 170 29.41 1993 313 5.96 36 6.23 Other 59 1.12 8 1.38 1994 438 8.34 43 7.44 Total 5,254 100.00 578 100.00 1995 350 6.66 39 6.75 1996 301 5.73 33 5.71 Panel C: By region 1997 488 9.29 49 8.48 Region (countries) Obs. % Firms % 1998 359 6.83 39 6.75 Africa & the Middle East (9) 196 3.73 22 3.81 1999 272 5.18 28 4.84 East and South Asia & the Pacific (15) 2,419 46.04 249 43.08 2000 214 4.07 26 4.50 Latin America & the Caribbean (8) 457 8.70 57 9.86 2001 117 2.23 13 2.25 Europe & Central Asia (27) 2,182 41.53 250 43.25 2002 101 1.92 10 1.73 Total (59) 5,254 100.00 578 100.00 2003 141 2.68 16 2.77 2004 85 1.62 12 2.08 2005 106 2.02 12 2.08 2006 228 4.34 28 4.84

2007 118 2.25 15 2.60 2008 44 0.84 6 1.04 2009 27 0.51 6 1.04 2010 56 1.07 14 2.42 2011 17 0.32 6 1.04 2012 33 0.63 5 0.87 2013 20 0.38 7 1.21 2014 7 0.13 5 0.87 Total 5,254 100.00 578 100.00 44

Table 2. 2. Descriptive Statistics and Pearson Correlation Matrix

This table presents descriptive statistics for the key variables in our analyses. Our sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. All financial variables are winsorized at the 1% level in both tails of the distribution. Variable definitions and sources are provided in the Appendix. Panel A reports summary statistics for the firm-level variables. Panel B reports pairwise correlations coefficients between firm-level variables. Panel C reports means of country-level variables across years. Superscripts a, b, and c indicate significance at the 1%, 5%, and 10% levels, respectively.

Panel A: Summary statistics Variables Obs. Min Mean Median Max Std.dev Tobin’s Q 5,254 0.37 1.34 1.13 4.88 0.74 Cash Holdings 5,254 0.00 0.12 0.09 0.53 0.11 State Ownership 5,254 0.00 0.25 0.12 1.00 0.28 Control 5,254 0.00 0.29 0.00 1.00 0.45 Political 2,943 0.00 0.33 0.00 1.00 0.47 Size 5,254 2.90 7.59 7.63 12.16 1.95 Leverage 5,254 0.00 0.17 0.14 0.59 0.15 Net Working Capital 5,254 -0.52 -0.05 -0.05 0.34 0.15 Cash Flow 5,254 -0.16 0.10 0.09 0.34 0.08 Capital Expenditures 5,254 0.00 0.07 0.06 0.25 0.05 Sales Growth 5,254 -0.45 0.12 0.08 1.27 0.25 Dividend 5,254 0.00 0.51 1.00 1.00 0.50 Cash Flow Volatility 5,254 0.00 0.04 0.03 0.21 0.04

45

Panel B: Pearson correlation matrix Variables 1 2 3 4 5 6 7 8 9 10 11 12 1. Tobin’s Q 1 2. Ln (Cash Holdings) 0.11 a 1 3. State Ownership -0.03 0.14 a 1 4. Control -0.03 0.11 a 0.87 a 1 5. Political -0.04 c 0.12 a 0.87 a 0.96 a 1 6. Size -0.10 a -0.10 a 0.05 c -0.00 -0.02 1 7. Leverage -0.08 a -0.34 a -0.15 a -0.14 a -0.08 a 0.33 a 1 8. Net Working Capital -0.07 a -0.07 a -0.06 a -0.02 -0.04 -0.13 a -0.14 a 1 9. Cash Flow 0.16 a 0.19 a -0.02 0.02 -0.03 0.11 a -0.14 a 0.21 a 1 10. Capital Expenditures 0.39 a -0.06 b 0.08 a 0.06 b 0.04 c 0.15 a 0.11 a -0.11 a 0.24 a 1 11. Sales Growth 0.07 a 0.10 a 0.09 a 0.08 a 0.08 a -0.07 a -0.05 c 0.01 0.17 a 0.08 a 1 12. Dividend 0.02 0.04 c 0.09 a 0.05 b 0.07 b 0.16 a -0.05 b 0.15 a 0.22 a 0.12 a -0.00 1 13. Cash Flow Volatility 0.18 a 0.06 b -0.09 a -0.09 a -0.10 a -0.23 a -0.07 a -0.06 * -0.04 c -0.12 a 0.00 -0.23 a 46

Panel C: Means of country-level variables across years Disclosure Shareholder Press Judicial Country Checks Democracy by Rights Freedom Independence Politician Argentina 2 55.13 4 4.55 1 1 Australia 4 83.7 4.58 6 1 1 Austria 2.5 79.16 4.1 5.45 0.67 1 Belgium 3 88.97 4.38 5.96 0.67 1 Brazil 5 60.47 4.44 4.74 0.67 1 Bulgaria 3 62 2 5.5 . 1 Chile 4 71.61 3.06 4.69 0.67 1 China 1 16.6 1 1.28 0 0 Colombia 3 42.68 4.04 4.38 0.33 1 Croatia 2.5 59.46 3.31 5.21 . 1 Czech Republic 4 79.27 5.33 5.29 . 1 Denmark 4 89.47 5.6 6 1 0 Egypt 3 32.65 2 1.96 0.67 1 Finland 3.5 89.18 4.1 6 1 0 France 3.5 78.46 4.18 5.79 0.33 1 Germany 3.5 83.73 4.65 5.69 1 1 Ghana 5 71.5 3 3 1 1 Greece 2 68.95 3 6 0.67 1 Hong Kong 5 19.61 . 2.18 . 1 Hungary 2 74.06 3.57 5.9 . 1 India 5 63.16 10.29 6 1 1 Indonesia 4 46.3 2.7 4.34 1 1 Ireland 5 83.87 5.58 6 1 1 Israel 4 70.82 4.05 5.99 1 1 Italy 2 68.77 3.42 5.28 0.67 1 Jamaica 4 83.3 4 4.03 . 1 Japan 4.5 79.51 3.12 5 0.67 1 Jordan 1 37.22 1 3.61 1 1 Kazakhstan 4 18.4 1 1.72 . 1 Kenya 2 40.5 3 4.76 1 1 Korea (South) 4.5 70 2.95 5.89 0.67 1

47

Panel C—Continued Disclosure Shareholder Press Judicial Country Checks Democracy by Rights Freedom Independence Politician Latvia 4 77.4 5.25 5 . 1 Lithuania 4 80.56 3.18 5.46 . 1 Malaysia 5 33.24 4 3.56 1 0 Mexico 3 47.59 4.8 5.78 0.33 1 Morocco 2 35.17 1 4.71 . 1 Netherlands 2.5 87.24 5.87 6 0.67 1 New Zealand 4 88.57 3.5 5.91 1 1 Nigeria 4 46.5 4 3.4 1 1 Norway 3.5 91.2 4.92 6 1 0 Pakistan 4 37.86 2.4 1.94 1 1 Peru 3.5 58.4 4 4.89 1 1 Philippines 4 57.87 4 5 1 1 Poland 2 77.15 4.18 6 . 1 Portugal 2.5 84.05 2.58 5.92 0.67 1 Romania 5 57.67 5.38 6 . 1 Russia 4 19.26 2.32 2.27 . 1 Singapore 5 32.91 2 2.06 1 0 Slovakia 3 78.07 4.46 6 . 1 South Africa 5 70.86 2 4.67 1 1 Spain 5 77.51 3.67 5.94 0.67 1 Sri Lanka 4 29.11 3.65 3.52 . 1 Sweden 3.5 90.08 4.38 6 1 1 Switzerland 3 89.06 3.13 6 0.67 1 Thailand 4 53.21 4.43 4.29 0.67 1 Turkey 3 44.52 2.85 4.25 1 1 U.K. 5 80.83 3.09 5.89 1 1 Venezuela 1 43.89 2.86 4.12 0.33 1 Zimbabwe 4 15.33 2 1.43 1 0 Average 3.13 54.96 3.49 4.28 0.65 0.71

48

Table 2.3 State Ownership and Cash Holdings

The dependent variable in Models 1, 2, and 3 is the natural log of Cash Holdings , which is calculated as cash and marketable securities divided by total sales. The main independent variable in Model 1 is State Ownershi p, which is defined as the percentage of shares held by a government. The main independent variable in Model 2 is Control, which is a dummy variable equal to 1 for firms in which the state maintains control (i.e., the government holds more than 50% of the firm’s shares) following partial privatization, and 0 otherwise. The main independent variable in Model 3 is Political, which is a dummy variable equal to 1 for firms that are politically connected as defined by Faccio (2006), and 0 otherwise. Models 4, 5, and 6 use three alternative measures of cash holdings. The dependent variable in Model 4 is the natural log of Cash/Assets , which is calculated as cash and marketable securities divided by total assets. The dependent variable in Model 5 is natural log of Cash/Net Assets , which is calculated as cash and marketable securities divided by assets less cash holdings. The dependent variable in Model 6 is Industry-Adjusted Cash , which is calculated as the difference between a firm’s actual cash holdings and its industry median, based on Campbell’s (1996) industry classification. Model 7 replaces country-level variables with country fixed effects. The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Three measures of state ownership Alternative dependent variable Ln (Cash Holdings) Industry- Country fixed Cash/Net State Control Political Cash/Assets Adjusted effects

49 Assets Cash (1) (2) (3) (4) (5) (6) (7) State Ownership 0.587*** 0.359*** 0.458*** 0.344*** 0.301** (4.500) (3.147) (3.817) (2.994) (2.225) Control 0.307*** (4.213) Political 0.225*** (2.708) Size -0.036 -0.031 -0.058** 0.051** -0.021 0.073*** -0.041* (-1.538) (-1.336) (-2.168) (2.420) (-0.949) (3.806) (-1.707) Leverage -1.225*** -1.260*** -1.380*** -1.744*** -0.579** -1.856*** -0.014 (-4.356) (-4.518) (-4.440) (-6.928) (-2.367) (-7.484) (-0.148) Net Working Capital -0.379*** -0.385*** -0.373** -0.993*** -1.250*** -0.042 -0.250** (-3.037) (-3.070) (-2.320) (-4.611) (-6.529) (-0.382) (-1.982) Cash Flow 2.376*** 2.393*** 2.133*** 3.309*** 4.868*** 1.291*** 2.494*** (8.613) (8.650) (6.737) (8.991) (15.451) (5.474) (9.455) Capital Expenditure 0.539** 0.570** 0.240 -1.605*** 0.194 -0.507** 0.273 (2.311) (2.483) (0.965) (-3.228) (0.412) (-2.541) (1.145) Sales Growth -0.009 -0.003 0.021 0.154** 0.014 0.178*** -0.136* (-0.114) (-0.040) (0.237) (2.254) (0.188) (2.723) (-1.737) Dividend -0.046 -0.042 -0.077 0.032 0.015 0.016 0.067 (-0.787) (-0.721) (-1.194) (0.602) (0.266) (0.311) (1.169)

Cash Flow Volatility 1.501* 1.578* 0.895 1.812** 1.153 2.552*** 1.651* (1.752) (1.840) (0.775) (2.356) (1.252) (3.488) (1.919) Shareholder Rights 0.003 -0.006 -0.020 -0.011 -0.021 -0.015 (0.115) (-0.201) (-0.620) (-0.427) (-0.778) (-0.592) Private Credit 0.003*** 0.003*** 0.002* 0.001 0.002 0.000 (3.136) (3.277) (1.877) (0.816) (1.621) (0.469) Constant -1.884*** -1.843*** -1.209*** -2.762*** -2.500*** -0.339 -1.760*** (-6.956) (-7.260) (-2.830) (-11.176) (-8.855) (-1.522) (-5.890) Country fixed effects No No No No No No Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Yes Observations 5,254 5,254 2,943 5,254 5,254 5,254 5,254 Adj. R 2 0.233 0.229 0.216 0.214 0.278 0.118 0.259

50

Table 2. 4 Endogeneity of State Ownership

The dependent variable is the natural log of Cash Holdings , which is calculated as cash and marketable securities divided by total sales. The main independent variable is State Ownership , which is defined as the percentage of shares held by a government. Model 1 reports results from a firm fixed effect regression. The remainder of the table reports results from additional tests that address the endogeneity of state ownership using instrumental variable regression (Model 2), Heckman selection model (Model 3), and Propensity score matching (Model 4). The first-stage regression results predicting state ownership are unreported. Model 2 reports the second-stage regression of the natural log of Cash Holdings on fitted-values of State Ownership . In the first-stage regression (unreported) the instrument for State Ownership is Collectivism , which is equal to 100 minus the value of Hofstede’s (2001) individualism index. In Model 3, the Heckman sample selection model controls for the inverse Mills ratio ( Lambda ). In Model 4, the propensity score matching analysis uses a sample of state-controlled firms matched to control firms with the closest propensity score. The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t- statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Firm Fixed Effects Instrumental Variable Heckman Sample Selection Propensity Score Ln ( Cash Holdings ) Regression Regression -- 2nd stage Model -- 2nd stage Matching (1) (2) (3) (4) State Ownership 0.360* 2.640*** 0.564*** 0.356*** 51 (1.862) (3.678) (3.830) (2.873) Size -0.073 0.036 0.068*** 0.096*** (-1.310) (1.373) (3.512) (4.428) Leverage -0.973*** -1.380*** -1.855*** -1.861*** (-4.191) (-3.922) (-7.436) (-6.960) Net Working Capital -0.166* 0.024 -0.026 -0.109 (-1.742) (0.195) (-0.230) (-0.751) Cash Flow 0.950*** 1.101*** 1.161*** 1.429*** (4.745) (3.937) (4.796) (5.379) Capital Expenditure -0.362** -0.764*** -0.478** -0.392 (-2.165) (-2.599) (-2.301) (-1.423) Sales Growth 0.183*** 0.094 0.230*** 0.217*** (3.069) (1.136) (3.331) (2.682) Dividend 0.103** -0.133 -0.009 -0.022 (2.560) (-1.642) (-0.177) (-0.345) Cash Flow Volatility 0.924 2.690*** 2.447*** 3.296*** (1.346) (2.904) (3.167) (3.957) Shareholder Rights 0.083* 0.003 0.020

(1.892) (0.103) (0.617) Private Credit -0.003** 0.001 0.001 -0.000 (-2.131) (0.549) (0.588) (-0.482) Lambda -0.109** (-2.085) Constant -1.856*** -2.910*** -2.837*** -2.935*** (-4.180) (-11.314) (-10.242) (-10.177) Industry fixed effects No Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 5,254 4,993 4,993 4,072 Adj. R 2 0.070 0.102 0.204 0.191

52

Table 2. 5 State Ownership and Cash Holdings: Disgorging Cash via Investm ent and Payout This table examines the relation between state ownership and investment and payout decisions using a logit regression in Model 1 to 4. Increase Capital Expenditures is a dummy variable equal to 1 if a firm increases capital expenditures in the next year, and 0 otherwise. Increase Acquisition is a dummy variable equal to 1 if a firm increases acquisitions in the next year, and 0 otherwise. Increase R&D is a dummy variable equal to 1 if a firm increases R&D investment in the next year. Increase Dividends & Repurchases is a dummy variable equal to 1 if a firm increases the sum of dividends and repurchases in the next year. Model 5 examines the impact of state ownership on investment efficiency captured with the sensitivity of firm investment to Q. The dependent variable in Model 5 is Investment Expenditure , measured as the sum of the yearly growth in property, plant, and equipment, plus growth in inventory, plus R&D expenditure, all deflated by the lagged book value of assets. The Appendix provides variables definitions and sources. Regressions include industry dummy variables based on industry groupings in Campbell (1996) and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Payout Investment Decisions Sensitivity Policy of Increase Increase Investment Capital Increase Increase Variables Dividends & Expenditures Expenditur Acquisition R&D Repurchases to Q es (1) (2) (3) (4) (5) State Ownership -0.108 -0.125 -0.261** 0.226 0.024** (-1.113) (-0.597) (-2.262) (1.555) (2.259) Cash Holdings 0.472* -0.240 0.695* 0.790** (1.746) (-0.520) (1.905) (2.435) State Ownership× 1.350** 2.414** -0.978 -0.798 Cash Holdings (2.115) (2.200) (-1.048) (-0.994)

State Ownership×Q -0.012** (-1.969) Size -0.007 0.169*** 0.154*** 0.061*** 0.004*** (-0.702) (6.271) (8.329) (3.661) (5.387) Cash Flows 0.369 -0.378 0.433 1.122*** 0.054*** (1.383) (-0.719) (1.042) (3.264) (7.403) Leverage 0.011 -0.354 0.317 -0.379* 0.043*** (0.081) (-1.189) (1.596) (-1.812) (3.855) MTB 0.038 -0.117 -0.154 0.058 0.009** (0.468) (-1.126) (-1.391) (0.693) (2.485) Net Working 0.017 1.451*** 0.249 0.512*** -0.021*** Capital (0.117) (4.372) (1.074) (2.682) (-4.433)

Shareholder Rights 0.039*** -0.026 -0.086*** 0.118*** 0.000 (3.184) (-0.803) (-3.588) (5.888) (0.388) Private Credit -0.001** 0.002** 0.001 0.004*** -0.000*** (-2.108) (2.103) (1.197) (5.348) (-9.353) Constant -0.220* -2.290*** -1.995*** -2.156*** 0.029*** (-1.875) (-9.353) (-10.701) (-12.076) (5.217) Industry fixed Yes Yes Yes Yes Yes effects Year fixed effects Yes Yes Yes Yes Yes Observations 4,351 4,351 4,351 4,351 5,254 Pseudo/Adj. R2 0.03 0.07 0.06 0.06 0.091

53

Table 2. 6 State Ownership and Cash Holdings: The Role of Market Power and the Banking Sector

This table shows regression results from examining the role of market power and access to state-owned banks on the relation between state ownership and cash holdings. The dependent variable is the natural log of Cash Holdings , calculated as cash and marketable securities divided by total sales. Market Share is defined as the ratio of a firm's sales to its industry sales. Government Ownership of Banks is the extent to which the banking system’s assets are owned by government. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include industry dummy variables based on industry groupings in Campbell (1996) and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively.

Dependent variable: Government Market Share Ownership Ln (Cash Holdings) of Bank (1) (2) State Ownership 0.589*** 0.152 (4.460) (1.306) Market Share 1.366*

(1.957) State Ownership×Market Share -1.077

(-1.127) Government Ownership of Bank -0.174* (-1.818) State Ownership×Government Ownership of Bank 0.713*** (4.123) Size -0.039* -0.027*** (-1.650) (-2.594) Leverage -1.222*** -1.054*** (-4.331) (-7.516) Net Working Capital -0.380*** -0.373*** (-3.045) (-5.518) Cash Flow 2.372*** 2.427*** (8.590) (17.438) Capital Expenditures 0.533** 0.549*** (2.282) (3.831) Sales Growth -0.025 -0.065 (-0.342) (-0.973) Dividend -0.045 -0.015 (-0.770) (-0.458) Cash Flow Volatility 1.460* 1.467*** (1.690) (2.925) Shareholder Rights 0.003 -0.009 (0.111) (-0.648)

Private Credit 0.003*** (3.222) Constant -2.323*** -2.070*** (-9.501) (-13.878) Industry fixed effects Yes Yes Year fixed effects Yes Yes Observations 5,254 5,159 Adj. R 2 0.234 0.228

54

Table 2.7 Comparison of Cash Holdings of U.S. Firms and NPFs

This table shows regression results comparing cash holdings of U.S. firms and newly privatized firms (NPFs). The dependent variable is the natural log of Cash Holdings , calculated as cash and marketable securities divided by total sales. The independent variable U.S. is a dummy variable equal to 1 for a U.S. firm, and 0 otherwise. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include industry dummy variables based on industry groupings in Campbell (1996) and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Sample Matched by Sample Matched by All Full Sample Ln (Cash Holdings) Size and Industry Explanatory Variables (1) (2) (3) (4) (5) (6) U.S. -0.537*** -0.839*** -0.824*** -0.747*** -0.466*** -0.770*** (-11.036) (-8.109) (-13.532) (-15.397) (-7.547) (-7.349) Size -0.176*** -0.046*** -0.113*** -0.016* -0.148*** 0.015 (-29.166) (-7.159) (-7.546) (-1.763) (-9.981) (0.890) Leverage -1.855*** -1.316*** -2.285*** (-35.036) (-13.906) (-11.694) Net Working Capital -0.238*** -1.244*** -0.531*** (-7.229) (-10.238) (-4.155) 55 Cash Flow 0.371*** 1.777*** 2.424*** (13.349) (9.573) (11.094) Capital Expenditures -1.007*** -1.152*** -2.727*** (-6.586) (-3.858) (-6.327) Sales Growth 0.139*** 0.273*** 0.210*** (16.575) (6.345) (4.450) Dividend -0.516*** -0.300*** -0.313*** (-18.229) (-10.142) (-6.292) Cash Flow Volatility 0.233*** 1.471*** 0.801*** (8.059) (5.553) (4.907) Shareholder Rights 0.079*** 0.053*** 0.052** (2.854) (4.674) (2.021) Private Credit 0.004*** 0.002*** 0.003*** (4.089) (5.735) (2.903) Constant -1.256*** -2.026*** -1.732*** -2.269*** -1.468*** -2.244*** (-20.215) (-12.576) (-14.300) (-17.202) (-12.205) (-10.127) Industry fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Observations 103,725 103,725 10,508 10,508 10,508 10,508 Adj. R 2 0.058 0.246 0.081 0.256 0.074 0.244

Table 2. 8 Time Trend of Cash Holdings of U.S. Firms and NPFs

This table shows regression results comparing cash holdings of U.S. firms and newly privatized firms (NPFs). The dependent variable is the natural log of Cash Holdings , calculated as cash and marketable securities divided by total sales. The independent variable U.S. is a dummy variable equal to 1 for a U.S. firm, and 0 otherwise. Time Trend is defined as the fiscal year minus 1994, which is the first year of our sample period. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include industry dummy variables based on industry groupings in Campbell (1996) and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Sample Matched by Sample Matched by All Full Sample Ln (Cash Holdings) Size and Industry Explanatory Variables (1) (2) (3) (4) (5) (6) U.S. -0.562*** -0.970*** -1.023*** -0.961*** -0.672*** -1.010*** (-5.630) (-7.306) (-9.934) (-7.788) (-7.954) (-6.836) Time Trend -0.019 -0.017 0.013** 0.011* 0.014*** 0.009 (-0.938) (-0.838) (2.125) (1.822) (3.923) (1.407) U.S×Time Trend 0.007 0.011 0.022*** 0.015** 0.015** 0.019** (1.037) (1.619) (2.808) (1.970) (2.363) (2.273) 56 Size -0.046*** -0.017 0.014 (-7.160) (-1.112) (0.882) Leverage -1.855*** -1.323*** -2.303*** (-35.035) (-9.086) (-11.833) Net Working Capital -0.238*** -1.255*** -0.533*** (-7.220) (-7.517) (-4.192) Cash Flow 0.371*** 1.809*** 2.419*** (13.339) (8.453) (11.115) Capital Expenditures -1.008*** -1.245*** -2.777*** (-6.595) (-3.197) (-6.466) Sales Growth 0.139*** 0.260*** 0.198*** (16.571) (5.754) (4.250) Dividend -0.517*** -0.303*** -0.318*** (-18.248) (-6.346) (-6.392) Cash Flow Volatility 0.233*** 1.498*** 0.799*** (8.056) (4.907) (4.895) Shareholder Rights 0.076*** 0.049* 0.049* (2.727) (1.863) (1.894) Private Credit 0.004*** 0.002*** 0.003***

(4.020) (2.856) (2.986) Constant -1.979*** -1.586*** -2.654*** -2.787*** -2.578*** -2.640*** (-5.088) (-3.805) (-19.270) (-15.853) (-31.008) (-15.205) Industry fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Observations 103,725 103,725 10,508 10,508 10,508 10,508 Adj. R 2 0.147 0.246 0.184 0.254 0.178 0.281

57

Table 2.9 State Ownership and Cash Holdings: Is China Different?

