9784431551270.Pdf

Advances in Japanese Business and Economics 8

Toshihiro Ihori Kimiko Terai Editors The Political Economy of Fiscal Consolidation in Japan Advances in Japanese Business and Economics 8

More information about this series at http://www.springer.com/series/11682 Advances in Japanese Business and Economics

Editor in Chief: RYUZO SATO C.V. Starr Professor Emeritus of Economics, Stern School of Business, New York University

Managing Editors: HAJIME HORI, Professor Emeritus, Tohoku University KAZUO MINO, Professor, Kyoto University MARIKO FUJII, Professor, The University of Tokyo

Editorial Board Members: TAKAHIRO FUJIMOTO MASAHIRO MATSUSHITA Professor, The University of Tokyo Professor Emeritus, Aoyama Gakuin YUZO HONDA University Professor Emeritus, Osaka University TAKASHI NEGISHI Professor, Kansai University Professor Emeritus, The University of TOSHIHIRO IHORI Tokyo, The Japan Academy Professor, The University of Tokyo KIYOHIKO NISHIMURA TAKENORI INOKI Professor, The University of Tokyo Professor Emeritus, Osaka University TETSUJI OKAZAKI Special University Professor, Professor, The University of Tokyo Aoyama Gakuin University YOSHIYASU ONO JOTA ISHIKAWA Professor, Osaka University Professor, Hitotsubashi University JUNJIRO SHINTAKU KUNIO ITO Professor, The University of Tokyo Professor, Hitotsubashi University KOTARO SUZUMURA KATSUHITO IWAI Professor Emeritus, Professor Emeritus, The University Hitotsubashi University, of Tokyo, Visiting Professor, The Japan Academy International Christian University HIROSHI YOSHIKAWA Professor, The University of Tokyo Advances in Japanese Business and Economics showcases the research of Japanese scholars. Published in English, the series highlights for a global readership the unique perspectives of Japan’s most distinguished and emerging scholars of business and economics. It covers research of either theoretical or empirical nature, in both authored and edited volumes, regardless of the subdiscipline or geographical coverage, including, but not limited to, such topics as macroeconomics, microeconomics, industrial relations, innovation, regional development, entrepreneurship, international trade, globalization, financial markets, technology management, and business strategy. At the same time, as a series of volumes written by Japanese scholars, it includes research on the issues of the Japanese economy, industry, management practice and policy, such as the economic policies and business innovations before and after the Japanese “bubble” burst in the 1990s. Overseen by a panel of renowned scholars led by Editor-in-Chief Professor Ryuzo Sato, the series endeavors to overcome a historical deficit in the dissemination of Japanese economic theory, research methodology, and analysis. The volumes in the series contribute not only to a deeper understanding of Japanese business and economics but to revealing underlying universal principles. Toshihiro Ihori • Kimiko Terai Editors

The Political Economy of Fiscal Consolidation in Japan Editors Toshihiro Ihori Kimiko Terai Professor Professor Faculty of Economics Faculty of Economics The University of Tokyo Keio University Tokyo, Japan Tokyo, Japan

ISSN 2197-8859 ISSN 2197-8867 (electronic) ISBN 978-4-431-55126-3 ISBN 978-4-431-55127-0 (eBook) DOI 10.1007/978-4-431-55127-0 Springer Tokyo Heidelberg New York Dordrecht London

Library of Congress Control Number: 2014954229

© Springer Japan 2015 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com) Preface

Japan’s accumulated public debt is over twice its GDP. This figure is high compared with that of other developed countries; it is even higher than that of Italy, which is experiencing a severe fiscal crisis. Remarkably, the level of Japan’s public debt was not so severe in the 1980s. The debt grew rapidly in the 1990s when the country suffered an economic downturn and conducted expansionary fiscal measures. Today, Japan is confronted with the pressing need to control fiscal deficits and the accumulation of public deficits. The puzzle is why Japan has not yet succeeded in its program of fiscal consolidation. This book investigates the reasons for Japan’s persistent fiscal deficits and delayed fiscal reform, placing a special emphasis on its political economy. Each chapter focuses on a particular policy field, such as public pensions, female labor participation, public works, and intergovernmental transfer schemes. By closely examining the reasons for increasing public expenditure or insufficient tax revenues, the authors identify the political elements that have hindered effective steps toward fiscal consolidation. In the course of our analyses, we came to understand the important reasons causing fiscal deficits in addition to the relatively low tax burden ratio compared with that of other countries: the recession of the 1990s and Japan’s rapidly changing demographics. In post-war Japan, the Liberal Democratic Party has formed the government except during the period of administration by the Democratic Party (August 2009– December 2012). The ruling party has frequently spent much on public works programs, as part of a Keynesian stimulus package. After the economic bubble in the 1980s, Japan experienced a severe recession. The government increased its expenditure on public works by drawing up supplementary budgets. These budgets can be considered as pork barrels provided for local interest groups and related politicians. We need to develop a view of special-interest politics to analyze why Japan’s government is compelled to rely on pork-barrel projects. In response to the rapid aging of Japan’s population, the government was urged to increase social security expenditures for the elderly. Today, the budget for social security expenditures represents approximately one-third of the general account’s entire budget. Baby boomers are now retiring, which will press the government

v vi Preface to further increase the expenditure for their pensions and medical expenses. The higher ratio of retired generations also means that their political power grows relative to that of working generations. Therefore, we cannot effectively control social security expenditures without considering the political economy of an aging population. This book illustrates that Japan’s political and bureaucratic character has worked to over-represent the demands of elderly generations and local interest groups. These groups worked to increase the expenditures from which they benefited. The results derived from our studies of Japanese fiscal problems suggest that institutional reforms are necessary because political factors aggravate economic problems. These results also have meaningful implications that can be applied to other democratic countries struggling with fiscal deficits. This book comprises three parts. The first part presents comprehensive views of Japan’s fiscal problems. Japan’s fiscal policies and fiscal situation in the 1990s and 2000s are summarized in Chap. 1, “Fiscal Consolidation in the Political Economy of Japan,” by Toshihiro Ihori. The chapter shows that macroeconomic measures to cope with business cycle risks are not always desirable in the face of privileged political pressures. It emphasizes the importance of the government’s commitment to fiscal consolidation, highlighting the incentive problems for the government and various interest groups. Chapter 2, “The Political Economy of Social Security Funding: Why Social VAT Reform?,” by Hideki Konishi, analyzes the political economy of social security funding. The chapter shows that demographic aging influences the equilibrium choice of wage or consumption tax funding due to a differential impact on inter- and intragenerational redistribution. The results derived may explain why some developed countries prefer financing social security expenditures with a consumption tax. Each chapter included in Part II is devoted to a particular fiscal problem in Japan. Chapter 3, “Female Labor Supply, Social Security, and Fiscal Consolidation,” by Ryuta Ray Kato and Masumi Kawade, examines the effect of expanding female labor supply on economic growth, as well as on fiscal revenues through increased tax revenues and contributions to the public pension scheme. Their simulation results indicate that even if the potential female labor force that cannot work due to childcare becomes a full-time workforce, its effect on the economy and fiscal situation is small. It is shown that reducing the gap in the wage profile between male and female workers is a key for increased female labor supply to have a significant impact. Chapter 4, “Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems Associated with Intergovernmental Transfers and Their Political Roles,” by Nobuo Akai, provides insights on the linkage between local public finance and fiscal consolidation in Japan, focusing on the incentive problems induced by intergovernmental transfers. He points out that with the lack of commitment in the political contract between central and local governments, ex post political discretion for bailing out local governments produces inefficiency. Preface vii

Chapter 5, “Public Policy and Economic Growth in the Integrating Japanese Economy,” by Hiroki Kondo, models fiscal competition among local governments and derives how economic activities are concentrated in a few regions. Conse- quently, the income gap among regions is serious and persistent, with many regions failing to attract industries despite over-providing public infrastructure. Thus, Chap. 5 provides another reason for local governments’ inefficient spending. Part III provides an overview of the Japanese electoral and budgetary systems and proposes the institutional reforms necessary for fiscal consolidation. Chapter 6, “Tax Policy Under the ‘Generational Election System,’” by Takero Doi, investigates the effects of introducing a new electoral system that defines electoral districts by region and age. The current system allocates seats in the legislative bodies among regions, without considering demographics. In an aging society under the majoritarian system, political participation of young voters cannot be proportionally reflected in the share of seats in the legislative bodies. The proposed system is expected to mitigate intergenerational conflict by representing the interests of young voters more than they are represented under the current system. Agency problems in the budgetary process, creating large and inefficient budgets, are explored in Chap. 7, “Budgets Under Delegation” by Kimiko Terai and Amihai Glazer. Japan’s budgetary process is divided into two stages: decisions on fiscal caps by the Ministry of Finance (MOF) and the allocation of budgets among projects by spending ministries. A spending ministry can exploit asymmetric information between the MOF and itself to obtain a larger budget. Public works spending and the increased total budget in the new administration are examined as two illustrations of theoretical results. This book is appropriate for advanced undergraduate students and graduate students who wish to learn about the latest research in the area of political economy, and for practitioners who wish to broaden their knowledge of the Japanese economy. We explore the background for each result and provide the reader with a feel for the development of each research area. We also provide technical analyses for the results, as well as more intuitive explanations. Finally, we present new results, making original contributions to the field. Each chapter is self-contained, so the reader can open the book to any chapter and begin reading without missing any of the notation or technique. The idea to publish this book emerged at the annual conferences on public policy at the University of California, Irvine. This project started when Hiroki Kondo, author of Chap. 5, spent his sabbatical year at the University. He had been conducting research with Professor Amihai Glazer, also of the University of California, Irvine and co-author of Chap. 7. The two scholars organized the first conference in February 2005. All the authors of this book have been participants in our previous conferences. For the 10th anniversary, each author presented his or her manuscript written on a topic within the field of Japan’s political economy and the issue of fiscal consolidation. Discussants who are experts in their corresponding fields provided suggestions in comments on the presentations. These comments are also included in the viii Preface book in separate comment papers. We should note that the authors were able to revise their manuscripts according to the discussants’ comments as they were preparing for publication. Furthermore, we are very thankful for the discussants’ efforts to participate in the tenth conference. We would like to express our deep gratitude to all the Japanese and U.S. participants who made presentations or provided comments in previous conferences, especially Matthew Beckmann, Shun-ichiro Bessho, Dan Bogart, Jan K. Brueckner, Sam Bucovetsky, Seiji Fujii, Tracy Gordon, Yoichi Hizen, Tomoya Ida, Ryo Ishida, Makoto Kakinaka, Masaru Kono, Joong Ho Kook, Ikuo Kume, Satoko Maekawa, Martin C. McGuire, Hiroaki Miyamoto, Tomomi Miyazaki, Takumi Naito, Shintaro Nakagawa, Yukihiro Nishimura, John M. Quigley, Masayo Sakata, Kenneth Small, Yumiko Taba, and Yuichiro Yoshida. We wish to thank the staff at the University of California, Irvine, for kindly assisting us. We are also grateful to Springer Japan and to Professor Ryuzo Sato, the Editor in Chief of the Advances in Japanese Business and Economics book series, for giving us effectual editorial advice and for guiding us toward the publication of this book. Last but not least, we would like to thank Professor Amihai Glazer for his cooperation, generous help, advice on our research, and encouragement to continue our conferences at the University of California, Irvine.

Tokyo, Japan Toshihiro Ihori May 2014 Kimiko Terai Contents

Part I Comprehensive Views of Japan’s Fiscal Policy

1 Fiscal Consolidation in the Political Economy of Japan ...... 3 Toshihiro Ihori Comment Paper to Chapter 1 Keigo Kameda 2 The Political Economy of Social Security Funding: Why Social VAT Reform? ...... 35 Hideki Konishi Comment Paper to Chapter 2 Naomi Miyazato

Part II Fiscal Problems in Japan

3 Female Labor Supply, Social Security, and Fiscal Consolidation .. 69 Ryuta Ray Kato and Masumi Kawade Comment Paper to Chapter 3 Masatoshi Jinno 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems Associated with Intergovernmental Transfers and Their Political Roles ...... 95 Nobuo Akai Comment Paper to Chapter 4 Takeshi Miyazaki

ix x Contents

5 Public Policy and Economic Growth in the Integrating Japanese Economy ...... 113 Hiroki Kondo Comment Paper to Chapter 5 Takashi Fukushima

Part III Institutional Reforms Necessary for Fiscal Consolidation

6 Tax Policy Under the “Generational Election System” ...... 145 Takero Doi Comment Paper to Chapter 6 Haruo Kondoh 7 Budgets Under Delegation ...... 167 Kimiko Terai and Amihai Glazer Comment Paper to Chapter 7 Keisuke Hattori

Index ...... 193 Contributors

Nobuo Akai Osaka School of International Public Policy, Toyonaka, Japan Takero Doi Keio University, Tokyo, Japan Takashi Fukushima National Graduate Institute for Policy Studies, Tokyo, Japan Amihai Glazer University of California, Irvine, Irvine, CA, USA Keisuke Hattori Osaka University of Economics, Osaka, Japan Masatoshi Jinno Toyo University, Tokyo, Japan Keigo Kameda Kwansei Gakuin University, Sanda, Japan Ryuta Ray Kato International University of Japan, Minami-Uonuma, Japan Masumi Kawade Nihon University, Tokyo, Japan Hiroki Kondo Sophia University, Tokyo, Japan Haruo Kondoh Seinan Gakuin University, Fukuoka, Japan Hideki Konishi Waseda University, Tokyo, Japan Takeshi Miyazaki Kyushu University, Fukuoka, Japan Naomi Miyazato Nihon University, Tokyo, Japan

About the Editors

Toshihiro Ihori is a professor of economics at The University of Tokyo. He has a B.A. and an M.A. from The University of Tokyo and a Ph.D. in economics from Johns Hopkins University. His major ﬁeld of research is public economics. Details are at http://www.e.u-tokyo.ac.jp/fservice/faculty/ihori/ihori.e/ihori01.e.html.

Kimiko Terai is a professor of economics at Keio University. She has a B.A. in education from Tokushima University and an M.A. and a Ph.D. in economics from The University of Tokyo. Her major ﬁeld of research is public economics. Details are at https://sites.google.com/site/kimikoteraiwebsite/home.

xiii Part I Comprehensive Views of Japan’s Fiscal Policy Chapter 1 Fiscal Consolidation in the Political Economy of Japan

Toshihiro Ihori

Abstract The present chapter highlights several important factors of fiscal consolidation in Japan by investigating incentives for governments and private agents to either implement or postpone consolidation and structural fiscal reforms in a political economy. This chapter shows that the counter-cyclical fiscal measures, which result in the accumulation of public debt, are not always desirable from a long-run perspective. The long-run commitment to fiscal consolidation may enhance fiscal structural reforms even if these reforms might produce short-run costs and, hence, cause severe resistance from interest groups.

Keywords Deﬁcit ceiling • Fiscal privilege • Income ﬂuctuation • Political effort

1 Introduction

Many countries, including Japan, are now confronted with the need to simultaneously pursue fiscal discipline for fiscal reconstruction (or consolidation) and implement structural reforms for reorganizing fiscal expenditures—such as improving the efficiency of public works and social welfare spending. These are macroeconomic and microeconomic reforms, respectively. However, in designing a reform package over time, the government faces a trade-off between enhancing such long-run structural and consolidation reforms and pursuing the short-run goal of improving existing macroeconomic circumstances in a political economy. Implementing short-term Keynesian stimulus policies at a time when the economy needs long-term consolidation measures and structural changes hurts fiscal sustainability and expenditure efficiency and discourages self-help efforts in the private sector and in local governments. In Japan, there have always been strong political pressures to conduct counter- cyclical Keynesian fiscal measures. Although deficit reduction and structural reforms such as raising consumption taxes and reallocating public works resources to cut wasteful and unproductive spending could well enhance the GDP in the future, most Japanese politicians have strongly argued that there is no immediate

T. Ihori (*) Department of Economics, University of Tokyo, Hongo, Tokyo 113-0033, Japan e-mail: [email protected]

© Springer Japan 2015 3 T. Ihori, K. Terai (eds.), The Political Economy of Fiscal Consolidation in Japan, Advances in Japanese Business and Economics 8, DOI 10.1007/978-4-431-55127-0_1 4 T. Ihori need for such burdensome measures as reducing public spending and/or raising taxes as long as government policies prevent macroeconomic activities from slipping into recession. In other words, it is often argued that good economic circumstances are necessary for attaining fiscal consolidation and successful fiscal reforms; this argument emphasizes the importance of existing macroeconomic situations and the benefit of counter-cyclical measures in the short run. Prime Minister Shinzo Abe is currently employing the so-called third arrow of “Abenomics,” a plan to pull Japan out of its long economic slump. The first arrow consists of an expansionary monetary policy, wherein vast quantities of money were pumped into the economy. The second arrow provides a dramatic fiscal stimulus package. The third arrow involves a growth strategy. Among the three arrows, the keenly awaited growth strategy should be the most important, since it seeks to boost Japan’s long-term economic performance. However, apart from the expansionary monetary policy conducted by the Bank of Japan, his actual economic policy has relied heavily on the second arrow, which involves nothing but conventional Keynesian fiscal measures. Although Abenomics is now popular among voters, his policy has not coped well with fiscal consolidation. The present chapter highlights several important factors of fiscal consolidation in Japan by investigating incentives for governments and private agents to either implement or postpone consolidation and structural fiscal reforms in a political economy. Needless to say, political factors such as the political stability of the government and the degree of lobbying activity of interest groups are important for determining the outcome. Politicians can accept, among other ideas, fiscal structural reform and a movement toward fiscal consolidation and reorganization only if the governing party enjoys a stable majority, and hence, sees the probability of losing power as being low. On the contrary, if the government is politically weak in the sense that it cannot control fiscal privileges directly, then macroeconomic and microeconomic reforms will not be well conducted. As a result, lobbying activities will be exaggerated. In particular, it becomes difficult to pursue fiscal consolidation and structural reforms during periods of recession. This is the case in Japan. Moreover, it should be stressed that there are several economic reasons for the delay in fiscal consolidation and structural reforms. First of all, cooperation with efforts towards fiscal reform has a feature of private provision of public goods in the sense that the benefits of fiscal reform derived from each agent’s cooperation spill over to the economy over time, even as the agent has to accept the burden of fiscal reform. Everyone welcomes the general idea of fiscal consolidation, but not everyone readily accepts an increase in individual burden. This attitude may create the problem of free riding, which may well result in the delay. Since the political support of all relevant agents is necessary for successful fiscal consolidation, incorporating a self-filling mechanism to overcome free riding becomes an important point of consideration. Secondly, everyone optimistically expects things to improve before such hard- hitting measures as cutting spending and/or raising taxes are implemented. In Japan, it was believed in the 1990s that the potential growth rate would somehow rise significantly again. Structural measures that would reduce the budget deficit were put off in the hope that the deficit would begin to shrink once the Japanese 1 Fiscal Consolidation in the Political Economy of Japan 5 economy recovered. A similar reaction emerged during the world financial crisis of 2008. The current 2014 three-arrow policy of “Abenomics” is based on a similar understanding. Recovering the current economic situation is always the top priority. However, since the potential growth rate has been declining for many years, the delay could not solve the fiscal problem in an aging Japan. This chapter shall discuss incentives for governments and private agents to either implement or postpone fiscal consolidations and structural fiscal reforms in a political economy. This chapter is organized as follows. In Sect. 2, we summarize the recent fiscal policy and fiscal situation in Japan and then examine the efficacy of recent fiscal measures in the country. In Sect. 3, we examine how the counter-cyclical fiscal policy should be used in theory if the government may conduct macroeconomic and microeconomic fiscal measures. In Sect. 4, we consider the politically quasi-strong government, which may impose a deficit ceiling to restore fiscal sustainability, but which is not strong enough in terms of implementing microeconomic measures to control fiscal privileges. We then explore a seemingly paradox- ical case of a pro-cyclical policy by developing a simple model of the optimal debt limit in a political economy. In Sect. 5, we discuss several policy implications of attempts at fiscal consolidation. Finally, we provide a conclusion in Sect. 6. This chapter will show that the counter-cyclical fiscal measures, which result in the accumulation of public debt, are not always desirable from a long-run perspective. The long-run commitment to fiscal consolidation may enhance fiscal structural reforms even if these reforms might produce short-run costs and, hence, cause severe resistance among interest groups.