The dependent variable in Models 1, 2, and 3 is the natural log of Cash Holdings , which is calculated as cash and marketable securities divided by total sales. The key independent variable in Model 1 is State Ownership , which is defined as the percentage of shares held by the state. The key independent variable in Model 2 is Control , which is a dummy variable equal to 1 for firms in which the state maintains control following partial privatization. The key independent variable in Model 3 is Political , which is a dummy variable equal to 1 for firms that are politically connected, as defined by Faccio (2006). Models 4 to 6 use propensity score matching. The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respe ctively. Dependent variable: Three measures of state ownership Propensity score matching Ln (Cash Holdings) State Control Political State Control Political (1) (2) (3) (4) (5) (6) State Ownership 0.607*** 0.704*** (4.418) (2.692) Control 0.364*** 0.522*** (4.303) (3.415) Political 0.264*** 0.520*** (2.781) (3.330) China 0.941*** 0.733*** 0.411* 0.809*** 0.570** 0.189 (4.699) (4.275) (1.915) (2.945) (2.328) (0.616) 58 China×State Ownership -1.089*** -1.224***

(-3.470) (-3.063) China×Control -0.509*** -0.667*** (-3.511) (-3.403) China×Political -0.295* -0.549*** (-1.864) (-2.745) Size -0.018 -0.016 -0.049* -0.061* -0.066** -0.071* (-0.723) (-0.678) (-1.730) (-1.823) (-2.020) (-1.679) Leverage -1.107*** -1.131*** -1.292*** -1.416*** -1.423*** -1.586*** (-4.040) (-4.165) (-4.244) (-3.353) (-3.441) (-3.005) Net Working Capital -0.321** -0.324** -0.350** -0.218 -0.226 -0.214 (-2.464) (-2.482) (-2.096) (-1.374) (-1.431) (-0.973) Cash Flow 2.439*** 2.463*** 2.178*** 1.134*** 1.159*** 0.906** (9.035) (9.086) (7.074) (3.455) (3.469) (2.061) Capital Expenditures 0.459* 0.476** 0.187 0.477* 0.490* 0.457 (1.929) (2.030) (0.738) (1.657) (1.723) (1.311) Sales Growth -0.039 -0.046 0.012 0.014 0.009 0.038 (-0.519) (-0.603) (0.134) (0.135) (0.085) (0.309) Dividend -0.037 -0.041 -0.075 0.125 0.120 0.148 (-0.630) (-0.708) (-1.163) (1.459) (1.440) (1.507) Cash Flow Volatility 1.494* 1.532* 0.871 -0.680 -0.609 -0.660 (1.724) (1.772) (0.747) (-0.527) (-0.469) (-0.462)

Shareholder Rights 0.073* 0.067 0.020 0.015 0.003 -0.073 (1.806) (1.640) (0.405) (0.223) (0.042) (-0.891) Private Credit 0.001 0.002 0.001 0.001 0.002 -0.000 (1.181) (1.358) (1.132) (0.511) (0.747) (-0.078) Constant -2.536*** -2.537*** -1.772*** -2.377*** -2.287*** -1.492** (-9.901) (-9.970) (-3.842) (-6.257) (-6.156) (-2.511) Industry fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Observations 5,254 5,254 2,943 2,036 2,036 1,182 Adj. R 2 0.246 0.243 0.220 0.243 0.242 0.226 59

Table 2.10 Investor Protection, State Ownership, and the Value of Cash

The dependent variable is Tobin's Q , computed as market value of equity minus book value of equity plus book value of assets, all scaled by book value of assets. Model 1 is the basic model. Models 2, 3, and 4 use State Ownership , Control , and Political as independent variables, respectively. Shareholder Rights is the revised anti-director rights index drawn from Djankov et al. (2008), which ranges from 0 to 6 (with lower scores indicating fewer shareholder rights). The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t- statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Measures of state ownership Basic Model Tobin's Q State Control Political (1) (2) (3) (4) Cash Holdings -0.122 0.972*** 0.836*** 0.574** (-0.586) (7.910) (3.464) (2.088) Cash Holdings×Shareholder Rights -0.010 (-0.182) State Ownership 0.154 (0.747) Cash Holdings×State Ownership -1.076*** (-5.598) Cash Holdings×State Ownership×Shareholder 0.164*** Rights (3.834) Control 0.083 (1.568) Cash Holdings×Control -0.786*** (-3.389) Cash Holdings×Control×Shareholder Rights 0.155*** (2.623) Political 0.081 (1.311) Cash Holdings×Political -0.794*** (-3.060) Cash Holdings×Political×Shareholder Rights 0.186** (2.504) Size -0.035** -0.040*** -0.037*** -0.024 (-2.571) (-6.972) (-2.701) (-1.484) Leverage -0.131*** -0.083*** -0.089** -0.103** (-3.370) (-3.988) (-2.255) (-2.229) Net Working Capital -0.260*** -0.237*** -0.228*** -0.211** (-3.224) (-5.938) (-2.846) (-2.300) Cash Flow 1.374*** 1.192*** 1.184*** 1.218*** (7.999) (15.092) (7.129) (6.414) Capital Expenditures 0.132** 0.122*** 0.127** 0.050 (2.410) (2.802) (2.359) (0.682) Sales Growth 0.006 -0.002 0.001 0.014 (0.186) (-0.092) (0.016) (0.402) Dividend 3.148*** 2.816*** 2.801*** 3.475*** (4.946) (8.191) (4.512) (4.694) Cash Flow Volatility -0.199 -0.112 -0.124 -0.097 (-1.523) (-1.259) (-0.971) (-0.578) Shareholder Rights 0.045** 0.032*** 0.028* 0.019

60

(2.463) (4.482) (1.715) (0.954) Private Credit 0.001* 0.001*** 0.001** 0.001* (1.878) (4.949) (2.087) (1.954) Constant 1.034*** 0.996*** 1.003*** 0.938*** (7.676) (15.617) (7.963) (4.198) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 5,254 5,254 5,254 2,943 Adj. R 2 0.183 0.187 0.191 0.201

61

Table 2. 11 Press Freedom, State Ownership, and the Value of Cash

The dependent variable is Tobin's Q , computed as market value of equity minus book value of equity plus book value of assets, all scaled by book value of assets. Model 1 is the basic model. Models 2, 3, and 4 use State Ownership , Control , and Political as independent variables, respectively. Press Freedom is 100 minus the original Freedom House index. Higher values of Press Freedom indicate that a country’s media are more independent. The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent variable: Basic Measures of state ownership Tobin's Q Model State Control Political (1) (2) (3) (4) Cash Holdings -0.056 0.734** 0.769*** 0.547* (-0.118) (2.316) (2.865) (1.714) Cash Holdings×Press Freedom 0.018** (1.977) State Ownership -0.040 (-0.396) Cash Holdings×State Ownership -2.016* (-1.844) Cash Holdings×State Ownership×Press Freedom 0.051*** (2.651) Control -0.022 (-0.230) Cash Holdings×Control -1.538*** (-2.640) Cash Holdings×Control×Press Freedom 0.038*** (3.482) Political 0.026 (0.220) Cash Holdings×Political -1.075 (-1.622) Cash Holdings×Political×Press Freedom 0.030** (2.431) Size -0.017 -0.015 -0.015 -0.005 (-1.100) (-0.958) (-0.931) (-0.292) Leverage -0.106 -0.119 -0.115 -0.120 (-0.683) (-0.761) (-0.739) (-0.600) Net Working Capital -0.221** -0.220** -0.216** -0.213** (-2.421) (-2.482) (-2.458) (-2.146) Cash Flow 0.953*** 0.940*** 0.952*** 1.004*** (6.188) (6.141) (6.289) (5.491) Capital Expenditures -0.179 -0.173 -0.178 -0.097 (-1.513) (-1.484) (-1.529) (-0.606) Sales Growth 0.160*** 0.163*** 0.163*** 0.091 (2.613) (2.692) (2.679) (1.064) Dividend 0.006 -0.001 -0.004 -0.002 (0.183) (-0.033) (-0.120) (-0.039) Cash Flow Volatility 3.407*** 3.400*** 3.395*** 4.034*** (4.560) (4.712) (4.730) (4.623) Shareholder Rights 0.015 0.009 0.010 -0.002 (0.865) (0.485) (0.602) (-0.095)

62

Private Credit 0.001 0.001 0.001 0.001 (1.189) (1.466) (1.482) (0.987) Press Freedom 0.000 0.001 0.001 0.002 (0.113) (0.567) (0.856) (1.620) Constant 0.709*** 0.682*** 0.654*** 0.593** (3.434) (3.310) (3.261) (2.178) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 5,254 5,254 5,254 2,943 Adj. R 2 0.191 0.195 0.197 0.211

63

Table 2. 12 Political Accountability, State Ownership, and the Value of Cash The dependent variable is Tobin's Q , computed as market value of equity minus book value of equity plus book value of assets, all scaled by book value of assets. Model 1 is the basic model. Models 2, 3, and 4 use State Ownership , Control , and Political as independent variables, respectively. Checks is defined as the number of checks and balances each in a country and is drawn from Beck et al. (2001). This index reflects the degree of restraint imposed on politicians by veto players in the government. Democracy measures how responsive government is to its people. The index ranges from 0 to 6 (with higher scores indicating more democracy). Judicial Independence is the average score of tenure of supreme court judges, tenure of administrative court judges, and whether case law is used. This index is drawn from La Porta et al. (2004) with higher values indicating more judicial independence. Disclosure requirement variable is from Djankov et al. (2010): Disclosure by Politicians , which is a dummy variable equal to 1 if the law or regulations of a country require MPs to disclose either financial or business interests, and 0 otherwise. The full sample contains 5,254 firm-year observations from 578 firms privatized in 59 countries from 1981 to 2014. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Panel A: Checks and balances Basic Measures of state ownership Dependent variable: Tobin's Q Model State Control Political (1) (2) (3) (4) Cash Holdings 0.405** 0.735*** 0.771*** 0.508*** (2.190) (4.235) (5.307) (2.676) Cash Holdings×Checks 0.120*** (3.139) State Ownership -0.007 (-0.130) Cash Holdings×State Ownership -0.513 (-1.144) Cash Holdings×State Ownership×Checks 0.204*** (3.712) Control 0.010 (0.196) Cash Holdings×Control -0.632*** (-2.647) Cash Holdings×Control×Checks 0.198*** (5.081) Political 0.014 (0.204) Cash Holdings×Political -0.365 (-1.195) Cash Holdings×Political×Checks 0.168*** (3.502) Size -0.007 -0.006 -0.006 0.006 (-0.996) (-0.843) (-0.854) (0.737) Leverage -0.099 -0.103 -0.100 -0.057 (-1.165) (-1.209) (-1.170) (-0.505) Net Working Capital -0.203*** -0.198*** -0.197*** -0.189*** (-4.205) (-4.108) (-4.108) (-3.192) Cash Flow 0.943*** 0.941*** 0.952*** 0.987*** (10.598) (10.632) (10.824) (8.809) Capital Expenditure -0.176** -0.174** -0.177** -0.160 (-2.006) (-1.965) (-2.003) (-1.408) Sales Growth 0.114** 0.115** 0.116** 0.056 (2.208) (2.226) (2.252) (0.838) Dividend 0.001 -0.000 -0.001 -0.001 (0.044) (-0.011) (-0.060) (-0.037) Cash Flow Volatility 3.510*** 3.529*** 3.539*** 4.164*** (8.348) (8.422) (8.470) (7.896) Shareholder Rights 0.027*** 0.029*** 0.030*** 0.024** (3.419) (3.559) (3.666) (2.149) Private Credit 0.000 0.000 0.000 0.000 (1.524) (1.357) (1.375) (1.214)

64

Checks -0.012* -0.013** -0.016** -0.013* (-1.745) (-2.037) (-2.577) (-1.699) Constant 2.090*** 2.120*** 2.110*** 1.735*** (8.636) (8.757) (8.621) (8.364) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 4,615 4,615 4,615 2,910 Adj. R 2 0.193 0.195 0.197 0.206 Panel B: Democracy Measures of state ownership Basic Model Dependent variable: Tobin's Q State Control Political (1) (2) (3) (4) Cash Holdings -0.224 0.789** 0.780*** 0.552* (-0.477) (2.490) (2.925) (1.715) Cash Holdings×Democracy 0.266** (2.376) State Ownership -0.013 (-0.131) Cash Holdings×State Ownership -2.096 (-1.625) Cash Holdings×State Ownership×Democracy 0.580** (2.306) Control -0.004 (-0.047) Cash Holdings×Control -1.451** (-2.252) Cash Holdings×Control×Democracy 0.410*** (2.843) Political 0.017 (0.145) Cash Holdings×Political -1.037 (-1.413) Cash Holdings×Political×Democracy 0.356** (2.178) Size -0.017 -0.014 -0.013 -0.006 (-1.075) (-0.912) (-0.840) (-0.341) Leverage -0.108 -0.120 -0.123 -0.109 (-0.677) (-0.752) (-0.768) (-0.537) Net Working Capital -0.231** -0.225** -0.222** -0.215** (-2.439) (-2.423) (-2.405) (-2.170) Cash Flow 0.947*** 0.943*** 0.954*** 0.998*** (6.175) (6.206) (6.353) (5.454) Capital Expenditures -0.179 -0.177 -0.180 -0.115 (-1.478) (-1.478) (-1.504) (-0.721) Sales Growth 0.155** 0.154** 0.152** 0.091 (2.468) (2.480) (2.439) (1.057) Dividend 0.010 0.003 0.001 -0.002 (0.286) (0.090) (0.042) (-0.044) Cash Flow Volatility 3.385*** 3.412*** 3.419*** 4.009*** (4.430) (4.595) (4.624) (4.580) Shareholder Rights 0.009 0.002 0.004 -0.009 (0.478) (0.091) (0.197) (-0.367) Private Credit 0.001 0.001 0.001 0.001 (1.388) (1.572) (1.561) (1.333) Democracy 0.004 0.015 0.019 0.034 (0.257) (0.964) (1.288) (1.616) Constant 0.870*** 0.827*** 0.772*** 0.541** (2.903) (2.753) (2.589) (1.994) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 4,979 4,979 4,979 2,935 Adj. R 2 0.198 0.201 0.202 0.211 Panel C: Judicial independence Dependent variable: Tobin's Q Measures of state ownership

65

Basic State Control Political Model (1) (2) (3) (4) Cash Holdings 0.107 0.719** 0.695** 0.571* (0.241) (2.016) (2.280) (1.676) Cash Holdings×Judicial Independence 1.147** (2.083) State Ownership -0.050 (-0.457) Cash Holdings×State Ownership -1.731* (-1.835) Cash Holdings×State Ownership×Judicial Independence 3.449*** (3.373) Control -0.015 (-0.234) Cash Holdings×Control -0.998** (-2.065) Cash Holdings×Control×Judicial Independence 2.247*** (3.650) Political 0.012 (0.163) Cash Holdings×Political -0.877* (-1.746) Cash Holdings×Political×Judicial Independence 2.073*** (3.210) Size -0.014 -0.013 -0.013 0.000 (-0.801) (-0.756) (-0.734) (0.010) Leverage -0.222 -0.222 -0.220 -0.095 (-1.288) (-1.258) (-1.252) (-0.480) Net Working Capital -0.193** -0.199** -0.200** -0.203** (-2.079) (-2.232) (-2.245) (-2.078) Cash Flow 0.859*** 0.826*** 0.847*** 0.878*** (5.253) (5.176) (5.386) (4.944) Capital Expenditures -0.070 -0.071 -0.078 -0.023 (-0.551) (-0.569) (-0.620) (-0.143) Sales Growth 0.101 0.114* 0.107* 0.048 (1.568) (1.786) (1.657) (0.569) Dividend -0.057 -0.050 -0.053 -0.023 (-1.516) (-1.295) (-1.389) (-0.596) Cash Flow Volatility 3.435*** 3.329*** 3.334*** 3.890*** (3.914) (3.932) (3.935) (4.210) Shareholder Rights -0.028 -0.037 -0.037 -0.033 (-1.117) (-1.489) (-1.467) (-1.237) Private Credit 0.001* 0.001** 0.002** 0.002** (1.937) (2.275) (2.330) (2.277) Judicial Independence 0.212* 0.230* 0.277** 0.262** (1.745) (1.887) (2.337) (2.083) Constant 0.699*** 0.729*** 0.670*** 1.689*** (2.848) (3.137) (2.977) (4.508) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 4,449 4,449 4,449 2,810 Adj. R 2 0.213 0.222 0.223 0.236 Panel D: Disclosure by politicians Measures of state ownership Basic Dependent variable: Tobin's Q Model State Control Politic al (1) (2) (3) (4) Cash Holdings -0.248 0.742** 0.720*** 0.463 (-0.688) (2.369) (2.612) (1.427) Cash Holdings×Disclosure by Politicians 1.483*** (3.346) State Ownership -0.056

66

(-0.563) Cash Holdings×State Ownership -2.355*** (-2.783) Cash Holdings×State Ownership×Disclosure by Politicians 3.981*** (4.352) Control -0.025 (-0.424) Cash Holdings×Control -1.333*** (-2.999) Cash Holdings×Control×Disclosure by Politicians 2.474*** (4.555) Political -0.010 (-0.130) Cash Holdings×Political -1.066** (-2.249) 2.283** Cash Holdings×Political×Disclosure by Politicians * (3.915) Size -0.015 -0.011 -0.011 0.004 (-1.004) (-0.750) (-0.784) (0.214) Leverage -0.120 -0.121 -0.115 -0.104 (-0.754) (-0.778) (-0.736) (-0.513) Net Working Capital -0.221** -0.215** -0.209** -0.200** (-2.455) (-2.483) (-2.416) (-2.047) 0.924** Cash Flow 0.906*** 0.860*** 0.884*** * (5.817) (5.658) (5.867) (5.048) Capital Expenditures -0.205* -0.181 -0.196* -0.133 (-1.722) (-1.575) (-1.685) (-0.845) Sales Growth 0.144** 0.153** 0.144** 0.057 (2.359) (2.544) (2.363) (0.669) Dividend 0.010 0.012 0.015 0.020 (0.292) (0.324) (0.410) (0.483) 3.940** Cash Flow Volatility 3.331*** 3.311*** 3.321*** * (4.491) (4.622) (4.628) (4.527) Shareholder Rights 0.015 0.000 0.004 -0.005 (0.836) (0.011) (0.234) (-0.234) Private Credit 0.001* 0.001* 0.001* 0.001 (1.699) (1.795) (1.766) (1.383) Disclosure by Politicians -0.067 -0.026 0.019 0.043 (-0.870) (-0.349) (0.272) (0.537) Constant 1.877*** 1.927*** 1.878*** 0.592** (5.130) (5.210) (5.067) (2.240) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 5,254 5,254 5,254 2,943 Adj. R 2 0.194 0.203 0.202 0.215

67

CHAPTER 3

STATE OWNERSHIP , LIQUIDITY AND VALUATION : EVIDENCE FROM PRIVATIZATION

3.1 Introduction

Despite substantial privatization efforts over the last three decades, governments around the world continue to be major players in stock markets, their ownership of equity accounting for nearly one-fifth of total stock market capitalization worldwide (Borisova et al., 2015; Megginson, 2016). Much of this increase in state capitalism occurred following the 2008 global financial crisis, as governments rushed to bail out “too big to fail” firms by partially nationalizing them. 15 This “reverse privatization” phenomenon has renewed the debate about the role of governments as shareholders, fostering research on the impact of state ownership on the valuation of corporate assets and equity (e.g., Holland, 2016), the cost of equity (Ben-Nasr, Boubakri, and Cosset, 2012), the cost of debt (Borisova and

Megginson, 2011; Borisova et al., 2015), cash holdings (Chen, El Ghoul, Guedhami, and

Nash, 2017), corporate risk-taking (Boubakri et al., 2013), governance quality (Borisova et al., 2012), and corporate investment efficiency (Jaslowitzer et al., 2016). 16

15 According to Megginson (2016, p. 18) “This policy emphasizes that governments can and should retain control over vital national economic assets and promote the development of national champions in various globally competitive industries, and invest some or most of the national savings through state-owned vehicles”. Nash (2017) investigates the balance between public and private ownership over the period 1996– 2015. He finds that although privatizations distinctly outpaced nationalizations in the late 1990s, the gap narrowed in the years post-2000, especially during the global financial crisis period when the number of nationalizations surpassed the number of privatizations. 16 See Megginson (2016) for a recent survey of the academic and professional literature examining various aspects of state ownership and the economic role of state-owned enterprises.

68

The purpose of this essay is to add to the literature by investigating the liquidity implications of state ownership in an international sample of newly privatized firms

(NPFs), relying on two measures of firm-level stock liquidity, namely zero-return days and

FTH (Fong, Holden, and Trzcinka, 2017). To date, evidence on the stock market implications of privatization have been very scarce, with the notable exceptions of

Boutchkova and Megginson (2000) and Bortolotti, De Jong, Nicodano, and Schindele

(2007) who focus on the link between privatization and aggregate liquidity. Boutchkova and Megginson (2000) find a positive relation between the turnover ratios for individual markets and the number of privatizations in the country. This result is later confirmed by

Bortolotti et al. (2007) who document an increased aggregate liquidity for privatized IPOs in a sample of 19 OECD countries between 1985 and 2002. The authors find that not only does privatization positively affect liquidity, but “it also has a positive externality on non- privatized companies by increasing their aggregate liquidity and turnover as well” (p. 229).

In this essay, we use an alternative perspective and examine at the stock level the link between state ownership and stock market liquidity using a large sample of partially and fully privatized firms from 54 countries as well as non-privatized matching firms, over the

1994–2014 period, which includes the recent financial crisis.

Focusing on liquidity is important in its own right: First, liquidity is material for the well-functioning capital markets and hence economic growth. According to Levine

(1991), more liquid stock markets provide investors with the opportunity to sell corporate securities quickly and cheaply, thus encouraging them to invest in capital markets’ newly issued securities. More liquid stock markets in turn provide investors with risk-sharing and portfolio diversification opportunities, which, as a consequence, increase investors’

69

demand for corporate securities. Rhee and Wang (2009) conclude that the lack of liquidity is a key determinant of stocks’ volatility, and constitutes a major obstacle to financial market development.

Second, liquidity is important as a determinant of the firms’ asset prices/valuation and cost of capital. The impact of liquidity on asset prices stems from the fact that more liquid stocks (given a level of cash flow) have higher prices. Amihud and Mendelson

(2000) assert that investors not only demand a risk premium while setting the stock price,

“but also a liquidity premium to compensate them for the costs they bear when buying and selling the firm’s claims” (p. 8). When liquidity risk decreases, the liquidity of the stock increases, which translates into a lower required return (or cost of capital). Investors are thus attracted to more liquid stocks and stock markets (Ginglinger and Hamon, 2007).

Examining stock liquidity in the context of privatization, a politically driven major economic reform, is particularly relevant because the development of stock markets and popular capitalism have often been put forward by governments and international donor agencies as primary objectives of privatization programs (e.g., Megginson and Netter,

2001; Megginson, Nash, Netter, and Poulsen, 2004). 17 More importantly, efficient and liquid capital markets promote economic growth and allow firms to fund investment opportunities they otherwise would have to forgo. Therefore, understanding the impact of privatization on stock liquidity is important given its essential role in promoting stock market development (Megginson and Netter, 2001) and developing “equity culture”

(Boutchkova and Megginson, 2000). To date, however, no evidence exists on the liquidity implications of state ownership at the firm level. Privatization provides us with an ideal

17 For example, Price Waterhouse (1989a, b) claims that the objectives of the British privatization program include promoting wider share ownership and developing the national capital market.

70

setting in which to test our conjectures: it allows us to observe differential levels of government ownership within and across countries, as the speed of implementation tends to vary between industries and countries. Privatization typically occurs gradually, allowing us to follow state divestiture as it changes over time. This change in the level of state ownership in turn captures changes in incentives and objective functions of the firm (from social/political ends to economic and profit maximization). Ownership changes associated with privatization are set in a context of extreme agency problems and information asymmetry (Shleifer and Vishny, 1997; Denis and McConnell, 2003; Dyck, 2001), allowing us to test their impact on firm outcomes.

The link between liquidity and ownership structure has been extensively examined in the literature. Studies have particularly focused on the impact of concentrated ownership, institutional investors, foreign investors, and family firms (e.g., Attig, Fong, Gadhoum, and

Lang, 2006; Brockman et al., 2009; Ginglinger and Hamon, 2012). In particular, related to our study, Brockman, Chung, and Yan (2009) assert that block ownership can affect firm’s stock liquidity through one of two mechanisms—altering the firm’s trading activity or changing its information environment. Block ownership is likely to discourage trading activities due to real friction effect . Blockholders trade significantly less than non- blockholders, and this reduction in trading will increase the costs of order processing and inventory, which would otherwise be spread over more trades. Moreover, block ownership can affect the firm’s liquidity through information friction effect . In the presence of informed traders, market makers tend to reduce depths and request higher spreads

(Copeland and Galai, 1983; Kyle, 1985). If blockholders frequently trade on private

71

information to the detriment of uninformed investors, stock liquidity will be lower in the presence of blockholders.

In this essay, we examine the impact of a particular type of block ownership, state ownership, on stock liquidity, and how investors perceive firms with the government as a residual owner. Understanding the relation between state ownership and stock liquidity is important for two reasons. First, government bailout programs in the recent financial crisis that were aimed at rescuing financially distressed firms led to a significant increase in state ownership around the world (Megginson, 2016). Second, the state, compared to other blockholders, is associated with unique characteristics such as political orientation and soft budget constraint, which have different implications for stock liquidity. In particular, state ownership is associated with implicit government guarantees and preferential access to credit, particularly during the financial crisis (Borisova et al., 2015; Borisova and

Megginson, 2011; Faccio et al., 2006). However, as asserted by Amihud and Mendelson

(2008, p. 32) “during the recent financial crisis, the shortage in funding and great uncertainty about asset values led to a dramatic reduction in the provision of liquidity services”. Therefore, the relation between state ownership and stock liquidity has significant policy implications and contributes to our understanding of the role of state as a large shareholder.

Privatization, because it causes former state-owned enterprises (SOEs) to be listed on the stock market, may increase liquidity by improving investors’ diversification opportunities and by spreading the cost of real friction over more traders. Indeed,

Boutchkova and Megginson (2000) show that privatizations have dramatically increased the number of shareholders in many countries and led to higher trading volume. Bortolotti

72

et al. (2007) argue that investors have the opportunity to diversify their portfolios only if many firms go public, encouraging them to enter the equity market, thus moving “the equilibrium on the market away from low liquidity–high risk premium”. The government

“as an owner of several listed firms, can better ‘internalize’ the benefits from additional listings” (Bortolotti et al., 2007).

The link between state ownership and liquidity is not as straightforward, depending on the views behind state ownership. There are two competing views of state ownership that result from governments imposing nonprofit-maximizing social and political objectives, while at the same time providing an implicit guarantee against failure (Borisova et al., 2015). The political view of government ownership suggests that governments pursue political objectives (extract political rents) that rarely coincide with profit maximization (Boycko, Shleifer, and Vishny, 1994). This is one of the main reasons behind

SOEs’ inefficiency compared to private firms. As a result, it is only after firms’ control is transferred to private owners that former SOEs can break away from the “grabbing hand” of the government, decrease minority shareholders’ expropriation, and overturn their performance. 18 The implication is that higher residual ownership should be related to lower liquidity, particularly since investors associate government ownership with policy risk and lack of monitoring, and are thus not willing to buy such stocks. Consistent with this view,

Chung, Elder, and Kim (2010) show that better corporate governance is associated with higher market liquidity. Similarly, Kyle (1985) shows that the price impact of trade is greater for firms with poor corporate governance. Moreover, continued state ownership

18 Empirical evidence is extensive about performance improvements observed post-privatization. For an extensive review of the literature, please refer to Megginson (2016).

73

also increases information asymmetry since there is less incentive for firms under government control to improve transparency. Guedhami et al. (2009) show, for instance, that high government ownership undermines accounting transparency in privatized firms.

Diamond (1985) shows that information disclosure will reduce traders’ incentive to search for private information and lead to less heterogeneity among trader beliefs. Lang, Lins, and

Maffett (2012) show that firm-level transparency is positively associated with stock liquidity. Therefore, we expect that state ownership is positively associated with information friction . This suggests that informed traders are likely to be discouraged by investing in NPFs, thus increasing illiquidity (decreasing liquidity). Consistent with this view, Ben-Nasr et al. (2012) and Borisova and Megginson (2011), respectively, show that residual state ownership increases the cost of equity and the cost of debt.

However, state ownership can also have benefits as it carries an implicit guarantee on the firm’s debt because “it is less likely that a firm with state ownership would be allowed to fail. This unwillingness of governments to allow firm default is due to several reasons: pursuit of political and socially desirable goals, such as low unemployment and domestic investment; the desire to maintain key industries providing crucial services to the country; and the reluctance to be associated with a failed investment” (Borisova and

Megginson, 2011). Faccio et al. (2006) and Borisova and Megginson (2011), among others, suggest that this soft budget constraint will likely reduce the risk premium required by investors against default risk, which, we anticipate, will put downward pressure on liquidity. Evidence for the soft budget constraint is documented in several papers. Faccio et al. (2006) show that politically connected firms are more likely to be bailed out than non-connected matching firms. In the same vein, Wang, Wong, and Xia (2008) argue that

74

SOEs can appeal to this budget constraint when facing financial distress, and this implicit guarantee provides them with no incentives to improve their information quality (Ben-

Nasr, Boubakri, and Cosset, 2015). Related evidence in Borisova and Megginson (2011) proves that the soft budget constraint is to a certain extent valued by creditors since government ownership and credit spreads are not linearly related. Consistent with high state ownership leading to soft budget constraints, Megginson et al. (2014) find that state ownership and cash holdings are negatively related for a sample of Chinese privatized firms. In sum, this soft budget constraint hypothesis suggests that higher residual ownership provides incentives for investors to invest in NPFs, which results in higher liquidity. The impact of state ownership on liquidity is quite complex: while the political view points to a lower liquidity with higher state ownership, the soft budget constraint hypothesis suggests that liquidity should in fact increase with state ownership.