2 Japan’s Fiscal Policy and Fiscal Situation

2.1 Japan’s Fiscal Policy in 1990s

Japan’s fiscal situation is now the worst of the G7 counties. This is partly due to a slowdown in economic growth since the 1990s and increases in public works and social welfare spending, owing to Keynesian fiscal measures in an aging society. On the one hand, when national income does not grow much, tax revenue will not increase either. On the other hand, government spending has gradually risen due to political pressures from interest groups, resulting in large budget deficits. In 1997, the Japanese government implemented the Fiscal Structural Reform (to reduce the budget deficit). However, in 1998, it stopped the reform and implemented both tax reductions and increases in public investment (the traditional Keynesian counter-cyclical policy) because of the severe economic and financial situations and the defeat of the governing Liberal Democratic Party (LDP) in the Upper House election. It is also noted that although Japanese government bonds were being issued largely, their yields were the lowest among the G7 countries in the bond market, which seemed to support the excessive fiscal measures. 6 T. Ihori

In particular, during the Obuchi and Mori administrations of the late 1990s, structural reforms including fiscal consolidation and reconstruction were put on the back burner because of the immediate fiscal needs of the coalition governments. In late 1999, the Komeito Party joined a coalition government with the LDP and the Liberal Party. Together, the three parties commanded a majority in both the Lower and Upper Houses. The overriding objective of the coalition was, as stated by Prime Minister Obuchi, to maintain a numerical advantage in the Diet. Given its low public approval ratings, however, the alliance faced a pressing need to produce results quickly to gain public support from the older population. This was particularly true of the Komeito, which even more urgently needed to deliver near-term achievements because of its emphasis on welfare-related spending. Such pressures set the stage for a free-spending policy. There was, indeed, a widespread feeling in the private sector and the local governments that the central government would come to their aid if the economic situation worsened. Since the late 1990s, government bonds outstanding have increased due to their continued issuance over the past years. Central government bonds outstanding at the end of the 2014 fiscal year would reach approximately 1,000 trillion yen. In terms of both the ratios of the general government deficit to GDP and the general government debt to GDP, Japan’s fiscal condition has become the worst among developed countries.

2.2 Fiscal Policy in the 2000s

In the face of a deteriorating fiscal situation in Japan, public opinion concerning the role of the government changed significantly from the 1980s to the 2000s. The number of business people favoring small government grew. This change was probably caused by the fear that further increases in fiscal burdens would fall on the business community. Such a concern was also shared by the household sector in urban areas. This is the background of the fiscal reconstruction and structural reform movement started by the Koizumi administration in the early 2000s. As to reorganizing fiscal expenditures, the Koizumi administration sought to reduce wasteful public projects. Regarding GDP distribution, comparatively more public works were implemented in rural areas than in urban areas in the 1990s, but the allocation was reversed in the early 2000s. The Koizumi administration also utilized cost-benefit analysis and outsourcing based on market testing to some extent, although the attempt was not sufficient enough to improve Japan’s fiscal situation. In the 2009 general election, Japan underwent a change in government. The Democratic Party (DP), for the first time, took office by obtaining a large majority in the Lower House. Voters at the 2009 election supported the DP’s assumption that huge wasteful spending exists in the government budget, therefore fiscal consolidation can easily be achieved by cutting such wasteful spending without raising consumption taxes. However, it turned out that the DP government could not identify large sources of wasteful spending. Although the new government intended 1 Fiscal Consolidation in the Political Economy of Japan 7 to conduct macroeconomic and microeconomic fiscal reforms, it could not attain its objectives. The DP government was finally forced to make the decision to raise the consumption tax rate from 5 to 10 % by 2015. This development helped to reduce the informational asymmetry between the Japanese government and the general voters with respect to the fiscal situation of Japan’s government. Shinzo Abe, Japan’s current Prime Minister, is now employing the so-called third arrow of Abenomics, the plan to pull the country out of its long economic slump. Since his concern is mainly the current macroeconomic situation, he has adopted the conventional Keynesian fiscal policy to stimulate aggregate demand through public works. As a result, fiscal consolidation is still not well handled. The Abe administration seemed reluctant to raise the consumption tax rate as scheduled, although he finally decided to raise it to 8 %, effective from April 2014.

2.3 Efﬁcacy of Counter-Cyclical Measures

Considering the above developments, we examine the macroeconomic effects of fiscal policy empirically. There exist competing arguments on the efficacy of the fiscal policy in Japan. One argument is that the effects of fiscal policy were highly significant, and hence recession would have deepened without fiscal expansion. Contrary to this view is the argument that the fiscal policy did not have enough of an expansionary effect to push up macroeconomic activity; hence, unlimited public expenditures simply worsened the fiscal crisis. These opposing arguments, which lead to different policy implications, are due mainly to different understandings of the macroeconomic analytical framework. Namely, the former argument is based on the conventional Keynesian model of liquidity-constrained agents, whereas the latter is based on the neoclassical model of rational agents. Using the VAR method, Ihori et al. (2002) showed that fiscal policies generated limited effects on output in Japan. Namely, tax policies did not have a stronger effect than changes in government expenditure on output. Furthermore, the effect of fiscal policies was too marginal to recover macroeconomic activities, which is consistent with the latter view based on the neoclassical model of rational agents. There are some recent studies that estimate fiscal multipliers in Japan, and these include Kato (2010), Watanabe et al. (2010) and Hirai and Nomura (2012). The multiplier effect of public works has become very low in recent years, and hence, the efficacy of stimulating aggregate demand by using public works is controversial. Let us then discuss the supply-side effect of public investment. Whether public capital is efficiently provided in Japan is a crucial question from a normative perspective. There are some empirical studies on the productivity of public capital in Japan. Iwamoto (1990) and Mitsui and Ota (1995) calculated the marginal productivity of public capital based on the estimated production function and concluded that public capital had been too low in Japan until the early 1980s. See also Ihori and Kondo (2001) for a similar conclusion. 8 T. Ihori

Ihori and Kondo (1998) investigated the effect of public investment on private consumption by estimating the consumption function for 1955 to 1996. In the 1960s, public investment had a great impact on private consumption, but its influence decreased after 1980. The evidence suggests that the total level of public capital was not too low in the 1990s. Kondo and Ihori (1999) and Ihori et al. (2001) found that the causality relationship between public works and private consumption was significant during the whole sample period (1957–1994). The causality was strong during 1958–1975 but then faded. The impulse response of private consumption to public works was large during 1958–1975, but became less in later years. Hence, the aggregate level of public capital might be sufficient or too high in the last part of that period. As shown in Asako et al. (1994), if the allocation of public works is appropriately revised, then it could stimulate macroeconomic activities and enhance economic welfare. To sum up, the supply-side effect of public investment has decreased in recent years. In these studies, the common conclusion is that public capital was productive, but its productivity has declined recently. See also Aso and Nakamoto (2008) for a similar conclusion. When the fiscal situation becomes very serious, fiscal reconstruction may stimulate private consumption and investment as a result of the “non-Keynesian” effect. This effect means that if fiscal spending is wasteful or if the fiscal condition is very bad, fiscal consolidation such as spending cuts and/or tax increases will instead stimulate private demand, contrary to the conventional Keynesian effect. If a non-Keynesian effect actually occurs, the government can attain both fiscal sustainability and economic recovery simultaneously. This argument is consistent with the analytical understanding that fiscal conditions affect the effectiveness of counter-cyclical policies. The deteriorating fiscal situation in Japan may suggest that the “non-Keynesian” effect has some relevancy in recent years. According to Nakazato (2002) and Kameda (2008), among others, during sustainable periods in Japan, when fiscal deficit and debt outstanding/GDP were smaller than a certain level, we can observe the standard Keynesian effect. How- ever, during unsustainable periods, when fiscal deficit and debt outstanding/GDP were much larger than a certain level, a non-Keynesian effect occurred. In such cases, expansionary fiscal measures such as increasing public spending and/or decreasing tax revenues depressed private demand, since these actions deteriorated the fiscal situation.

2.4 Needs of Fiscal Consolidation

It seems logical to put fiscal consolidation into effect as soon as possible if the fiscal system is no longer sustainable. Thus, the need for fiscal consolidation depends on how sustainable the fiscal system would be in the near future. A few empirical analyses have been conducted on the sustainability of government debt in Japan. Fukuda and Teruyama (1994) and Ihori et al. (2002), among others, have attempted to test the fiscal sustainability condition using the methodology of Hamilton and 1 Fiscal Consolidation in the Political Economy of Japan 9

Flavin (1986). Ihori et al. (2002) conducted an empirical analysis for Japan using data for fiscal years 1957–1958 to 1998–1999. Overall, these studies found that the Japanese government debt did not satisfy the transversality condition for the period 1965–2000. After the Koizumi administration implemented structural reforms, Japan’s primary deficits decreased in the early 2000s. However, the deficits increased rapidly after the world financial crisis of 2008. On the basis of Bohn’s(1998) condition, Doi and Ihori (2009) could not reject the hypothesis that the consolidated government debt was not sustainable in the 2000s. Sakuragawa and Hosono (2010, 2012) showed that if the government reacts to the fiscal crisis, the debt-to-GDP ratio will increase without bound, and the fiscal policy will not be sustainable. A delayed fiscal consolidation will result in a higher debt-to-GDP ratio and require a larger primary surplus to restore sustainability. See also Doi et al. (2011). There are several simulation studies on future outcomes in the event that fiscal structural reform is not attained but fiscal sustainability is somehow maintained. By proposing an alternative policy rule that incorporates the reaction of the surplus to the debt, the simulation study of Sakuragawa and Hosono showed that if the primary deficit turns into a surplus in 10 years and increases up to 2.2 % surplus of the GDP in 20 years, then the debt-to-GDP ratio will stabilize at 230 %. Ihori et al. (2011) estimated that the national burden to GDP, defined as the sum of the tax burden and the social security burden to GDP, will be approximately 60 % in 2050 in the benchmark case so as to maintain fiscal sustainability. If the current scheme is maintained, an aging population will result in an increase in the total amount of public pension benefits as well as in the total amount of public health insurance benefits. Public health insurance benefits are expected to increase by approximately 1 percentage point of the GDP every 10 years, reaching 13 % in 2050. Had it not been for the 2004 pension reform, the tax burden would have increased to approximately 36 % of the GDP, and the social security burden would have risen to 23.3 % in 2050. As explained in Sect. 2, the former Prime Minister Noda Yoshihiko of the DP and the opposing LDP and Komeito parties finally agreed in 2012 to achieve the integrated reform of the social security and tax systems, with a focus on raising the consumption tax to 10 % (from 5 % in 2013) by fiscal year 2015. This agreement also aimed to strengthen the function of the social security system. Although such an increase in consumption tax would be favorable to future generations, it seems that it would be difficult to achieve a primary balance surplus by the 2020s with this increase alone. Moreover, intergenerational inequalities still remain in an aging society with the pay-as-you-go social security system. Namely, even if the government attains a positive primary balance in the near future by raising taxes further, the burden on future generations will still be very high in Japan, suggesting that the current financial situation facing the Japanese government in terms of intergenerational conflicts is very serious. In an aging society, any reforms of the pay-as-you-go system will produce intergenerational conflicts. If the government postpones reducing its deficits, the situation will be even worse because of increased interest payments incurred by the huge amount of outstanding government debt. 10 T. Ihori

3 Counter-Cyclical Macro Policy

3.1 Ordinary Fluctuations and Serious Crises

In this section, we first consider the normative macro fiscal policy of a benevolent government as the first best solution, wherein the government may conduct both macroeconomic and microeconomic measures. Macro-fiscal measures to cope with business cycle risks have two objectives. The first is to alleviate the damage of negative shocks and the second is to stimulate aggregate demand in a recession. Moreover, the objective of maintaining a boom as long as possible or raising the potential GDP as much as possible could also be regarded as an objective of countercyclical macro-fiscal measures in a broader sense. If the potential growth path is maintained at a high level by such a policy, it will be beneficial for future generations in the long run. For each policy objective, it is useful to conduct a discretionary policy and use automatic built-in stabilizers appropriately. To make countercyclical policies more effective, it is important to collect information on business cycle risks and compare benefits of ex ante and ex post policies. The automatic stabilizing function mainly prepares for risks in advance and works well even if the negative shock is not well anticipated. However, it may not be able to fully cope with large-size risks. If a negative shock is large, we need to conduct a discretionary policy as well. That is, if the macroeconomic situation reflects a serious crisis, GDP or wage income will decline significantly. Many people will suffer badly from the recession. Hence, it becomes a top priority to improve the macroeconomic situation as soon as possible. An active countercyclical policy may be justified, even if the burden will be shifted to future generations. In reality, it is necessary to determine if the decline in GDP or wages is truly significant. Even if a recession occurs temporally as just one stage in ordinary fluctuations, it may well be perceived as a serious and permanent crisis in a political economy. In Japan, the political pressure to pursue massive fiscal measures is always strong, even though these measures can be justified only in a serious crisis. Furthermore, when the economy actually faces a serious crisis, all kinds of fiscal stimulation measures may be easily conducted in a political economy. It is therefore crucial to distinguish between ordinary fluctuations and serious crises in evaluating the usefulness of countercyclical policies. In the case of a normal business cycle, output fluctuations cause recessions temporally. So long as the macroeconomic condition is sustainable, the government may still attain the potential growth path in the long run. Although expansionary fiscal measures might be needed, built-in stabilizers can handle macroeconomic stabilization better. It is normally desirable to have a small size of deficits in a recession in the case of ordinary fluctuations. This policy implication holds in both the Keynesian and neoclassical models. The Keynesian argument utilizes fiscal deficits as a tool for stabilizing measures, whereas the neoclassical argument utilizes fiscal deficits as a tool for smoothing tax revenues over time. 1 Fiscal Consolidation in the Political Economy of Japan 11

As stressed by public choice literature, expansionary measures are taken under political pressure in the second best world. Expansionary measures are often conducted with little attention to fiscal sustainability. Therefore, on the one hand, fiscal deficits easily increase in a recession. On the other hand, it is hard to reduce deficits by conducting restrictive fiscal policies during a boom. In response to the world financial crisis in 2008, the Japanese government conducted a large-scale fiscal stimulus plan to stimulate the economy, which was badly affected by the crisis. In 2011, Japan’s government also conducted massive fiscal measures to cope with the Great East Japan Earthquake. As a result, Japan’s fiscal condition became worse and worse. Even in the case of a serious emergency, the costs and benefits of fiscal measures should still be considered appropriately. We cannot automatically justify all expansionary fiscal measures even during a serious crisis. As to intergovernmental financing, discretionary policies are usually conducted by the central government. Since a negative shock normally hurts the overall economy, the central government has the comparative advantage in conducting expansionary fiscal measures. However, business cycle risks do not often occur uniformly across the country. If the shocks in ordinary fluctuations are relatively small and affect regions differently, then the role of local governments would become important. If a negative shock hurts specific regions badly, such as the Great East Japan Earthquake of March 11, 2011 did, local governments in the affected areas should conduct fiscal measures. Considering informational asymmetry between the central government and regional individuals with respect to region- specific needs, local governments have a comparative advantage in conducting discretionary policies such as social welfare programs. Overall, both the central and local governments should conduct discretionary policies consistent with their fiscal capabilities. To sum up, we should use an automatic stabilizing measure mainly for ordinary fluctuations, whereas we should also implement an expansionary discretionary policy in the case of serious emergencies. At the same time, the discretionary policy measure should be consistent with long-run fiscal sustainability. From the viewpoint of fiscal consolidation, it might be useful to impose automatic fiscal stabilizing functions into the budget system in that if deficits increase, spending automatically decreases and taxes automatically increase. Moreover, it is desirable to improve the fiscal condition before another recession occurs. In Sect. 4, we shall point out that a pro-cyclical policy may be desirable in a second-best case of the political economy.

3.2 Explanations of Delay

There are several explanations as to why fiscal consolidation and structural reforms have been postponed or put into effect so slowly. First of all, the political circumstance is important for the successful outcome of fiscal consolidation attempts. The government cannot pursue fiscal reconstruction without stable political conditions. 12 T. Ihori

Specifically, politicians can accept the idea of fiscal reform for fiscal reconstruction only if the governing party enjoys a stable majority in the Diet and, hence, perceive a low probability of losing power. Among others, Alesina and Tabellini (2005) found that a stable government has an incentive to reduce government deficits. Alesina and Perotti (1995) also reported that coalition governments in the OECD countries delayed reducing fiscal deficits. This explanation may well be applicable to Japan. In Japan, the traditional governing party, the LDP, had been politically weak, and budget deficits had increased, starting in the mid-1970s. Then, in the early 1980s, the LDP swept the general elections for the House of Representatives and began to implement what was described as a fiscal reconstruction—the reduction of fiscal deficits. In contrast, in the 1990s, especially after 1993, several parties formed coalition governments. Since these governments were not politically strong enough to impose the deficit ceiling effectively, deficits increased. The Koizumi administration in the early 2000s was politically strong enough to impose the ceiling, and the deficit decreased. Thus, the situation in Japan fits the overall findings of theoretical and empirical works on fiscal deficits. Secondly, everyone optimistically expected the economic situation to improve before such hard-hitting measures were implemented. In particular, it was widely believed in the 1990s that the potential growth rate would rise again. In the face of the financial crisis caused by the bursting of the bubble economy, draconian efforts such as bad-debt disposal were postponed in the hope that land and stock prices would begin to rise. Such procrastination had earlier led to a full-blown crisis in the financial sector in November 1997. Consolidation measures that would reduce the budget deficit were put off in the hope that the deficit would begin to shrink once the Japanese economy recovered. A similar policy response occurred in the world financial crisis of 2008. However, since the potential growth rate had already been declining, the delay could not solve the problem in Japan. Another reason was the several scandals swirling around the central and local governments in the 1990s. These undermined public confidence in the central government and the ruling political parties. Moreover, poor communication due to information asymmetry between the public and politicians delayed reforms. Even if policy makers were correctly informed about the merits of reform, the voting public did not share that information and, therefore, could not properly evaluate their policies. Drastic reforms could not get off the ground because voters did not trust the ruling government or the bureaucracy. As explained above, the change in government in 2009 helped to reduce the informational asymmetry by allowing the public to recognize the serious fiscal situation. Many voters now understand how bad the fiscal situation in Japan is. The delay is also attributable to the free-riding incentive. When the debt limit is imposed, cooperation with attempts at fiscal reform may be regarded as a private provision of public goods. In other words, the benefit of fiscal reform derived from each agent’s cooperation spills over to the economy over time, while the agent has to accept the burden of fiscal reform. This may create the problem of free-riding. These explanations are plausible and well applicable to Japan’s case. It should also be noted that consideration of the macroeconomic situation has been an 1 Fiscal Consolidation in the Political Economy of Japan 13 important factor of the delay. In Sect. 4, we will explore the impact of macroeconomic shock on fiscal consolidation attempts more fully. It is not always desirable that bad economic circumstances should postpone fiscal reforms in a political economy. Some could argue that the fiscal situation is not bad, and hence, there is no need to conduct fiscal consolidation as a top priority. They may emphasize that the central and local governments, although heavily indebted, also have credits and assets. That is, the total value of the government-held tangible and financial assets—those of the central government, local governments, and social security funds—is very large. It is therefore argued that government debt is not a great concern. However, the argument that debt is not a serious problem in net terms raises two questions. One question is just how many of the government assets could actually be sold. Many government-held tangible assets exist in the form of public infrastructure, such as roads. These would be hard to sell. By the same token, many of the financial assets, held in pension funds, are also unsaleable. The pension fund is different from tax revenues, which the government can use freely. Pension reserves, of course, are intended to be dedicated to future payments to pensioners. Pension insurance premiums collected from working people must be paid sometime in the future in the same way that public bonds must be redeemed as they mature. Moreover, the balance of pension funds will deteriorate as the birthrate declines and the population ages. Perhaps 20 or 30 years from now, this could lead the overall government deficit to assume even more serious proportions.

3.3 Political Efforts and Micro Fiscal Policy

Even if the government is benevolent, it cannot always conduct the first best policy in a political economy. When the government is politically weak, it may not conduct either macroeconomic or microeconomic measures effectively. In reality, the political strength of a government is affected by the rent-seeking activities of interest groups. Therefore, it is important to consider the role of political efforts in micro-fiscal policies. Since the 1990s, government deficits in Japan have increased rapidly partly because the Japanese government has been politically weakened under the political pressure of many interest groups. We first examine the outcome of the microeconomic fiscal policy in Japan. The microeconomic measure is to revamp the fiscal system drastically. Public-sector operations must be made efficient. Ihori and Kondo (2001) investigated the efficacy of public works for the Japanese economy. We investigated the productivity of disaggregated public capital goods by estimating the productivity of public capital- related income, or labor income, based on the aggregate production function approach. While not definitive due to the restriction of data availability, our results suggest that the allocation of public works was not optimal in Japan. Namely, there still existed large differences among the marginal productivity levels of the various types of public capital. Infrastructure for railways, telephone networks, and postal 14 T. Ihori services were not large enough until the early 1990s. Therefore, if public works spending were reallocated to projects to improve economic efficiency, it could have stimulated private consumption and enhanced economic welfare in Japan. One important reason why funds for public works have not been efficiently allocated arises from political pressures from local interest groups. Many local interest groups (or politicians) seek to obtain more money from the central and local governments through a variety of lobbying activities. In particular, local interest groups living in the rural and agricultural areas got a lot of transfers mainly in the form of public works. These interest groups may be regarded as one of the most powerful in Japan, a plausible explanation for which is as follows. The ruling party exerts an influence to decide the national budget. Getting more grants is important for it to be reelected. More representatives in the ruling party, the LDP for postwar period, have been seated for the rural regions. Hence, people in the rural regions have more representatives in the ruling party than people in the urban regions. A region where more representatives are elected from the ruling party is distributed more subsidies from the central government throughout the period of the party’s rule. Therefore, representatives of the Diet appeal to the cabinet or the central bureaucrats to distribute more funds to their own regions. The Japanese government is politically weak in microeconomic measures and therefore cannot control fiscal privileges such as wasteful public works. Comparing the data on Japan’s public works with those of other countries, we may say that local residents in Japan have greater privileges than those in other countries, reflecting the influential role of their interest groups. Moreover, although expansionary fiscal measures have often been conducted, these measures have not produced any expansionary effects on the economy. Agriculture-related public capitals and fishing ports and measures for flood control and conservation of forests are being allocated too many privileges/subsidies due to lobbying activities of local interest groups. Therefore, representatives of the rural regions, influenced by local interest groups and voters, put political pressure on the central government to distribute more grants to the rural regions. As a result, the allocation of region-specific privileges in the form of subsidies and public works from the central government has been determined mainly by political factors. In fact, lobbying activities of local interest groups and local governments were exaggerated during the 1990s, as shown in Ihori et al. (2001) and in the empirical evidence of Doi and Ihori (2002). This is one of the main reasons for the lack of progress of fiscal reconstruction in the 1990s. As a result, in intergovernmental financing in Japan, many transfers are made from the central government to local governments. The central government subsidizes local governments by the amount of approximately 5 % of the GDP every year. Since local governments depend heavily on subsidies from the central government, they may try to obtain as much money as possible from the central government, irrespective of their economic conditions. When the central government is politically weak, it may respond to these pressures by simply giving subsidies to local governments. This soft budget mechanism further stimulates rent-seeking behavior on the part of local governments and 1 Fiscal Consolidation in the Political Economy of Japan 15 politicians. That is, even when the economy is not in a recession, too much discretionary policy is conducted by local governments, resulting in a huge amount of wasteful public works. Local governments may free ride on subsidies from the central government, without imposing sufficient taxes on their residents. As a result, the overall government deficit may increase. In particular, in Japan, the criterion of basic fiscal need in the local allocation tax formula has not been explicitly specified. The amount of local allocation tax is actually determined by political negotiation among various interest groups and politicians, wherein local politicians have a strong bargaining power. Furthermore, local governments do not determine their own local tax rates with their own will. Instead, they seek heavy subsidies from the central government. When the government is politically weak in both macroeconomic and microeconomic measures, the outcome is the worst. However, even if the government is politically weak in microeconomic measures and cannot control the political activities of interest groups, it could still be strong enough in macroeconomic terms to impose a deficit ceiling. In Sect. 4, we consider this quasi-strong case in order to examine the normative aspect of macro-fiscal policies by investigating the impact of income fluctuations on the deficit ceiling. In a recession, it is normally necessary to stabilize output fluctuations. It does not seem a good idea to reduce deficits in a recession as the first best solution. However, if rent-seeking political activities are strong but not efficient in the second best case, it may not be desirable for the government to stick to the counter-cyclical fiscal policy from a long-term perspective. We explore the possibility that a pro-cyclical policy could be desirable as a second best solution of a political economy.

4 Pro-Cyclical Policy and Macroeconomic Situation

4.1 Debt Ceiling

To avoid a fiscal crisis, the government may impose a deficit ceiling. This section investigates the impact of income fluctuations on the deficit ceiling in a simple political economy model of a second best world. Suppose the government is politically strong enough to impose a deficit ceiling, but it is weak in the sense that it cannot control the fiscal privileges of interest groups directly. The government can only use the deficit ceiling to cope with fiscal worsening. Hence, in the face of fiscal instability, it becomes important to know how to impose target levels for the deficit ceiling. For example, the Stability and Growth Pact (SGP) is an agreement among the 27 member states of the European Union (EU) to facilitate and maintain the stability of the Economic and Monetary Union. As is well known, the actual budget criteria that member states must respect are: (1) an annual budget deficit of no higher than 3 % of the GDP and (2) a national debt lower than 60 % of the GDP or 16 T. Ihori approaching that value. In March 2005, the EU Council relaxed the rules due to criticisms of insufficient flexibility and to make the pact more enforceable. That is, the ceilings of 3 % for the budget deficit and 60 % for public debt were maintained, but the decision to declare a country in excessive deficit can now depend on certain parameters: the behavior of the cyclically adjusted budget, the level of debt, the duration of the slow growth period, and the possibility that the deficit is related to productivity-enhancing procedures. As explained in Sect. 2, the Fiscal Structural Reform Act in Japan was implemented in early 1997 to achieve consolidation targets similar to the ones in the EU. However, when the severe recession occurred in late 1997, the Japanese government, prompted by political pressure from various interest groups, aban- doned the consolidation attempt by conducting massive fiscal measures to stimulate aggregate demand. As such, the Act was no longer regarded as a legal constraint. In 2013, the Japanese government decided to raise the sales tax in April 2014 from 5 % to the current 8 % and to 10 % in October 2015, aiming to cover swelling social security costs amid the increasing proportion of older people in Japan. However, it seems that Prime Minister Abe was reluctant to raise the consumption tax rate as scheduled. Before deciding whether to increase the consumption tax to 8 % in April 2014, the government held a meeting from August 26 through 31 on the sales tax hike to hear the opinions of a total of 60 experts and other leading figures. Some experts have argued, however, that the tax hike should be postponed or carried out at a slower pace to mitigate its impact on the recuperating economy. On October 1, 2013, Prime Minister Abe finally decided to raise the tax rate to 8 % with massive fiscal measures. Although his popularity is high and he is politically strong, he does not impose the deficit ceiling effectively. These developments suggest the importance of macroeconomic considerations in maintaining a commitment to the debt limit in a political economy. A key issue that has arisen is how long-term objectives should be modified to accommodate economic fluctuations and to maintain some flexibility with respect to fiscal stabilization policies. The conventional wisdom is that a counter-cyclical fiscal policy is needed in a recession. Section 1.3 discussed how a benevolent and politically strong government should conduct counter-fiscal measures in response to income fluctuations in a first best solution. However, in a second best solution, the size of the public deficit is endogenously determined within the political process and affected by the community’s privileges and consolidation activities. Although countercyclical policies are dominant in many developed countries, pro-cyclical fiscal policies are sometimes observed when the fiscal situation is serious. It is important to explore how the ceiling should respond to income fluctuations in a political economy. As explained in previous sections, the macroeconomic effect of fiscal policy is marginal and political efforts are strong but inefficient in Japan. In such a second best case, even if the government may impose the deficit ceiling effectively, it cannot employ the first best solution in a political economy. This section explores the possibility that a pro-cyclical policy might be desirable from the long-run perspective. 1 Fiscal Consolidation in the Political Economy of Japan 17

4.2 Theoretical Analysis in a Political Economy

It is, hence, important to incorporate the political influence of local interest groups explicitly into the analytical framework. Interest groups can seek fiscal privileges through political efforts, which are normally stimulated during periods of recession. Therefore, it is argued that in order to attain successful outcomes for fiscal consolidation, a good macroeconomic situation is needed. In fact, Japan’s politically weak government has conducted countercyclical fiscal measures for many years. How- ever, this does not necessarily suggest that a benevolent and politically strong government should raise the debt limit in a recession. In this section, we investigate how the optimal deficit ceiling varies in response to income fluctuations, based on Ihori (2013, 2014). We assume that the government is politically strong enough in the macro-policy setting to impose and commit itself to a debt limit. However, it is politically weak in the micro-policy setting in that it cannot control fiscal privileges or reorganize the allocation of spending directly. Analytically, by imposing the debt limit or spending ceiling, the fiscal consolidation efforts made by private agents assume a public good nature of improving the overall fiscal situation. Hence, as shown in Ihori and Itaya (2002, 2004) and Ihori (2011), among others, the analytical framework of the private provision of public goods is useful for examining the outcome of fiscal reconstruction and consolidation attempts. In this section, we explicitly incorporate the opposite of fiscal consolidation, that is, political efforts to seek fiscal privileges, which hurt the fiscal situation because of the public bad nature of raising fiscal deficits. When the ceiling on fiscal deficits is employed, an increase in fiscal privileges results in a decrease in useful public goods. Hence, the interest groups may recognize the additional cost of their political efforts. See also Cornes and Sandler (1996), Velasco (2000), and Battaglini and Coates (2008), among others. We suppose that there are two generations: the present generation and the future generation. Each agent lives for only one period, and she is selfish in the sense that she does not leave any bequests. We assume that n agents are born in period 1. Furthermore, n agents also live in period 2. They are identical. The benevolent government discounts the welfare of the future generation by ρ (< 1). The government budget constraint for each period is given as

T1 À G1 À Z1 ¼ÀD, ð1:1:1Þ

T2 À G2 À Z2 ¼ ðÞ1 þ r D, ð1:1:2Þ

Xn where Ti is the exogenously given total tax revenue in period i, i ¼ 1,2. T1 τ1j, j¼1 Xn T2 τ2j, and τij > 0 is the exogenously given tax burden for each agent j in j¼1 period i. Let Zi ∑ zij be the total fiscal privileges in period i. zij is the sum of the agent-specific transfer eij and agent-specific public spending hij that may be 18 T. Ihori politically determined, and which benefits only interest group j in period i. Although hij is useful for group j, we call it wasteful spending in the sense that its size becomes too large as a result of political pressure.

Zi ¼ Ei þ Hi ðÞi ¼ 1, 2 ð1:2Þ where Ei ∑ eij and Hi ∑ hij. Finally, let Gi be the amount of useful public good in period i, which beneﬁts all agents. The present-value budget constraint is given by

T G Z T þ 2 ¼ G þ 2 þ Z þ 2 : ð1:3Þ 1 1 þ r 1 1 þ r 1 1 þ r

The ceiling on fiscal deficits is denoted by D, which is determined by the government in period 1, whereas r is the exogenously given interest rate. The government can impose the deficit ceiling as a consolidation rule, but it cannot control the fiscal privileges directly at the micro-policy level. In other words, the government may impose a ceiling on the total spending G + Z, but it cannot control the distribution between G and Z. In fact, many countries, including Japan, have recently imposed such a ceiling on total spending. As Nerlich and Reuter (2012) explained, many EU countries have introduced some kind of scale rule; in particular, expenditure rules and balanced budget rules are the most common scale rules in place among EU countries, whereas only a few revenue rules are in place, generally. Since taxes are fixed in our second best model, expenditure rules are equivalent to deficit rules here. The (lifetime) utility function of the representative agent of generation i (i ¼ 1, 2) i i is given as U ¼ U (ci, hi, Gi). The utility is dependent on private consumption ci, the quantity of (wasteful) agent-specific spending hi, and the quantity of (useful) pure i public good, Gi. In addition, ci, hi and, Gi would be an increasing function of U , since private consumption, wasteful fiscal spending, and useful public goods are normal goods. By substituting the budget constraints (1.1.1) and (1.1.2) into the utility function, the social welfare function of a benevolent government may be written as ÀÁÀÁ 1 2 W ¼ U c1, h1, T1 À Z1 þ D þ ρU c2, h2, T2 À Z2 À ðÞ1 þ r D : ð1:4Þ

The utility is dependent on the amounts of two ﬁscal privileges, hi, ei, as well as on the quantity of private goods consumed, ci. The utilitarian government has discounted preferences on the future period. We may regard δ (1 À ρ)/ρ as the discount rate. Greater political effort by each agent leads to more privileges and more income. Hence, it is plausible to formulate, for simplicity, that agent j’s production of overall disposable income is given by wij À τij + eij and, to some extent, is associated with a cost of obtaining privileges, denoted by ϕeij + Φhij, which increases with 1 Fiscal Consolidation in the Political Economy of Japan 19 political efforts. The disposable income includes the cost of ﬁscal privileges, ϕeij + Φhij. Thus, each agent’s (lifetime) budget constraint is given as

cij ¼ wij À τij þ eij À ϕeij À Φhij: ð1:5Þ

Here, the before-tax income, wij, is exogenously given. In order to obtain zij, some resources are needed. The given parameters ϕ, Φ summarize the degree of inefficiency of political efforts. In other words, the larger ϕ, Φ are, the greater the political efforts must be to produce a given amount of privilege and, hence, the less economically efficient the technology of the political effort is. To explore the implications of political efforts with regard to fiscal policy, let us first investigate the first-best solution. Suppose that the government can control fiscal privileges as well as taxes and that political efforts are thus absent in the first- best benchmark case. The political effort activities equal zero: Φ ¼ 0. ei is just a transfer and hi is not regarded as wasteful spending since the government can now control the size of hi. In the first best case, taxes and transfers are perfect substitutes. So long as the government can control a lump-sum transfer ei and, hence, disposable income, it may optimize private consumption over time, even if taxes are exogenously fixed. Then, the government should increase subsidies whenever before-tax income declines in recessions, so as to maintain after-tax disposable income at the original level. This policy can attain the intertemporal smoothing allocation of private consumption and public goods at the original values.