To test our conjectures, we use a sample of 3,934 firm-year observations representing 478 NPFs from 54 countries over the period 1994–2014. We start by comparing NPFs to other publicly listed firms, using two alternative proxies of liquidity.

We find that NPFs on average have higher liquidity than publicly listed firms. This confirms country-level evidence that aggregate liquidity improves after privatization

(Boutchkova and Megginson, 2000; Bortolotti et al., 2007). However, when we focus on partially privatized firms in comparison to fully privatized firms, we find that the former are more liquid than the latter. The higher liquidity for NPFs and partially privatized firms may at first seem at odds, but this can potentially be explained by the existence of a non- monotonic relation between state ownership and liquidity. We test this conjecture and find, after controlling for firm- and country-level variables, an inversely U-shaped relation

75

between state ownership and our two proxies for liquidity. The results are statistically and economically significant: moving state ownership from the 25 th percentile (0.00) to the 75 th percentile (0.51) results in a 10.2% (=0.51× (-0.021) / 0.105) decrease in zero-return days

(with all other variables constant). The results from the quadratic regression show an inflection point of 42% government ownership at which liquidity is highest. This suggests that when the government retains more than 42% of shares, the market responds negatively with lower stock liquidity, suggesting that investors fear the “grabbing hand” of the government.

One major concern in our analysis is endogeneity. The documented relation between state ownership and liquidity could be driven by omitted variables related to residual state ownership and stock liquidity. In addition, governments may keep higher

(lower) stakes in better (poorly) governed firms. To address these concerns we employ instrumental variables, the Heckman selection model, and propensity score matching. Our results remain unchanged.

We adopt two approaches to further investigate the soft-budget constraint explanation underlying the relation between state ownership and stock liquidity. First, we examine whether country-level factors related to government ownership of banks, restrictions on foreign banks, and extent of political influence affect the relation between state ownership and liquidity. We find that this relation is stronger in countries with higher levels of state ownership of banks, and weaker in countries with fewer limits on foreign banks and little or no political influence as measured by the independence of the supervisory authority. Second, we examine the role of the global financial crisis on this

76

relation and find that the role of state ownership on liquidity is stronger during the crisis period.

Finally, we examine the impact of stock liquidity on valuation and cost of capital, of which liquidity is a determinant, and hence a potential channel of influence. We show that stock liquidity is associated with higher valuation and lower cost of equity capital.

These results suggest that liquidity is a channel through which residual state ownership in

NPFs affects valuation and cost of capital. Our findings on liquidity effects are closer in spirit to Borisova and Megginson (2011) who document a non-linear relation between state ownership and cost of debt. Our findings thus document that Ben-Nasr et al. (2012) provide only a partial explanation of the relation between state ownership and the cost of equity in

NPFs.

This essay makes several contributions to the literature. First, our results are related to the literature on the effects of privatization on financial market development [see

Megginson (2005, 2016) for an excellent survey]. This essay complements previous country-level studies (e.g., Boutchkova and Megginson, 2000; Bortolotti et al., 2007) by using firm-level liquidity measures to show that state ownership affects liquidity in a nonlinear fashion. In addition, we examine one channel through which state ownership affects cost of capital and thus firm value and find that superior access to state-owned banks, which leads to soft budget constraints, explains the high stock liquidity in NPFs. Second, we contribute to the literature on the link between ownership structure and liquidity by focusing on a particular stakeholder, the State. As one of the most important large shareholders, the state is not only related to poor corporate governance, but is also unique, compared to other blockholders, in terms of soft budget constraint. We show that state

77

ownership is associated with high stock liquidity, in particular during the financial crisis.

Third, we further add to the privatization literature by showing how the distortion between economic and political/social objectives associated with state ownership plays out in determining stock liquidity.

This essay has policy implications related to the process of privatization: our evidence that high residual ownership decreases liquidity suggests that continued ownership is suboptimal. Of particular interest, we show that the benefits of privatization disappear at high levels of state ownership suggesting that, as argued by Megginson (2016), the links with government should be severed for positive outcomes to materialize. The economic distortions introduced by state ownership can be very costly to the economy when governments dominate as owners and market players.

The remainder of This essay is organized as follows. In Section 2, we review the literature and derive our hypotheses. In Section 3 we present our sample and the descriptive statistics. We then elaborate and discuss our results and the empirical analysis in Section

4, before we conclude in Section 5.

3.2 Literature Review and Hypothesis Development

3.2.1 Ownership structure, corporate governance, and liquidity

Given its theoretical importance to asset pricing, liquidity has been the subject of an abundant literature on its determinants and consequences. This essay fits in the strand of literature on the link between liquidity and corporate ownership, putting an emphasis on the role of state ownership, which to date has remained unexplored.

Empirical studies focusing on institutional, foreign, and block ownership abound.

Bolton and Von Thadden (1998) and Brockman et al. (2009) show that ownership

78

concentration adversely affects market liquidity. Furthermore, Brockman et al. (2009) argue that this negative relation is due to the lack of trading. Focusing on ultimate ownership, Attig et al. (2006) and Ginglinger and Hamon (2012) find that greater control- ownership divergence—which captures the expropriation incentives of the largest shareholder—lowers liquidity. These results suggest that controlling shareholders have incentives to increase firm opacity so that their expropriation remains undetected.

Anticipating these incentives, minority shareholders are reluctant to trade, which results in lower liquidity.

Prior studies suggest that block ownership affects liquidity through one of two main mechanisms: trading activity (real friction effect) or informed trading (informational friction effect) (Bhide, 1993; Bolton and Von Thadden, 1998; Maug, 1998; Brockman et al., 2009). The first mechanism revolves around the extent of trading of the stock. The existence of blockholders, who generally are reluctant to trade, leaves few shares available on the market, which discourages other shareholders from trading. In addition, according to Holmström and Tirole (1993), concentrated ownership reduces the benefit of monitoring of the firm by stock market participants, thereby reducing the amount of public information available about the firm. In such a context, the acquisition of information is more costly, which further dampens any incentive to trade, resulting in lower liquidity. The second mechanism driving the relation between concentrated ownership and liquidity is based on the general assertion that blockholders typically have private information about firms’ value, leading to potential losses for uninformed traders (informational friction). If these large shareholders are likely to trade based on their private information (informed trading),

79

their informational advantage translates into higher informational friction costs, hence lower liquidity.

We focus in this essay on the role of state owners and examine their effect on market liquidity of NPFs, an issue yet to be explored in the literature. We develop our arguments below.

3.2.2 State ownership, liquidity, and valuation

The privatization literature 19 justifies the necessity of the reform by emphasizing the costs of continued government ownership, seen as the main source of inefficiency in

SOEs. In particular, these inefficiencies are compounded because (1) the firms pursue political objectives that differ from profit maximization, and (2) managers in SOEs are typically entrenched bureaucrats that are not adequately monitored. This is consistent with the political view of government ownership, holding that politicians extract rents from the firms under their control (e.g., Shleifer and Vishny, 1994) and use them primarily to build political support among voters rather than to maximize profits and efficiency. There is theoretical and empirical evidence showing the extent of association between government ownership and ineffective corporate governance, and the extent of information asymmetry and extreme agency problems that plague firms with government ownership (Shleifer and

Vishny, 1997; Megginson and Netter, 2001; Megginson, 2016). For instance, residual state ownership is related to lower earnings quality, investment efficiency, and performance

(e.g., Guedhami et al., 2009; Ben-Nasr et al., 2012; Boubakri et al., 2005; Chen, El Ghoul,

Guedhami, and Nash, 2017). The privatization literature also shows significant performance and governance improvements postprivatization (e.g., Megginson et al., 2004;

19 Please refer to Nash (2017) for a discussion of the costs and advantages of government ownership.

80

Boubakri et al., 2005), thus providing support for the benefits of relinquishing state ownership.

Effective corporate governance mitigates management’s ability and incentive to distort financial information (Leuz, Nanda, and Wysocki, 2003) and improve financial and operational transparency, which decreases information asymmetry between blockholders and investors. Indeed, Chung et al. (2010) show that corporate governance is positively associated with stock liquidity. Similarly, Kyle (1985) shows that the price impact of trade is greater for firms with poor corporate governance. Given the weak internal and external governance that characterizes NPFs, we expect that investors are likely to shy away from these stocks. Moreover, Guedhami et al. (2009) find that government ownership undermines financial transparency in NPF. Therefore, there is information asymmetry between management and outsider investors (e.g., outsider owners and liquidity provider) given there is less incentive for firms under government control to improve transparency.

Diamond (1985) shows that information disclosure will reduce traders’ incentive to search for private information and lead to less heterogeneity among trader beliefs. Lang et al.

(2012) show that firm-level transparency is positively associated with stock liquidity.

Taken together, we expect that state ownership is negatively associated with stock liquidity due to the information asymmetry problem ( informational friction effect ). In firms with high ownership stakes, these problems are compounded by the policy risk (i.e., government intervention in the firms and their environment) inherent in government ownership.

Moreover, these large shareholders, according to the micro-structure literature, do not trade much. Demsetz (1968) shows that liquidity increases with the number of stockholders.

Indeed, in privatizations, if the government floats only a few portions of the firm in

81

successive sales, the stock will be held by fewer investors and the trading activities and depth of the market is impaired ( real friction effect ). The political view of government ownership and the trading activity explanation developed above suggest that state ownership should have the same liquidity effects as blockholders if their equity stake is large. In such cases, governments’ large stakes will discourage other shareholders from trading, leading to a negative association between state ownership and market liquidity.

The soft budget constraint view implies that government ownership carries an implicit government guarantee and protection in times of financial distress (Faccio et al.,

2006; Borisova and Megginson, 2011). Faccio et al. (2006) show that politically connected firms are more likely than non-politically connected firms to be bailed out. Supporting the preferential access to credit for government-owned firms, Cull, Xu, and Zhu (2009) find that credits are heavily tilted towards SOEs in China. In the same vein, Boubakri et al.

(2012) show that politically connected firms enjoy a lower cost of equity. This relation is stronger in countries where the likelihood of government bailouts is higher. Finally,

Chaney et al. (2011) find that investors do not penalize politically connected firms by requiring higher returns, although their earnings’ quality is lower, suggesting that investors value the benefits the firms can extract by being linked to the government. Under the soft budget constraint hypothesis, state ownership is valuable to shareholders who are willing to trade the NPF’s stocks, hence increasing liquidity.

These competing views suggest that the relation between liquidity and state ownership is quite complex, ranging from more illiquidity under the political view to more liquidity under the soft budget constraint view. More formally, this discussion leads us to expect the following:

82

H1: Privatization is positively related to stock liquidity

H1a: Under the political view, residual ownership is negatively related to stock

liquidity.

H1b: Under the budget constraint view, residual state ownership is positively

related to stock liquidity.

3.3 Sample and Variables

3.3.1 Sample

To empirically test the relationship between state ownership and liquidity, we construct a sample of 478 NPFs from 54 countries over the period 1994–2014. Our sample comes mainly from Boubakri et al. (2013), which we update using Privatization Barometer,

Thomson Reuters, SDC Platinum Global New Issues database, and SDC Platinum Mergers

& Acquisitions database. This sample provides an ideal setting in which to investigate the role of state ownership on stock liquidity. We track the change in ownership after the first privatization, which allows us to analyze the time-varying effect of state ownership on stock liquidity. We obtain stock liquidity and other financial statement information from

Compustat Global and ownership data from Boubakri et al. (2013), firms’ annual reports,

Bureau van Dijk’s Osiris database, and Bloomberg. Since the behavior of financial firms

(SIC codes between 6000 and 6999) is heavily influenced by a country’s regulatory environment, we exclude these firms from our sample. After excluding missing observations, our final sample contains 3,934 firm-year observations.

Table 3.1 summarizes the sample distribution by country, year, and industry. Panel

A shows that the sample firms are widely distributed across both developing and developed countries, with 16.29% from China, 5.08% from India, 4.91% from Poland, and 4.68%

83

from France and Italy. 20 Panel B reports the sample distribution by year. More than 90% of our sample firms were privatized in the 1990s and 2000s, peaking between 2011 and

2012. Panel C shows that the sample firms are widely distributed across industries [defined according to Fama and French’s (1997) classification], with 18.63% in manufacturing,

16.37% in utilities, and 15.1% in telecom. 21

3.3.2 Variables

3.3.2.1 Liquidity measures

Following prior literature (e.g., Lesmond, Ogden, and Trzcinka, 1999), we first calculate stock liquidity as the proportion of days with zero returns. Stocks with lower liquidity are more likely to have zero-volume days and thus are more likely to have zero- return days. Moreover, stocks with higher transaction costs have less private information acquisition. Therefore, even on positive-volume days, these stocks are more likely to have no-information-revelation, which translates into zero-return days. Lesmond (2005) shows that this liquidity measure performs better at representing cross-country liquidity effects than do other liquidity measures for studies in emerging markets. Lesmond et al. (1999) define the proportion of days with zero returns as ZERO RETURN DAYS = (# of days with zero returns) / T, where T is the number of trading days in a year.

We also use another measure of stock liquidity, FHT , developed by Fong et al.

(2017). FHT is a percent-cost proxy that simplifies the existing LOT Mixed measure. Fong et al. (2017) finds that this measure is useful in global research. Following Fong et al.

20 Our main findings are not sensitive to sequentially excluding each country from our analysis. 21 All our inferences remain when we sequentially exclude each industry from our analysis.

84

(2017), we define FHT as 2*Sigma*Probit((1+Zeros)/2) , where Sigma=Std

(NonZeroReturns) .

For robustness checks, we use AMIHUD (Amihud, 2002) to proxy stock liquidity.

AMIHUD captures the daily price response associated with one dollar of trading volume, which is calculated as Average (|r|/Volume) , where r is the stock return on day t and Volume is the dollar volume on day t. The average is calculated over all positive-volume days.

Note that the variables described above measure stock illiquidity . Therefore, the higher the value of these variables, the lower (higher) the stock liquidity (illiquidity). The

Appendix summarizes the definitions for these and all other variables that we use in our analyses.

3.3.2.2 Independent and control variables

We capture state ownership using the percentage of shares held by a government

(STATE ). Our regressions also include several firm- and country-level control variables to ensure that the relation between state ownership and stock liquidity is not driven by confounding factors. In particular, following prior literature (e.g., Lang, Lins, and Miller,

2004; Lang et al., 2012; Stoll, 2000), we control for firm size as measured by the log of a firm’s market value of equity (LOG MV ), book-to-market (BM ), return variability

(STDRET ), transparency as measured by earnings smoothness (EM ), analyst coverage as measured by the number of analysts forecasting current year earnings (ANALYST ), and whether the firm had a loss (LOSS ). We further include indicator variables for whether the stock trades in the U.S., either on an exchange (ADR EX) or on the OTC or PORTAL markets (ADR NEX). U.S. trading is likely to improve transparency (Lang, Lins, and Miller,

2003), but it may also draw liquidity from local markets to the extent that shares are less

85

costly to trade in the U.S. (Baruch, Karolyi, and Lemmon, 2007). We also control whether the firm reports under IFRS or U.S. GAAP (INTGAAP ). Leuz and Verrecchia (2000) show that firms that convert to IAS or U.S. GAAP are associated with higher liquidity. Therefore, we expect that INTGAAP is positively associated with stock liquidity.

Additionally, we control for country-level institutions that are likely to influence the extent to which firm-level transparency affects liquidity (e.g., Lesmond, 2005; Lang et al., 2012). Specifically, we include a country’s legal origin as a proxy for investor protection (COMMON ), the number of listed firms to control for stock market development

(LISTED ), press freedom to capture media penetration (MEDIA ), and income level measured as log of GDP per capita (LGDPC ). 22

3.3.3 Descriptive statistics

Panel A of Table 3.2 provides summary statistics. To mitigate the influence of outliers, we winsorize all financial variables at the 1% level, on both sides of the sample distribution. ZERO RETURN DAYS has a mean (median) of 0.105 (0.073), while FHT has a mean (median) of 0.006 (0.004). Residual state ownership (STATE ) has a mean (median) of 0.265 (0.166), in line with a sharp decline in state ownership after privatization

(Boubakri et al., 2005).

Panel B of Table 3.2 presents Pearson correlation coefficients between key variables. State ownership is negatively correlated with all three stock liquidity measures, suggesting that higher state ownership is associated with higher stock liquidity.

22 We expect that in countries with lower investor protection, liquidity providers increase the spread to protect themselves from insiders (Bhattacharya and Daouk, 2002), resulting in lower stock liquidity. Similarly, we expect that when media penetration is poor, stock liquidity is lower (e.g., Lang et al., 2012). Moreover, we expect that in countries with better stock market development and higher income level, stock liquidity tends to be higher.

86

Meanwhile, state ownership is negatively correlated with Tobin’s Q and positively associated with cost of equity capital. In the next session, we present multivariate results.

3.4 Results

3.4.1 Preliminary findings

To have a preliminary understanding of the relation between state ownership and stock liquidity, in Figure 1 we plot state ownership against stock liquidity as measured by

ZERO RETURN DAYS and FHT . We find that stock illiquidity as measured by zero-return days and FHT is at its peak when state ownership is 0, while stock illiquidity gradually decreases as state ownership increases before state ownership is more than 40%. This plot provides some preliminary evidence on a non-linear relation between state ownership and stock liquidity.

In Table 3.3, we perform a univariate analysis on the relation between state ownership and stock liquidity. In Table 3.3A, we start by comparing the stock liquidity of privatized firms and that of publicly listed firms. The mean (median) value of ZERO

RETURN DAYS for privatized firms is 0.135 (0.088), while it is 0.217 (0.149) for other publicly listed firms. The t-statistic (z-statistic) for the difference is 33.7 (38.4), suggesting that privatized firms have significantly higher stock liquidity than other publicly listed firms. The univariate differences between privatized firms and other publicly listed firms remain similar when it comes to FHT .

To address the concern that the differences may come from firm characteristics, we match privatized firms to other publicly listed firms by size and industry, and use the matched sample to estimate the univariate differences in liquidity. Interestingly, using the matched sample shows barely any significant difference between privatized firms and

87

matching publicly listed firms in terms of stock liquidity. This suggests that privatized firms and other publicly listed firms are indifferent in terms of stock liquidity after controlling for size and industry.

We then split the privatized firms into two groups: partially privatized firms and fully privatized firms. We find that partially privatized firms have a significantly higher stock liquidity than publicly listed firms whether we use the full sample or the matched sample. These results, which suggest that state ownership (prevalent in partially privatized firms) is positively related to liquidity, are consistent with the soft budget constraint hypothesis. When we compare fully privatized firms to publicly listed firms, we find that in the full sample, fully privatized firms have higher stock liquidity. Interestingly, once we control for firm size and industry, fully privatized firms exhibit lower stock liquidity than other publicly listed firms. Not surprisingly, partially privatized firms consistently show higher stock liquidity than fully privatized firms. All in all, the above findings provide univariate evidence in support of the soft budget constraint hypothesis.

In Table 3.3B, using the matched sample, we explore the relation between privatized firms and other publicly listed firms during the 2008–2010 financial crisis. Our results show that compared with matched publicly listed firms, privatized firms have fewer zero-return days before the financial crisis but not during and after the crisis. However, privatized firms enjoy higher stock liquidity in terms of FHT than other publicly listed firms during the financial crisis, which supports the idea that firms suffer less from financial constraints and adverse economic shocks particularly during times of financial distress (Borisova et al., 2015; Borisova and Megginson, 2011; Faccio et al., 2006).

88

Partially privatized firms in general are more liquid than other publicly listed firms before, during, and after the financial crisis, while fully privatized firms have lower stock liquidity than publicly listed firms, in particular during the financial crisis. Nevertheless, partially privatized firms consistently have higher stock liquidity than fully privatized firms, regardless of the financial crisis.

Taken together, the above univariate results provide primary evidence that state ownership is associated with a soft budget constraint, and thus with higher stock liquidity.

The apparently conflicting results between our test on NPFs (higher liquidity than other listed firms), and partially privatized firms (higher liquidity in partial divestitures where the state is still a stakeholder) suggest that state ownership might be non-linearly related to stock liquidity. In the next section, we analyze the relation between state ownership and stock liquidity in a multivariate setting.

3.4.2 Partial privatization and liquidity

Before the test on the relation between state ownership and liquidity, we investigate the impact of partial privatization on liquidity in the multivariate analysis. To do so, we first construct a dummy variable PARTIAL to indicate whether governments retain shares after privatization (i.e. STATE > 0) as our independent variable. We use two measures of liquidity and two regression specifications. In Models (1) to (2), we use ZERO RETURN

DAYS as the dependent variable and in Models (3) to (4), we use FHT as the dependent variable. Specifically, we estimate the following regression (superscripts omitted for simplicity):

89

= + + + + +

+ + + + +

+ + +

+ . 1

To control for within-firm correlation, we present significance levels based on robust standard errors adjusted for clustering at the firm level.

In Model (1) of Table 3.4, we find that the coefficient of PARTIAL is negative and statistically significant at the 1% level, suggesting that partially privatized firms are associated more with higher stock liquidity than fully privatized firms. Economically, when a firm is fully privatized from partial privatization (i.e. PARTIAL from 1 to 0), it will result in a 14% (=(0-1)× (-0.015) / 0.105) increase in zero-return days (with all other variables constant), and therefore a significant decrease in stock liquidity. In Model (2), we include numerous country-level characteristics rather than country fixed effects: legal origin

(COMMON ), stock market development (LISTED ), press freedom (MEDIA ), and income level (LGDPC ). We continue to find that the coefficient of PARTIAL loads negatively and is statistically significant at the 1% level.

In Models (3) and (4), we use FHT as our dependent variable with two specifications: one with country fixed effects and one with numerous country-level variables. We find that the coefficients on PARTIAL continue to be negative and statistically significant at the 1% level, with the magnitude of the economic impact of state ownership on stock liquidity comparable to that in Model (1).

Taken together, we find in Table 3.4 that partial privatization is associated with higher stock liquidity, which is consistent with the soft budget constraint view of state

90

ownership. These results suggest that firms have higher liquidity when government retains some shares.

3.4.3 State ownership and liquidity

In Table 3.5, we present the regression results investigating the impact of state ownership on stock liquidity. In Models (1) to (4), we use ZERO RETURN DAYS as the dependent variable, and in Models (5) to (8), we use FHT as the dependent variable and test for the effect of state ownership. Specifically, we estimate the following regression

(superscripts omitted for simplicity):

= + + + + +

+ + + + +

+ + +

+ . 2

In Model (1), we find that STATE is negatively associated with ZERO RETURN

DAYS . This relation is statistically significant at the 5% level. It is also economically significant, with the coefficient on STATE suggesting that moving state ownership from the 25 th percentile (0.00) to the 75 th percentile (0.51) results in a 10.2% (=0.51× (-0.021) /

0.105) decrease in zero-return days (with all other variables constant). In line with the univariate results, these findings support the soft budget constraint view of state ownership.

In Model (2), we explore the non-monotonic relation between state ownership and stock liquidity, using a quadratic model. We continue to find that STATE loads with a negative and statistically significant coefficient at the 1% level. More importantly, we find that the quadratic term STATESQR has a positive and statistically significant coefficient at the 1% level. The results suggest an inversely U-shaped relationship between state

91

ownership and liquidity. Relatedly, Borisova and Megginson (2011) show that the cost of debt is also non-monotonically related to residual state ownership. The results from the quadratic regression Model (2) in Table 3.4 (illustrated in Figure 2) show an inflection point of 42% government ownership at which stock liquidity is highest. With around half of the shares, the government continues to be a significant owner and is still able to exercise some control over the firm, although the majority of shares of NPFs are open to the public.

Additional privatization sales will reduce the firm’s stock liquidity, bringing some unwanted volatility to its potential future shareholder.

In Models (3) and (4), when including country-level characteristics rather than country fixed effects, we continue to find that the coefficient of STATE loads negatively and statistically significantly at the 1% level, while the coefficient of the quadratic term,

STATESQR, remains positive and statistically significant.

In Models (5) through (8), we employ FHT, our alternative measure of stock liquidity. In Model (5), we find that the coefficient on STATE continues to be negative and statistically significant at the 5% level, with the magnitude of the economic impact of state ownership on stock liquidity comparable to that in Model (1). In Model (6), we also include a quadratic form of state ownership, and continue to find that state ownership is negatively associated with FHT , while the square term of state ownership is positively associated with

FHT . The results from the quadratic regression Model (6) in Table 3.4 show an inflection point of 44% government ownership, at which stock liquidity is highest. These results echo our finding above. Models (7) and (8) replicate the specification used in Models (3) and

(4), while removing country fixed effects and controlling for country-level variables, and we observe similar findings.

92

In summary, our results in Table 3.5 are consistent with the soft budget constraint hypothesis of state ownership, indicating that firms are likely to be more liquid when government ownership increases. However, this relationship is non-monotonic: when the government retains more than 42–44% of shares, the market responds negatively with lower stock liquidity, suggesting that investors fear the “grabbing hand” of the government.

3.4.4 Endogeneity

In the context of privatization, a major econometric concern is selection bias. As

Megginson and Netter (2001, p. 346) point out, “sample selection bias can arise from several sources, including the desire of governments to make privatization look good by privatizing the healthiest firms first”. Further, governments may retain higher stakes in firms with higher liquidity to extract greater private/political benefits or because few shareholders may be interested in investing in privatized firms. In addition, the relation between state ownership and stock liquidity could be driven by unobserved determinants of liquidity that also explain residual state ownership. We address these issues using several approaches in Table 3.6.

We first estimate instrumental variable regressions. In particular, we use LIEC , which measures legislative electoral competitiveness, and DEFICITS , which is measured as deficits/GDP as an instrument for state ownership. The choice of these variables follows prior literature (Bortolotti and Faccio, 2009; Borisova and Megginson, 2011) documenting a relation between state ownership and LIEC and DEFICITS . In the first-stage regression

(unreported), we regress STATE on LIEC and DEFICITS along with the full set of control variables. We find that LIEC and DEFICITS are both negatively and significantly associated with STATE . We use three tests to check the validity of our instrument. We

93

conduct an F-test of the excluded exogenous variable. The results reject the null hypothesis of the F-test that the instrument does not explain state ownership. We also conduct a

Kleibergen-Paap rk LM test and reject the null hypothesis that the model is under-identified at the 1% level. Moreover, we find that the Hansen J statistic for over-identification test is

0.466, suggesting that our model is not over-identified. Model (1) shows the results of the second-stage regression. We again find that STATE is significantly negatively associated with ZERO RETURN DAYS .

In Model (2), we treat both state ownership and state ownership squared terms as endogenous variables. Again, in the first-stage regression (unreported), we regress STATE on LIEC and DEFICITS along with the full set of control variables. We then use the predicted state ownership, and the squared term of the predicted state ownership as instrumental variables. In Model (2), we show that state ownership is associated with fewer zero-return days, while state ownership square is associated with more zero-return days, confirming the existence if a nonlinear relation between liquidity and state ownership.