4.3 Second-Best Outcome

We now consider the second best case, where the government cannot control either taxes or fiscal privileges. The structure of the second-best game in a political economy is as follows. In the first stage, the government imposes a deficit ceiling. In the second stage, a representative agent conducts political efforts to obtain privileges in each period.

4.3.1 Political Efforts by Two Generations in the Second Stage

In the second stage of the game, we formulate the optimizing behaviors of private agents in each period. For simplicity, we do not always explicitly provide subscript j for agent j, since the agent of each interest group behaves in the same way. Note that in this optimization problem, the quantity of other privileges associated with others’ political efforts is taken as a given. The representative agent chooses his or her own political effort in each period, in a similar manner. 20 T. Ihori

First of all, the ﬁrst-order condition with respect to h2 of the future generation in period 2 is given as

2 ¼ 2 þ 2Φ ð : : Þ Uh UG Uc , 1 6 1

2 ∂U2 2 ∂U2 2 ∂U2 where U , U , and U . We assume an interior solution of h2 > 0. c ∂c2 h ∂h2 G ∂G2 In (1.6.1), h2 likely decreases with D. In fact, the reaction function of h2 is given as ÀÁ 2 h2 ¼ h D; w2; T2 , ð1:7:1Þ where

∂h 1 ÂÃ h2 2 ¼À ðÞ1 þ r U2 Φ À U2 þ U2 , D ∂D Δ cG hG GG ∂ ÂÃ 2 h2 ¼À1 À 2 Φ þ 2 À 2 hw Ucc Uch UGc , ∂w2 Δ ∂ ÂÃ 2 h2 ¼À1 À 2 Φ þ 2 À 2 hT UcG UhG UGG , ∂T2 Δ

∂ 2 ∂U2 ∂U2 and Δ U2 À U2 Φ2 + U2 À 2U2 , where U2 Uc , U2 G, and U2 G. hh cc GG hG cc ∂c2 cG ∂c2 GG ∂G2 From the second-order condition, Δ < 0. Suppose that private consumption is additively separable from two public goods. UcG ¼ Uch ¼ 0. Then, we know that 2 > hw 0. A smaller amount of disposable income, w2, results in a smaller amount of political efforts to obtain h2 in the future period. This effort becomes a normal good. The related intuition is as follows: When w2 is low, c2 is low, so that an increase in h2 raises the cost of h2 significantly, as the marginal utility of c2 is already high. > 2 < 2 > In addition, if UhG 0, then hD 0, hT 0. When the government imposes a deficit ceiling at a higher level in period 1, it will produce a smaller amount of fiscal privilege by the agent in period 2. In other words, a weaker deficit ceiling in period 1 results in a smaller amount of political effort in period 2. The related intuition is as follows: When the ceiling is weak and D is high, G2 is small, so that the marginal cost of h2, UG,is large. Hence, h2 is depressed. In other words, when the fiscal situation worsens, the agent in period 2 does not have a stronger incentive to obtain more fiscal privileges. Moreover, a smaller amount of tax revenue T2 results in a smaller amount of political efforts in the future period. When T2 is low, G2 is small, so that an increase in h2 raises the cost of h2 significantly if UhG > 0. Even if UhG 0 but |UhG| were not < 2 < 2 > large, we still have similar responses since UGG 0: hD 0, hT 0 Second, the first-order condition with respect to e2 of the future generation in period 2 is given as

2 ¼ 2 ð : : Þ Ucm UG, 1 6 2 1 Fiscal Consolidation in the Political Economy of Japan 21

where m 1 À ϕ. We assume an interior solution of e2 > 0. In (1.6.2), e2 likely decreases with D. Actually, the reaction function of e2 is given as ÀÁ 2 e2 ¼ e ; D; w2; T2 , ð1:7:2Þ where

∂e 1 ÂÃ e2 2 ¼À ðÞÀ1 þ r U2 m þ U2 , D ∂D Δ cG GG ∂ ÂÃ 2 e2 ¼À1 2 À 2 ew Uccm UGc , ∂w2 Δ ∂ ÂÃ 2 e2 ¼À1 2 À 2 eT UcGm UGG , ∂T2 Δ

∂ 2 ∂U2 ∂U2 and Δ U2 mm + U2 À 2U2 m, where U2 Uc , U2 G, and U2 G. cc GG cG cc ∂c2 cG ∂c2 GG ∂G2 From the second-order condition, Δ < 0. If UcG ¼ 0, this condition is easily 2 < 2 < 2 > satisfied. Then, we know that eD 0, ew 0, eT 0. Similar to (1.7.1), when the government imposes a deficit ceiling at a higher level in period 1, it will produce a smaller amount of fiscal privilege by the agent in period 2. In other words, a weaker deficit ceiling in period 1 results in a smaller amount of political effort in period 2. Now, contrary to (1.7.1), a larger amount of disposable income w2 results in a smaller amount of political efforts to obtain agent-specific transfer e2 in the future period. This effort becomes an inferior good. The related intuition is as follows: When w2 is high, c2 is large, so that an increase in e2 does not increase the benefit of e2 significantly, as the marginal utility of c2 is already low. Next, we consider the optimizing behavior of the representative agent of the present generation in period 1. Similar to (1.6.1), the optimality condition with respect to h1 reduces to

1 ¼ 1 þ 1Φ ð : : Þ Uh UG Uc , 1 8 1

1 ∂U1 1 ∂U1 where U and U . The reaction function of h1 is given as c ∂c1 G ∂G1 ÀÁ 1 h1 ¼ h D; w1; T1 , ð1:9:1Þ where

∂h 1ÂÃ h1 1 ¼À U1 À ΦU1 À U1 , D ∂D Λ hG cG GG 22 T. Ihori

∂ ÂÃ 1 h1 ¼À1 À 1 Φ þ 1 À 1 hw Ucc UcG UGh , ∂w1 Λ ∂ ÂÃ 1 h1 ¼À1 À 1 Φ þ 1 À 1 hT UcG UGh UGG , ∂T1 Λ

Λ 1 À 1 Φ2 1 À 1 and Uhh Ucc + UGG 2UhG. From the second-order condition, Λ < 0. As before, suppose that private consumption is additively separable from public goods. UcG ¼ Uch ¼ 0. Then, we know 1 > > 1 > 1 > 1 that hw 0. In addition, if UhG 0, then hD 0, hT 0. Now, the sign of hD is the 2 opposite of the sign of hD.In(1.9.1), h1 likely increases with D. When the government imposes a deficit ceiling at a higher level in period 1, it produces a larger amount of fiscal privileges in period 1—namely, a weaker deficit ceiling likely results in a larger amount of political effort being expended in the present period. As in the second period, either a larger amount of income w1 or a larger amount of tax T1 results in a larger amount of political effort in the present period, h1. Second, we consider the optimizing behavior of the representative agent of the present generation in period 1. Similar to (1.6.2), the optimality condition with respect to e1 reduces to

1 ¼ 1 : ð : : Þ Uc m UG 1 8 2

The reaction function of e1 is given as ÀÁ 1 e1 ¼ e D; w1; T1 , ð1:9:2Þ where

∂e 1ÂÃ e1 1 ¼À U1 m À U1 , D ∂D Λ cG GG ∂ ÂÃ 1 e1 ¼À1 1 À 1 ew Uccm UGc , ∂w1 Λ ∂ ÂÃ 1 e1 ¼À1 1 À 1 eT UcGm UGG , ∂T1 Λ

Λ 1 1 À 1 and Uccmm + UGG 2UcGm. Λ < ¼ 1 > From the second-order condition, 0. Then, if UcG 0, we have eD 0, 1 < 1 > 1 2 ew 0, eT 0. Now, the sign of eD is the opposite of the sign of eD.In(1.9.2), e1 likely increases with D. As in the second period, either a larger amount of income w1 or a smaller amount of tax T1 results in a smaller amount of political effort to obtain agent-speciﬁc transfer in the present period, e1. 1 Fiscal Consolidation in the Political Economy of Japan 23

4.3.2 Debt Ceiling by the Government in the First Stage

It is important to know how the objective of the government in the first stage of the game is specified. Since the agent is identical, it is plausible to assume that the government is benevolent—namely, the government determines the deficit ceiling by considering the response of agents in both periods—given by reaction functions (1.7.1 and 1.7.2) and (1.9.1 and 1.9.2), respectively—so as to maximize the weighted sum of welfare of the two generations (1.4). An increase in D corresponds to an increase in G1 and a decrease in G2, benefiting the present welfare, while hurting the future welfare. The present-value government budget constraint is given as (1.3). Unlike the situation in the first-best case, the government cannot directly choose zi (and Zi) in the second-best case. zi (and Zi) is determined by the political efforts of private agents. The benevolent government maximizes W by choosing D (and hence, effectively, G1, G2) subject to (1.7.1 and 1.7.2), (1.9.1 and 1.9.2), and (1.3), and exogenously given T1, T2. Hence, the optimal level of D is determined to satisfy the following equation to be zero.

dW ÂÃÀÁ ÂÃÀÁ ¼ U1 1 À h1 þ e1 ðÞn À 1 À ðÞ1 þ r ρU2 1 þ h2 þ e2 ðÞn À 1 : ð1:10Þ dD G D D G D D

Note that the effects of z1, z2 through changes in D on the optimality condition (1.10) vanish from the envelope theorem. However, the indirect spillover effect of 1 1 À 2 2 À z1, z2 on other agents, (hD + eD)(n 1) or (hD + eD)(n 1), does not vanish. Since À 1 1 À 2 2 À the sign of 1 (hD + eD)(n 1) or (1 + (hD + eD)(n 1)) is generally ambiguous, let us consider two cases: 1 À 1 1 À À ρ 2 2 2 À < Case (i): UG[1 (hD + eD)(n 1)] (1 + r) UG[1 + (hD + eD)(n 1)] 0. When an increase in D leads to a reduction in W, we have a corner solution of D ¼ 0. 1 À 1 1 À À ρ 2 2 2 À ¼ Case (ii) UG[1 (hD + eD)(n 1)] (1 + r) UG[1 + (hD + eD)(n 1)] 0. When the direct effect is greater than the indirect effect in the first term, we have an interior solution of D > 0, where (1.10) equals zero. From now on, we will focus on an interior solution. We do not consider the case of (1.10) > 0, since this case is À 1 1 À unlikely to occur during consolidation attempts. We assume 1 (hD + eD)(n 1) > 2 2 À > 0 and 1 + (hD + eD)(n 1) 0. The first term in (1.10) is the marginal benefit of increasing D (or the marginal gain for the present generation), whereas the second term is the marginal cost of increasing D (or the marginal gain for the future generation). Either an increase in r or ρ or a decrease in δ will increase the cost of D, so that the optimal level of D declines. The optimality condition is a condition by which useful public spending is smoothed over time at the marginal utility level. 24 T. Ihori

4.4 Income Fluctuations and Deﬁcit Ceiling

4.4.1 Wage Fluctuation

Suppose w1 declines and w2 increases. Note that the after-tax income, w^ w À τ, and the before-tax income, w, change in the same way, since taxes are ﬁxed. We have ÂÃÀÁÃÂ ÀÁÀÁ dD 1 À h1 þ e1 ðÞn À 1 U1 1 þ me1 À Φh1 À U1 e1 þ h1 þ h1 U1 ¼À ÂÃDÀÁD Gc w ÂÃw ÀÁGG w w w Gh , dw þ 2 þ 2 ðÞÀ ðÞþ 2ρ 2 þ À 1 þ 1 ðÞÀ 1 1 1 hD eD n 1 1 r UGG 1 hD eD n 1 UGG ð1:11:1Þ ÂÃÀÁÂÃÀÁÀÁ dD ðÞ1 þ r ρ 1 þ h2 þ e2 ðÞn À 1 U2 1 þ me2 À Φh2 À U2 e2 þ h2 þ h2 U2 ¼À ÂÃÀÁD D Gc w ÂÃwÀÁGG w w w Gh , dw þ 2 þ 2 ðÞÀ ðÞþ 2ρ 2 þ À 1 þ 1 ðÞÀ 1 2 1 hD eD n 1 1 r UGG 1 hD eD n 1 UGG ð1:11:2Þ

2 where U ∂ U and so on. These two equations may capture the effect of GG ∂2G disposable income. The sign of (1.11.1) (or of (1.11.2)) is generally ambiguous. We can show that if private consumption c is separable from the two public goods h, G (UcG ¼ Uch ¼ 0), then the term ÂÃÀÁ 1 þ 1 À Φ 1 À 1 1 ¼À 1 1 UGc 1 mew hw UGGew UGGew becomes negative, whereas if, in addition, h and G are not strong substitutes (UhG > UGG), then the term

À 1 1 þ 1 1 UGGhw hwUGh becomes positive. If the latter effect is greater than the former effect, such a change in disposable income reduces D. Namely, the sign of (1.11.1) (or (1.11.2)) becomes positive (or negative) under these conditions. When the political effort for transfers is small but for wasteful spending is large, the second term would likely dominate the first term and, hence, a pro-cyclical policy can more likely be justified. In other words, the combination of a worse macroeconomic situation in the present period and a better macroeconomic situation in the future period does not necessarily justify the counter-cyclical policy. It might be desirable to reduceD in times of recession. Under such a case, the government conducts a restrictive fiscal policy in times of recession. The related intuition is as follows. Firstly, a decrease in w1 directly raises e1.An increase in e1 reduces G1 at the given ceiling level, increasing the marginal utility of 1 public goods. However, if the absolute value of ew is small, an increase in w1 does not 1 > increase e1 much, making this effect small. Secondly, if hw 0 is large, it reduces h1, raising G1 to a great extent. This increase in G1 much reduces the marginal utility of public good, G1, in the present period. At the same time, suppose h1 and G1 are not 1 Fiscal Consolidation in the Political Economy of Japan 25

strong substitutes (UhG > UGG). Then, a decrease in h1 would reduce the marginal utility of public good, G1, as the additional effect of UhG. Hence, combining both effects, the marginal utility of public spending declines in period 1 under these conditions. Thus, it becomes optimal to reduce G1 and to increase G2 so as to maximize the expected sum of welfare in each of the two generations. We do not argue that the counter-cyclical case is unlikely to occur. It is true that many governments actually do conduct counter-cyclical fiscal policies. If the government can control fiscal privileges optimally, this case becomes desirable. Even in the second best case, if UhG < UGG (i.e., h1 and G1 are strong substitutes) and the political effort for transfer is relatively large, (1.11.1) could become negative and (1.11.2) positive. Fluctuations in disposable income w1,w2 (or the macroeconomic situation) could affect the deficit ceiling, D, counter-cyclically. Conventional wisdom is justified in this case. Rather, our analysis suggests that this policy may not be justified when h1 and G1 are not strong substitutes within a political economy where the political effort for agent-specific transfer is relatively small.

4.4.2 Tax Revenue Fluctuation

Next, we suppose that after-tax incomes w1 À τ1, w2 À τ2 remain ﬁxed, but that tax revenues T1, T2 change exogenously. Then, under normal circumstances, we may expect that either a decrease in T1 or an increase in T2 would increase the optimal level of D, since the present ﬁscal situation becomes worse, while that of the future improves. However, either a decrease in T1 or an increase in T2 could paradoxically reduce the optimal level of D. We have ÂÃÀÁÃÂ ÀÁÀÁ dD 1 À h1 þ e1 ðÞn À 1 U1 1 À e1 À h1 þ U1 À1 þ me1 À Φh1 þ U1 h1 ¼À ÂÃD ÀÁD GG T T ÂÃGc ÀÁT T Gh T , dT þ 2 þ 2 ðÞÀ ðÞþ 2ρ 2 þ À 1 þ 1 ðÞÀ 1 1 1 hD eD n 1 1 r UGG 1 hD eD n 1 UGG ð1:11:3Þ ÂÃÀÁÃÂ ÀÁÀÁ þ 1 þ h2 þ e2 ðÞn À 1 U2 1 À e2 À h2 þ U2 À1 þ me2 À Φh2 þ U2 h2 dD ¼ 1 r ÂÃD ÀÁD GG T T ÂÃGc ÀÁT T Gh T : dT 1 þ ρ þ 2 þ 2 ðÞÀ ðÞþ 2ρ 2 þ À 1 þ 1 ðÞÀ 1 2 1 hD eD n 1 1 r UGG 1 hD eD n 1 UGG ð1:11:4Þ

As before, suppose UcG ¼ Uch ¼ 0. The term

1 < UGG 0 is negative. The term ÀÁ À 1 1 þ 1 > UGG eT hT 0 26 T. Ihori is positive. The term

1 1 > hTUGh 0 is positive if UhG > 0. If eT and hTwere large, (1.11.3) could become positive and (1.11.4) negative. In such a case, when tax revenue declines in the present period and increases in the future period, it would be optimal to reduce the initial debt limit in the present period. This debt management tack corresponds to a pro-cyclical policy. The related intuition is as follows. On the one hand, a decrease in T1 at a given level of w1 À τ1 directly reduces G1, raising the marginal utility of G1. On the other hand, it directly depresses e1 and hT, increasing useful public spending G1 at a given ceiling level, which decreases the marginal utility of public spending. Additionally, if UGz > 0, it indirectly reduces the marginal utility of public spending as well. If the former effect were dominated by the latter effect, then the marginal utility of public spending would decline in period 1, at which point it would become optimal to reduce G1.

4.5 Policy Implications

Suppose now that τ ¼ tw, where t is the tax rate; hence, the after-tax income w^ i is given as w^ i ¼ ðÞ1 À t wi. When the economy is in recession and before-tax income declines, both after-tax income and tax revenue decline. Thus, it is necessary to consider both the effect of disposable income on D and the effect of tax revenue on D simultaneously. If the sum of the effect of after-tax income on D and the effect of revenue on D were positive, then it would be optimal for the government to reduce the deficit ceiling in a recession. If h and G were not strong substitutes, we could justify the pro-cyclical policy as before. Fiscal policy is generally counter-cyclical in many OECD countries. We may justify this policy if fiscal privileges and/or taxes are optimally chosen or two public goods are strong substitutes. Interestingly, Alesina and Tabellini (2005) pointed out that, in contrast, fiscal policy tends to be pro-cyclical in many developing countries. In particular, government spending as a share of GDP goes up during booms and goes down in recessions, whereas deficits increase during booms and decrease in recessions. Their explanation is that voters do not trust corrupt governments with resources and, hence, demand tax cuts or increases in productive government spending or transfers when positive shocks hit the economy; otherwise, they fear that available resources will be “wasted” in rents. Although these studies are interesting in terms of explaining the pro-cyclical policy of a corrupt government in a political economy, they do not investigate whether a benevolent government should conduct a pro-cyclical policy during a recession. Actually, many developed countries employ pro-cyclical policies when their fiscal deficits are very large. Our study provides a plausible and new 1 Fiscal Consolidation in the Political Economy of Japan 27 explanation as to whether benevolent governments should control the deficit ceiling counter-cyclically from a long-run perspective. In some countries, agents conduct political activities and the government is politically quasi-strong enough in macroeconomic measures to impose a ceiling. Our analytical framework may be relevant. Then, if private consumption is separable from two public goods, which are not strong substitutes for each other, a pro-cyclical policy may be desirable from the long-run perspective. We may regard our theoretical model as a formulation of the limit imposed by the central government on local public debt issued by the local government. Suppose that the central government is politically strong enough to impose a limit on local public debt. Local governments are politically weak, and therefore, they cannot directly control fiscal privileges, as above. Then, (1.1.1) and (1.1.2) are the budget constraints of the representative local government and (1.4) is the objective function of the central government. The central government may set D in order to maximize (1.4). In an urban area, we would expect that the local government may choose taxes optimally, or two public goods are strong substitutes. In such a case, we could say that local governments should conduct a counter-cyclical policy. On the other hand, in a rural area, we would expect that the local government may not choose taxes optimally, or two public goods are not strong substitutes. Then, local governments should conduct a pro-cyclical policy.

5 Consolidation Policy

5.1 Reform of Intergovernmental Financing

In order to attain fiscal consolidation, it is necessary to restrain political activities that seek fiscal privileges and to reform intergovernmental financing. See Doi and Ihori (2009), among others. Reforming intergenerational financing may also restore the efficacy of counter-cyclical measures. The main direction of fiscal federalism in Japan has been that of transferring the tax base from the central government to local governments. Actually, some steps have been implemented. However, when the deficit of local governments decreased as a result of this reform, the deficit of the central government increased by the same amount. The overall deficit in the public sector in Japan did not change; therefore, these steps did not improve the fiscal condition of the public sector. Local governments should refrain from relying on the central government under hard budget constraints. It is important to revise the local allocation tax formula so that local governments and residents may take the burden of tax revenues without relying on subsidies from the central government under soft budget constraints. Seeking to enhance efficiency and transparency by a politically independent re-assessment system of public works is important for reducing local privileges. 28 T. Ihori

5.2 Commitment in a Political Economy

Without credible commitments, the government has to raise taxes ex post after fiscal conditions become worse. This may result in an increase in wasteful spending in a political economy, since an ex post tax increase would only have an income effect, and interest groups would not have an incentive to cooperate with fiscal consolidation attempts as the fiscal situation improves. In order to attain a sustainable consolidation policy, it might also be useful to impose the conditional consolidation policy rule. For example, we may impose an institutional setting for both countercyclical fiscal policy and procyclical fiscal consolidation. In the standard budget system, built-in-stabilizers have been imposed such that spending is automatically raised and taxes are automatically reduced in a recession as a result of progressive income tax, the social welfare system, unemployment benefits, and so on. Automatic stabilizers normally improve the macroeconomic situation. In addition to this, we may impose automatic fiscal stabilizers on the budget system. Namely, if fiscal conditions worsened, the budget system may automatically improve fiscal conditions by reducing spending and raising taxes. Automatic fiscal stabilizers may work at the stage when the economy is recovering after the worst of the recession is over. For example, if the fiscal condition becomes bad to a certain level, the rule should imply that a tax increase will be implemented in the near future. In such a case, anticipation of a future tax rate increase will stimulate consumption demand, owing to the intertemporal substitution effect. Stimulating aggregate demand will thus help the economy to recover. When a recession is serious, the economy will likely soon recover. Therefore, automatic lagged fiscal stabilizers could be consistent with instantaneous automatic macroeconomic stabilizers. On the other hand, fiscal stabilizers are fiscal rules as commitments, such as for raising taxes in a recession, which are built into the budget system. Such commitments are effective for fiscal consolidation in a political economy. If an increase in deficits automatically corresponds to an increase in taxes, and the interest groups can anticipate this rule in advance, it has a substitution effect to support fiscal consolidation. Namely, when the deficit rises and the tax burden is shifted to the far future, the tax burden will automatically be raised further in the near future. Hence, interest groups will have an incentive to cooperate immediately with efforts towards fiscal consolidation. In order to restore fiscal sustainability, a large-sized tax increase and/or spending reduction will be necessary sooner or later. The commitment of an ex ante fiscal stabilization rule is a more feasible and credible method than discretionary measures such as an ex post tax increase for promoting fiscal consolidation. This rule is more effective for reducing wasteful spending and attaining fiscal sustainability than discretionary fiscal consolidation in a political economy. 1 Fiscal Consolidation in the Political Economy of Japan 29

Conclusion The Japanese government has intended to implement fiscal consolidation and structural reforms in the public sector, but the job has been only partly done. Japan’s fiscal policy since the 1990s has created a serious sustainability problem because of its tendency to postpone attempts at consolidation and structural reform. Reducing outstanding government debt and promoting efficiency in the public sector are both urgent tasks. In addition to the concern that the accumulation of public debts may well be unsustainable, the deteriorated fiscal situation since the 1990s has another problem. Prolonged excessive budget deficits are harmful to the economy in the sense that excessive deficits today mean higher interest rates tomorrow, which results in crowding out private capital accumulation in the long run. Fiscal reforms should definitely be promoted in two ways using both macroeconomic and microeconomic measures. The first way, in terms of macroeconomic reforms, is to reduce the massive deficit. Although giving top priority to deficit reduction as an end in itself is undoubtedly not a rational course, deficit reduction is nevertheless an important policy objective, given Japan’s deteriorating fiscal health. The question is how long it would take to cut the deficit. Considering the serious problems that can arise from delays, a reduction program should be implemented as soon as possible. The pro-cyclical fiscal policy might be needed from the long-term perspective. The Japanese government intends to raise the consumption tax rate from 8 % to 10 % by 2015. However, this increase in taxes would not be enough to attain sustainability. A further increase in taxes and/or a decrease in public spending would be necessary. To avoid the fiscal crisis, deficit growth must be tuned with fiscal consolidation. An important and powerful tool is to impose a debt or spending ceiling. Suppose the government is politically quasi-strong enough to impose a deficit ceiling, but it is not strong enough in the sense that it cannot control fiscal privileges directly. This chapter has developed a simple theoretical model that incorporates the political behavior of private agents into the analysis of debt limit under a scenario wherein the government imposes a ceiling on the deficit and agents are selfish. The intertemporal smoothing condition normally suggests the conventional counter-cyclical fiscal policy as the first best solution. When the macroeconomic situation becomes bad, the government normally conducts a counter-cyclical fiscal policy. However, we cannot exclude pro-cyclical measures if wasteful fiscal privileges and useful public spending are not strong substitutes and political efforts for wasteful spending are relatively large in a political economy. Japan’s central government is so weak politically that it cannot impose the deficit ceiling effectively. Moreover, many interest groups direct political efforts to obtaining region-specific public works. As a result, the central

(continued) 30 T. Ihori

(continued) government provides numerous subsidies to local governments for public investment, and these subsidies have been disbursed largely to rural regions. In this sense, even though local governments implement public investments in rural regions, their investment are financed from national taxes, paid largely by taxpayers in urban areas. These facts imply that the central government has implemented more public investment in rural regions because of political pressures from local governments, although fiscal reorganization in the 2000s by the Koizumi administration changed the allocation to some extent. In order to achieve successful consolidation, the government should be politically strong enough to impose the deficit limit effectively. It is also important to reform intergovernmental financing more efficiently. The second way is to conduct microeconomic reforms to revamp the fiscal system drastically. Public-sector operations must be made more efficient. An important policy objective is to achieve an efficient government through reforms based on the principles “from public sector to private sector” and “from the state to the regions.” To ensure efficiency of public spending, both privatization and decentralization reforms are useful for closing the gap between benefits and financial burdens. These reforms need to be put forth in order to reduce unnecessary public projects. In view of the severe fiscal situation, public spending that has few benefits compared to the burden it imposes is not sustainable. As was explained in Doi and Ihori (2009), the Japanese government reformed the Fiscal Investment and Loan Program (FILP) and privatized, merged, or abolished several state-owned companies. These steps are desirable, but only in the first phase of structural reforms. The speed of reform has not been great due to resistance from interest groups. Further decisive efforts are needed to make the public sector, including public pension and medical insurance systems, more efficient. Confidence in future fiscal management should be enhanced by implementing structural reforms more fervently. A successful outcome in fiscal reorganization, including the reform of social security systems, may increase overall political support for the drastic fiscal reforms the country needs. Through continued deficit reduction and a revamping of the public sector using macroeconomic and microeconomic measures, Japan can build an efficient and equitable public sector and bequeath to future generations a strong base that will enable them to achieve a dynamic economy and society and improve their standards of living in aging Japan.

Acknowledgments An earlier version of this chapter was presented at the Tenth Irvine-Japan Conference on Public Policy. The author wishes to acknowledge the helpful comments of Professor Kameda and participants. Should any errors remain, they are the sole responsibility of the author. 1 Fiscal Consolidation in the Political Economy of Japan 31

Comment Paper to Chapter 1

Keigo Kameda Kwansei Gakuin University, 2-1 Gakuen, Sanda, Hyogo, Japan e-mail: [email protected] The first half of Chapter 1 provides a thorough survey of the Japanese fiscal situation and an extensive summary of the vast body of literature on Japanese fiscal policy. The level of detail provided means that it seems complete. Therefore, these comments concentrate on the theoretical model developed in Sect. 4, which we consider to have insight on pro-cyclical fiscal policy. The mechanism that affirms pro-cyclical fiscal policy is as follows. When the wages of an interest group decrease (usually in a recession), they demand more transfer and less public goods of the type specific to them due to the shortage of rent-seeking activity costs. When the latter effect dominates the former, the government will increase its general public good provision because outstanding government debt and tax revenue are exogenous in the model. This results in a decrease in the current marginal utility of general public goods. Thus, the government should transfer its resources to the future by reducing their debt ceiling. This reasoning behind pro-cyclical fiscal policy is new and valuable. However, as mentioned in the chapter, we must make some assumptions to obtain these results. In addition, even if these assumptions are satisfied, the effect of after-tax income, which would be more important, is ambiguous when the tax is positively related with the wage. Thus, the applicability of this model to the actual economy should be tested empirically. My back-of-the-envelope calculations show that the correlation between the real total income of a farming household and the ratio of agricultural government spending to other government spending changes from negative to positive around 1990.1 Moreover, while the correlation is positive for one rural area, it is negative for the other.2 These findings show the necessity of further development of the model: When and why did the correlation change? In what types of rural areas can we find a positive correlation? While raising these questions does not decrease the value of this study, we should try to resolve these issues in greater depth.

1 À0.833 before 1989 and 0.914 after 1990. The data sources are as follows. Income: Statistics on Trend of Management. Central government: General Accounts—Settlement of Expenditure by Purpose. Local government: Ordinary Accounts of Local Governments—Settlement of Expendi- ture by Purpose and Function. Deﬂator: Annual Report on National Accounts of 2000 and 2011 (we adjust the latter using the ratio of the former to the latter in 1980 and connect this with the former). 2 0.935 in Hokuriku and À0.200 in Shikoku. The income data source is Statistical Survey on Farm Management and Economy. The source of the governmental data and the price deﬂator are the same as in footnote 1. 32 T. Ihori

References

Alesina A, Perotti R (1995) The political economy of budget deficits. IMF Staff Papers 42:1–31 Alesina A, Tabellini G (2005) Why is fiscal policy often procyclical? IGIER Working Paper, p 297 Asako K, Tsuneki A, Fukuda S, Teruyama H, Tsukamoto T, Sugiyama M (1994) Shakai shihon no seisanryoku kouka to koukyou toushi seisaku no keizai kousei hyouka (Productivity effect of public capital and welfare evaluation of public investment policy). Keizai Bunseki 135 Aso Y, Nakamoto A (2008) Shakai shihon seibi no makuro keizai kouka (Macroeconomic effects of public investment based on OLG model). Financ Rev 89:6–29 Battaglini M, Coate S (2008) A dynamic theory of public spending, taxation and debt. Am Econ Rev 98:201–236 Bohn H (1998) The behavior of U.S. public debt and deficits. Quart J Econ 113:949–963 Cornes R, Sandler T (1996) The theory of externalities, public goods and club goods. Cambridge University Press, Cambridge Doi T, Ihori T (2002) Fiscal reconstruction and local interest groups in Japan. J Jpn and Int Econ 16:492–511 Doi T, Ihori T (2009) The public sector in Japan: past developments and future prospects. Edward Elgar, Cheltenham Doi T, Hoshi T, Okimoto T (2011) Japanese government debt and sustainability of fiscal policy. J Jpn Int Econ 25:414–443 Fukuda S, Teruyama H (1994) The sustainability of budget deficits in Japan. Hitotsubashi J Econ 35:109–119 Hamilton J, Flavin M (1986) On the limitations of government borrowing: a framework for empirical testing. Am Econ Rev 76:808–816 Hirai T, Nomura M (2012) Wagakuni ni okeru seifu shisyutsu to seifu syunyu no inga kankei (Causality between government spending and revenue in Japan). Kagawa Univ Keizai Ronsou 74:259–282 Ihori T (2011) Fiscal structural reform and economic activity: public spending and private effort. Jpn Econ Rev 62:1–24 Ihori T (2013) Flexibility of deficit ceiling and income fluctuation: long run perspective. Mimeo Ihori T (2014) Commitment, ceficit ceiling, and fiscal privilege. FinanzArchiv/Public Finance Anal, forthcoming Ihori T, Itaya J (2002) A dynamic model of fiscal reconstruction. Eur J Polit Econ 17:1057–1097 Ihori T, Itaya J (2004) Fiscal reconstruction and local government financing. Int Tax Public Financ 11(1):1–13 Ihori T, Kondo H (1998) Koukyou toushi to minkan shouhi: zaisei akaji to jousu no bunseki (Public investment and private consumption: government deficits and multiplier). Financ Rev 47:106–133 Ihori T, Kondo H (2001) Efficiency of disaggregate public capital provision in Japan. Public Financ Manag 1:161–182 Ihori T, Doi T, Kondo H (2001) Japanese fiscal reform: fiscal reconstruction and fiscal policy. Jpn World Econ 13:351–370 Ihori T, Nakazato T, Kawade M (2002) Japan’s fiscal policies in the 1990s’. World Econ 26:325–338 Ihorr T, Kato RR, Kawade M, Bessho S (2011) Health insurance reform and economic growth: simulation analysis in Japan. Jpn World Econ 23:227–239 Iwamoto Y (1990) Nihon no koukyou toushi seisaku no hyouka ni tsuite (An evaluation of public investment policy in postwar Japan). Keizai Kenkyu 41:250–261 Kameda K (2008) Wagakuni no minkan shouhi ni taisuru hi-Keynes kouka no jisshou bunseki (An empirical analysis of non-Keynesian effects on private demand in Japan). Working paper, Kwansei Gakuin University Kato H (2010) Seifu shishutsu ga minkan juyou ni oyobosu kouka no kenshou (An examination of the impact of public spending on private demand). Meiji Univ Seikei Ronsou 78:167–206 1 Fiscal Consolidation in the Political Economy of Japan 33

Kondo H, Ihori T (1999) Saiteki shakai shihon, koukyou toushi no kibo to minkan shouhi no doukou (The dynamic effects of public investment on private consumption and the welfare implication). Nihon Keizai Kenkyu 39:55–75 Mitsui K, Ota K (eds) (1995) Productivity of social capital and government financial intermedi- ation. Nihon Hyoron Sha, Tokyo Nakazato T (2002) Zaisei saiken no hi-Keynes kouka wo meguru ronten seiri (Issues on non-Keynesian effects in Japan). Keizai Bunseki 163:71–90 Nerlich C, Reuter WH (2012) The design of national fiscal frameworks and their budgetary impact. Paper presented at EPCS Meetings, Zurich, 2013 Sakuragawa M, Hosono K (2010) Fiscal sustainability of Japan: a dynamic stochastic general equilibrium approach. Jpn Econ Rev 61:517–537 Sakuragawa M, Hosono K (2012) Fiscal sustainability in Japan. TCER Working paper E-40 Velasco A (2000) Debts and deficits with fragmented fiscal policymaking. J Public Econ 76:105–125 Watanabe T, Yabu T, Ito A (2010) Seido jouhou wo mochiita zaisei jousu no keisoku (An estimation of fiscal multipliers using institutional information). In: Ihori T (ed) Zaisei seisaku to shakai hoshou (Fiscal policy and social security). Keio University Press, Tokyo, pp 143–177 Chapter 2 The Political Economy of Social Security Funding: Why Social VAT Reform?