In Models (3) and (4), we perform a Heckman two-stage analysis to address sample selection bias concerns. In the first stage, we use a Probit model to predict whether governments retain control over the privatized firms. In particular, we regress Control, a dummy variable indicating whether governments retain more than 50% of the privatized firms, on LIEC and DEFICITS , the full set of control variables, and country, industry, and year fixed effects (as in Model 1 of Table 3.4). This allows us to estimate the inverse Mills ratio (Lambda ). In the second stage, we include LAMBDA as an additional independent variable in the liquidity regression. The results in Model (3) show that STATE is significantly positively associated with stock liquidity, while STATESQR is negatively

94

associated with stock liquidity in Model (4). In addition, LAMBDA loads positively and is statistically significant at the 1% level.

In Models (5) and (6), we employ propensity score matching. Developed by

Rosenbaum and Rubin (1983), propensity score matching allows us to randomize the sample selection procedure by using observable firm characteristics to match firms that are under government control with firms that are not. 23 In the first stage, we use the same Probit model as in the Heckman first-stage analysis. We then match state-controlled firms with the closest propensity score. In the second stage (Model 5), we estimate the regression using the matched sample. Consistent with our main analysis, the instrumental variable analysis, and the Heckman analysis, we continue to find that the coefficient on STATE loads positively and significantly at the 5% level, and that state ownership is non-linearly related to stock liquidity.

In Models (7) to (12), we re-estimate the instrumental variable analysis, the

Heckman analysis, and the propensity score matching analysis using FHT as the dependent variable, and continue to find similar results.

3.4.5 Robustness

An additional concern is that the relation between state ownership and stock liquidity has alternative explanations. We investigate several possibilities in Table 3.7.

First, Lesmond (2005) finds that political risk is an important factor in determining stock liquidity in emerging markets. Therefore, our results could be driven by an omitted variable—political risk. We test this explanation in Models (1) and (2) by including the

23 We obtain statistically similar results when we match firms that are partially privatized (i.e., State Ownership > 0) with firms that are fully privatized (i.e., State Ownership = 0).

95

political risk (POLRISK ) as an additional control variable. Political risk measure is obtained from the International Country Risk Guide (ICRG). A higher value of this measure indicates less political risk. We find that POLRISK is negatively associated with zero-return days, suggesting that less political risk increases stock liquidity. More importantly, the coefficient on STATE remains negative and statistically significant at the

5% level. In Model (2), we continue to find that STATE is negatively and STATESQR is positively associated with zero-return days. In Models (3) and (4), we replicate the estimations in Models (1) and (2) with FHT as a dependent variable, and our results remain statistically unchanged.

Another concern is that our results are driven by measurement errors. Therefore, we adopt another liquidity measure that is commonly used in the liquidity literature, namely AMIHUD . Using AMIHUD as our dependent variable, we show in Models (5) and

(6) that our results are not sensitive to an alternative dependent variable.

3.4.6 Additional analysis

3.4.6.1 Soft budget constraint

In this section, we explore the channel through which state ownership affects stock liquidity: soft budget constraint. In the following empirical analysis, we interact state ownership with proxies for soft budget constraint at the country level. Table 3.8 reports the results. In Model 1, we explore whether superior access to state-owned banks explains the high stock liquidity documented in NPFs with higher residual state ownership. We measure government ownership of banks (GOVBANK) at the country level using data obtained from

Barth et al. (2013). We expect state-owned firms to benefit from preferential access to financing in countries with higher government ownership of banks (e.g., Chen, El Ghoul,

96

Guedhami, and Nash, 2017; Jaslowitzer et al., 2016). Therefore, our focus is on the interaction term between STATE and GOVBANK . We find that the coefficient of STATE is negative and significant at the 1% level. More importantly, we find that the coefficient of

STATE × GOVBANK is negative and statistically significant at the 5% level. This suggests that stated-owned firms have higher stock liquidity in countries with greater government ownership of banks, and confirms that these firms are likely to have superior access to bank financing, leading to higher stock liquidity.

In Model 2, we use (LIMFOREIGN ) to measure the extent to which foreign banks may own domestic banks and whether foreign banks may enter a country’s banking industry. A higher value indicates less limitation on foreign entry, therefore less comparative advantage of state ownership. In Model 3, we measure the soft budget constraint as the degree to which the supervisory authority within the government is independent of political influence ( POLITICAL INDP ). In both regressions, we show that when there is comparative advantage of state ownership in terms of access to finance, state ownership is associated with higher stock liquidity.

3.4.6.2 Global financial crisis

Amihud and Mendelson (2008) assert that during the recent financial crisis, the shortage in funding and great uncertainty about asset values led to a dramatic reduction in the provision of liquidity services by market participants. To better understand the relation between state ownership and stock liquidity, we investigate in Table 3.9 the role of the recent global financial crisis. Faccio et al. (2006) find that politically connected firms are more likely to be bailed out by the government during times of financial distress. Similarly,

Boubakri et al. (2012) show that firms increase leverage after a politician joins the board

97

of directors. Chaney et al. (2011) further document that politically connected firms have a lower cost of borrowing even though their reporting quality is poorer. If the soft budget constraint hypothesis is true, we should observe a stronger relation between state ownership and stock liquidity during the financial crisis period. To test this conjecture, we include in

Table 3.8 two dummy variables (BEFORE CRISIS and DURING CRISIS ) and their interaction with STATE. We find that the coefficient on the interaction term between

STATE and DURING CRISIS is negative and statistically significant, suggesting that when state ownership increases, firms tend to be more liquid during financial crises.

3.4.7 The link between liquidity, cost of capital, and valuation

Following Lang et al. (2012), we test the effects of liquidity using two modeling approaches: cost of capital and Tobin’s Q. Table 3.10 presents the regression results relating state ownership, liquidity, and cost of capital. We use the implied cost of capital as the dependent variable in Models (1) and (2).

In Model (1) we use ZERO RETURN DAYS as our independent variable, and find that the coefficient is positive and statistically significant at the 1% level, which indicates that as predicted, firms with higher levels of illiquidity tend to face a higher cost of equity capital. It is also economically significant, with the coefficient on ILLIQUIDITY suggesting that moving ZERO RETURN DAYS from the 25 th percentile (0.035) to the 75 th percentile

(0.139) results in a 3% (=0.104× 0.042 / 0.105) increase in cost of equity (with all other variables constant). In Model (2), we use FHT as our independent variable, and continue to find that illiquidity is positively associated with cost of equity capital.

Models (3) and (4) present the regression results relating liquidity to valuation, measured as Tobin’s Q (TOBIN’S Q). Model (3) shows that ZERO RETURN DAYS is

98

negatively related to firm value, while Model (4) finds that as FHT increases, the market tends to discount the value of the firm.

3.5 Conclusion

In this study, we investigate the impact of state ownership on liquidity. Using a unique sample of 478 NPFs from 54 countries over the period 1994–2014, we find evidence consistent with the soft budget constraint associated with state ownership. Specifically, we show that state ownership is positively related to stock liquidity. However, this relationship is non-monotonic. As state ownership increases beyond a certain point, the market perceives state ownership as a negative sign. This result is robust to endogeneity tests, alternative measures of liquidity, and additional control variables. We further find that the relation between state ownership and liquidity is stronger in countries with higher levels of state ownership of banks, and is weaker in countries with fewer limits on foreign banks and little or no political influence as measured by the independence of the supervisory authority. Finally, we find that stock liquidity is related to firm value and cost of equity capital, suggesting that liquidity is a channel through which residual state ownership in

NPFs affects valuation. Our study contributes to the privatization literature by providing the first firm-level evidence on the liquidity implications of the reform, and by showing that extensive continued state ownership could dissuade investors who fear the “grabbing hand” of the government. This in turn dampens the liquidity of privatized stocks and puts upward pressure on the cost of capital and downward pressure on the value of NPFs. This also suggests that the shorter the process, the greater the privatization benefits.

99

Table 3.1 Sample Distribution

This table reports sample distribution by country, year, and industry. Panel A: By Country Country N % Country N % Country N % Argentina 40 1.02 Indonesia 148 3.76 Romania 9 0.23 Australia 52 1.32 Ireland 40 1.02 Russia 22 0.56 Austria 121 3.08 Israel 27 0.69 South Africa 28 0.71 Belgium 30 0.76 Italy 184 4.68 Singapore 89 2.26 Brazil 171 4.35 Japan 35 0.89 Slovakia 1 0.03 Bulgaria 2 0.05 Jordan 18 0.46 South Korea 78 1.99 Chile 30 0.76 Kazakhstan 8 0.2 Spain 125 3.18 China 641 16.29 Kenya 4 0.1 Sri Lanka 13 0.33 Colombia 11 0.28 Latvia 4 0.1 Sweden 81 2.06 Croatia 25 0.64 Malaysia 127 3.23 Switzerland 17 0.43 Czech 18 0.46 Mexico 16 0.41 Thailand 49 1.25

100 Denmark 28 0.71 Morocco 14 0.36 Turkey 117 2.97 Egypt 12 0.31 Netherlands 22 0.56 U.K. 135 3.43 Germany 155 3.94 New Zealand 74 1.88 Venezuela 6 0.15 Finland 144 3.66 Norway 52 1.32 France 184 4.68 Pakistan 51 1.3 Greece 86 2.19 Peru 11 0.28 Hong Kong 15 0.38 Philippines 25 0.64 Hungary 80 2.03 Poland 193 4.91 India 200 5.08 Portugal 66 1.68

Table 3.1—continued

Panel B: By Year Panel C: By Industry Year N % Industry N % 1994 8 0.2 Consumer Nondurables 153 3.89 1995 12 0.31 Consumer Durables 115 2.92 1996 25 0.64 Manufacturing 733 18.63 1997 45 1.14 Energy 334 8.49 1998 79 2.01 Chemistry 161 4.09 1999 134 3.41 Business Equipment 111 2.82 2000 174 4.42 Telecom 594 15.1 2001 198 5.03 Utilities 644 16.37 2002 197 5.01 Wholesale 88 2.24 2003 218 5.54 Health 37 0.94 2004 234 5.95 Others 964 24.5 2005 235 5.97

101 2006 252 6.41 2007 260 6.61 2008 273 6.94 2009 270 6.86 2010 305 7.75 2011 298 7.57 2012 295 7.5 2013 233 5.92 2014 189 4.8 Total 3,934 100

Table 3.2. Descriptive Statistics and Correlation Matrix

This table reports summary statistics and correlation matrix for key variables. Panel A: Descriptive Statistics Variable N Mean P25 Median P75 Std TOBIN'S Q 3,934 1.319 0.931 1.146 1.512 0.643 COST OF EQUITY 2,866 0.127 0.094 0.116 0.148 0.052 ZERO RETURN DAYS 3,934 0.105 0.035 0.073 0.139 0.112 FHT 3,934 0.006 0.002 0.004 0.008 0.007 AMIHUD 3,934 0.144 0.000 0.001 0.014 0.535 STATE 3,934 0.265 0.000 0.166 0.511 0.278 STATESQR 3,934 0.148 0.000 0.028 0.261 0.197 LOG MV 3,934 7.585 6.405 7.655 8.821 1.759 BM 3,934 0.907 0.661 0.872 1.074 0.379 STDRET 3,934 0.024 0.016 0.021 0.028 0.019 EM 3,934 0.772 0.285 0.543 0.936 1.226 ANALYST 3,934 12.655 5.000 11.000 19.000 9.128

102 LOSS 3,934 0.003 0.000 0.000 0.000 0.053 ADR_EX 3,934 0.109 0.000 0.000 0.000 0.311 ADR_NEX 3,934 0.421 0.000 0.000 1.000 0.494 INTGAAP 3,934 0.515 0.000 1.000 1.000 0.500 COMMON 3,934 0.258 0.000 0.000 1.000 0.437 LISTED 3,934 0.017 0.002 0.006 0.026 0.024 MEDIA 3,934 59.759 37.000 69.000 80.000 25.290 LGDPC 3,934 9.249 8.243 9.473 10.389 1.277 Panel B: Correlation Matrix TOBIN'S Q COST OF EQUITY ZERO RETURN DAYS FHT AMIHUD STATE TOBIN'S Q 1 COST OF EQUITY -0.30*** 1 ZERO RETURN DAYS -0.08*** 0.10*** 1 FHT -0.14*** 0.23*** 0.83*** 1 AMIHUD -0.03 0.04* 0.30*** 0.33*** 1 STATE -0.07*** 0.09*** -0.10*** -0.06** -0.13*** 1 STATESQR -0.05* 0.08*** -0.07*** -0.04* -0.11*** 0.95***

Table 3. 3A Univariate Analysis

Full Sample Matched Sample ZERO RETURN DAYS FHT ZERO RETURN DAYS FHT Mean Median Mean Median Mean Median Mean Median Privatized firms vs. Other publicly listed firms Privatized firms 0.135 0.088 0.010 0.005 0.140 0.088 0.011 0.005 Other publicly listed firms 0.217 0.149 0.023 0.010 0.135 0.088 0.010 0.005 t-statistics / z-statistics 33.772 38.425 30.825 46.946 1.968 -0.691 1.400 -1.06 Partially privatized firms vs. Other publicly listed firms Partially privatized firms 0.119 0.080 0.009 0.005 0.119 0.080 0.009 0.005 Other publicly listed firms 0.217 0.149 0.023 0.010 0.137 0.085 0.010 0.005 103 t-statistics / z-statistics 33.470 38.614 28.063 44.721 6.250 2.982 3.381 2.02 Fully privatized firms vs. Other publicly listed firms Fully privatized firms 0.170 0.115 0.013 0.006 0.170 0.115 0.013 0.006 Other publicly listed firms 0.217 0.149 0.023 0.010 0.150 0.097 0.012 0.006 t-statistics / z-statistics 11.009 11.812 13.634 18.02 -3.689 -4.671 -1.527 -3.536 Fully privatized firms vs. Partially privatized firms Fully privatized firms 0.170 0.115 0.013 0.006 0.170 0.115 0.013 0.006 Partially privatized firms 0.119 0.080 0.009 0.005 0.128 0.084 0.010 0.005 t-statistics / z-statistics -13.843 -12.692 -8.097 -11.05 -8.558 -8.78 -4.656 -7.566

Table 3.3B Univariate Analysis: The Role of Financial Crisis

Sample: matched sample by firm size and industry Variables ZERO RETURN DAYS FHT t-statistics t-statistics

Before During After During After Before During After During After vs. vs. vs. vs. Before During Before During

Privatized firms vs. Other publicly listed firms Privatized firms 0.114 0.140 0.188 6.181 7.109 0.010 0.011 0.012 1.434 1.104 Other publicly listed firms 0.125 0.133 0.185 1.700 6.625 0.010 0.013 0.013 4.803 -0.168 t-statistics (Other - Privatized firms) 3.947 -0.973 -0.458 -0.543 2.204 1.420

104 Partially privatized firms vs. Other publicly listed firms Partially privatized firms 0.106 0.114 0.163 1.727 6.754 0.009 0.008 0.010 -0.982 1.310 Other publicly listed firms 0.122 0.125 0.184 0.448 6.615 0.009 0.012 0.012 3.635 0.372 t-statistics / z-statistics 5.127 1.369 2.569 0.050 3.277 3.268 Fully privatized firms vs. Other publicly listed firms Fully privatized firms 0.136 0.184 0.229 6.073 3.425 0.012 0.015 0.015 2.768 0.145 Other publicly listed firms 0.134 0.150 0.197 1.756 3.013 0.011 0.014 0.014 2.809 -0.310 t-statistics / z-statistics -0.305 -2.391 -2.241 -0.982 -0.145 -0.718 Fully privatized firms vs. Partially privatized firms Fully privatized firms 0.136 0.184 0.229 6.073 3.425 0.012 0.015 0.015 2.768 0.145 Partially privatized firms 0.114 0.135 0.172 2.640 2.944 0.010 0.011 0.010 -0.461 -0.162 t-statistics / z-statistics -4.504 -3.8135 -4.1341 -1.994 -2.5432 -3.373

Table 3.4 Partial Privatization and Liquidity

This table reports regression results relating partial privatization to stock liquidity. The sample comprises firm-year observations between 1994 and 2014. The dependent variable in Models (1) and (2) is ZERO RETURN DAYS , which is calculated as the percentage of days in the fiscal year for which the stock price does not change. The dependent variable in Models (3) to (4) is FHT , which is the liquidity proxied based on low-frequency data. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t- statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent Variables ZERO RETURN DAYS FHT (1) (2) (3) (4) PARTIAL -0.015** -0.020*** -0.001*** -0.002*** (-2.141) (-2.724) (-2.898) (-3.788) LOG MV -0.010*** -0.012*** -0.001*** -0.001*** (-3.726) (-3.507) (-4.512) (-4.256) BM 0.025*** 0.007 0.002*** 0.001 (3.147) (0.666) (3.349) (1.473) STDRET 0.042 0.394* 0.112*** 0.136*** (0.212) (1.671) (4.623) (5.428) EM -0.001 0.000 -0.000** -0.000 (-1.106) (0.218) (-2.001) (-0.261) ANALYST -0.001** -0.001*** -0.000 -0.000** (-1.990) (-3.537) (-0.435) (-2.187) LOSS 0.072 0.060 0.003 0.002 (1.292) (1.355) (1.415) (1.188) ADR_EX -0.028** -0.004 -0.002** -0.000 (-2.446) (-0.375) (-2.379) (-0.012) ADR_NEX -0.010 0.006 -0.000 0.001 (-1.392) (0.617) (-0.781) (1.325) INTGAAP -0.008 -0.018*** -0.001* -0.001*** (-1.575) (-3.161) (-1.930) (-4.253) COMMON 0.016 -0.001 (0.957) (-0.845) LISTED 0.280 0.025 (0.906) (1.638) MEDIA -0.001*** -0.000 (-3.585) (-1.586) LGDPC 0.007 -0.000 (0.955) (-0.644) Constant 0.226*** 0.233*** 0.014*** 0.016*** (7.432) (3.920) (6.807) (4.593) Country Fixed Effects Yes No Yes No Year Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Observations 3,934 3,766 3,934 3,766 Adjusted R 2 0.229 0.361 0.180 0.179

105

Table 3. 5 State Ownership and Liquidity

This table reports regression results relating state ownership to stock liquidity. The sample comprises firm -year observations between 1994 and 2014. The dependent variable in Models (1) to (4) is ZERO RETURN DAYS , which is calculated as the percentage of days in the fiscal year for which the stock price does not change. The dependent variable in Models (5) to (8) is FHT , which is the liquidity proxied based on low-frequency data. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively . Dependent Variables ZERO RETURN DAYS FHT (1) (2) (3) (4) (5) (6) (7) (8) STATE -0.022** -0.119*** -0.039*** -0.111*** -0.001** -0.008*** -0.003*** -0.008*** (-2.020) (-3.726) (-2.620) (-2.866) (-2.356) (-4.343) (-3.135) (-3.668) STATESQR 0.141*** 0.106** 0.009*** 0.008** (3.355) (1.979) (3.768) (2.574) LOG MV -0.010*** -0.011*** -0.011*** -0.012*** -0.001*** -0.001*** -0.001*** -0.001*** (-3.624) (-3.868) (-3.279) (-3.646) (-4.428) (-4.645) (-4.049) (-4.209)

106 BM 0.026*** 0.025*** 0.008 0.009 0.002*** 0.002*** 0.001 0.001 (3.210) (3.113) (0.850) (0.902) (3.342) (3.260) (1.599) (1.533) STDRET 0.034 0.030 0.376 -0.227 0.112*** 0.112*** 0.135*** 0.135*** (0.171) (0.154) (1.628) (-0.543) (4.565) (4.563) (5.318) (5.336) EM -0.001 -0.001 0.000 0.000 -0.000** -0.000** -0.000 -0.000 (-1.092) (-1.094) (0.291) (0.251) (-1.977) (-1.988) (-0.134) (-0.162) ANALYST -0.001** -0.001* -0.001*** -0.001*** -0.000 -0.000 -0.000** -0.000** (-2.040) (-1.700) (-3.738) (-3.468) (-0.478) (-0.125) (-2.461) (-2.197) LOSS 0.071 0.070 0.056 0.057 0.003 0.003 0.002 0.002 (1.264) (1.265) (1.230) (1.304) (1.373) (1.373) (0.972) (1.090) ADR_EX -0.029** -0.029** -0.006 -0.005 -0.002** -0.002** -0.000 -0.000 (-2.461) (-2.544) (-0.536) (-0.455) (-2.369) (-2.443) (-0.156) (-0.167) ADR_NEX -0.010 -0.010 0.007 0.008 -0.000 -0.000 0.001 0.001 (-1.387) (-1.299) (0.668) (0.768) (-0.789) (-0.677) (1.399) (1.451) INTGAAP -0.008 -0.008 -0.019*** -0.019*** -0.001* -0.001* -0.002*** -0.001*** (-1.554) (-1.594) (-3.304) (-3.450) (-1.902) (-1.943) (-4.344) (-4.288)

COMMON 0.016 0.011 -0.001 -0.001 (1.006) (0.681) (-0.729) (-0.996) LISTED 0.247 0.248 0.022 0.022 (0.823) (0.846) (1.513) (1.529) MEDIA -0.001*** -0.001*** -0.000 -0.000* (-3.578) (-3.777) (-1.540) (-1.884) LGDPC 0.006 0.005 -0.000 -0.000 (0.776) (0.749) (-0.792) (-0.591) Constant 0.227*** 0.229*** 0.241*** 0.271*** 0.015*** 0.015*** 0.017*** 0.017*** (7.428) (7.650) (3.994) (4.417) (6.787) (6.958) (4.487) (4.507) Country Fixed Effects Yes Yes No No Yes Yes No No Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Observations 3,934 3,934 3,766 3,766 3,934 3,934 3,766 3,766 107 Adjusted R 2 0.229 0.361 0.180 0.179 0.491 0.495 0.307 0.310

Table 3.6. Endogeneity Tests

This table reports regression results addressing endogeneity of state ownership using instrumental variable regression, Heckman two-stage selection analysis, and propensity score matching (PSM). In the first-stage regression (unreported), we regress state ownership ( STATE ) on legislative electoral competitiveness ( LIEC ) and a country's deficits over GDP ( DEFICITS/GDP ), along with all control variables and year and industry effects. The sample comprises firm-year observations between 1994 and 2014. The dependent variable in Model (1) to (6) is ZERO RETURN DAYS , which is calculated as the percent of days in the fiscal year for which the stock price does not change. The dependent variable in Model (7) to (12) is FHT, which is the liquidity proxied based on low-frequency data. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997) and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Panel A: ZERO RETURN DAYS Variables IV Regression Heckman PSM (1) (2) (3) (4) (5) (6) STATE -0.555*** -1.015*** -0.088*** -0.160*** -0.025** -0.128*** (-4.687) (-4.090) (-3.987) (-4.134) (-2.563) (-4.712) STATESQR 0.755** 0.123** 0.144*** (1.977) (2.264) (4.255) LOG MV 0.014 0.006 -0.013*** -0.014*** -0.009*** -0.010*** (1.577) (1.177) (-3.427) (-3.659) (-2.777) (-3.209) BM 0.060*** 0.048*** 0.009 0.008 0.013 0.011 (2.687) (3.786) (0.849) (0.716) (1.427) (1.231) STDRET 0.110 0.099 0.414* 0.408* 0.443 0.408 (0.430) (0.472) (1.925) (1.931) (1.332) (1.256) EM -0.000 -0.000 0.001 0.001 0.002 0.002 (-0.061) (-0.181) (0.476) (0.436) (0.795) (0.765) ANALYST -0.002** -0.001** -0.001*** -0.001*** -0.001*** -0.001*** (-2.313) (-2.536) (-3.208) (-2.930) (-4.220) (-3.723) LOSS 0.004 0.031 -0.008 -0.004 -0.008 -0.012 (0.098) (0.984) (-0.761) (-0.380) (-0.348) (-0.536) ADR_EX -0.084** -0.078*** -0.009 -0.009 -0.005 -0.006 (-2.567) (-5.961) (-0.840) (-0.833) (-0.656) (-0.811) ADR_NEX -0.008 -0.005 -0.001 -0.001 0.015*** 0.016*** (-0.446) (-0.673) (-0.129) (-0.080) (2.765) (2.794) INTGAAP -0.019 -0.013* -0.015** -0.014** -0.018*** -0.017*** (-1.438) (-1.684) (-2.449) (-2.338) (-3.097) (-3.000) COMMON -0.023 -0.046*** 0.012 0.008 -0.015 -0.020 (-0.856) (-2.934) (0.655) (0.429) (-1.256) (-1.604) LISTED -0.084 0.033 0.559* 0.573* 1.402*** 1.436*** (-0.170) (0.151) (1.845) (1.885) (5.017) (5.054) MEDIA -0.001** -0.001*** -0.001*** -0.001*** -0.000** -0.001*** (-2.352) (-5.172) (-2.622) (-2.867) (-2.549) (-3.130) LGDPC -0.035** -0.022*** 0.004 0.005 -0.004 -0.002 (-2.420) (-2.691) (0.480) (0.690) (-0.912) (-0.450) LAMBDA 0.020*** 0.015** (2.793) (1.991) Constant 0.585*** 0.556*** 0.314*** 0.310*** 0.271*** 0.279*** (4.714) (10.340) (4.553) (4.534) (5.214) (5.434) Country Fixed Effects No No No No No No Year Fixed Effects Yes Yes Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Yes Yes Observations 3,143 3,143 3,012 3,012 1,952 1,952

108

Adjusted R 2 0.031 0.176 0.217 0.221 0.176 0.183 Panel B: FHT Variables IV Regression Heckman PSM (7) (8) (9) (10) (11) (12) STATE -0.034*** -0.059*** -0.006*** -0.011*** -0.002** -0.010*** (-4.851) (-3.910) (-4.597) (-4.812) (-2.329) (-3.851) STATESQR 0.041* 0.009*** 0.011*** (1.799) (2.675) (3.390) LOG MV 0.001 0.000 -0.001*** -0.001*** -0.000 -0.000* (0.971) (0.336) (-3.850) (-4.071) (-1.366) (-1.814) BM 0.004*** 0.004*** 0.001* 0.001 0.002* 0.001 (2.982) (4.645) (1.662) (1.525) (1.855) (1.632) STDRET 0.112*** 0.112*** 0.129*** 0.129*** 0.212*** 0.209*** (4.799) (5.030) (5.234) (5.256) (10.263) (10.463) EM -0.000 -0.000 0.000 0.000 0.000 0.000 (-0.346) (-0.577) (0.288) (0.234) (1.454) (1.569) ANALYST -0.000* -0.000 -0.000** -0.000* -0.000*** -0.000** (-1.686) (-1.626) (-2.204) (-1.859) (-2.789) (-2.231) LOSS 0.000 0.002 -0.000 -0.000 -0.002 -0.002* (0.099) (0.928) (-0.531) (-0.123) (-1.447) (-1.678) ADR_EX -0.005** -0.004*** -0.000 -0.000 -0.001 -0.001 (-2.408) (-5.711) (-0.381) (-0.379) (-0.951) (-1.045) ADR_NEX -0.000 0.000 0.000 0.000 0.000 0.000 (-0.034) (0.319) (0.582) (0.652) (0.517) (0.540) INTGAAP -0.002* -0.001*** -0.001*** -0.001*** -0.001*** -0.001*** (-1.940) (-2.727) (-3.899) (-3.794) (-2.882) (-2.797) COMMON -0.003* -0.004*** -0.000 -0.001 -0.001 -0.002 (-1.685) (-4.181) (-0.489) (-0.771) (-1.206) (-1.564) LISTED -0.006 -0.000 0.027* 0.028* 0.066*** 0.069*** (-0.229) (-0.001) (1.678) (1.714) (3.198) (3.254) MEDIA -0.000 -0.000*** -0.000 -0.000* -0.000 -0.000 (-1.363) (-3.405) (-1.331) (-1.707) (-0.910) (-1.341) LGDPC -0.003*** -0.002*** -0.000 -0.000 -0.000 0.000 (-3.056) (-3.830) (-0.547) (-0.250) (-0.275) (0.030) LAMBDA 0.001*** 0.001** (3.317) (2.347) Constant 0.037*** 0.035*** 0.018*** 0.017*** 0.010** 0.010*** (4.873) (10.432) (5.158) (5.050) (2.508) (2.699) Country Fixed Effects No No No No No No Year Fixed Effects Yes Yes Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Yes Yes Observations 3,143 3,143 3,012 3,012 1,952 1,952 Adjusted R 2 0.118 0.273 0.341 0.345 0.306 0.315

109

Table 3. 7 Robustness Checks

This table reports regression results relating state ownership to stock liquidity with additional controls and using alternative dependent variables. The dependent variable in Models (1) to (2) is ZERO RETURN DAYS , which is calculated as the percentage of days in the fiscal year for which the stock price does not change. The dependent variable in Models (3) to (4) is FHT, which is the liquidity proxied based on low- frequency data. The dependent variable in Models (5) to (6) is AMIHUD , which is the average stock return over trading volume. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Alternative Dependent Additional controls Variable Dependent Variables ZERO RETURN DAYS FHT AMIHUD (1) (2) (3) (4) (5) (6) STATE -0.030** -0.097** -0.002*** -0.007*** -0.101** -0.450*** (-2.094) (-2.473) (-2.591) (-3.233) (-2.148) (-2.883) STATESQR 0.099* 0.007** 0.517** (1.827) (2.373) (2.388) LOG MV -0.012*** -0.012*** -0.001*** -0.001*** -0.066*** -0.068*** (-3.444) (-3.593) (-4.132) (-4.278) (-3.343) (-3.411) BM 0.012 0.011 0.001* 0.001* -0.040 -0.043 (1.258) (1.193) (1.917) (1.849) (-0.841) (-0.906) STDRET 0.397* 0.395* 0.136*** 0.136*** 3.426*** 3.417*** (1.744) (1.767) (5.307) (5.323) (3.894) (3.934) EM 0.000 0.000 -0.000 -0.000 0.004 0.004 (0.296) (0.274) (-0.111) (-0.139) (0.864) (0.818) ANALYST -0.001*** -0.001*** -0.000** -0.000** -0.002 -0.002 (-3.698) (-3.522) (-2.354) (-2.108) (-1.311) (-0.992) LOSS 0.054 0.056 0.002 0.002 -0.102** -0.092** (1.118) (1.189) (0.883) (0.982) (-2.529) (-2.198) ADR_EX -0.007 -0.007 -0.000 -0.000 -0.012 -0.012 (-0.638) (-0.643) (-0.320) (-0.322) (-0.305) (-0.320) ADR_NEX 0.008 0.008 0.001 0.001 -0.043 -0.042 (0.754) (0.793) (1.488) (1.539) (-1.218) (-1.178) INTGAAP -0.017*** -0.017*** -0.001*** -0.001*** 0.023 0.026 (-3.166) (-3.129) (-4.104) (-4.074) (0.896) (1.034) COMMON 0.009 0.006 -0.001 -0.001 0.075 0.059 (0.526) (0.345) (-1.030) (-1.265) (1.347) (1.064) LISTED 0.524 0.529 0.034** 0.035** -0.931 -0.908 (1.570) (1.586) (2.038) (2.057) (-0.849) (-0.827) MEDIA -0.002*** -0.002*** -0.000*** -0.000*** 0.001* 0.001 (-3.078) (-3.085) (-3.075) (-3.089) (1.655) (1.365) LGDPC 0.005 0.006 -0.000 -0.000 0.074*** 0.079*** (0.662) (0.799) (-0.795) (-0.613) (2.601) (2.741) POLRISK -0.012* -0.011* -0.001*** -0.001** (-1.931) (-1.833) (-2.618) (-2.499) Constant 0.335*** 0.331*** 0.024*** 0.023*** -0.050 -0.054 (4.091) (4.035) (4.767) (4.708) (-0.175) (-0.190) Country Fixed Effects No No No No No No Year Fixed Effects Yes Yes Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Yes Yes Observations 3,751 3,751 3,751 3,751 3,766 3,766 Adjusted R 2 0.188 0.190 0.314 0.316 0.088 0.091

110

Table 3.8 Soft Budget Constraint, State Ownership and Liquidity

This table reports regression results relating soft budget constraint, state ownership, and stock liquidity. The dependent variable is ZERO RETURN DAYS , which is calculated as the percentage of days in the fiscal year for which the stock price does not change. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent Variables ZERO RETURN DAYS (1) (2) (3) STATE 0.053 -0.451** -0.068*** (1.261) (-2.555) (-3.872) GOVBANK -0.102*** (-4.062) STATE*GOVBANK -0.115** (-2.227) LIMFOREIGN -0.176*** (-6.807) STATE*LIMFOREIGN 0.107** (2.356) POLITICAL INDP -0.003 (-0.214) STATE*POLITICAL INDP 0.072** (2.166) LOG MV -0.011*** -0.012*** -0.009*** (-3.150) (-3.288) (-2.737) BM 0.017* 0.014 0.014 (1.809) (1.283) (1.489) STDRET 0.325 0.458** 0.393* (1.319) (2.396) (1.738) EM 0.001 0.001 0.000 (0.643) (0.374) (0.054) ANALYST -0.001*** -0.001*** -0.001*** (-3.424) (-3.016) (-3.115) LOSS 0.036 0.070 0.059 (0.732) (1.099) (1.318) ADR_EX -0.020* 0.008 -0.010 (-1.864) (0.766) (-0.995) ADR_NEX -0.000 0.015 0.005 (-0.039) (1.359) (0.479) INTGAAP -0.018*** -0.023*** -0.017*** (-3.225) (-4.045) (-3.044) COMMON -0.031** -0.018 0.017 (-2.002) (-0.947) (1.151) LISTED 0.320 0.551 0.226 (1.291) (1.496) (0.759) MEDIA -0.001*** 0.000 -0.001*** (-3.462) (0.082) (-3.936) LGDPC -0.013* -0.002 0.001 (-1.891) (-0.261) (0.085) Constant 0.426*** 0.950*** 0.270*** (6.250) (7.312) (4.858) Country Fixed Effects Yes Yes No Year Fixed Effects Yes Yes Yes Industry Fixed Effects Yes Yes Yes Observations 3,934 3,934 3,766 Adjusted R 2 0.229 0.361 0.180

111

Table 3. 9 State Ownership and Liquidity: The Role of Financial Crisis

This table reports regression results investigating the role of financial crisis. The sample comprises firm - year observations between 1994 and 2014. The dependent variable in Models (1) to (2) is ZERO RETURN DAYS , which is calculated as the percentage of days in the fiscal year for which the stock price does not change. The dependent variable in Models (3) to (4) is FHT , which is the liquidity proxied based on low- frequency data. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t- statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent Variables ZERO RETURN DAYS FHT (1) (2) (3) (4) STATE -0.003 -0.016 -0.000 -0.001 (-0.197) (-0.788) (-0.066) (-0.965) BEFORE CRISIS -0.002 0.019 0.001 0.002 (-0.086) (0.744) (0.940) (1.102) DURING CRISIS 0.008 0.004 0.001* 0.000 (0.903) (0.443) (1.938) (0.933) STATE * BEFORE CRISIS -0.024 -0.025 -0.002 -0.002* (-1.307) (-1.141) (-1.629) (-1.658) STATE * DURING CRISIS -0.030* -0.045** -0.002** -0.003** (-1.933) (-2.579) (-2.125) (-2.554) LOG MV -0.010*** -0.011*** -0.001*** -0.001*** (-3.628) (-3.245) (-4.420) (-4.038) BM 0.025*** 0.008 0.002*** 0.001 (3.144) (0.791) (3.274) (1.523) STDRET 0.040 0.379* 0.112*** 0.135*** (0.205) (1.673) (4.528) (5.277) EM -0.001 0.000 -0.000** -0.000 (-1.106) (0.295) (-1.987) (-0.162) ANALYST -0.001** -0.001*** -0.000 -0.000** (-2.133) (-3.813) (-0.579) (-2.519) LOSS 0.072 0.058 0.004 0.002 (1.272) (1.251) (1.386) (1.030) ADR_EX -0.028** -0.006 -0.002** -0.000 (-2.421) (-0.526) (-2.335) (-0.134) ADR_NEX -0.010 0.007 -0.000 0.001 (-1.378) (0.669) (-0.768) (1.419) INTGAAP -0.007 -0.017*** -0.000 -0.001*** (-1.328) (-3.040) (-1.556) (-3.868) COMMON 0.016 -0.001 (0.995) (-0.746) LISTED 0.253 0.023 (0.841) (1.543) MEDIA -0.001*** -0.000 (-3.606) (-1.573) LGDPC 0.005 -0.000 (0.713) (-0.866) Constant 0.231*** 0.227*** 0.014*** 0.016*** (7.743) (3.794) (6.182) (4.312) Country Fixed Effects Yes No Yes No Year Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Observations 3,934 3,766 3,934 3,766 Adjusted R 2 0.404 0.181 0.492 0.309

112

Table 3.10 Liquidity, Cost of Capital, and Valuation

This table reports regression results relating liquidity to cost of equity and valuation. The sample comprises firm-year observations between 1994 and 2014. The dependent variable for Models (1) and (2) is average implied cost of capital, which is calculated as the yearly average of four commonly used empirical estimation techniques as described in the Appendix. The dependent variable for Models (3) and (4) is Tobin's Q, which is calculated as total assets less book value plus market value scaled by total assets. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummies, industry dummy variables based on industry groupings in Fama and French (1997), and year dummy variables. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent Variables COST OF EQUITY TOBIN'S Q ZERO RETURN DAYS FHT ZERO RETURN DAYS FHT (1) (2) (3) (4) ILLIQUIDITY 0.042*** 1.042*** -0.711*** -13.906*** (3.651) (4.715) (-5.724) (-7.383) LOG ASSETS 0.003*** 0.004*** -0.094*** -0.098*** (2.629) (2.777) (-5.187) (-5.451) LEV 0.047*** 0.046*** 0.296** 0.312** (3.729) (3.645) (2.142) (2.320) CASH 0.571*** 0.572*** (3.086) (3.119) CAPX 0.369 0.316 (1.259) (1.089) DV DUMMY -0.098*** -0.102*** (-3.639) (-3.777) CASH FLOW 4.343*** 4.245*** (12.909) (12.745) NWCAP -0.402*** -0.422*** (-2.958) (-3.146) ADR_EX -0.007 -0.006 -0.079 -0.090 (-1.171) (-0.961) (-1.187) (-1.392) ADR_NEX -0.009** -0.009** 0.009 0.012 (-2.060) (-2.029) (0.195) (0.269) STDRET 0.235*** 0.137** (3.399) (2.568) BIAS 0.508 0.508 (0.965) (0.958) INFLATION 0.017 0.020 (0.356) (0.421) Constant 0.124*** 0.119*** 1.488*** 1.592*** (6.548) (6.433) (5.057) (5.414) Country Fixed Effects Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes Yes Yes Yes Observations 2,776 2,776 3,856 3,856 Adjusted R 2 0.260 0.265 0.465 0.470

113

Figure 3.1 Scatter Plot of the Relation between State Ownership and Illiquidity

114

Figure 3.2 Quadratic and Linear Relationship between State Ownership and Stock Liquidity

115

Figure 3.2 –continued

116

CHAPTER 4

STATE OWNERSHIP AND TRADE CREDIT : EVIDENCE FROM PRIVATIZATION

4.1 Introduction

Trade credit, which arises when a supplier allows a customer to delay payment, is an important source of external financing for both small and large firms around the world

(Petersen and Rajan, 1997; Maksimovic, 2001; Giannetti, Burkart, and Ellingsen, 2011).

According to the U.S. Flow of Funds Accounts, as of 2012, accountable payables are three times as large as bank loans for U.S. firms. COMPUSTAT data indicate that trade receivables account for around 15% of corporations’ total sales, and finance about 90% of the world merchandise trade (Williams, 2008). Despite its economic significance, there is little attention paid to trade credit in the literature relative to firms’ other financial activities.

In this essay, we add to the literature by examining the effect of state ownership on trade credit in an international sample of newly privatized firms (NPFs). Privatization provides us with an ideal setting in which to isolate the importance of state ownership to trade credit provisions: ownership structure undergoes dramatic changes during the privatization process as government ownership is transferred to new private owners. Privatization typically occurs gradually, allowing us to track state ownership as it changes over time.

This change in the level of state ownership in turn captures changes in incentives and objective functions of the firm from social/political ends to economic and profit maximization (e.g., Boubakri et al., 2005; Denis and McConnell, 2003; Guedhami et al.,

2009), and thus shifts in agency problems and information asymmetry (Shleifer and Vishny ，

117

1997; Denis and McConnell, 2003; Dyck, 2001), allowing us to test their impact on firms along the supply chain. Moreover, the speed of implementation of privatization tends to vary across industries and countries, allowing us to observe different levels of state ownership within and across various institutional environments.

Focusing on state ownership is important in its own right: following the recent financial crisis, government bailout programs aimed at rescuing financially distressed firms led to a significant increase in state ownership around the world, apparently reversing three decades of privatization that decreased the role of the state in the economy. State ownership of equity accounts for nearly one-fifth of total stock market capitalization worldwide

(Borisova et al., 2015; Megginson, 2016). This “reverse privatization” phenomenon has renewed the debate about the role of governments as shareholders, fostering research on the impact of state ownership on corporate policies and firm value.24 Moreover, examining trade credit in the context of privatization is particularly relevant because it allows us to understand whether reduction in state ownership affects firms along the supply chain. For firms in poorly developed financial markets, trade credit provides an alternative source of funds and therefore contributes to higher economic growth (Fisman and Love, 2003).

Therefore, understanding the relation between state ownership and trade credit has significant policy implications. The success of privatization reform depends not only on the post-privatization performance of privatized firms, but also on its impact on other firms.

24 See Megginson (2016) for a comprehensive survey. Prior literature examines the role of state ownership on the valuation of corporate assets and equity (e.g., Holland, 2016), the cost of equity (Ben-Nasr et al., 2012), the cost of debt (Borisova and Megginson, 2011; Borisova et al., 2015), corporate risk-taking (Boubakri et al., 2013), governance quality (Borisova et al., 2012), and corporate investment efficiency (Jaslowitzer et al., 2016).

118

The redistribution view of trade credit (Meltzer, 1960; Petersen and Rajan, 1997;

Nilsen, 2002) asserts that firms with better access to external finance will redistribute the credit to less advantaged firms in the form of trade credit (Love, Preve, and Sarria-Allende,

2007; Schwartz, 1997; Boissay and Gropp, 2007). State ownership, on the other hand, is associated with implicit government guarantees, preferential access to credit, and soft budget constraints, particularly during times of financial distress (Borisova et al., 2015;

Borisova and Megginson, 2011; Faccio et al., 2006). A government can also often relax a state-owned firm’s budget constraints by providing tax discounts, preferential access to credit, and other forms of support (Kornai et al., 2003). This suggests that firms with high state ownership are likely to pass on their financing advantage to less privileged customers in the form of trade credit.

The market power view of trade credit (Brennan, Maksimovic, and Zechner, 1988) suggests another explanation as to why suppliers to extend credit to customers. Low competition among suppliers in an input market may create incentives to use trade credit as a tool for price discrimination. The rationale is that sellers with higher profit margins have an incentive to sell additional units on credit as long as the profits from additional sales outweigh the cost of providing trade credit. This suggests that firms in low competition industries tend to extend more trade credit for price discrimination. On the other hand, trade credit can be used as a strategic instrument in the oligopolistic supplier market (Love, 2011). For example, Singh (2017) shows that trade credit is used by incumbent firms to competitors from entering the market. However, recent empirical evidence shows that firms with less market power extend more credit (Fabbri and Klapper,

2008) and that a customer that generates a large share of its supplier’s profits tends to

119

receive more credit (Giannetti, Burkart, and Ellingsen, 2011). Privatized firms are usually the largest and most powerful companies producing a vital good or service. These firms with a high level of retained state ownership are expected to have more market power to discriminate between cash and credit customers, or use trade credit to deter competition.

However, privatized firms with little state ownership may lose bargaining power over large buyers that request trade credit provisions. Overall, the impact of state ownership on the supply of trade credit is thus an open question.

To test our conjecture, we use a unique sample of 552 NPFs from 60 countries to investigate how state ownership affects the supply of trade credit. This sample provides an ideal setting for isolating the real effect of state ownership on trade credit. The dramatic change in ownership structure in the course of privatization is associated with changes in agency problem, information asymmetry, and implicit government guarantees making it suitable for addressing questions on how state ownership affects trade credit. We find strong and economically significant evidence that state ownership is positively related to the supply of trade credit. Economically, an increase in government ownership from the

25 th to the 75 th percentile leads to an 18% increase in trade credit, while all other variables remain constant. This result is robust to alternative measures of state ownership, alternative measures of trade credit, and additional control variables.

A key concern in the privatization literature is the endogeneity of state ownership.

The first source of endogeneity is sample selection bias, as suggested by Megginson and

Netter (2001). Governments may intentionally privatize the healthiest firms to make privatization look good or, alternatively, they may divest firms that perform poorly.

Therefore, it is possible that the observed relation between state ownership and trade credit

120

simply reflects the choice by governments. Another source of endogeneity is that the relation between state ownership and trade credit is driven by unobserved determinants of trade credit that also explain residual state ownership following privatization. To address this concern, we employ instrumental variables, the Heckman selection model, and propensity score matching to address the endogeneity issue. Consistent with our main results, we find that the coefficient of state ownership continues to load positively and is statistically significant.

We also investigate how state ownership affects trade credit in countries with different levels of development of financial markets. Fisman and Love (2003) show that industries with higher dependence on trade credit financing exhibit higher growth in countries with weaker financial institutions. Consistent with their findings, we find that the relation between state ownership and trade credit supply is stronger in countries with a poorly developed financial market, suggesting that SOEs pass their financing advantage to buyers and provide an alternative source of financing in those countries. In addition, we examine how product market competition affects the relation between state ownership and the supply of trade credit. Interestingly, we find that state ownership is associated with more trade credit in industries with higher competition, lower price-cost margin, smaller market size, and low entry cost. The results are consistent with Barrot’s (2016) findings that financially stronger firms extend trade credit to expose their financially weaker rivals to liquidity shocks.

We further investigate the soft-budget constraint explanation underlying the relation between state ownership and trade credit. Specifically, we examine whether country-level factors related to government ownership of banks, and restrictions on foreign

121

banks, affect the relation between state ownership and trade credit. We find that this relation is stronger in countries with higher levels of state ownership of banks and weaker in countries with fewer limits on foreign banks.

This essay contributes to the privatization literature in several ways. First, our study of the impact of state ownership on trade credit makes an important contribution to the privatization literature. Prior literature provides evidence on how state ownership in NPFs is associated with financial and operating performance after privatization. However, there is little evidence on how post-privatization state ownership affects firms along the supply chain of the NPFs. This essay extends this literature by providing preliminary insights on this issue. Second, our findings of how state ownership facilitates reallocating funds through trade credit in countries with a weak financial market extend our understanding of privatization. In particular, our results suggest that a gradual reduction of state ownership is recommended in countries without a well-established financial market. Firms along the supply chain of the NPFs are likely to experience a dramatic shock in financing if both trade credit and external financing are no longer in place.

Further, this essay contributes to the trade credit literature by showing that ownership is one of the key determinants of trade credit. We shed light on how ownership structure is associated with different incentives to provide trade credit. By considering the impact of state ownership, we broaden our understanding of how ownership structure affects the supply of trade credit.

The remainder of this essay is organized as follows. In Section 4.2, we review the related literature and develop our main hypotheses. In 4.3, we describe the sample and variables, and provide descriptive statistics. Section 4.4 reports our main empirical results

122

and robustness tests. Section 4.5 presents results on the impact of financial market development and product market competition on the relation between state ownership and trade credit. Section 4.6 concludes.

4.2 Literature Review and Hypothesis Development

4.2.1 Trade Credit

There are several theories providing explanations for trade credit by suppliers. We outline these theories of trade credit, which fall into five categories: (1) financing advantage view, (2) market power view, (3) implicit warranty view, (4) transaction cost view, and (5) redistribution view.

4.2.1.1 Financing Advantage View

The first set of theories suggest that trade credit will be more likely in circumstances where suppliers have an advantage over financial institutions in investigating the credit quality of buyers, as well as the ability to monitor and force repayment (Schwartz, 1974;

Mian and Smith, 1992; Frank and Maksimovic, 1998). Suppliers have cost advantage over financial institutions because the collateral is more valuable in case of default. Although financial institutions can claim the collateral to pay off loans, suppliers can repossess the collateral and resell it on more favorable terms than financial institutions can. Suppliers also have an advantage over traditional lenders in terms of assessing credit quality of buyers and forcing repayment. Monitoring of buyers’ credit quality can occur when suppliers visit the buyers, while the size and timing of the buyers’ orders also reveal information about the buyers’ creditworthiness. Moreover, if the goods being supplied are unique with few economical alternative sources, suppliers can force repayment of trade

123

credit by threatening to cut off further supplies. A traditional lender, however, has less immediate power by threatening to withdraw further financing.

4.2.1.2 Market Power View

Suppliers may provide trade credit even if they do not have financing advantage over traditional lenders. Suppliers with a high price-cost margin can use trade credit as a tool for price discrimination. The rationale is that sellers with a higher profit margin have the incentive to sell additional units on credit as long as the profits from additional sales outweigh the cost of providing trade credit. As a result, customers that are credit rationed could afford to buy more products from suppliers. Therefore, trade credit offered to buyers who are credit rationed permits them to increase their demand (Petersen and Rajan, 1997).

Indeed, suppliers have an incentive to extend more trade credit because it allows additional selling of the product while not affecting their existing sales. On the other hand, trade credit can be used as a strategic instrument in the oligopolistic supplier market (Love, 2011).

Singh (2017) shows that trade credit is used by incumbent firms to deter competitors from entering the market. Recent empirical evidence shows that firms with less market power, however, extend more credit (Fabbri and Klapper, 2009) and that a customer that generates a large share of its supplier’s profits tends to receive more credit (Giannetti, Burkart, and

Ellingsen, 2011).

4.2.1.3 Implicit Warranty View

Information asymmetry between suppliers and buyers is another reason for trade credit. Buyers are exposed to a moral hazard problem if they make payment before the quality of products can be verified. Therefore, sellers can use trade credit as a tool to certify product quality (Emery and Nayar, 1998; Lee and Stowe, 1993; Long, Malitz, and Ravid,

124

1993). If the product’s quality is poor, the customer can withhold payment and return the product to the seller.

4.2.1.4 Transaction Cost View

Trade credit may be used to reduce the transaction cost because it enables buyers to separate the payment cycle from the delivery schedule (Ferris, 1981). Using trade credit to maintain smooth production cycles, firms are able to build up large inventories and manage their inventory position better.

4.2.1.5 Redistribution View

The redistribution view of trade credit (Meltzer, 1960; Petersen and Rajan, 1997;

Nilsen, 2002) assets that firms with better access to external finance will redistribute the credit to less advantaged firms in form of trade credit (Love et al., 2007; Schwartz, 1997;

Boissay and Gropp, 2007).

4.2.2 State Ownership and Trade Credit

4.2.2.1 State Ownership and Market Power View of Trade Credit

Some of the largest and most powerful companies producing a vital good or service are newly privatized firms. For instance, the 13 biggest oil firms are all state-owned or state-backed ( Economist , 2012). Retaining a high stake of state ownership allows NPFs to exert monopoly power over their buyers, which has several implications for the supply of trade credit. Monopoly power enables NPFs to have financing advantage over traditional lenders because they can threaten to cut off buyers’ supply and they have higher liquidity value in case of default. Moreover, monopoly power provides an incentive for suppliers to use trade credit as a tool for price discrimination given their high profit margin. However, prior literature also documents that firms with financial troubles or weak bargaining power

125

are more likely to extend trade credit compared to firms with strong bargaining power

(Petersen and Rajan, 1997). Consistent with this view, Fabbri and Klapper (2008) find that firms with relatively weaker market power are more likely to extend trade credit.

4.2.2.2 State Ownership and the Redistribution View of Trade Credit

Proponents of state ownership suggest that it allows politicians to pursue socially desirable objectives such as investing in infrastructure in rural or undeveloped regions that would otherwise be unable to attract private investors. Yarrow, King, Mairesse, and Melitz

(1986) present a theoretical model of the justification for state ownership, which suggests that it reduces the public-good problem, and can take immediate action once there is any deviation between the social and private returns in goods and factor markets. Similarly,

Sappington and Stiglitz (1987) argue that state ownership reduces transaction costs when government intervention in economic affairs is needed and such intervention is particularly important in the presence of market failure and weak capital markets. Cook and Kirkpatrick

(1988) suggest that market failures are especially damaging in developing countries with a poorly developed financial market, where private firms have limited access to external finance. A well-functioning financial market is important to effectively allocate resources and is therefore important for economic growth. As a response to financial market failure,

SOEs may provide an alternative source of funds, namely trade credit, to firms in countries with a poorly developed financial market. Fisman and Love (2003) find that industries with higher dependence on trade credit exhibit stronger growth in countries with a weak financial market. Therefore, SOEs are likely to extend more trade credit in response to market failure.

126

The soft budget constraint theory asserts that government can relax an SOE’s budget constraint by using tax discounts, easy access to finance, and other forms of support

(Kornai et al., 2003). Consistent with this view, Anderson et al. (2000) find that privatization and decentralization are essential in reducing soft budget constraints because when the central government retains ownership in privatized firms, more than two-thirds of enterprises still have soft budget constraints. Petersen and Rajan (1997) find that firms with better access to credit usually extend more trade credit. Therefore, we expect that as government stakes increase, NPFs are likely to extend more trade credit.

The above discussion leads to our two competing hypotheses on the relation between state ownership and trade credit of NPFs.

Hypothesis 1a: Higher state ownership in NPFs is associated with less trade credit.

Hypothesis 1b: Higher state ownership in NPFs is associated with more trade

credit.

4.3 Sample, Variables, and Descriptive Statistics

4.3.1 Sample Selection

To empirically investigate how state ownership affects trade credit, we construct a sample of 552 NPFs from 60 countries over the period 1981–2008. The ownership data, from Boubakri et al. (2013), track the change of ownership for seven years after the first privatization. The sample of newly privatized firms provides us an ideal setting in which to investigate the role of state ownership in trade credit. It allows us to analyze the time- varying effect of state ownership on trade credit given that state ownership changes around privatizations. Moreover, these data cover firms from countries with diverse financial

127

markets and institutional environments, which allows us to investigate the heterogeneous effects of state ownership on trade credit conditional on various institutional environments.