Hideki Konishi

Abstract Recently, taxation reforms entailing a “social” valued-added tax (VAT), i.e., a social security reform shifting funding from traditional wage-based taxation to consumption taxation, have been obtaining political support in some developed countries, e.g., Japan, France, Denmark, and Germany. This paper analyzes the political economy of social security funding in an overlapping-generations economy. In particular, we consider how population aging inﬂuences the choice of wage or consumption tax ﬁnancing by focusing on their differential impact on inter- and intragenerational redistribution. Our results show that population aging may drastically alter the political equilibrium: if the population growth rate is higher than the interest rate, wage taxation is the only equilibrium choice, but if it is lower, multiple equilibria are likely to emerge, in which the introduction of consumption taxation emerges as an alternative equilibrium choice.

Keywords Consumption tax • Political economy of social security • Structure- induced equilibrium

1 Introduction

This paper examines the political economy of a change in social security funding. Recently, countries such as Japan, France, Germany, and Denmark, among others, have considered the idea of a “social” value-added tax (VAT) as a way to shift ﬁnancial sources from traditional wage-based social security contributions to consumption taxes. Denmark has already implemented a social VAT reform, while Germany has raised its VAT by 3 % as a way to lower the labor costs of domestic ﬁrms. In France, a social VAT reform, although not yet implemented, has remained

The author would like to thank Toshihiro Ihori and Naomi Miyazato for their helpful comments and suggestions. Financial support from the Zengin Foundation for Studies on Economics and Finance and Waseda University Grant for Special Research Projects (#2013A-001) is gratefully acknowledged. JEL classiﬁcation numbers: D78, H55 H. Konishi (*) School of Political Science and Economics, 1-6-5 Nishi-Waseda, Shinjuku-ku 169-8050, Japan e-mail: [email protected]

© Springer Japan 2015 35 T. Ihori, K. Terai (eds.), The Political Economy of Fiscal Consolidation in Japan, Advances in Japanese Business and Economics 8, DOI 10.1007/978-4-431-55127-0_2 36 H. Konishi controversial at least since the last presidential election. Lastly, Japan increased its consumption tax rate from 5 to 8 % as of April 1, 2014, with a pledge to place all additional revenues in the public pension budget. The increasing popularity of consumption tax financing is the result of several social and economic motives. One is the international competitiveness of domestic industries. For example, one argument has been that the international competitiveness of domestic industries has weakened or will become weaker through increasing labor costs associated with the rise in social security contributions. Concerns for intergenerational fairness appears to have more, or at least equal, importance.1 This motive emphasizes that retirees, especially richer retirees, should share the burden of population aging with the working generation. In fact, this appears to have played a role already in shaping tax policies on social security transfer in developed countries. As Adema et al. (2011) argue, most OECD countries employ the indirect taxation of consumption out of benefit income. This enables governments to receive back a considerable amount of public social expenditure as well as the direct taxation of benefit income.2 Table 2.1 illustrates the extent of diversification in the composition of funding for social protection expenditure across several European countries and Japan.3 Looking at the share of social contributions in total receipts, we can see that these differ significantly, ranging from only 23.5 % in Denmark to 66.5 % in the Netherlands. It is certain that most social contributions are from wage-based taxes such as payroll taxes. As the reverse, the share of government contributions is also very variable, ranging from just 24.3 % in the Netherlands to 74.2 % in Denmark. Naturally, government contributions consist of various revenue sources, including consumption, individual income, corporate income, capital income taxes, bond issues, and so on. It is important to note that these components include taxes that are theoretically expected to have excise tax effects; that is, part of their burden is shifted to commodity prices and effectively paid by consumers.4 To the extent this excise tax effect prevails, the share of government contributions has to be included in any rough estimate of consumption tax financing. Table 2.1 also shows how differently social contributions are allocated between employers and protected persons across these countries. In Sweden, for example, employers pay more than three times the amount that protected persons do, whereas

1 In Japan, the concern for intragenerational fairness is also emphasized in the call for consumption tax funding. The reason is that a substantial share of the working population, especially those in their 20s or 30s, are not paying national pension contributions, despite their being mandated. 2 According to their estimates of net total social expenditure including taxation and private social spending, international differences in the ratio of social expenditure to GDP are less than what we usually observe for gross social expenditure. For example, in terms of gross social expenditure, France, Sweden, and Denmark are the three largest social spenders in the OECD, but in terms of net total social expenditure, Sweden ranks fourth and Denmark ninth. 3 The social protection programs underlying the ﬁgures in Table 2.1 include not only old-age pension programs but also health care, unemployment, housing, and social assistance programs. 4 See Atkinson and Stiglitz (1980, Ch. 6) for the excise tax effect of corporate income taxes. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 37

Table 2.1 The composition of social protection receipts in 2011, % Social contributions Employer Protected persons’ Government Other Nation Total contributions contributions contributions receipts* Netherlands 66.5 32.4 34.1 24.3 9.1 Austria 64.3 37.6 26.7 34.0 1.7 France 63.3 43.0 20.3 34.7 2.0 Germany 63.1 33.5 29.6 35.2 1.7 Hungary 56.0 35.9 20.1 40.8 3.2 Japan** 52.0 26.9 25.1 37.6 10.5 Italy 53.1 38.2 14.8 45.3 1.6 Finland 47.4 35.4 12.0 46.1 6.6 Sweden 45.2 35.6 9.6 52.6 2.2 United Kingdom 44.0 31.2 12.8 47.9 8.1 Denmark 23.5 11.8 11.7 74.2 2.4 Data source: Eurostat for European countries and Financial Statistics of Social Security (published by the National Institute of Population and Social Security Research) for Japan Note: (*) Other receipts include transfers from reserves. (**) The data for Japan are based on the International Labor Organization (ILO) criterion, which differs somewhat from that used by Eurostat

employers and protected persons share payments almost equally in countries such as the Netherlands, Germany, and Denmark. Japan also belongs to this latter category of countries. If we accept the conventional, but theoretically somewhat questionable, argument that social contributions paid by employers but not those paid by employees are likely to be shifted to commodity prices, then we may also include the share of employer contributions when estimating the size of consumption tax financing.5 The question is, why does the composition of social protection receipts differ so much across countries? In addition, why has consumption tax financing become so popular of late in some countries? Lastly, why are the social contributions of employees and employers shared so differently across countries? This paper attempts to respond to these questions from a political economy perspective. We place an analytical focus on inter- and intragenerational conflict over the choice of social security funding. For this purpose, we construct an overlapping-generations model of workers and retirees with different levels of income and wealth. Using this model, we analyze the outcome of majority voting taking place over the choice of wage and consumption taxation to finance social security benefits.

5 Theoretically, when markets are perfectly competitive, the share-out ratios should make no difference in equilibrium prices and resource allocation. This argument, however, seems to be hardly acceptable in practice. 38 H. Konishi

The choice of wage and consumption tax financing causes different redistributive effects within the same generation as well as across generations. To highlight these effects, suppose that all tax revenues go into a social security program providing a flat benefit for each retiree. Consumption tax financing then redistributes income not only among workers but also among retirees. The beneficiaries are poor workers and poor retirees, who receive more than they pay, whereas rich retirees and rich workers pay more than they receive. In contrast, wage tax financing redistributes income only across workers. This is because all retirees receive benefits without paying wage taxes. The beneficiaries from wage tax financing thus consist of poor workers and all retirees. Therefore, as far as the intragenerational redistribution effects are concerned, a majority coalition including rich retirees will support a proposal for replacing consumption tax financing with wage tax financing. As a result, wage tax financing appears to be a unique outcome in the political choice of social security funding. However, the differential effects of the two taxes on the cost of intergenerational transfers also play a critical role in this political choice. Wage tax financing entails intergenerational transfers in transferring tax revenues from workers to retirees. The amount of intergenerational transfers also increases with a higher wage tax. Con- sumption tax financing, on the other hand, transfers a smaller amount across generations than wage tax financing, even when they yield the same revenues. This is because both workers and retirees pay consumption taxes, but only those taxes paid by workers are transferred to retirees.6 Moreover, a higher consumption tax decreases the wage tax revenues transferred to retirees to the extent that it discourages labor supply. Importantly, in a dynamically efficient economy, where the population growth rate is lower than the interest rate, intergenerational transfers are more expensive than private savings for workers to finance their postretirement consumption. As demographic aging advances, a larger population of workers may prefer replacing wage tax financing with consumption tax financing because the latter reduces the amount of costly intergenerational transfers. In this paper, we consider a voting model that determines a wage tax, a consumption tax, and the size of the social security benefit per retiree financed by these taxes. Generally, no Condorcet winner arises in majority voting over multiple issues. Following Conde-Ruiz and Galasso (2005), we employ the concept of a structure- induced equilibrium invented by Shepsle (1979) to aggregate policy preferences over wage and consumption taxes and to describe the political equilibrium. Specifically, we suppose that a wage tax rate and a consumption tax rate are put separately to the vote and each Condorcet winner is selected with the other tax rate taken as given. We find the decisive worker–voter is poorer in the voting over a wage tax than in the voting over a consumption tax. This is because all retirees always support a proposal for increasing the size of benefits under wage tax financing. As poorer workers gain more from redistribution in the social security program, the majority

6 This further implies that under consumption tax financing, workers have to save more for their postretirement consumption to pay for consumption taxes. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 39 coalition allows the social security benefit per retiree to grow under wage tax financing. Given that a wage tax finances such a large benefit, a majority coalition is less likely to emerge to support a proposal of introducing a consumption tax to increase the benefit further. We show that in a dynamically efficient economy with a small margin prevailing between population growth and the interest rate, as well as in a dynamically inefficient economy, wage tax financing is a unique structure- induced equilibrium outcome. The result is that as conventionally observed in many countries, only wage taxes finance social security benefits. As population aging proceeds, however, the policy preferences for social security funding change. Regarding wage tax financing, even poor workers prefer a lower wage tax because wage taxation yields smaller marginal revenues due to a smaller labor force. On the other hand, because of the reduced population of worker–voters with an aging population, the political influence of retiree–voters strengthens to the extent that a proposal for a higher wage tax tends to gain more support. Although these two changes tend to counteract each other, we show that if the density of the median worker–voter is sufficiently large, which seems quite natural in a standard wage income distribution, the effect of the political influence of retirees is outweighed; that is, population aging shifts aggregate policy preferences toward a lower wage tax. In terms of financing through a consumption tax, we find that poor and middle- class workers are inclined to prefer a higher consumption tax as population aging proceeds. This is because for them, funding social security with intergenerational transfers becomes more expensive. Poor and middle-class retirees also prefer a higher consumption tax to compensate them for wage tax revenues lost because of the smaller labor force. Thus, aging shifts the aggregate policy preferences toward a higher consumption tax. Combining these effects, we argue that population aging may drastically change the political equilibrium. We show that in a dynamically efficient economy with a large margin between the population growth rate and the interest rate, multiple structure-induced equilibria potentially arise to finance social security: only the wage taxation, only the consumption taxation, or their combination may fund social security. Consumption tax financing thus emerges as an equilibrium outcome in response to population aging. This result explains the recent increase in the political popularity of “social” VAT reform in some countries. The coexistence of three types of equilibrium also explains the observed international diversity in the composition of social security funding, as shown earlier in Table 2.1. We also show that with a sufficiently large margin between the population growth rate and the interest rate, there is no majority coalition that supports wage tax financing. As a result, in this situation, only the consumption taxation funds social security. To the best of our knowledge, no existing study addresses the public choice of social security funding with a combination of different taxes. In the literature on the political economy of income redistribution and social security, including Brow- ning (1975), Meltzer and Richard (1981), Hu (1982), Boadway and Wildasin (1989), Cooley and Soares (1999), Tabellini (2000), and Razin et al. (2002) 40 H. Konishi among others, almost all employ models with only wage taxation to finance government spending for redistribution. As an exception, the model in Conde- Ruiz and Galasso (2005), on which we base our model, employs two different wage taxes—namely, a social security tax for transfers to retirees and an income redistribution tax for transfers to poor workers—for explaining the coexistence of the two social transfer programs. The remainder of the paper is organized as follows. Section 2 sets out the model. Section 3 examines majority voting over wage tax rates, taking the consumption tax rate as given. Here we show that under a plausible condition, population aging lowers the Condorcet-winner wage tax rate despite retirees becoming more politically influential. Section 4 considers majority voting over consumption tax rates, taking the wage tax rate as given. Here we show that population aging increases the Condorcet-winner consumption tax rate. Using the results of the two previous sections, Sect. 5 analyzes the structure-induced equilibria in which wage and consumption tax rates are Condorcet winners given the other rate, and examines the impact of population aging on the equilibrium outcomes. Section 6 concludes the paper.

2 The Model

2.1 Brief Description of the Economy

We consider an economy with two overlapping generations, workers and retirees, in which the population grows at the rate of n > 0. To highlight the redistribution effects of a social security program, we assume that every individual works only when young and consumes only after retirement. Worker i has Ei units of leisure as her initial endowment, out of which she supplies Ni units of labor. She consumes Ci units of numeraire goods after retirement, financing it out of her savings, interest income, and social security benefits. Every retiree receives a flat social security benefit, the amount of which is denoted by B in real terms. The funding of social security benefits is in a pay-as-you-go fashion by either wage tax revenues, consumption tax revenues, or both. The wage tax rate is denoted as τw, and the net consumption tax rate is bτ. We convert the latter into a gross consumption tax rate τc :¼ bτ∕ð1 þ bτÞ. There are no intrafamily transfers, such as bequests. The interest rate and the wage rate are assumed to be constant, the former denoted as r and the latter normalized to unity.

2.2 Workers and Retirees

Worker i faces a lifetime budget constraint,

ð1 þ bτÞCi ¼ð1 þ rÞð1 À τwÞNi þð1 þ bτÞB, 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 41 where 0 Ni Ei and the term ð1 þ bτÞB represents nominal social security beneﬁts. Using a net consumption tax rate τc :¼ bτ∕ð1 þ bτÞ, we can rewrite the budget constraint as

Ci ¼ð1 þ rÞð1 À τcÞð1 À τwÞNi þ B:

The result is that consumption taxation affects a worker’s budget constraint in the same way as wage taxation despite their differences in the timing of payment. The worker’s leisure endowment, Ei, which reflects her innate ability, depends on her luck, which follows a cumulative distribution function, F(Ei), withR support ½E, EÞ. The mean E and the median E are respectively defined by E ¼ and m EExf ðxÞdx FðEmÞ¼1∕2, where f(x) denotes the probability density function. The ability distribution is left skewed, as in Fig. 2.1, yielding Em < E. Worker i’s utility function is specified as

C u ¼ i þ lnðE À N Þ, i 1 þ δ i i where δ is the rate of time preference. We set δ ¼ r for simplicity. We also assume that E is sufﬁciently large to ensure Ni > 0 for every individual in equilibria. We then respectively obtain her labor supply and indirect utility as follows:

1 Ni ¼ Ei À ð1 À τwÞð1 À τcÞ and

B U ¼ð1 À τ Þð1 À τ ÞE À lnð1 À τ Þð1 À τ Þþ : ð2:1Þ i c w i w c 1 þ r

Her savings are written as Ai ¼ð1 À τwÞNi. The distribution of before-tax wage income, which is equal to the labor supply, is as skewed as the ability distribution.

f (Ei)

_ Ei Fig. 2.1 Distribution 0 E_ E E E of ability m 42 H. Konishi

Suppose that worker i has retired after saving Ai. Her utility in the retirement period comes only from her consumption. We write this as

Vi ¼ð1 À τcÞð1 þ rÞAi þ B: ð2:2Þ

2.3 Social Security System

Wage taxes and consumption taxes ﬁnance social security expenditures. No revenues remain for the future. From the budget constraint, we express the social security beneﬁt per retiree in real terms as

B ¼ τcð1 þ rÞA þ τwð1 þ nÞN, ð2:3Þ where

1 N ¼ Nðτw, τcÞ :¼ E À ð2:4Þ ð1 À τwÞð1 À τcÞ and

A ¼ð1 À τwÞNðτw, τcÞð2:5Þ are labor supply per worker and savings per retiree, respectively. The ﬁrst term in (2.3) is consumption tax revenue (except for taxes paid by consumption out of social security beneﬁts), and the second term is wage tax revenue. If the tax rates are constant over time, we can write (2.3)as

B ¼ Bðτw, τcÞ :¼½ð1 þ rÞτcð1 À τwÞþð1 þ nÞτwNðτw, τcÞ: ð2:6Þ

3 Wage Tax Financing

Society has to determine a wage tax rate, a consumption tax rate, and the size of a social security beneﬁt per retiree funded by these taxes. Considering the government’s budget, the public choice reduces to determining the two tax rates. We assume that society puts each tax rate separately to majority voting, where every voter votes taking the other tax rate as given. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 43

3.1 Policy Preferences of Retirees

Let us ﬁrst consider retirees’ preferences over wage tax rates, taking the consumption tax rate as given. Plugging (2.3)into(2.2), we obtain retiree i’s utility function as