We manually match the sample of NPFs with financial data obtained from

Compustat Global. To rule out the possible effect of regulations on trade credit, we exclude financial firms (i.e., four-digit SIC codes between 6000 and 6999) from our sample. We also exclude observations with missing key variables and with missing Standard Industrial

Classification (SIC) codes. These procedures yield a sample of 4,406 firm-year observations.

Table 4.1 presents sample distribution by year, industry, and region in Panels A, B, and C, respectively. Panel A shows the number of firm-year observations and firms privatized from 1981 to 2008. There was significant growth in privatizations through the

1980s and the 1990s. Panel B shows the sample distribution across industries as defined by Campbell’s (1996) classification. Not surprisingly, around 30% of our sample firms are in utilities, 16% in basic industries, and 12% in transportation. 25 Panel C reveals how the sample is widely distributed across regions. Around 44% of our sample firms are located in Europe and Central Asia, 40% in East and South Asia and the Pacific, 11% in Latin

America and the Caribbean, and 5% in Africa and the Middle East.

4.3.2 Variables

4.3.2.1 Trade Credit

Following prior literature (Petersen and Rajan, 1997), we define trade credit ( Trade

Credit) as the amount of trade receivables scaled by total sales obtained from Compustat

(Receivables/Sales ). Trade receivables is the amount owned by customers for goods and

25 Our main findings are not sensitive to sequentially excluding each industry from our analysis.

128

services sold in the ordinary course of business. 26 In all the regressions the dependent variable is multiplied by 100. In addition to using trade receivables over total sale as our dependent variable, we use the log (1+trade receivables) as alternative dependent variable.

The Appendix summarizes the definitions for these and all other variables we use in our analyses.

4.3.2.2 State Ownership

We define state ownership ( STATEOWN ) as the percentage of shares held by a government. For robustness, we also use Control , an indicator variable equal to one for firms in which the government retains control (i.e., more than 50% of a firm’s shares) as an alternative dependent variable. Moreover, we follow prior literature (Boubakri et al.,

2013) and use the averaged state ownership over the sample period ( AVG_STATEOWN ) as another alternative independent variable.

4.3.2.3 Control Variables

We also include several firm- and country-level variables following prior literature

(Giannetti et al., 2011; Love et al., 2007; Petersen and Rajan, 1997) that are related to trade credit. In particular, we include firm size ( LOG (SALES) ), the natural logarithm of total sales in $US millions; profits ( ROS ), measured as the ratio of income before extraordinary items to total sales; cash holdings ( CASH HOLDINGS ), defined as the ratio of cash and short-term investments to total assets; firm’s fixed assets ( FIXED ASSETS ), defined as the ratio of total property, plant, and equipment to total assets; sales growth ( SALES GROWTH ), measured as the growth rate of total sales; and gross profit margins ( GROSS PROFIT

MARGIN ), which is the ratio of total sales less cost of goods sold to total sales. We also

26 Compustat separates accounts receivable due to trade from other receivables but does not do the same for accounts payable.

129

control several country-level variables: GDP per capita ( GDP PER CAPITA ), which is the natural logarithm of GDP per capita in constant 2002 U.S. dollars, and private credit

(PRIVATE CREDIT ), which is the financial resources provided to the private sector by financial corporations. Both variables are from World Development Indicators. To further ensure the causal relation between state ownership and trade credit, we lag all independent variables and control variables one year, and include country, industry, and year dummies in all regressions. We also winsorize all financial variables at the 1% and 99% percentiles.

4.3.3 Descriptive Statistics

Panel A of Table 4.2 presents summary statistics for the key variables used in our regression. On average, the ratio of trade credit to sales is around 0.17, with a standard deviation of 0.14. It varies from 0.00 to 0.85. Residual state ownership ( STATEOWN ) has a mean (median) of 0.26 (0.12), in line with a sharp decline in state ownership after privatization (Boubakri et al., 2005). Panel B of Table 4.2 presents Pearson correlation coefficients between the firm-level variables. We find that our primary proxy for trade credit provision ( TRADE CREDIT ) is negatively associated with firm size, profitability, cash holdings, and fixed assets, but positively associated with gross profit margin and GDP per capita.

4.4 Results

4.4.1 The Relation between State Ownership and Trade Credit

Table 4.3 reports regression results on the impact of state ownership on trade credit using a pooled multivariate regression framework. We estimate the regressions using ordinary least squares (OLS) and calculate robust standard errors clustered at the firm level.

In our baseline specification (Model 1), we use the Accounts Receivable/Sales ( TRADE

130

CREDIT ), multiplied by 100 times, as the dependent variable, and test for the effect of

State Ownership . Specifically, we estimate the following regression (superscripts omitted for simplicity):

= + + +

+ + +

+ +

+ +

+ + + .

Our focus is on the coefficient , which measures the sensitivity of trade credit to the level of state ownership.

In Model 1, the results suggest that STATEOWN is positively associated with trade credit in newly privatized firms. 27 This association is statistically significant at the 1% level.

These results support hypothesis H 2, which posits that governments provide more trade credit. Our results are also economically significant. Indeed, the coefficients on

STATEOWN suggest that moving state ownership from the 25 th percentile (0.00) to the 75 th percentile (0.51) results in an 18% decline in the trade credit proxy (from 0.17 to 0.139), holding all other variables at their mean values. Turning to the control variables, we find that the coefficient of LOG (SALES) loads negatively and statistically significant at the 1% level, suggesting that smaller firms tend to offer a higher proportion of sales on credit.

Consistent with Giannetti et al. (2001), we find that FIXED ASSETS is negatively associated with trade credit. We also show that CASH HOLDINGS and SALES GROWTH are negatively associated with trade credit, suggesting that when firms hold less internal

27 In unreported results, we use alternative measures of trade credit ( LOG (1+ACCOUNTS RECEIVABLE) ). We find that STATEOWN continues to load positively and statistically significant at the 5% level.

131

cash, or have a decline in sales, they tend to offer more trade credit. The coefficient on

GROSS PROFIT MARGIN is positive and significant at the 1% level, indicating that firms with larger profit margins tend to offer more trade credit to expand customer bases.

One potential concern with the benchmark regression results as shown in Models 1 and 2 is that STATEOWN may not be exogenous. In the context of privatization, as

Megginson and Netter (2001) point out, sample selection bias can arise from several sources. Governments may intentionally choose to privatize the healthiest firms first.

Therefore, poorly performing firms are more likely to exhibit higher state ownership and extend more trade credit to preserve sales. 28 Moreover, the relation between state ownership and trade credit could be driven by unobserved determinants of trade credit that also explain residual state ownership.

In Model 2 we confront the issue of endogeneity using two-stage instrumental variable estimations, following prior privatization literature (Borisova and Megginson,

2011; Boubakri et al., 2013; Guedhami et al., 2009). We use a country’s political orientation, LEFT , and the voting share of government parties, NUMVOTE , as instruments for state ownership. Both variables are derived from the World Bank’s Database of

Political Institutions by Beck et al. (2001). LEFT is a dummy variable indicating whether the party orientation with respect to economic policy can be defined as communist, socialist, social democratic, or left wing, while NUMVOTE is the voting share of controlling parties.

The choice of instrument is motivated by prior privatization that shows a country’s political system is associated with residual state ownership (Bortolotti, Fantini, and Siniscalco, 2004;

Boubakri, Cosset, Guedhami, and Saffar, 2011; Boubakri, Cosset, and Saffar, 2013). In the

28 Petersen and Rajan (1997) find that firms in financial trouble may extend trade credit to preserve sales and customers are reluctant to repay these suppliers.

132

unreported first-stage regression, we find Hansen’s J statistics for overidentification test fail to reject the null hypothesis that the chosen instrument variables are properly excluded from the main model, with p value being 0.91. For the underidentification test, we find that the Kleibergen and Paap (2006) rk LM statistic is 42, with its p value being 0.00, indicating that it is appropriate to exclude instruments for the endogenous variables. Finally, the Wald

F statistic is 21.74, which is larger than Staiger and Stock’s (1997) 10% size threshold of

19.9. Model 2 shows regression results of the second-stage instrumental estimation. We find that STATEOWN enters positively and is statistically significant at the 5% level, suggesting that our results, using two-staged instrumental regressions, are not affected by the endogeneity issue.

Another concern with our analysis, as argued above, is that the residual government ownership is influenced by firm characteristics, which in turn affect firms’ trade credit policy. Therefore, in Models 3 and 4, we use Heckman two-stage selection analysis and propensity score matching (PSM) to address these concerns. In the Heckman two-stage analysis, we first use a Probit model to predict whether governments retain control over the privatized firms. 29 To be specific, we regress CONTROL on LEFT , NUMVOTE , the full set of control variables, and country, industry, and year fixed effects (as in Model 1 of

Table 4.3). In the second stage, we include the inverse Mills ratio ( LAMBDA ), which is estimated from the first-stage regression, as an additional control variable. The results in

Model 3 show that STATEOWN is positively associated with trade credit at the 1% level,

29 Alternatively, we use whether government retains any share of the privatized firms (i.e. STATEOWN > 0) as the dependent variable in the first stage and find our results are unchanged.

133

confirming the results shown in Models 1 and 2. In addition, LAMBDA loads negatively and is statistically significant at the 1% level.

In Model 4, we use PSM analysis to randomize the sample selection procedure by using observed firm- and country-level characteristics to match firms under government control with firms not under government control. 30 In the first stage, we use the same Probit model as in the Heckman first-stage analysis. We then match state-controlled firms with the closest propensity score. In the second stage (Model 4), we estimate the regression using the matched sample. Consistent with our main analysis, the instrumental variable analysis, and the Heckman analysis we continue to find that the coefficient on STATEOWN loads positively and significantly at the 5% level.

In Models 5 and 6, we use alternative proxies of state control variables. In Model

5, we follow Boubakri et al. (2005) and Guedhami et al. (2009) and replace STATEOWN with a dummy variable CONTROL equal to one if the government retains control over the privatized firm (i.e., owns more than 50% of its shares). We find that the coefficient of

CONTROL is positive and statistically significant at the 1% level. It is also economically significant. Firms with government control provide 18% more trade credit than firms without government control. In Model 6, we replace STATEOWN with the averaged percentage of state ownership ( AVG _STATEOWN ) over the sample period. As argued in

Boubakri et al. (2013), given the staggered nature of government sales after privatization, using STATEOWN may overestimate the impact of state ownership. The results remain statistically unchanged, with AVG _STATEOWN loading negatively and significantly at the

1% level.

30 Similarly , we obtain statistically similar results when we match firms that are partially privatized (i.e., State Ownership > 0) with firms that are fully privatized (i.e., State Ownership = 0).

134

In summary, the results in Table 4.4 show that state ownership is positively associated with trade credit. These results hold when we address the endogeneity issue, and when we use alternative measures of state control.

4.4.2. Robustness Checks

Table 4.4 presents additional tests to ensure the robustness of our results. In Model

1, we include state ownership and foreign ownership simultaneously and find that

STATEOWN continues to enter the regression positively and statistically at the 1% level, while FOREIGNOWN loads positively but is statistically insignificant. Boubakri et al.

(2005) find that local investors and employees absorb the stake divested by the government; therefore it is possible that the effect of state ownership on trade credit is driven by local

(LOCALOWN ) and employee ownership ( EMPLOYEEOWN ). In Model 2 we include these two variables and, indeed, we find LOCALOWN is positively associated with trade credit.

Interestingly and more importantly, we find STATEOWN continues to load positively and statistically significant at the 1% level.

In Model 3, we further include leverage ( LEVERAGE ) and capital expenditure

(CAPEX ) to account for the effect of firms’ leverage and investment on trade credit. We find that CAPEX loads positively and is statistically significant at the 5% level. More importantly, the coefficient on STATEOWN remains positive and statistically significant at the 1% level. Interestingly, the coefficient of STATEOWN becomes even larger after including more control variables than in Model 1.

Models 4 to 7 use alternative estimation methods to ensure that our results are not driven by estimation techniques. In Model 4, we adjust standard error clustered at the firm and year as suggested by Petersen (2009). The results show that STATEOWN enters

135

positively and statistically significant at the 1% level. In Models 5 and 6 we use Newey-

West regression and Prais-Winsten, respectively, to account for autocorrelation and heteroscedasticity in standard error. In both regressions, we show that STATEOWN continues to be positively associated with trade credit. In Model 7, given that our dependent variable is always positive, we use Tobit regression to account for the censored dependent variable. We show that STATEOWN is positive and statistically significant.

4.4.3 Trade Credit from State Ownership: Additional Analysis

In this section, we further explore the cross-sectional determinants of trade credit.

In particular, we investigate how financial market development and product market competition shape corporate policy of NPFs in terms of trade credit.

4.4.3.1 Trade Credit as an Alternative Source of Funds

Fisman and Love (2003) find that implicit borrowing in the form of trade credit can be an alternative source of funds for firms in countries with a poorly developed financial market. Petersen and Rajan (1997) find that firms with better access to finance offer more trade credit. Taken together, we expect that more state ownership is associated with access to external finance, and therefore supply more trade credit to fund firms with limited access to financial markets.

Table 4.5 presents results of relating state ownership, financial development, and trade credit. In particular, we use three measures of financial development. In Model 1, we use private credit ( PRIVATE CREDIT ), which is the credit provided to the private sectors by financial corporations. A higher value of PRIVATE CREDIT means a more developed financial market. We regress TRADE CREDIT on STATEOWN and its interaction with

PRIVATE CREDIT , and the controls. Consistent with our expectation, we find the

136

coefficient on the interaction between STATEOWN and PRIVATE CREDIT loads negatively and is statistically significant at the 5% level, suggesting that state ownership is associated with more trade credit in countries with weak financial market development.

These results are also economically significant. For a firm with mean state ownership of

0.26, a decrease of private credit from the 25 th percentile (34.2) to the 75 th percentile (110.9) is associated with a 0.078 increase in trade credit. Since the mean TRADE CREDIT is 0.17, this corresponds to a 46% (0.078/0.17) increase.

In Model 2, we use government ownership of banks ( GOVERNMENT

OWNERSHIP OF BANK ) as an alternative measure of financial market development.

While private credit is a measure of credit availability to the private sector, government ownership of banks is a proxy to measure credit to SOEs. We obtain this measure from La

Porta et al. (2002). We re-run the specification in Model 2 using GOVERNMENT

OWNERSHIP OF BANK rather than PRIVATE CREDIT . Consistent with the results in

Model 2, we find that the coefficient on STATEOWN×GOVERNMENT OWNERSHIP OF

BANK is positive and statistically significant at the 5% level. These results are also economically significant. For a firm with mean state ownership of 0.26, an increase of government ownership of bank from the 25 th percentile (2.5) to the 75 th percentile (44) is associated with a 0.083 increase in trade credit. Since the mean TRADE CREDIT is 0.17, this corresponds to a 48.8% (0.083/0.17) increase.

In Model 3, we replace GOVERNMENT OWNERSHIP OF BANK in Model 2 with

LIMITATIONS OF FOREIGN BANK ENTRY, which is obtained from Barth et al. (2013).

It measures stringency of rules governing foreign banks owning domestic banks and whether foreign banks may enter a country’s banking industry. A higher value indicates a

137

system more protective of domestic banks and less openness in the banking industry. We continue to find that the interaction between LIMITATIONS OF FOREIGN BANK ENTRY and state ownership is statistically positively related to trade credit.

Taken together, the results in Table 4.5 show that state ownership plays an important role in reallocating funds in countries with poorly developed financial markets, with higher government ownership of banks, and with a less open banking industry. These findings highlight the importance of gradual reduction of state ownership, particularly when a well-developed financial market is not in place. In the absence of an open, well- functioning financial market, state ownership provides an alternative source of financing for firms along the supply chain.

4.4.3.2 Trade Credit as a Tool to Deter Competitors

One of the key goals of privatization is to make SOEs subject to product market competition and thus improve efficiency. In Table 4.6, we assess how SOEs respond to product market competition in adjusting trade credit policy. In Model 1 we use the

Herfindahl index ( HERFINDAHL INDEX ) to measure product market competition. We construct this index using the full Compustat Global sample. A higher value of

HERFINDAHL INDEX indicates a more concentrated product market. The results show that STATEOWN continues to load positively and is statistically significant at the 1% level.

More importantly, we find the interaction term between HERFINDAHL INDEX and

STATEOWN is negative and statistically significant at the 5% level, suggesting that state ownership is likely to extend more trade credit in the face of competition. In Model 2, we use price-cost margin ( PRICE-COST MARGIN ) to measure the product competition. In a highly competitive industry, the price-cost margin should be relatively lower than in a less

138

competitive industry. Consistent with the findings in Model 1, the results show that the coefficient of STATEOWN×PRICE-COST MARGIN is negative and statistically significant at the 5% level.

Following Karuna (2007), we employ two alternative proxies to measure competition in the product market. In Model 3 we use market size ( MARKET SIZE ) to measure the density of consumers in the industry, which is defined as the ratio of an industry’s market size to industry sales. A higher value of MARKET SIZE indicates a less competitive product market. In Model 4, we use entry cost as an alternative measure, which is defined as the minimal level of investment for entrant firms to an industry. We weight this variable by each firm’s market share in this industry. Regression results in Models 3 and 4 both show consistent findings in Model 1, suggesting that state ownership responds to competition by extending more trade credit to deter potential entrants and competition from the product market.

Interestingly, according to price discrimination theories, firms with a monopoly position could extend more trade credit for price discrimination, thus earning more profit.

However, our results show that state ownership takes advantage of its privilege in borrowing and uses it to deter competitors in competitive industries, while not using it as a means for price discrimination and profit maximization.

We show that, on the one hand, state ownership provides an alternative source of funds when there is a poorly developed financial market, while, on the other hand, it uses trade credit as a tool to deter competition.

139

4.4.4 Is it Efficient for State Ownership to Supply Trade Credit?

So far we have nothing to say about whether trade credit policy is an efficient corporate policy for state ownership. In this section, we address this question by exploring the valuation effect of trade credit in SOEs. Table 4.7 shows the regression results. In

Model 1, we regress Tobin’s Q on TRADE CREDIT , and STATEOWN, along with controls, and find that the coefficient of TRADE CREDIT is negative but statistically insignificant.

In Model 2, we also include the interaction term between TRADE CREDIT and

STATEOWN. We find that the coefficient of STATEOWN×TRADE CREDIT is negative and statistically significant at the 5% level, suggesting that outside investors discount the value of trade credit in SOEs.

4.4.5 Alternative Explanations of Trade Credit from State Ownership

In this section, we explore an alternative explanation of trade credit from state ownership, namely, implicit warranty of product quality. It is possible that state ownership tends to extend more trade credit when there is information asymmetry between buyers and sellers about the quality of the goods supplied. Table 4.8 presents results of regressing trade credit on product quality measures: production time ( PRODUCTION TIME ) and age

(AGE ). We use these two variables following Long et al. (1993). We find that

PRODUCTION TIME is negatively associated with trade credit, supporting the idea that trade credit can be used as a tool for quality guarantee. However, the interaction term between STATEOWN and PRODUCTION TIME is not significant, as shown in Model 1.

Similarly, we find that the interaction term between STATEOWN and AGE is not significant either. In summary, Table 4.8 rules out the possibility that state ownership extends more trade credit to provide implicit warranties to buyers.

140

4.5 Conclusion

In this study, we investigate the impact of state ownership on trade credit. Using a unique sample of 552 NPFs from 60 countries over the period 1981–2008, we find that state ownership is positively related to the supply of trade credit. This result is robust to endogeneity tests, alternative measures of trade credit, various measures of state control, and additional control variables. We further show that state ownership plays an important role in reallocating funds in countries with a poorly developed financial market, with higher government ownership of banks, and with a less open banking industry. These findings highlight the importance of gradual reduction of state ownership, particularly when a well- developed financial market is not in place. In the absence of an open, well-functioning financial market, state ownership provides an alternative source of financing for firms along the supply chain. However, we also show that state ownership responds to competition by extending more trade credit to deter potential entrants and competition from the product market. Our results show that state ownership takes advantage of its privilege in borrowing and uses it to deter competitors in competitive industries, while not using it as a means of price discrimination and profit maximization. Overall, this essay contributes to the literature by showing how state ownership affects firms along the supply chain of the NPFs and contributes to the trade credit literature by showing ownership structure is a key determinant of trade credit.

141

Table 4.1 Distribution of the Sample of Newly Privatized Firms

This table presents the distribution of the sample of 523 privatized firms. The full sample contains 4,406 firm-year observations. Panel A reports sample distribution by privatization year. Panel B reports sample distribution by industry based on industry groupings in Campbell (1996). Panel C reports sample distribution by region based on World Bank group classification. Panel A: By year Panel B: By industry Year Obs. % Firms % Industry Obs. % Firms % 1981 2 0.05 1 0.18 Basic industry 733 16.64 88 15.94 1983 6 0.14 1 0.18 Capital goods 219 4.97 28 5.07 1984 5 0.11 1 0.18 Construction 330 7.49 46 8.33 1985 16 0.36 2 0.36 Consumer durable 332 7.54 39 7.07 1986 38 0.86 4 0.72 Food and tobacco 226 5.13 35 6.34 1987 77 1.75 7 1.27 Leisure 111 2.52 13 2.36 1988 82 1.86 8 1.45 Petroleum 325 7.38 39 7.07 1989 147 3.34 12 2.17 Services 67 1.52 8 1.45 1990 121 2.75 14 2.54 Textiles and trade 123 2.79 17 3.08

142 1991 298 6.76 33 5.98 Transportation 536 12.17 66 11.96 1992 357 8.10 49 8.88 Utilities 1348 30.59 165 29.89

1993 288 6.54 36 6.52 Other 56 1.27 8 1.45 1994 368 8.35 46 8.33 Total 4,406 100.00 552 100.00 1995 326 7.40 45 8.15 1996 288 6.54 36 6.52 Panel C: By region 1997 425 9.65 52 9.42 By Region (countries) Obs. Percentage Firms Percentage 1998 328 7.44 37 6.70 Africa & the Middle East (10) 207 4.70 29 5.25 1999 251 5.70 27 4.89 East and South Asia & the Pacific (14) 1,781 40.42 219 39.67 2000 180 4.09 23 4.17 Latin America & the Caribbean (8) 481 10.92 63 11.41 2001 101 2.29 14 2.54 Europe & Central Asia (28) 1,937 43.96 241 43.66 2002 77 1.75 10 1.81 Total (60) 4,406 100.00 552 100.00 2003 111 2.52 16 2.90 2004 65 1.48 11 1.99 2005 105 2.38 13 2.36 2006 185 4.20 28 5.07 2007 115 2.61 16 2.90 2008 44 1.00 10 1.81 Total 4,406 100.00 552 100.00

Table 4.2 Summary Statistics and Correlation Matrix

This table presents descriptive statistics for key variables in our analyses. Our full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries All financial variables are winsorized at the 1% level in both tails of the distribution. Variable definitions and sources are provided in the Appendix. Panel A reports summary statistics for the firm-level variables. Panel B reports pairwise correlations matrix among variables. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Panel A: Summary Statistics Variables N Min Mean Median Max Std. Dev TRADE CREDIT 4,406 0.00 0.17 0.14 0.85 0.14 STATEOWN 4,406 0.00 0.26 0.12 1.00 0.29 LOG(SALES) 4,406 -9.00 6.94 6.92 12.65 1.99 ROS 4,406 -0.65 0.08 0.07 0.54 0.15 CASH HOLDINGS 4,406 0.00 0.12 0.08 0.53 0.11 FIXED ASSETS 4,406 0.01 0.46 0.47 0.91 0.23 SALES GROWTH 4,406 -0.98 0.13 0.09 1.76 0.30 GROSS PROFIT MARGIN 4,406 -0.01 0.45 0.39 1.00 0.29 GDP PER CAPITA 4,406 6.11 8.79 8.75 11.12 1.33

143 PRIVATE CREDIT 4,406 1.12 76.72 75.94 227.75 43.37 Panel B: Correlation Matrix

1 2 3 4 5 6 7 8 9 1. TRADECREDIT 1.00 2. STATEOWN 0.01 1.00 3. LOG(SALES) -0.07*** 0.09*** 1.00 4. ROS -0.05** 0.05*** -0.01 1.00 5. CASH HOLDINGS -0.03* 0.12*** -0.07*** 0.20*** 1.00 6. FIXED ASSETS -0.27*** 0.05*** -0.02 0.07*** -0.41*** 1.00 7. SALES GROWTH -0.07*** 0.05*** 0.00 0.09*** 0.06*** 0.01 1.00 8. GROSS PROFIT MARGIN 0.08*** 0.01 0.04* 0.30*** -0.06*** 0.20*** -0.05** 1.00 9. GDP PER CAPITA 0.06*** -0.27*** 0.41*** -0.04* -0.15*** -0.07*** -0.17*** 0.21*** 1.00 10. PRIVATE CREDIT 0.01 0.09*** 0.27*** 0.00 0.06*** -0.09*** -0.08*** -0.01 0.31***

Table 4.3 State Ownership and Trade Credit

This table reports regression results relating state ownership to trade credit. The dependent variable is measured as 100 times the ratio of trade receivables to total sales. Trade receivable equals amounts on open account (net of applicable reserves) owed by customers for goods and services sold in the ordinary course of business. The Appendix provides variables definitions and sources. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. Regressions include country dummy variables, industry dummy variables based on Campbell (1996), and year dummy variables. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Endogeneity of state ownership Basic IV Alternative state control Variables Heckman model regression PSM variables 2nd stage 2nd stage (1) (2) (3) (4) (5) (6) STATEOWN 0.061*** 0.754** 0.127*** 0.067** (2.945) (2.540) (4.810) (2.504) CONTROL 0.030*** (2.598) AVG_STATEOWN 0.072*** (2.868) LOG(SALES) -0.009** -0.023** -0.012** -0.011 -0.008** -0.009** (-2.213) (-2.498) (-2.221) (-1.615) (-2.063) (-2.268) ROS -0.003 0.039 0.008 0.011 -0.005 -0.004 (-0.113) (0.573) (0.198) (0.219) (-0.168) (-0.118) CASH HOLDINGS -0.175*** -0.142* -0.146** -0.138** -0.171*** -0.174*** (-3.877) (-1.773) (-2.539) (-2.128) (-3.810) (-3.874) FIXED ASSETS -0.191*** -0.226*** -0.192*** -0.179*** -0.189*** -0.192*** (-6.659) (-3.495) (-4.842) (-3.428) (-6.625) (-6.692) SALES GROWTH -0.019** -0.023 -0.024*** -0.029*** -0.019** -0.018** (-2.366) (-1.611) (-2.614) (-2.595) (-2.339) (-2.217) GROSS PROFIT 0.054*** 0.012 0.048** 0.048** 0.054*** 0.053*** MARGIN (3.265) (0.401) (2.515) (2.181) (3.279) (3.204) GDP PER CAPITA 0.001 0.044 -0.022 -0.043 0.001 0.000 (0.038) (0.729) (-0.609) (-1.081) (0.031) (0.005) PRIVATE CREDIT 0.000 0.000 -0.000 -0.000 0.000 0.000 (0.922) (0.264) (-0.038) (-0.398) (1.046) (0.789) LAMBDA -0.033*** (-3.730) Constant 0.430* 0.060 0.387 0.708** 0.417* 0.440* (1.913) (0.160) (1.035) (2.084) (1.841) (1.958) Country fixed effects Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Observations 4,406 2,751 2,638 1,962 4,406 4,406 Adjusted R 2 0.328 0.059 0.364 0.312 0.325 0.329

144

Table 4.4 Robustness Checks

This table reports regression results of robustness checks relating state ownership to trade credit. Models 1 to 3 report regressions including additional control variables. Models 4 to 7 report regression results using alternative estimation methods. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. Regressions include country dummy variables, industry dummy variables based on 48 industry groupings in Fama and French (1997), and year dummy variables. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Alternative estimation methods Additional controls variables Two-way cluster Newey West Prais-Winsten Tobit regression Variables (1) (2) (3) (4) (5) (6) (7) STATEOWN 0.060*** 0.086*** 0.094*** 0.061*** 0.061*** 0.028** 0.061*** (2.863) (3.732) (3.938) (3.088) (5.126) (2.378) (2.997) LOG(SALES) -0.009** -0.009** -0.010** -0.009** -0.009*** -0.002 -0.008** (-2.323) (-2.193) (-2.583) (-2.319) (-4.000) (-0.970) (-2.215) ROS -0.006 -0.012 -0.024 -0.003 -0.003 0.021** -0.004 (-0.208) (-0.400) (-0.782) (-0.118) (-0.157) (2.401) (-0.132)

145 CASH HOLDINGS -0.178*** -0.173*** -0.173*** -0.175*** -0.175*** 0.009 -0.174*** (-3.950) (-3.838) (-3.661) (-4.119) (-6.105) (0.525) (-3.901) FIXED ASSETS -0.193*** -0.190*** -0.209*** -0.191*** -0.191*** -0.015 -0.192*** (-6.721) (-6.575) (-7.135) (-6.890) (-11.042) (-1.170) (-6.766) SALES GROWTH -0.021*** -0.022*** -0.024*** -0.019** -0.019** -0.005 -0.019** (-2.598) (-2.689) (-2.849) (-2.213) (-2.504) (-1.556) (-2.406) GROSS PROFIT MARGIN 0.054*** 0.056*** 0.051*** 0.054*** 0.054*** 0.010 0.054*** (3.270) (3.381) (2.898) (3.203) (4.867) (1.250) (3.303) GDP PER CAPITA 0.008 0.007 0.023 0.001 0.001 0.014 0.001 (0.314) (0.284) (0.888) (0.037) (0.040) (0.692) (0.044) PRIVATE CREDIT 0.000 0.000 0.000 0.000 0.000 -0.000 0.000 (0.831) (0.809) (0.867) (1.041) (1.256) (-0.601) (0.903) FOREIGNOWN 0.004 0.027 0.033 (0.244) (1.262) (1.509) LOCALOWN 0.044** 0.049*** (2.484) (2.708) EMPLOYEEOWN -0.139 -0.152 (-1.208) (-1.242) LEVERAGE -0.005 (-0.163) CAPEX 0.077**

(2.576) Constant 0.232 0.228 -0.011 0.190 0.121 -0.012 0.120 (1.281) (1.227) (-0.053) (0.816) (0.752) (-0.069) (0.731) Country fixed effects Yes Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Yes Observations 4,365 4,280 4,118 4,406 4,406 4,406 4,406 Adjusted R 2 0.328 0.345 0.356 0.328 . 0.196 . 146

Table 4.5 State Ownership, Financial Market Development, and Trade Credit

This table reports regression results relating to state ownership, financial market development, and trade credit. The dependent variable is measured as 100 times the ratio of trade receivables to total sales. Trade receivable equals amounts on open account (net of applicable reserves) owed by customers for goods and services sold in the ordinary course of business. The Appendix provides variables definitions and sources. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. GOVERNMENT LIMITATIONS PRIVATE Variables OWNERSHIP OF ON FOREIGN CREDIT BANK BANK ENTRY (1) (2) (3) STATEOWN 0.124*** -0.023 0.012 (3.648) (-0.615) (0.408) BANK DEVELOPMENT MEASURE 0.000 -0.031** (0.113) (-2.355) STATEOWN*BANK DEVELOPMENT MEASURE -0.001** 0.002** 0.130*** (-2.230) (2.437) (2.804) LOG(SALES) -0.008** -0.010** -0.003 (-2.151) (-2.334) (-0.775) ROS -0.003 -0.017 -0.027 (-0.116) (-0.522) (-0.852) CASH HOLDINGS -0.173*** -0.191*** -0.233*** (-3.859) (-3.552) (-4.777) FIXED ASSETS -0.190*** -0.185*** -0.183*** (-6.545) (-5.576) (-5.265) SALES GROWTH -0.018** -0.021** -0.017* (-2.253) (-2.360) (-1.764) GROSS PROFIT MARGIN 0.051*** 0.025 0.032* (3.135) (1.323) (1.672) GDP PER CAPITA 0.003 0.019** 0.008 (0.111) (2.528) (0.943) PRIVATE CREDIT 0.000* -0.000 -0.000 (1.676) (-0.248) (-0.935) Constant 0.093 0.090 0.184** (0.561) (0.989) (2.255) Country fixed effects Yes No No Industry fixed effects Yes Yes Yes Year fixed effects Yes Yes Yes Observations 4,406 3,129 2,904 Adjusted R 2 0.331 0.222 0.267

147

Table 4.6 State Ownership, Product Market Competition, and Trade Credit

This table reports regression results relating to state ownership, product market competition, and trade credit. The dependent variable is measured as 100 times the ratio of trade receivables to total sales. Trade receivable equals amounts on open account (net of applicable reserves) owed by customers for goods and services sold in the ordinary course of business. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. PRICE- HERFINDAHL MARKET COST ENTRY COST INDEX SIZE MARGIN (1) (2) (3) (4) STATEOWN 0.130*** 0.059*** 0.194*** 0.200*** (3.326) (2.608) (3.647) (3.627) COMPETITION MEASURE 0.015 0.000 0.000 -0.000 (0.736) (0.982) (0.061) (-0.053) STATEOWN*COMPETITION MEASURE -0.109** -0.001** -0.010*** -0.010*** (-2.393) (-2.057) (-3.169) (-3.059) LOG(SALES) -0.009** -0.013*** -0.006 -0.007 (-2.256) (-3.603) (-1.517) (-1.622) ROS 0.000 -0.023 -0.002 -0.001 (0.003) (-0.761) (-0.062) (-0.044) CASH HOLDINGS -0.180*** -0.200*** -0.177*** -0.177*** (-3.914) (-4.227) (-3.968) (-3.973) FIXED ASSETS -0.195*** -0.233*** -0.187*** -0.183*** (-6.900) (-7.893) (-6.697) (-6.567) SALES GROWTH -0.019** -0.017** -0.019** -0.020** (-2.305) (-2.011) (-2.400) (-2.431) GROSS PROFIT MARGIN 0.056*** 0.069*** 0.055*** 0.057*** (3.472) (3.773) (3.374) (3.433) GDP PER CAPITA -0.002 -0.004 0.009 0.009 (-0.098) (-0.167) (0.365) (0.374) PRIVATE CREDIT 0.000 0.000 0.000 0.000 (0.987) (0.347) (0.635) (0.718) Constant 0.123 0.202 0.062 0.062 (0.717) (1.176) (0.371) (0.374) Country fixed effects Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 4,406 4,282 4,406 4,356 Adjusted R 2 0.332 0.245 0.336 0.337

148

Table 4.7 The Valuation Effect of State Ownership and Trade Credit

This table reports regression results relating state ownership, trade credit, and valuation. The dependent variable is Tobin’s Q, which is computed as market value of equity minus book value of equity plus book value of assets, all scaled by book value of assets. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Dependent Variable: Tobin’s Q Variables (1) (2) STATEOWN 0.0002 0.0031 (0.101) (1.384) TRADE CREDIT -0.3525 0.1484 (-1.187) (0.352) STATEOWN*TRADE CREDIT -0.0163** (-2.175) LOG(SALES) -0.0923*** -0.0942*** (-2.692) (-2.745) ROS 0.8090*** 0.8106*** (3.703) (3.680) CASH HOLDINGS 0.3707 0.4067 (1.133) (1.239) FIXED ASSETS -0.3851** -0.3736** (-2.096) (-2.031) SALES GROWTH 0.0703 0.0716 (1.026) (1.050) GROSS PROFIT MARGIN 0.3164*** 0.2987*** (2.687) (2.646) GDP PER CAPITA 0.2981 0.3037 (1.434) (1.463) PRIVATE CREDIT -0.0010 -0.0010 (-0.823) (-0.837) Constant 1.0392 0.9980 (0.804) (0.773) Country fixed effects Yes Yes Industry fixed effects Yes Yes Year fixed effects Yes Yes Observations 3,971 3,971 Adjusted R-squared 0.252 0.257

149

Table 4.8 State Ownership, Product Quality, and Trade Credit

This table reports regression results relating state ownership, product quality, and trade credit. The dependent variable is measured as 100 times the ratio of trade receivables to total sales. Trade receivable equals amounts on open account (net of applicable reserves) owed by customers for goods and services sold in the ordinary course of business. The full sample contains 4,406 firm-year observations from 552 firms privatized in 60 countries. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variables definitions and sources. t-statistics are reported in parentheses below each coefficient. Standard errors are White-corrected for heteroscedasticity and clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Product quality measures Variables PRODUCTION TIME AGE (1) (2) STATEOWN 0.073*** 0.100* (2.819) (1.678) QUALITY MEASURE -0.015** -0.008 (-2.275) (-0.506) STATEOWN*QUALITY MEASURE -0.013 -0.019 (-1.015) (-0.759) LOG(SALES) -0.007* -0.008** (-1.776) (-2.012) ROA -0.008 -0.005 (-0.255) (-0.170) CASH HOLDINGS -0.183*** -0.175*** (-4.111) (-3.894) FIXED ASSETS -0.214*** -0.190*** (-6.986) (-6.642) SALES GROWTH -0.015* -0.020** (-1.948) (-2.580) GROSS PROFIT MARGIN 0.041** 0.054*** (2.496) (3.244) GDP PER CAPITA 0.011 0.002 (0.435) (0.068) PRIVATE CREDIT 0.000 0.000 (0.780) (0.849) Constant 0.082 0.137 (0.491) (0.738) Country fixed effects Yes Yes Industry fixed effects Yes Yes Year fixed effects Yes Yes Observations 4,406 4,406 Adjusted R 2 0.339 0.329

150

CHAPTER 5 OVERALL CONCLUSION

This dissertation investigates the role of state ownership in three related essays. All three essays examine state ownership in the context of privatization. Specifically, the first essay investigates the impact of state ownership on corporate cash holding. The second one assesses how state ownership influences firms’ liquidity and valuation. The third essay examines the relation between state ownership and trade credit. The robust findings of three essays add to the literature of privatization, state ownership, and corporate finance.

The first essay in Chapter 2 investigates the impact of state ownership on corporate cash holding. Using a unique sample of newly privatized firms from 59 countries, this study provides new evidence about the agency costs of state ownership and new insight into the corporate governance role of country-level institutions. Consistent with agency theory, we find strong and robust evidence that state ownership is positively related to corporate cash holdings. Moreover, we find that the strength of country-level institutions affects the relation between state ownership and the value of cash holdings. In particular, as state ownership increases, markets discount the value of cash holdings more in countries with weaker institutions.

This essay makes an important contribution to the privatization literature. Extant research [summarized in Megginson and Netter (2001)] generally identifies significant

151

improvements in the financial and operating performance of NPFs. However, there is limited empirical evidence on the sources of these improvements. Our paper extends this literature by providing preliminary insight into the sources of the performance improvements of NPFs. As we previously noted, the financial and operating performance of SOEs can be reduced by state misappropriation of corporate resources (i.e., the agency costs of state ownership). Privatization should mitigate such costs. One area where a reduction in the agency costs of state ownership could be directly measured is a firm’s cash management policies. Specifically, since cash is the most liquid asset, Caprio, Faccio, and

McConnell (2013), Kusnadi et al. (2015), and Myers and Rajan (1998) argue that cash is most vulnerable to political extraction. Therefore, cash is an especially tempting target for state expropriation. Because privatization should dampen that temptation, one source of performance improvement following privatization is likely to be an increase in the value of cash holdings. By documenting that the value of cash holdings increases as residual state ownership decreases, we provide evidence of a direct source of the value accretion associated with privatization.

Further, since we examine cash holdings across a large sample of institutionally diverse countries, our study provides interesting evidence regarding the effect of the institutional environment on financial decision-making. We know that institutions matter.

In this essay, we attempt to shed light on which institutions matter more and why certain institutions may matter more under certain circumstances.

152

The second essay, in Chapter 3, investigates the impact of state ownership on liquidity and valuation. Using a unique sample of 478 newly privatized firms (NPFs) from

54 countries over the period 1994–2014, we examine the relation between state ownership, stock market liquidity, and firm value. We first find that NPFs, on average, are more liquid than other listed firms. However, partially privatized firms are more liquid than fully privatized counterparts. Both of these results suggest a non-linear relation between state ownership and stock liquidity. We find empirical support for this conjecture and the soft budget constraint associated with state ownership. We further find that the relation between state ownership and liquidity is stronger in countries with higher levels of state ownership of banks, and is weaker in countries with fewer limits on foreign banks and little or no political influence as measured by the independence of the supervisory authority. Finally, we show that stock liquidity is related to firm value and cost of equity capital, suggesting that liquidity is a channel through which residual state ownership in NPFs affects valuation.

This essay makes several contributions to the literature. First, our results are related to the literature on the effects of privatization on financial market development [see

Megginson (2005, 2016) for an excellent survey]. This essay complements previous country-level studies (e.g., Boutchkova and Megginson, 2000; Bortolotti et al., 2007) by using firm-level liquidity measures to show that state ownership affects liquidity in a nonlinear fashion. In addition, we examine one channel through which state ownership affects cost of capital and thus firm value and find that superior access to state-owned banks,

153

which leads to soft budget constraints, explains the high stock liquidity in NPFs. Second, we contribute to the literature on the link between ownership structure and liquidity by focusing on a particular stakeholder, the State. As one of the most important large shareholders, the state is not only related to poor corporate governance, but is also unique, compared to other blockholders, in terms of soft budget constraint. We show that state ownership is associated with high stock liquidity, in particular during the financial crisis.

Third, we further add to the privatization literature by showing how the distortion between economic and political/social objectives associated with state ownership plays out in determining stock liquidity.

This essay has policy implications related to the process of privatization: the evidence that high residual ownership decreases liquidity suggests that continued ownership is suboptimal. Of particular interest, we show that the benefits of privatization disappear at high levels of state ownership suggesting that, as argued by Megginson (2016), the links with government should be severed for positive outcomes to materialize. The economic distortions introduced by state ownership can be very costly to the economy when governments dominate as owners and market players.

The third essay, in Chapter 4, investigates the relation between state ownership and trade credit. Using a unique sample of 552 firms privatized in 60 countries, we find strong evidence that state ownership is positively associated with trade credit. This positive relation between state ownership and trade credit is stronger in countries with poorly

154

developed financial markets, suggesting that implicit borrowing in forms of trade credit from state-owned enterprises provides an alternative source of funds for firms with little access to finance. Contrary to the price discrimination theory, which predicts that firms extend trade credit to extract marginal profit, we find that state ownership is more likely to provide trade credit in competitive industries. Moreover, we find that the market discounts the value of trade credit in the presence of state ownership.

This essay contributes to the privatization literature in several ways. First, prior literature provides evidence on how state ownership in NPFs is associated with financial and operating performance after privatization. However, there is little evidence on how post-privatization state ownership affects firms along the supply chain of the NPFs. This essay extends this literature by providing preliminary insights on this issue. Second, the findings of how state ownership facilitates reallocating funds through trade credit in countries with a weak financial market extend our understanding of privatization. In particular, our results suggest that a gradual reduction of state ownership is recommended in countries without a well-established financial market. Firms along the supply chain of the NPFs are likely to experience a dramatic shock in financing if both trade credit and external financing are no longer in place. Further, this essay contributes to the trade credit literature by showing that ownership is one of the key determinants of trade credit. We shed light on how ownership structure is associated with different incentives to provide

155

trade credit. By considering the impact of state ownership, we broaden our understanding of how ownership structure affects the supply of trade credit.

156

REFERENCES

Amihud, Y. 2002. Illiquidity and stock returns: Cross-section and time-series effects. Journal of Financial Markets , 5(1): 31-56.

Amihud, Y., & Mendelson, H. 2000. The liquidity route to a lower cost of capital. Journal of Applied Corporate Finance , 12(4): 8-25.

Amihud, Y., & Mendelson, H. 2008. Liquidity, the value of the firm, and corporate finance. Journal of Applied Corporate Finance , 20(2): 32-45.

Anderson, J. H., Korsun, G., & Murrell, P. 2000. Which enterprises (believe they) have soft budgets after mass privatization? Evidence from Mongolia. Journal of Comparative Economics , 28(2): 219-246.

Attig, N., Fong, W. M., Gadhoum, Y., & Lang, L. H. 2006. Effects of large shareholding on information asymmetry and stock liquidity. Journal of Banking & Finance , 30(10): 2875-2892.

Barrot, J. N. 2016. Trade credit and industry dynamics: evidence from trucking firms. Journal of Finance, 71(5): 1975-2016. Barth, J. R., Caprio Jr, G., & Levine, R. 2013. Bank regulation and supervision in 180 countries from 1999 to 2011. Journal of Financial Economic Policy , 5(2): 111-219.

Baruch, S., Karolyi, G. A., & Lemmon, M. L. 2007. Multimarket trading and liquidity: Theory and evidence. Journal of Finance , 62(5): 2169-2200.

Bates, T. W., Kahle, K. M., & Stulz, R. M. 2009. Why do US firms hold so much more cash than they used to? Journal of Finance , 64(5): 1985-2021

Beck, T., Clarke, G., Groff, A., Keefer, P., & Walsh, P. 2001. New tools in comparative political economy: The database of political institutions. World Bank Economic Review , 15(1): 165-176.

Becker, G. 1968. Crime and punishment: An economic approach. Journal of Political Economy , 76(2): 169-217.

Ben-Nasr, H., Boubakri, N., & Cosset, J. C. 2012. The political determinants of the cost of equity: Evidence from newly privatized firms. Journal of Accounting Research , 50(3): 605-646.

157

Ben-Nasr, H., Boubakri, N., & Cosset, J. C. 2015. Earnings quality in privatized firms: The role of state and foreign owners. Journal of Accounting and Public Policy, 34(4): 392-416 .

Bhattacharya, U., & Daouk, H. 2002. The world price of insider trading. Journal of Finance , 57(1): 75-108.

Bhide, A. 1993. The hidden costs of stock market liquidity. Journal of Financial Economics , 34(1): 31-51.

Boissay, F., and Gropp, R. 2007. Trade credit defaults and liquidity provision by firms. Working Paper No. 753, European Central Bank. Bolton, P., & Von Thadden, E.-L. 1998. Blocks, liquidity, and corporate control. Journal of Finance, 53(1): 1-25. Booth, J. R., & Smith, R. L. 1986. Capital raising, underwriting and the certification hypothesis. Journal of Financial Economics , 15(1): 261-281.

Borisova, G., Brockman, P., Salas, J. M., & Zagorchev, A. 2012. Government ownership and corporate governance: Evidence from the EU. Journal of Banking and Finance , 36(11): 2917-2934.

Borisova, G., Fotak, V., Holland, K., & Megginson, W. L. 2015. Government ownership and the cost of debt: Evidence from government investments in publicly traded firms. Journal of Financial Economics , 118(1): 168-191.

Borisova, G., & Megginson, W. L. 2011. Does government ownership affect the cost of debt? Evidence from privatization. Review of Financial Studies , 24(8): 2693-2737.

Bortolotti, B., De Jong, F., Nicodano, G., & Schindele, I. 2007. Privatization and stock market liquidity. Journal of Banking & Finance , 31(2): 297-316.

Bortolotti, B., & Faccio, M. 2009. Government control of privatized firms. Review of Financial Studies , 22(8): 2907-2939.

Bortolotti, B., Fantini, M., and Siniscalco, D. 2004. Privatisation around the world: Evidence from panel data. Journal of Public Economics 88: 305-332. Bortolotti, B., & Pinotti, P. 2003. The political economy of privatization. Working paper No. 2003.45, FEEM.

Boubakri, N., & Cosset, J. C. 1998. The financial and operating performance of newly privatized firms: Evidence from developing countries. Journal of Finance , 53(3): 1081-1110.

Boubakri, N., Cosset, J. C., & Guedhami, O. 2005. Postprivatization corporate governance: The role of ownership structure and investor protection. Journal of Financial Economics , 76(2): 369-399.

158

Boubakri, N., Cosset, J.-C., Guedhami, O., and Saffar, W. 2011. The political economy of residual state ownership in privatized firms: Evidence from emerging markets. Journal of Corporate Finance, 17: 244-258.

Boubakri, N., Cosset, J. C., & Saffar, W. 2013. The role of state and foreign owners in corporate risk-taking: Evidence from privatization. Journal of Financial Economics , 108(3): 641-658.

Boubakri, N., Guedhami, O., Kwok, C., & Saffar, W. 2016. National culture and privatization: The relationship between collectivism and residual state ownership. Journal of International Business Studies , 47(2): 170-190.

Boubakri, N., Guedhami, O., Mishra, D., & Saffar, W. 2012. Political connections and the cost of equity capital. Journal of Corporate Finance , 18(3): 541-559.

Boutchkova, M. K., & Megginson, W. L. 2000. Privatization and the rise of global capital markets. Financial Management 29(4): 31-75.

Boycko, M., Shleifer, A., & Vishny, R. W. 1994. Voucher privatization. Journal of Financial Economics, 35(2): 249-266.

Brennan, M. J., Maksimovic, V., & Zechner, J. 1988. Vendor financing. Journal of Finance , 43(5): 1127-1141.

Brockman, P., Chung, D. Y, & Yan, S. 2009. Block ownership, trading activity, and market liquidity. Journal of Financial and Quantitative Analysis, 44(6): 1403-1426.

Brunetti, A., & Weder, B. 2003. A free press is bad news for corruption. Journal of Public Economics , 87(7): 1801-1824.

Bushman, R., Piotroski, J., & Smith, A. 2004. What determines corporate transparency? Journal of Accounting Research , 42(2): 207-252.

Campbell, J. Y. 1996. Understanding risk and return. Journal of Political Economy , 104(2): 298-345.

Caprio, L., Faccio, M., & McConnell, J. 2013. Sheltering corporate assets from political extraction. Journal of Law, Economics, and Organizations , 29(2): 332-354.

Carter, R. B., Dark, F. H., & Singh, A. K. 1998. Underwriter reputation, initial returns, and the long-run performance of IPO stocks. Journal of Finance , 53(1): 285-311.

Carter, R., & Manaster, S. 1990. Initial public offerings and underwriter reputation. Journal of Finance , 45(4): 1045-1067.

Chaney, P. K., Faccio, M., & Parsley, D. 2011. The quality of accounting information in politically connected firms. Journal of Accounting and Economics , 51(1): 58-76.

159

Chemmanur, T. J., & Fulghieri, P. 1994. Investment bank reputation, information production, and financial intermediation. Journal of Finance , 49(1): 57-79.

Chen, H. C., & Ritter, J. R. 2000. The seven percent solution. Journal of Finance , 55(3): 1105-1131.

Chen, Q., Chen, X., Schipper, K., Xu, Y., & Xue, J. 2012. The sensitivity of corporate cash holdings to corporate governance. Review of Financial Studies , 25(12): 3610-3644.

Chen, R., El Ghoul, S., Guedhami, O., & Nash, R. 2017. State ownership and corporate cash holdings: Evidence from privatization. Journal of Financial and Quantitative Analysis . forthcoming.

Chen, R., El Ghoul, S., Guedhami, O., & Wang, H. 2017. Do state and foreign ownership affect investment efficiency? Evidence from privatizations. Journal of Corporate Finance , 42: 408-421.

Chung, K. H., Elder, J., & Kim, J. C. 2010. Corporate governance and liquidity. Journal of Financial and Quantitative Analysis , 45(02): 265-291.

Claus, J., & Thomas, J. 2001. Equity premia as low as three percent? Evidence from analysts' earnings forecasts for domestic and international stock markets. Journal of Finance , 56(5): 1629-1666.

Cook, P., and Kirkpatrick, C. H. 1988. Privatisation in less developed countries . New York: St. Martin’s Press.

Copeland, T. E., & Galai, D. 1983. Information effects on the bid-ask spread. Journal of Finance , 38(5): 1457-1469.

Cull, R., Xu, L. C., & Zhu, T. 2009. Formal finance and trade credit during China’s transition. Journal of Financial Intermediation , 18(2): 173-192.

D’Souza, J., & Megginson, W. L. 1999. The financial and operating performance of privatized firms during the 1990s. Journal of Finance , 54(4): 1397-1438.

Daske, H., Hail, L., Leuz, C., & Verdi, R. 2008. Mandatory IFRS reporting around the world: Early evidence on the economic consequences. Journal of Accounting Research , 46(5): 1085-1142.

Demsetz, H. 1968. The cost of transacting. Quarterly Journal of Economics , 1: 33-53.

Denis, D. K., & McConnell, J. J. 2003. International corporate governance. Journal of Financial and Quantitative Analysis , 38(1): 1-36.

Diamond, D. W. 1985. Optimal release of information by firms. Journal of Finance , 40(4): 1071-1094.

160

Dinc, S., & Gupta, N. 2011. The decision to privatize: Finance and politics. Journal of Finance , 66(1): 241-269.

Dittmar, A., & Mahrt-Smith, J. 2007. Corporate governance and the value of cash holdings. Journal of Financial Economics , 83(3): 599-634.

Dittmar, A., Mahrt-Smith, J., & Servaes, H. 2003. International corporate governance and corporate cash holdings. Journal of Financial and Quantitative Analysis , 38(1): 111- 133.

Djankov, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. 2002. The regulation of entry. Quarterly Journal of Economics , 117(1): 1-37.

Djankov, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. 2008. The law and economics of self-dealing. Journal of Financial Economics , 88(3): 430-465.

Djankov, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. 2010. Disclosure by politicians. American Economic Journal: Applied Economics , 2(2): 179-209.

Duchin, R. 2010. Cash holdings and corporate diversification. Journal of Finance , 65(3): 955-992.

Dyck, A. 2001. Privatization and corporate governance: Principles, evidence, and future challenges. World Bank Research Observer , 16(1): 59-84.

Dyck, A., Volchkova, N., & Zingales, L. 2008. The corporate governance role of the media: Evidence from Russia. Journal of Finance , 63(3): 1093-1135.