Viðτw, τcÞ :¼ð1 þ rÞ½ð1 À τcÞAi þ τcAþð1 þ nÞτwNðτw, τcÞ: ð2:7Þ

Given that Ai and A are predetermined, every retiree’s most preferred wage tax rate τo τ ðτ τ Þ is w, at which the per-worker wage tax revenue, wN w, c , is maximized. Retirees share the same preferences because wage tax financing causes no redistribution among them. More specifically, from the first-order condition, ∂Vi ∂N ¼ð1 þ nÞ Nðτw, τcÞþτw ¼ 0, ð2:8Þ ∂τw ∂τw their most preferred wage tax rate is obtained as

1 τo ¼ τo ðτ Þ :¼ À pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : w w c 1 ð2:9Þ ð1 À τcÞE

This tax rate decreases as the consumption tax rate increases. We also ﬁnd ’ ∂2 ∕∂τ2 ¼ that retirees preferences are single peaked because Vi w 3 À2∕½ð1 À τcÞð1 À τwÞ < 0:

3.2 Policy Preferences of Workers

Next, consider workers’ policy preferences. In contrast to the situation of retirees, wage tax ﬁnancing redistributes income from rich to poor workers. Let us examine how a higher wage tax affects worker i’s utility. Following the literature, we assume policy commitment in that workers expect that the tax rate determined in voting today will not change in their retirement period.7 From (2.1), worker i’s utility is then written as

Bðτ , τ Þ U ðτ , τ Þ :¼ð1 À τ Þð1 À τ ÞE À lnð1 À τ Þð1 À τ Þþ w c : ð2:10Þ i w c c w i w c 1 þ r

7 In Appendix 3 we relax this assumption following Conde-Ruiz and Galasso (2005) and show that the same equilibrium outcomes are realized in the subgame-perfect equilibria of an inﬁnitely repeated voting game. 44 H. Konishi

Differentiating (2.6) and (2.10), we have

∂ ∂ Ui 1 B ð : Þ ¼Àð1 À τcÞNi þ 2 11 ∂τw 1 þ r ∂τw and ∂B ∂N ∂N ¼ð1 þ nÞ N þ τw þð1 þ rÞτc À N þð1 À τwÞ : ð2:12Þ ∂τw ∂τw ∂τw

From (2.12), we observe that a higher wage tax produces two effects on the amount of social security beneﬁt that workers expect to receive after retirement. First, as shown by the ﬁrst term in (2.12), a higher wage tax rate increases wage tax revenues in the next period when workers today retire. Second, as shown by the second term, this also reduces consumption tax revenues in the next period by reducing savings today. Combining (2.11) and (2.12), we obtain

∂Ui ∂N r À n ∂N ¼ð1 À τcÞðE À EiÞþ½1 Àð1 À τwÞð1 À τcÞ À N þ τw : ð2:13Þ ∂τw ∂τw 1 þ r ∂τw

The first term represents the intragenerational redistribution effect. This favors workers with less ability than the mean (and hence, workers poorer than the average). The second term, which is negative for all workers, represents the loss due to aggravated distortions in labor supply.8 The third term shows the loss arising from the intergenerational redistribution. In fact, this is negative in as far as a higher wage tax rate increases wage tax revenues. As is well known, this is because intergenerational income transfers are dynamically inefficient in an economy with r > n. τy ’ Let w be worker i s most-preferred wage tax rate. Developing and rearranging (2.13) yields the first-order condition, ∂Ui 1 þ n 1 1 þ n ¼Àð1 À τcÞEi þ À τc E þ 1 À 0: ∂τw 1 þ r 1 À τw ð1 þ rÞð1 À τcÞð1 À τwÞ ð2:14Þ

8 Because 1 Àð1 À τwÞð1 À τcÞ is the difference between the before- and after-tax wage rates, multiplying it by the change in the labor supply yields the change in the deadweight loss in the labor supply. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 45

τy ¼ With the strict inequality, w 0. This condition allows us to write this as a function of Ei, τc, and n:

τy ¼ τy ð τ Þ: ð : Þ w w Ei, c, n 2 15

τy > ∂2 ∕∂τ2 < Whenever w 0, the second-order condition, Ui w 0, must be met. The condition reduces to

2ð1 þ nÞ 1 À < 0, ð2:16Þ ð1 þ rÞð1 À τwÞð1 À τcÞ which is guaranteed if ð1 þ nÞ∕ð1 þ rÞ > 1∕2. We assume this throughout the analysis to ensure that workers’ preferences, as well as those of retirees, are single peaked. Workers’ most-preferred wage tax rates have the following properties. First, as τy ¼ shown by (2.13), w 0 for Ei E whenever r n. Rich workers are then better off without a social security program if it makes for a dynamically inefﬁcient transfer. τy Second, w is decreasing in Ei. That is, richer workers have lower most-preferred 2 wage tax rates. This follows from ∂ Ui∕∂Ei∂τw ¼Àð1 À τcÞ < 0:

3.3 Voting Equilibrium

Suppose that majority voting takes place to determine a wage tax rate with the consumption tax rate taken as given. We can invoke the median voter theorem to ﬁnd a Condorcet-winner wage tax rate given that all voters’ preferences are single peaked. Comparing the policy preferences between retirees and workers, we have τy ð τ Þ< τo ðτ Þ τo w Ei, c, n w c because w is the tax rate that maximizes wage tax revenues. Combined with the assumption of n > 0, it results that the decisive voter is among the workers. If we denote her ability by Ew, this is determined by 1 þð1 þ nÞFðEwÞ¼ð2 þ nÞ∕2, or n FðEwÞ¼ : ð2:17Þ 2ð1 þ nÞ

From this condition, it is clear that Ew < Em. Taking account of the fact that Ew τ∗ increases with n, we write the Condorcet-winner wage tax rate, w , as a function of τc and n:

τ∗ ¼ τy ð τ Þ :¼ τ∗ðτ Þ: ð : Þ w w Ew, c, n w c, n 2 18 46 H. Konishi

Fig. 2.2 Condorcet-winner Poor Workers wage tax rate Rich Workers Retirees

τw * o 0 w τw

Figure 2.2 depicts the distribution of the most-preferred wage tax rates to explain diagrammatically the nature of the Condorcet winner. The rectangular areas represent τ ¼ τ ¼ τo the population share of workers and retirees who most prefer w 0 and w w, τo respectively. The height of the graph at each tax rate between zero and w corresponds to the density of workers who prefer it most. The hump-shaped schedule is drawn by reversing the right and left of the ability distribution and extending it upward by a factor of 1 + n. The Condorcet-winner wage tax rate is determined at the level that equalizes the areas to the left and to the right under the graph.

3.4 Comparative Statics

Let us consider the effects of population aging and a higher consumption tax on the τ∗ > Condorcet-winner wage tax rate, assuming that w 0. Consider ﬁrst the effect of population aging, which is identiﬁed as a decrease in n in our model. Population aging affects the Condorcet-winner wage tax rate in opposite directions:

∂τ∗ ∂τy ∂τy ∂ w ¼ w þ w Ew : ∂n ∂n ∂Ei ∂n

τ∗ First, it reduces w by lowering the most-preferred tax rate of the decisive worker– τy > 9 voter. This effect is captured by the ﬁrst term, which is positive if w 0. The intuitive reason is that aging increases the costs of intergenerational transfer. τ∗ Second, aging increases w by strengthening the political inﬂuence of retirees. τy > 10 This effect is captured by the second term, which is negative if w 0. Because the population of retirees increases relative to workers, the poorer worker–voter, who prefers a higher wage tax rate, becomes decisive in the voting.

9 2 Differentiating (2.11)withrespectton and making use of (2.14)yields∂ Ui∕∂n∂τw ¼½ð1 À τcÞ þ τ ∕ð þ Þ > τ ¼ τy Ni cE 1 n 0when w w. 10 ∂τy ∕∂ < ∂ ∕∂ ¼ ∕½ ð þ Þ2 ð Þ > This follows given w Ei 0 and Ew n 1 2 1 n f Ew 0. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 47

Because these two effects counteract each other, the way that population aging affects the Condorcet-winner wage tax rate is generally ambiguous. Nonetheless, if the density of the decisive worker–voter, f ðEwÞ, is sufficiently large, which is quite natural in the most commonly observed wage income distribution, such as in Fig. 2.1, the first effect is likely to outweigh the second. As a result, population aging decreases the Condorcet-winner wage tax rate.11 Consider next the effect of a higher consumption tax rate. The effect has three channels, and in total, it turns out to decrease unambiguously the Condorcet-winner wage tax rate. First, a higher consumption tax induces poor workers to prefer a lower wage tax because the extent of income redistribution across workers through wage tax financing effectively becomes smaller. In other words, a higher consumption tax leads wage tax financing to be less beneficial to poor workers because they have to pay a larger part of their benefits back to the government through consumption taxation. Second, a higher consumption tax induces every worker to prefer a lower wage tax because it aggravates labor market distortions. Third, and conversely, a higher consumption tax induces every worker to prefer a higher wage tax because it decreases the size of costly intergenerational transfers owing to the smaller wage tax revenue collected. As a matter of fact, the second effect is shown to outweigh the third, and a higher consumption tax reduces the Condorcet-winner wage tax rate;

∂τ∗ ∂τy w ¼ w < 0 ∂τc ∂τc

τ∗ > 12 whenever w 0. The following proposition summarizes the observations obtained in this section. τ∗ > Proposition 1. Suppose that w 0. (i) The way that population aging affects the Condorcet-winner wage tax rate is then generally ambiguous. Provided that the density of the decisive worker–voter is sufﬁciently large, population aging then

11 This observation is similar to that by Razin et al. (2002). They used an overlapping generations model with human capital formation and showed that population aging may lead to a downsizing in the size of the welfare state. They also tested this hypothesis with data for the US and 12 European countries over the period 1965–1992 to obtain a positive empirical result. See also Disney (2007), Simonovits (2007), and Galasso and Profeta (2007), among others, for the controversies that their paper initiated. 12 If we look at (2.13), the three channels correspond to the changes in the three terms on the right- hand side. Differentiating (2.11) with respect to τc, we have

∂2U 1 þ n i ¼ÀðE À E ÞÀ : ∂τ ∂τ i 2 2 c w ð1 þ rÞð1 À τwÞ ð1 À τcÞ

The first term reflects the effect through the first channel, and the second term reflects the net effect through the second and third channels. Given that the decisive worker–voter has Ei < E, both terms are negative. 48 H. Konishi lowers the Condorcet-winner wage tax rate. (ii) A higher consumption tax rate unambiguously decreases the Condorcet-winner wage tax rate.

4 Financing by Consumption Tax

4.1 Policy Preferences of Retirees

Consider ﬁrst retirees’ preferences over consumption tax rates. Given a wage tax rate, (2.3) and (2.7), we show that retiree j’s most-preferred consumption tax rate, τo c, satisﬁes ∂ ∂ Vj B ð : Þ ¼Àð1 þ rÞAj þ 0, 2 19 ∂τc ∂τc

τo ¼ and c 0 with the strict inequality. Retirees, whose asset holdings are predetermined, expect the effect of a higher consumption tax on the size of a beneﬁt per retiree to be

∂ ∂ B N ð : Þ ¼ð1 þ rÞA þð1 þ nÞτw : 2 20 ∂τc ∂τc

The ﬁrst term shows the increase in consumption tax revenues per retiree, and the second term shows the reduction in wage tax revenues per retiree. Substitut- ing (2.20) into (2.19) yields

1 ∂V ð1 þ nÞτ j ¼ A À A À w 0: ð : Þ þ ∂τ j 2 2 21 1 r c ð1 þ rÞð1 À τcÞ ð1 À τwÞ

τo ¼ ∂2 ∕∂ Thus, c 0 for every retiree wealthier than the average. Furthermore, as Vj τ2 < ’ c 0 is guaranteed, each retiree s policy preferences are single peaked. Suppose that the same wage tax rate was applied to retiree j’s earnings when young. If retiree j was a worker with ability Ej, thenA À Aj ¼ð1 À τwÞðE À EjÞ, and τo hence from (2.21), her most-preferred consumption tax rate, c, is determined as a function of Ej, τw, and n:

τo ¼ τoð τ Þ: ð : Þ c c Ej, w, n 2 22

τo It is intuitively straightforward that c is decreasing in Ei; the richer retiree has the lower most-preferred consumption tax rate.13

13 ∂2 ∕∂ ∂τ ¼Àð þ Þð À τ Þ < ∂2 ∕∂τ2 < This observation follows from Vj Ej c 1 r 1 w 0 and Vj c 0. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 49

4.2 Policy Preferences of Workers

Consider next workers’ preferences over consumption tax rates, assuming policy commitment as in the case of wage tax ﬁnancing. Differentiation of (2.10) shows that a higher consumption tax rate affects worker i’s utility and her social security beneﬁt as follows:

∂ ∂ Ui 1 B ð : Þ ¼Àð1 À τwÞNi þ 2 23 ∂τc 1 þ r ∂τc and

∂ ∂ ∂ B N N ð : Þ ¼ð1 þ rÞA þð1 þ rÞð1 À τwÞτc þð1 þ nÞτw : 2 24 ∂τc ∂τc ∂τc

Comparing (2.20) with (2.24), the only difference is the second term in (2.24). This reﬂects workers’ expectations about how consumption tax revenues in their retirement period will respond to the change in their current labor supply. Plugging (2.24) into (2.23) yields

∂ ∂ ð À Þτ ∂ Ui N r n w N ð : Þ ¼ð1 À τwÞðE À EiÞþ½1 Àð1 À τwÞð1 À τcÞ À : 2 25 ∂τc ∂τc 1 þ r ∂τc

Similar to (2.14), the first term represents the intragenerational redistribution, the second represents the distortionary effect on the labor supply, and the third represents the effect associated with the intergenerational transfer. The second term is always negative. The third term is positive when r > n, because a higher consumption tax decreases intergenerational transfers by reducing wage tax revenues. However, the second term outweighs the third, and thus their net effect is negative, as shown below. Arranging the terms in (2.25), we find that worker i’s most-preferred consump- τy tion tax rate, c, satisfies

∂U ð1 þ nÞτ τ i ¼ð1 À τ ÞðE À E ÞÀ w À c 0, ð : Þ ∂τ w i 2 2 2 26 c ð1 þ rÞð1 À τcÞ ð1 À τwÞ ð1 À τcÞ

τy ¼ ∂2 ∕∂τ2 < and c 0with the strict inequality. Note that Ui c 0is guaranteed, and thus ’ τy workers preferences are single peaked. Equation (2.26) allows us to write c as a function of Ei, τw, and n:

τy ¼ τyð τ Þ: ð : Þ c c Ei, w, n 2 27 50 H. Konishi

τy ¼ From (2.26), c 0 for workers richer than the average. Moreover, it is intuitively τy straightforward that c is decreasing in Ei; the richer worker has the lower most- preferred consumption tax rate.14

4.3 Voting Equilibrium

Suppose that majority voting takes place to determine a consumption tax rate, given a wage tax rate. To ﬁnd a Condorcet winner, we ﬁrst consider the pairing of a worker and a retiree whose most-preferred consumption tax rates coincide. Comparing (2.21) and (2.26), we observe that the most-preferred consumption tax rate is the same for worker i and retiree j if and only if their abilities satisfy

τy ¼ oð τ Þ :¼ þ c : ð : Þ Ej E Ei, w, n Ei y 2 2 28 ð1 À τwÞð1 À τcÞ

o Combining a worker of ability Ei and a retiree of ability E ðEi, τw, nÞ, we can aggregate the policy preferences. Importantly, the abilities matched satisfy

o E ðEi, τw, nÞEi, ð2:29Þ

τyð τ Þ¼ with the equality holding if and only if c Ei, w, n 0. Recall here that the most- preferred consumption tax rate is decreasing in ability for workers as well as for retirees. The above inequality then implies that retirees prefer a higher consumption tax rate than workers if they have the same ability. In other words, for any given consumption tax rate, the proportion of retirees who prefer raising it is greater than that of workers. This is because retirees do not care about how a higher consumption tax today affects consumption tax revenues in the next period. Figure 2.3 diagrammatically explains how to identify a Condorcet winner. The upper graph in Fig. 2.3 represents the distribution of retirees’ most-preferred consumption tax rates, and the lower graph represents that of workers. The respective rectangles show the proportions of retirees and workers who most prefer a zero tax rate. Though both graphs have a similar shape, the retirees’ graph is located more to the right than the workers’ graph. This reﬂects the fact that retirees prefer a higher consumption tax than workers with the same ability. In addition, the height of the retirees’ graph is contracted by 1∕ð1 þ nÞ, compared with that of the workers τ∗ owing to population growth. The Condorcet-winner consumption tax rate, c ,is determined at the level separating the two graphs such that the sum of the areas on each side is equalized.

14 2 This observation follows from ∂ Ui∕∂Ei∂τc ¼Àð1 À τwÞ < 0. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 51

Fig. 2.3 Voting Retirees equilibrium with Poor consumption taxes Rich

o τc 0

Workers Poor

Rich

y τc * 0 τc

To formalize the above diagrammatic exposition, aggregate the population of workers and retirees with the same preferences, making use of (2.28). The ability of ∗ the median worker–voter, Ec , is then determined by 2 þ n FðEoðE∗, τ , nÞÞ þ ð1 þ nÞFðE∗Þ¼ : ð2:30Þ c w c 2

The ﬁrst term on the left-hand side is the proportion of retirees voting for increasing ∗ consumption taxes and the second term is that of workers. Generally, Ec depend on τ ∗ ¼ ∗ðτ Þ n and w. Hence, we denote it by Ec Ec w, n . The Condorcet-winner consumption tax rate is written as

τ∗ ¼ τyð ∗ τ Þ :¼ τ∗ðτ Þ: ð : Þ c c Ec , w, n c w, n 2 31

∗ < τ∗ > Note that because of (2.29), Ec Em whenever c 0.