Dyck, A., & Zingales, L. 2002. The corporate governance role of the media. In R. Islam (Ed.), The right to tell: The role of the media in development . Washington, DC: World Bank.

Dyck, A., & Zingales, L. 2004. Private benefits of control: An international comparison. Journal of Finance , 59(2): 537-600.

Easton, P. 2004. PE Ratios, PEG ratios, and estimating the implied expected rate of return on equity capital. Accounting Review, 79(1): 73-95.

Eberhart, A. C., Maxwell, W. F., & Siddique, A. R. 2004. An examination of long-term abnormal stock returns and operating performance following R&D increases. Journal of Finance , 59(2): 623-650.

Economist , 2012 (January 21). The visible hand.

Emery, G., and Nayar, N. 1998. Product quality and payment policy. Review of Quantitative Finance and Accounting, 10: 269-284.

Fabbri, D., and Klapper, L. F. 2008. Market power and the matching of trade credit terms. World Bank Policy Research Working Paper 4754.

161

Faccio, M. 2006. Politically connected firms. American Economic Review , 96(1): 369- 386.

Faccio, M., Masulis, R. W., & McConnell, J. 2006. Political connections and corporate bailouts. Journal of Finance , 61(6): 2597-2635.

Fama, E. F., & French, K. R. 1997. Industry costs of equity. Journal of Financial Economics , 43(2): 153-193.

Fang, L. H. 2005. Investment bank reputation and the price and quality of underwriting services. Journal of Finance , 60(6): 2729-2761.

Ferris, J. S. 1981. A transactions theory of trade credit use. Quarterly Journal of Economics, 96: 243-270.

Firth, M., Malatesta, P., Xin, Q., & Xu, L. 2012. Corporate investment, government control, and financing channels: Evidence from China’s listed companies. Journal of Corporate Finance , 18(3): 433-450.

Fisman, R., and Love, I. 2003. Trade credit, financial intermediary development, and industry growth. Journal of Finance 58: 353-374.

Fong, K. Y. L., Holden, C. W., & Trzcinka, C. A. 2017. What are the best liquidity proxies for global research? Review of Finance , forthcoming. Available at SSRN: https://ssrn.com/abstract=1558447.

Frank, M., and Maksimovic, V. 1998. Trade credit, collateral, and adverse selection. Unpublished manuscript, University of Maryland.

Frydman, R., Gray, C., Hessel, M., & Rapaczynski, A. 2000. The limits of discipline: Ownership and hard budget constraints in the transition economies. Economics of Transition , 8(3): 577-601.

Gao, H., Harford, J., & Li, K. 2013. Determinants of corporate cash policy: Insights from private firms. Journal of Financial Economics , 109(3): 623-639.

Gao, X., & Ritter, J. R. 2010. The marketing of seasoned equity offerings. Journal of Financial Economics , 97(1): 33-52.

Gebhardt, W. R., Lee, C., & Swaminathan, B. 2001. Toward an implied cost of capital. Journal of Accounting Research , 39(1): 135-176.

Giannetti, M., Burkart, M., and Ellingsen, T. 2011. What you sell is what you lend? Explaining trade credit contracts. Review of Financial Studies, 24: 1261-1298.

Ginglinger, E., & Hamon, J. 2007. Actual share repurchases, timing and liquidity. Journal of Banking & Finance , 31(3): 915-938.

162

Ginglinger, E., & Hamon, J. 2012. Ownership, control and market liquidity. Finance , 33(2): 61-99.

Glaeser, E., La Porta, R., López-de-Silanes, F., & Shleifer, A. 2004. Do institutions cause growth? Journal of Economic Growth , 9(3): 271-303.

Golubov, A., Petmezas, D., & Travlos, N. G. 2012. When it pays to pay your investment banker: New evidence on the role of financial advisors in M&As. Journal of Finance , 67(1): 271-311.

Graham, J., & Leary, M. 2016. The evolution of corporate cash. Working paper. Available at SSRN: https://ssrn.com/abstract=2805505.

Guedhami, O., Pittman, J. A., & Saffar, W. 2009. Auditor choice in privatized firms: Empirical evidence on the role of state and foreign owners. Journal of Accounting and Economics , 48(2): 151-171.

Harford, J. 1999. Corporate cash reserves and acquisitions. Journal of Finance , 54(6): 1969-1997.

Harford, J., Mansi, S. A., & Maxwell, W. F. 2008. Corporate governance and firm cash holdings in the US. Journal of Financial Economics , 87(3): 535-555.

Haw, I., Hu, B., Hwang, L., & Wu, W. 2004. Ultimate ownership, income management, and legal and extra-legal institutions. Journal of Accounting Research , 42(2): 423- 462.

Hellman, J., Jones, G., & Kaufman, D. 2003. Seize the state, seize the day: State capture, corruption, and influence in transition. Journal of Comparative Economics , 31(4): 751-773.

Henisz, W. 2002. The institutional environment for infrastructure investment. Industrial and Corporate Change , 11(2): 355-389.

Hofstede, G. 2001. Culture’s consequences: Comparing values, behaviors, institutions, and organizations across nations, 2nd edition . Thousand Oaks, CA: Sage.

Holland, K. V. 2016. Are all government owners viewed the same? Evidence from government acquisitions of publicly traded firms. Available at SSRN: https://ssrn.com/abstract=2810341.

Holmström, B., & Tirole, J. 1993. Market liquidity and performance monitoring. Journal of Political Economy , 101(4): 678-709.

Huang, R., & Zhang, D. 2011. Managing underwriters and the marketing of seasoned equity offerings. Journal of Financial and Quantitative Analysis , 46(01): 141-170.

163

Jaslowitzer, P., Megginson, W. L., & Rapp, M. 2016. Disentangling the effects of state ownership on investment – Evidence from Europe. Working paper. Available at SSRN: https://ssrn.com/abstract=2735698.

Jensen, M. C. 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review , 76(2): 323-329.

Johnson, J. M., & Miller, R. E. 1988. Investment banker prestige and the underpricing of initial public offerings. Financial Management , 17(2): 19-29.

Johnson, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. 2000. Tunneling. American Economic Review Papers and Proceedings , 90(2): 22-27.

Johnston, M. 1997. Public officials, private interests, and sustainable democracy: When politics and corruption meet. In K. Elliot (Ed.), Corruption and the global economy . Washington, DC: Instithdute for International Economics.

Kalcheva, I., & Lins, K. V. 2007. International evidence on cash holdings and expected managerial agency problems. Review of Financial Studies , 20(4): 1087-1112.

Karuna, C. 2007. Industry product market competition and managerial incentives. Journal of Accounting and Economics, 43: 275-297.

Keefer, P. 1999. When do special interests run rampant? Disentangling the role of elections, incomplete information, and checks and balances in banking crises. Working paper, World Bank. Available: http://dx.doi.org/10.1596/1813-9450- 2543

Keefer, P., & Knack, S. 1997. Why don’t poor countries catch up? A cross-national test of an institutional explanation. Economic Inquiry , 35(3): 590-602.

Kleibergen, F., and Paap, R. 2006. Generalized reduced rank tests using the singular value decomposition. Journal of Econometrics 133: 97-126.

Kornai, J. 1979. Resource-constrained versus demand-constrained systems. Econometrica , 47(4): 801-819.

Kornai, J. 1980. Economics of shortage (Vol. 2) . Amsterdam: North Holland.

Kornai, J., Maskin, E., & Roland, G. 2003. Understanding the soft budget constraint. Journal of Economic Literature , 41(4): 1095-1136.

Kusnadi, Y., Yang, Z., & Zhou, Y. 2015. Institutional development, state ownership, and corporate cash holdings: Evidence from China. Journal of Business Research , 68(2): 351-359.

Kyle, A. S. 1985. Continuous auctions and insider trading. Econometrica , 53(6): 1315- 1335.

164

La Porta, R., Lopez-de-Silanes, F., Pop-Eleches, C., & Shleifer, A. 2004. Judicial checks and balances. Journal of Political Economy , 112(2): 445-470.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 1997. Legal determinants of external finance. Journal of Finance , 52(3): 1131-1150.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 1998. Law and finance. Journal of Political Economy , 106(6): 1113-1155.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 2002. Investor protection and corporate valuation. Journal of Finance , 57(3): 1147-1170.

Laffont, J. J., & Tirole, J. 1993. A theory of incentives in procurement and regulation . Cambridge, MA: MIT Press.

Lang, M., Lins, K. V., & Maffett, M. 2012. Transparency, liquidity, and valuation: International evidence on when transparency matters most. Journal of Accounting Research , 50(3): 729-774.

Lang, M. H., Lins, K. V., & Miller, D. P. 2003. ADRs, analysts, and accuracy: Does cross listing in the United States improve a firm's information environment and increase market value? Journal of Accounting Research , 41(2): 317-345.

Lang, M. H., Lins, K. V., & Miller, D. P. 2004. Concentrated control, analyst following, and valuation: Do analysts matter most when investors are protected least? Journal of Accounting Research , 42(3): 589-623.

Lee, Y. W., and Stowe, J. D. 1993. Product risk, asymmetric information, and trade credit. Journal of Financial and Quantitative Analysis, 28: 285-300.

Leland, H. E., & Pyle, D. H. 1977. Informational asymmetries, financial structure, and financial intermediation. Journal of Finance , 32(2): 371-387.

Lesmond, D. A. 2005. Liquidity of emerging markets. Journal of Financial Economics , 77(2): 411-452.

Lesmond, D. A., Ogden, J. P., & Trzcinka, C. A. 1999. A new estimate of transaction costs. Review of Financial Studies , 12(5): 1113-1141.

Leuz, C., Nanda, D., & Wysocki, P. 2003. Investor protection and earnings management: An international comparison. Journal of Financial Economics , 69(3): 505-527.

Leuz, C., & Verrecchia, R. E. 2000. The economic consequences of increased disclosure (digest summary). Journal of Accounting Research , 38: 91-124.

Levine, R. 1991. Stock markets, growth, and tax policy. Journal of Finance , 46(4): 1445- 1465.

165

Liu, Q., & Siu, A. 2011. Institutional and corporate investment: Evidence from investment- implied return on capital in China. Journal of Financial and Quantitative Analysis , 46(6): 1831-1864.

Liu, X., & Ritter, J. R. 2011. Local underwriter oligopolies and IPO underpricing. Journal of Financial Economics , 102(3): 579-601.

Logue, D. E. 1973. On the pricing of unseasoned equity issues: 1965–1969. Journal of Financial and Quantitative Analysis , 8(01): 91-103.

Long, M. S., Malitz, I. B., and Ravid, S. A. 1993. Trade credit, quality guarantees, and product marketability. Financial Management , 22: 117-127.

Love, I. 2011. Trade credit versus bank credit during financial crises. Trade finance during the great trade collapse . The World Bank, Washington, DC, 27-39.

Love, I., Preve, L. A., and Sarria-Allende, V. 2007. Trade credit and bank credit: Evidence from recent financial crises. Journal of Financial Economics, 83: 453-469.

Maksimovic, V. 2001. Firms as financial intermediaries: Evidence from trade credit data. World Bank Policy Research Working Paper . Available at: http://dx.doi.org/10.1596/1813-9450-2696

Masulis, R. W., Wang, C., & Xie, F. 2007. Corporate governance and acquirer returns. Journal of Finance , 62(4): 1851-1889.

Maug, E. 1998. Large shareholders as monitors: Is there a trade-off between liquidity and control? Journal of Finance , 53(1): 65-98.

McLean, R. D. 2011. Share issuance and cash savings. Journal of Financial Economics , 99(3): 693-715.

Megginson, W. L. 2005. The economics of bank privatization. Journal of Banking & Finance , 29(8): 1931-1980.

Megginson, W. L. 2016. Privatization, state capitalism, and state ownership of business in the 21st century. Foundations and Trends in Finance , forthcoming. Available at: https://ssrn.com/abstract=2846784

Megginson, W. L., Nash, R. C., Netter, J. M., & Poulsen, A. B. 2004. The choice of private versus public capital markets: Evidence from privatizations. Journal of Finance , 59(6): 2835-2870.

Megginson, W. L., Nash, R. C., & Randenborgh, M. 1994. The financial and operating performance of newly privatized firms: An international empirical analysis. Journal of Finance , 49(2): 403-452.

Megginson, W. L., & Netter, J. M. 2001. From state to market: A survey of empirical studies on privatization. Journal of Economic Literature , 39(2): 321-389.

166

Megginson, W. L., Ullah, B., & Wei, Z. 2014. State ownership, soft-budget constraints, and cash holdings: Evidence from China’s privatized firms. Journal of Banking & Finance , 48(1): 276-291.

Megginson, W. L., & Weiss, K. A. 1991. Venture capitalist certification in initial public offerings. Journal of Finance , 46(3): 879-903.

Meltzer, A. H. 1960. Mercantile credit, monetary policy, and size of firms. Review of Economics and Statistics , 42(4): 429-437.

Meltzer, A. H. 1963. The demand for money: The evidence from the time series. Journal of Political Economy , 71(3): 219-246.

Mian, S. L., and Smith, C. W. 1992. Accounts receivable management policy: Theory and evidence. Journal of Finance 47: 169-200.

Miller, M. H., & Orr, D. 1966. A model of the demand for money by firms. Quarterly Journal of Economics , 80(3): 413-435.

Mola, S., & Loughran, T. 2004. Discounting and clustering in seasoned equity offering prices. Journal of Financial and Quantitative Analysis , 39(01): 1-23.

Mulligan, C. B. 1997. Scale economies, the value of time, and the demand for money: Longitudinal evidence from firms. Journal of Political Economy , 105(5): 1061-1079.

Myers, S. C. 1984. The capital structure puzzle. Journal of Finance , 39(3): 574-592.

Myers, S. C., & Majluf, N. S. 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics , 13(2): 187-221.

Myers, S. C., & Rajan, R. 1998. The paradox of liquidity. Quarterly Journal of Economics , 113(3): 733-771.

Nash, R. 2017. Contracting issues at the intersection of the public and private sectors: New data and new insights. Journal of Corporate Finance , 42: 357-366

Nilsen, J. H. 2002. Trade credit and the bank lending channel. Journal of Money, Credit, and Banking , 34(1): 226-253.

Ohlson, J. A., & Juettner-Nauroth, B. E. 2005. Expected EPS and EPS growth as determinants of value. Review of Accounting Studies , 10: 349-365.

Opler, T., Pinkowitz, L., Stulz, R., & Williamson, R. 1999. The determinants and implications of corporate cash holdings. Journal of Financial Economics , 52(1): 3- 46.

Petersen, M. A. 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22: 435-480.

167

Petersen, M. A., and Rajan, R. G. 1997. Trade credit: Theories and evidence. Review of Financial Studies, 10: 661-691.

Pinkowitz, L., Stulz, R., & Williamson, R. 2006. Does the contribution of corporate cash holdings and dividends to firm value depend on governance? A cross-country analysis. Journal of Finance , 61(6): 2725-2751.

Price Waterhouse. 1989a. Privatization: Learning the lessons from the U.K. experience . London: Price Waterhouse.

Price Waterhouse. 1989b. Privatization: The facts. London: Price Waterhouse.

Rhee, S. G., & Wang, J. 2009. Foreign institutional ownership and stock market liquidity: Evidence from Indonesia. Journal of Banking & Finance , 33(7): 1312-1324.

Rosenbaum, R. R., & Rubin, D. B. 1983. The central role of the propensity score in observational studies for causal effects. Biometrika , 70(1): 41-55.

Sappington, D. E., and Stiglitz, J. E. 1987. Privatization, information and incentives. Journal of Policy Analysis and Management , 6(4): 567-585.

Schwartz, E. S. 1997. The stochastic behavior of commodity prices: Implications for valuation and hedging. Journal of Finance, 52: 923-973.

Schwartz, R. A. 1974. An economic model of trade credit. Journal of Financial and Quantitative Analysis, 9: 643-657.

Shleifer, A., & Vishny, R. W. 1994. Politicians and firms. Quarterly Journal of Economics , 109(4): 995-1025.

Shleifer, A., & Vishny, R. W. 1997. A survey of corporate governance. Journal of Finance , 52(2): 737-783.

Singh, M. 2017. Financial constraints and trade credit as a strategic tool: Evidence from small-scale reservation reforms in India. Working Paper . Available: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2665032

Staiger, D., and Stock, J. 1997. Instrumental variables regression with weak instruments. Econometrica, 65: 557-586.

Stoll, H. R. 2000. Presidential address: friction. Journal of Finance , 55(4): 1479-1514.

Titman, S., & Trueman, B. 1986. Information quality and the valuation of new issues. Journal of Accounting and Economics , 8(2): 159-172.

Vickers, J., & Yarrow, G. 1988. Privatization: An economic analysis . Cambridge, MA: MIT Press.

Vickers, J., & Yarrow, G. 1991. Economic perspectives on privatization. Journal of Economic Perspectives , 5(2): 111-132.

168

Wang, Q., Wong, T. J., & Xia, L. 2008. State ownership, the institutional environment, and auditor choice: Evidence from China. Journal of Accounting and Economics , 46(1): 112-134.

Williams, F. (2008, Nov. 11) World Bank urged to lift trade credit finance. Financial Times.

Yarrow, G., King, M., Mairesse, J., and Melitz, J. 1986. Privatization in theory and practice . Economic Policy , 1(2): 324-377.

169

APPENDIX A –CHAPTER 2 VARIABLE DEFINITIONS AND DATA SOURCES

Variable Definition Source Dependent variables Cash Holdings Cash and marketable securities to total sales Compustat Global Cash/Assets Cash and marketable securities to total assets As above

Cash/Net assets Cash and marketable securities to net assets, where net assets As above equal total assets minus cash holdings Industry-Adjusted Difference between a firm’s Cash Holdings and its industry As above cash median Cash Holdings . Industries are defined according to Campbell’s (1996) 12-industry grouping Investment The sum of the yearly growth in property, plant, and As above Expenditure equipment, plus growth in inventory, plus R&D expenditure, all deflated by lagged book value of assets Tobin's Q Market value of equity minus book value of equity plus book As above value of assets, all scaled by book value of assets

Explanatory variables State Ownership Percentage of shares held by a government Firms' annual reports and offering prospectuses Control Dummy variable equal to 1 for firms in which the state As above maintains control (i.e., the government holds more than 50% of the firm’s shares) following partial privatization, and 0 otherwise Political Dummy variable equal to 1 for politically connected firms, Faccio (2006) and 0 otherwise. Local Ownership Percentage of shares held by local investors Firms' annual reports and offering prospectuses Size Log of total sales in USD million Compustat Leverage The ratio of total debt to total assets As above Net Working Capital The ratio of current assets net of cash minus current As above liabilities, all scaled by total sales Cash Flow Earnings before extraordinary items plus depreciation and As above amortization deflated by lagged total sales Capital Expenditures The ratio of capital expenditures to total sales As above Sales Growth The growth rate of sales As above Dividend Dummy variable equal to 1 if a firm pays dividends, and 0 As above otherwise Cash Flow Volatility The standard deviation of cash flows over the previous 5 As above years

170

Shareholder Rights The revised anti-director rights index, which ranges Djankov et al. from 0 to 6 (with lower scores indicating fewer (2008) shareholder rights) Private Credit The ratio of credit from financial institutions to the World Bank private sector to GDP R&D The ratio of research and development expenditures As above to total sales Acquisition The ratio of acquisition expenditures to total sales As above Market Capitalization The ratio of stock market capitalization to GDP World Bank Lagged ROA The lagged ratio of return of assets Compustat

Asset Sales The change in property, plant, and equipment As above

Debt Issuance The long-term debt issuance minus long-term debt As above reduction, both scaled by total assets Increase Capital A dummy variable equal to 1 if a firm increases As above Expenditures capital expenditures in the next year, 0 otherwise Increase Acquisitions A dummy variable equal to 1 if a firm increases As above acquisitions in the next year, 0 otherwise Increase R&D A dummy variable equal to 1 if a firm increases As above R&D expenses in the next year, 0 otherwise Increase Dividends & A dummy variable equal to 1 if a firm increases the As above Repurchases sum of dividends and repurchases Market Share The ratio of a firm’s sales to its industry sales Compustat

Government Ownership of The extent to which the banking system's assets are Barth et al. Banks government owned (2013) U.S. A dummy variable equal to 1 if a firm is a U.S. firm, Compustat 0 otherwise Time Trend The number of years since 1994, which is the first As above year in our sample period China A dummy variable equal to 1 if a firm is a Chinese As above firm, 0 otherwise Collectivism An indicator of collectivism, which is equal to 100 Hofstede (2001) minus Hofstede’s cultural index of individualism Checks The number of checks and balances in a country Beck et al. (2001) Press Freedom 100 minus the original Freedom House index. Freedom House Higher values of Press Freedom indicate that a country’s media are more independent Democracy A measure of how responsive government is to its International people, on the basis that the less responsive it is, the Country Risk more likely it is that the government will fall, Guide peacefully in a democratic society, but possibly violently in a non-democratic one. The index ranges from 0 to 6 (with higher scores indicating more democracy) Judicial Independence The average of three variables: (1) Tenure of La Porta et al. supreme court judges; and (2) Tenure of the (2004) administrative court judges; and (3) Case law. This index ranges from 0 to 1 with higher scores indicating more judicial independence

171

Disclosure by Politician Dummy variable equal to 1 if the law or regulations Djankov et al. of the country require MPs to provide either (2010) financial and/or business interests disclosures, and 0 otherwise

172

APPENDIX B –CHAPTER 3 VARIABLE DEFINITIONS AND DATA SOURCES

Variables Definition Source Firm-level variables TOBIN'S Q Total assets less book value plus market value scaled by total Compustat assets. Global COST OF The average implied cost of capital calculated as the yearly average Author's EQUITY of four commonly used empirical estimation techniques: 1) the calculation modified PEG ratio model by Easton (2004), 2) the Ohlson and Juettner-Nauroth (2005) model; 3) the Gebhardt, Lee, and Swaminathan (2001) model; 4) the Claus and Thomas (2001) model. Following Daske et al. (2008), we use the yearly average from these four estimation techniques as our measure of implied cost of capital. ZERO The percentage of days in the fiscal year for which the stock price Lesmond et al. RETURN DAYS does not change. (1999) FHT The liquidity proxy based on low-frequency data calculated as: Fong et al. FHT = 2*Sigma*Probit((1+Zeros)/2), where Sigma=Std(Non Zero (2017) Returns); Zeros=Zero Return Days/Total Days. AMIHUD The liquidity proxy developed by Amihud (2002), which is Amihud (2002) calculated as: Illiquidity=Average (|r|/Volume). The average is calculated over all positive-volume days, since the ratio is undefined for zero-volume days. PARTIAL A dummy variable equal to one if government retains shares (i.e. Firm's annual STATE >0) in a privatized firm. report STATE The percentage of state ownership. As above STATESQR The squared term of state ownership. Author's calculation LOG MV The log of the market value of equity at year end. Compustat Global BM The book value of common equity divided by the market value of As above equity. STDRET The annual standard deviation of monthly stock returns. As above EM The standard deviation of income over the standard deviation of As above cash flows. ANALYST The number of analysts. IBES LOSS A dummy variable equal to one if net income before extraordinary Compustat items is negative, and zero otherwise. Global ADR_EX A dummy variable equal to one if the firm trades on a U.S. As above exchange during the year, and zero otherwise. ADR_NEX A dummy variable equal to one if the firm has an ADR but is not As above traded on a U.S. exchange during the year, and zero otherwise.

173

INTGAAP A dummy variable equal to one if the firm reports under IFRS As above or U.S. GAAP during the year, and zero otherwise. Country-level variables COMMON A dummy variable equal to one for common law countries, La Porta et al. (1998) and zero otherwise. LISTED The number of listed firms. World Bank MEDIA A variable that rates each country's media from 0 to 100. It is Freedom House converted by 100 minus the original Freedom House index. Higher values of Press Freedom indicate that a country’s media are more independent. LGDPC Log of GDP per capita. World Bank POLRISK A variable measured as an amalgamation of 12 country International Country elements and which ranges from zero to 100. Risk Guide (ICRG) GOVBANK Government ownership of banks. Barth et al. (2013) LIMFOREI The extent to which foreign banks may own domestic banks As above GN and whether foreign banks may enter a country's banking industry. Higher number indicates less limitation. POLITICA The degree to which the supervisory authority within the As above L INDP government is independent of political influence. Higher values indicate greater independence.

174

APPENDIX C –CHAPTER 4 VARIABLE DEFINITIONS AND DATA SOURCES

Appendix Variables, definitions, and sources. Variable Definition Source Panel A: Dependent variable 100 times the ratio of trade receivables to total sales. Trade TRADE receivable equals amounts on open account (net of applicable CREDIT reserves) owed by customers for goods and services sold in Compustat Global the ordinary course of business.

The market value of equity minus book value of equity plus TOBIN’S Q book value of assets, all scaled by book value of assets. As above

Panel B: Ownership and state control variables Firms' annual reports STATEOWN The percentage of shares held by a government. and offering prospectuses A dummy variable equal to one for firms in which the state CONTROL maintains control following privatization. As above AVG_STATE The average percentage of shares held by a government in a OWN firm. As above FOREIGNOW N The percentage of shares held by foreigners. As above LOCALOWN The percentage of shares held by local investors. As above EMPLOYEEO WN The percentage of shares held by employees. As above Panel C: Firm-level control variables LOG(SALES) The natural logarithm of total assets in $US millions. Compustat Global ROS Ratio of income before extraordinary items to total sales. As above CASH HOLDINGS Ratio of cash and short-term investments to total assets. As above FIXED Ratio of total (net) property plant and equipment to the book ASSETS value of total assets. As above Growth ratio in total sales in year t, defined as the ratio of the SALES difference between total sale in year t and total sales in year As above GROWTH t-1 to total sales in year t -1. GROSS PROFIT Ratio of the difference between total sales and cost of goods As above MARGIN sold to total sales. LEVERAGE The ratio of long-term debt to the book value of assets. As above CAPEX The ratio of capital expenditure to total sales. As above Panel D: Industry- and country-level control variables GDP PER The natural logarithm of GDP per capita in constant 2000 World Development CAPITA U.S. dollars. Indicators

175

International Financial PRIVATE Private credit by deposit money banks divided by GDP. Statistics; CREDIT International Monetary Fund Database of Political NUMVOTE Vote shares of government parties. Institutions (2012) A dummy variable indicating whether the party orientation with LEFT respect to economic policy can be defined as communist, As above socialist, social democratic, or left wing HERFINDAHL Compustat INDEX The sum of squared market shares for all firms in an industry Global PRICE -COST Sales divided by operating costs, all at the four -digit SIC code MARGIN level As above MARKET SIZE The ratio of an industry’s market size by industry sales As above The weighted average gross value of the cost of property, plant, and equipment for firms for which this is the primary industry (at ENTRY SIZE the four-digit SIC code level), weighted by each firm’s market As above share in this industry. GOVERNMENT OWNERSHIP OF Government ownership of banks. La Porta (2000) BANK LIMITATIONS A variable measures stringency of foreign banks may own ON FOREIGN domestic banks and whether foreign banks may enter a country's Barth et al. BANK ENTRY banking industry. (2013)

176