4.4 Comparative Statics

Consider first the effects of population aging on the preferences of retirees and workers, and then on the Condorcet-winner consumption tax rate. Population aging induces retirees to favor a higher consumption tax rate irrespective of how wealthy they are. This is owing to a smaller labor force. Every retiree expects to receive a larger amount of benefits from a consumption tax increase than before because it causes a smaller reduction in wage tax revenues. Population aging thus leads workers to prefer a higher consumption tax rate. Recall that aging makes intergenerational transfers more costly for workers. A higher 52 H. Konishi consumption tax is then beneficial to every worker because it decreases the amount of intergenerational transfers by reducing the labor supply and thus the wage tax revenues transferred to retirees. Moreover, aging changes the position of the decisive voter and makes a poorer worker–voter decisive. Recall from the inequality in (2.29) that retirees have more intense preferences for a consumption tax increase than workers do. As a result, population aging, or strengthening retirees’ ∗ political influence, decreases Ec . With these three effects combined, the Condorcet- winner consumption tax rate becomes unambiguously higher as population ages. Consider next the effect of a higher wage tax rate. First, retirees prefer a lower consumption tax rate when there is a higher wage tax rate. The reason is that a consumption tax increase reduces wage tax revenues more with a higher wage tax. A wage tax increase also induces workers to prefer a lower consumption tax, symmetrically as a consumption tax increase induces them to prefer a lower wage tax.15 Overall, as all voters change their preferences in the same direction, the Condorcet-winner consumption tax rate decreases in response to a higher wage tax rate. The following proposition summarizes our observations in this section. τ∗ > Proposition 2. Suppose that c 0. (i) Population aging increases the Condorcet- winner consumption tax rate. (ii) A wage tax increase lowers the Condorcet-winner consumption tax rate. Proof. See Appendix 1.

5 Political Economy of Social Security Funding

5.1 Structure-Induced Equilibria

We now analyze the public choice of the pair of the two tax rates. No Condorcet winner generally exists in voting over multiple issues without restricting either voters’ policy preferences or the structure of political decision making.16 Taking the latter course, we use the notion of structure-induced equilibrium, ﬁrst introduced by Shepsle (1979). The focus of the analysis is on how the equilibrium outcome responds to a change in the population growth rate. Suppose that a wage tax rate and a consumption tax rate are within the jurisdictions of separate committees in a legislature.17 We assume that the members of each committee reﬂect all voters’ policy preferences in society with no biases—the case

15 See footnote 12. 16 See, e.g., Persson and Tabellini (2000). 17 In Japan, the Committee on Health, Welfare, and Labor in the lower house (as well as that in the upper house) has jurisdiction over wage-based social security contributions, while the Committee on Financial Affairs has jurisdiction over consumption taxes. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 53

τc

τw O C2 W1

Fig. 2.4 A unique structure-induced equilibrium that Shepsle (1979) refers to as “the committee of the whole.” Each committee determines a tax rate within its jurisdiction by majority voting, taking the other tax ðτe τe Þ rate as given. A pair of the two tax rates, c, w , is a structure-induced equilibrium τe τe τe ¼ τ∗ðτe Þ if and only if w is a Condorcet winner given c and vice versa; that is, c c w, n τe ¼ τ∗ðτe Þ and w w c, n . Let us start with a diagrammatic analysis. Figures 2.4 and 2.5 depict the wage–tax reaction curveW1W2τc, which plots the relationship of the two taxes satisfying (2.18) with a given population growth rate. The Condorcet-winner wage tax rate decreases in response to a higher consumption tax rate and is equal to zero when the consumption tax rate exceeds the threshold rate at W2. The reaction curve coincides with the schedule of the most-preferred wage tax rates of a worker with ability Ew. Similarly, the consumption tax reaction curve C1C2τw depicts the relationship satisfying (2.31) with the same population growth rate. The Condorcet-winner consumption tax rate decreases in response to a higher wage tax rate and is equal to zero when the wage tax rate exceeds the threshold rate at C2. In contrast to the wage tax reaction curve, the consumption tax reaction curve does not coincide with the schedule of a particular worker’s most-preferred consumption tax rate, because the decisive worker–voter changes, depending on the wage tax rates, as shown in (2.28). If r > n at a small margin or if n r, the wage tax reaction curve is located above the consumption tax reaction curve, as depicted in Fig. 2.4. The structure- induced equilibrium is then uniquely determined at W1, where the social security benefits are financed only by wage taxation. The equilibrium size of social security benefits is driven so large that a coalition of poor workers and poor 54 H. Konishi

τc

τw O C2 W1

Fig. 2.5 Multiple structure-induced equilibria retirees who may support the introduction of consumption tax financing cannot constitute a majority. The economic advantage of consumption tax financing is limited because the costs of intergenerational transfer are not very large for workers.Conversely,ifonlyconsumptiontaxfinancingisemployedtofund social security benefits, the size is so unsatisfactory for poor workers and all retirees that they can form a majority coalition to support a proposal introducing wage tax financing. As the population growth rate decreases, the positions of the two schedules change. If the effect of retirees’ strengthened political influence does not change the median worker–voter’s ability very much, the wage tax reaction curve shifts to the left. The consumption tax reaction curve unambiguously moves upward. If the population growth rate becomes sufficiently small, and thus r > n holds with a sufficiently large margin, the reaction curves are then likely to cross each other, as illustrated in Fig. 2.5. There then occur three different types of structure-induced equilibrium. The social security program in equilibrium employs wage taxation only at W1, consumption taxation only at C1, and both at C3. Comparing Figs. 2.4 and 2.5, we demonstrate that consumption tax financing may emerge as a political equilibrium outcome in an aging society. This is because population aging induces the society to shift the revenue source of its social security program from wage taxation to consumption taxation, totally or partially, in equilibria like C1 and C3 in Fig. 2.5. The economic condition behind such a shift is that with slower population growth, a wage tax increase contributes less to the social security budget, and the costs of intergenerational transfer become higher at the same time. Politically, this leads to an increasing population of poor workers 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 55 and poor retirees that support a proposal of introducing consumption tax financing. It also induces an increasing population of middle-class and rich workers to oppose a wage tax increase. In spite of all retirees’ preferring a wage tax increase, the population of poor workers whose preferences align with their own is too small to form a majority coalition. Backed by such a politico-economic interaction, the society introduces consumption taxes to fund its social security program. Nonetheless, as shown in W1 in Fig. 2.5, it is also possible that a social security program continues to be financed only with wage taxation, even if population aging proceeds. Wage tax financing is one of the political equilibrium outcomes unless the population growth rate becomes sufficiently low to make the two reaction curves intersect only on the vertical axis. If a wage tax rate increases to the median worker–voter’s ideal level, no majority coalition can support the introduction of consumption tax financing, although it does help society to save the costs of intergenerational transfer.

5.2 Formal Presentation and Simulation

To present these findings formally, we require some new notation. First, taking r and τc τw n as given, let w be the wage tax rate at C2, and let c be the consumption tax rate at τc :¼ τ jτ∗ðτ Þ¼ W2 in Figs. 2.4 and 2.5. Precisely, these are defined by w min w c w, n 0 τw :¼ τ jτ∗ðτ Þ¼ :¼ ∗ð Þ and c min c w c, n 0. Second, let Ec Ec 0, n be the ability held by the median worker–voter in voting over consumption taxes in the absence of wage taxation. It is already clear that Ew < Ec < Em. Finally, define two threshold ability levels, r À n 1 E :¼ E À E À H m þ À τc 1 r 1 w and r À n 1 E :¼ E À E À : L c ð þ Þð À τcÞ À τc 1 r 1 c 1 c

Then the results of the diagrammatic analysis above are formally stated in the following proposition. τe > ¼ τe Proposition 3. (i) A structure-induced equilibrium with w 0 c exists if and τe ¼ < τe only if Ew EH. (ii) A structure-induced equilibrium with w 0 c exists if and τe > τe > only if Ew EL. (iii) A structure-induced equilibrium with w 0 and c 0 exists simultaneously with the equilibria in (i) and (ii) if and only if EL < Ew < EH. 56 H. Konishi

Table 2.2 Simulation results Case 1: r ¼ 1:0325 À 1, n ¼ 1:0225 À 1

ðEL, Ew, EHÞ¼ð2:34, 2:42, 2:45Þ ðτe τeÞ¼ð : Þ ð : : Þ ð : Þ w, c 0 137, 0 , 0 077, 0 038 , 0, 0 126 Case 2: r ¼ 1:0325 À 1, n ¼ 1:02525 À 1

ðEL, Ew, EHÞ¼ð2:56, 2:47, 2:64Þ ðτe τeÞ¼ð : Þ w, c 0 204, 0 Case 3: r ¼ 1:0325 À 1, n ¼ 1:01525 À 1

ðEL, Ew, EHÞ¼ð2:14, 2:36, 2:29Þ ðτe τeÞ¼ð : Þ w, c 0, 0 130

Proof. See Appendix 2. As illustrated in Fig. 2.4, a social security program financed entirely by wage taxation is a unique political equilibrium outcome in a society with n r, where EH Em > Ew and EL Ec > Ew hold. Conversely, population aging must pro- ceed to establish n < r for consumption tax financing to emerge as a political equilibrium outcome. The situation illustrated in Fig. 2.5 occurs if and only if 18 EL < Ew < EH. A social security program whose funding is only by consumption taxation is a unique political outcome if r is sufficiently high relative to n to establish EL < EH < Ew. Finally, we simulate the three patterns of equilibria to verify the above findings. Suppose that one period corresponds to 25 years and write worker i’s ability as Ei ¼ E þ zi. We let E ¼ 1:2 and zi follow a log-normal distribution with mean 0.5 19 and standard deviation 0.35. This distribution produces E ¼ 2. 95 and Em ¼ 2. 85. The annual interest rate is held fixed at 3 %, and the annual population growth rate is chosen out of three options, 1.5, 2, and 2.5 %. Table 2.2 provides the simulation results. In Case 1, where the annual population growth rate is 2 %, we obtain EL < Ew < EH, and the three types of equilibrium arise. In the respective equilibria, funding is through a 13.7 % wage tax, a 12.6 % consumption tax, and a combination of a 7.7 % wage tax and a 3.8 % consumption tax. In Case 2, where the annual population growth rate is 2.5 %, we obtain Ew < EL < EH, and the funding in the unique equilibrium is via a 20.4 % wage tax. In Case 3, where the annual population growth rate is 1.5 %, we obtain EL < EH < Ew, and the funding in the unique equilibrium is via a 13.0 % consumption tax.

18 τe > τe > As shown in Appendix 2, the equilibrium with w 0 and c 0 is unique if it exists. 19 This ability distribution yields a before-tax wage distribution with quartile dispersion coefﬁcient (3rd quartile – 1st quartile)/(2Âmedian) ¼ 0.23 and decile dispersion coefﬁcient (9th decile – 1st decile)/(2Â median) ¼ 0.45. These are close to the corresponding values in the Japanese wage distribution for males aged 40 to 44 years. According to the Basic Survey on Wage Structure 2013, they are 0.23 and 0.48, respectively. 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 57

Concluding Remarks This chapter analyzed the political economy of social security funding by wage and consumption taxes using a model of majority voting in an overlapping-generations framework. We placed the analytical focus on the difference in the distributional impacts between the two taxes as well as the costs of intergenerational transfer. We showed that as population aging proceeds, consumption tax financing tends to be included as part of an equilibrium outcome, while wage tax financing also continues to be used. Such equilibrium multiplicity explains why the revenue sources for social security vary across countries as well as why “social” VAT reform has recently become so popular in some countries. Our theoretical insights provide some interesting testable hypotheses. First, a social security program is more likely to be funded by taxes whose burdens are expected to be shifted to consumers as population aging proceeds in a society. Second, admitting the conventional argument that employer- pays social security contributions shift to prices more than employee-pays contributions, a society with slower population growth tends to increase the share-out ratio of the former. We leave the empirical analysis of these hypotheses to future research. Our analysis also has an implication for the recent debates on the effect of aging on the size of the welfare state. Recently, Razin et al. (2002) theoretically argued that population aging may serve to downsize the size of the welfare state, by taking account of the trade-off between a political and an economic effect; namely, on the one hand, aging makes stronger the political power of the old, and on the other hand, it increases the cost of redistribution. They also tested this hypothesis with data for the US and 12 European countries over the period 1965–1992 and provided some empirical evidence. Their analysis initiated theoretical and empirical debates by several studies, including Disney (2007), Simonovits (2007), and Galasso and Profeta (2007), among others. These studies, however, do not take account of the shift in financing methods. Our analysis suggests that the shift from wage tax financing to consumption tax financing will make the impact of population aging on the size of the welfare state more ambiguous than these analyses concur. This is because in the steady state of a dynamically efficient economy, consumption taxation can collect larger revenues than wage taxation when imposed at the same rate because the former has a broader tax base. As argued in this analysis, however, we have to consider a political factor in that wage tax financing makes a poorer worker decisive; other things being equal, wage tax financing leads to a larger social security benefit per retiree. With these counteracting effects, the shift to consumption tax financing adds an additional ambiguity to the overall effect of population aging on the size of the welfare state.

(continued) 58 H. Konishi

(continued) We also leave several other extensions of the model to future research. First, our model assumes policy commitment in that the tax policies determined today will not change in the future. One way to relax this assumption is to utilize a dynamic voting game and solve its subgame-perfect structure- induced equilibrium, a solution concept introduced by Conde-Ruiz and Galasso (2005). In Appendix 3, we show that every equilibrium outcome described in the main text is realized in the equilibrium of an inﬁnitely repeated voting game. Second, we have assumed a small open economy with ﬁxed rates of wages and interest. By extending the framework to a model with endogenous growth, we would be able to obtain richer insights about the relationship between social security funding and economic growth.

Appendix 1: Proof of Proposition 2

τyð τ Þ > τ > Suppose c Ei, w, n 0 and w 0. Then, differentiating (2.26) yields

∂τy 1 À τ c ¼ w < 0, ∂Ei Ucc ∂τy 1 1 þ n c ¼ ð1 þ τ Þþτyð1 À τ Þ < 0, ∂τ y 2 2 þ w c w w Uccð1 À τcÞ ð1 À τwÞ 1 r and

∂τy τ c ¼ w < 0, ∂ y 2 n ð1 þ rÞUccð1 À τcÞ ð1 À τwÞ where ∂2U 1 2ð1 þ nÞ τ U :¼ i ¼À w þ 1 þ τ < 0: cc ∂τ2 y 3 þ À τ c c ð1 À τcÞ 1 r 1 w

To prove (i), differentiate (2.30) with respect to n, and we have ∂ o ∂ ∗ ∂ o ð oÞ E þð þ Þ ð ∗Þ Ec ¼ 1 À ð ∗ÞÀ ð oÞ E > : ð : Þ f E 1 n f Ec F Ec f E 0 2 32 ∂Ei ∂n 2 ∂n

ð ∗Þ < ð Þ¼ ∕ The sign follows because F Ec F Em 1 2 and 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 59

∂ o ∂ ∂τy E ¼ Ej c < y 0, ∂n ∂τc ∂n

∂ ∕∂τy > ∂τy∕∂ the latter of which comes from (2.28) given Ej c 0 and c n 0. From (2.28), on the other hand, we have

∂Eo 1 þ τy ∂τy 2ð1 þ nÞτ ¼ 1 þ c c ¼À w > 0: ∂ y 3 ∂ y 3 Ei ð1 À τwÞð1 À τcÞ Ei ð1 þ rÞUccð1 À τcÞ ð1 À τwÞ

Thus, the bracketed term on the left-hand side of (2.32) must be positive, and hence ∂ ∗∕∂ > Ec n 0. From (2.31), then,

∂τ∗ ∂τy ∂E∗ ∂τy c ¼ c c þ c < 0 ∂n ∂Ei ∂n ∂n

τ∗ > whenever c 0. The proof of (ii) is quite similar. Differentiating (2.30) with respect to τw yields ∂ o ∂ ∗ ∂ o ð oÞ E þð þ Þ ð ∗Þ Ec ¼À ð oÞ E > ð : Þ f E 1 n f Ec f E 0 2 33 ∂Ei ∂τw ∂τw

τ∗ > whenever c 0, because from (2.28) "# ∂ o τy τy ∂τy E ¼ c þ 1 1 þ 2 c c ∂τ y 2 2 À τ y 2 y 3 ∂τ w ð1 À τcÞ ð1 À τwÞ 1 w ð1 À τcÞ ð1 À τcÞ w ð þ Þ½ þ τ þ τyð À τ Þ ¼ 1 n 1 w c 1 w < y 5 3 0 ð1 þ rÞð1 À τcÞ ð1 À τwÞ Ucc

τy > whenever c 0. Given that the bracketed term on the left-hand side of (2.33)is ∂ ∗∕∂τ > positive, we obtain Ec w 0. From (2.31), then,

∂τ∗ ∂τ ∂E∗ ∂τ c ¼ c c þ c < 0 ∂τw ∂Ei ∂τw ∂τw τ∗ > jj whenever c 0.

Appendix 2: Proof of Proposition 3

τw τc Let w and w be the wage tax rates at W1 and C2 in Fig. 2.4, respectively. Formally, τw :¼ τ∗ð Þ τc :¼ τ jτ∗ðτ Þ¼ w :¼ ∕ they are deﬁned by w w 0, n and w min w c w, n 0. If we let Tw 1 ð À τwÞ c :¼ ∕ð À τc Þ 1 w and Tw 1 1 w to simplify the notations, then the equilibrium conditions, (2.14), (2.26), (2.28), and (2.30) yield 60 H. Konishi

À ∗ þ wð À wÞ¼ ð : Þ kE Ew Tw 1 kTw 0 2 34 and

À þ c ð À c Þ¼ ð : Þ E Em kTw 1 Tw 0, 2 35

:¼ð þ Þ∕ð þ Þ ∗ ¼ where k 1 n 1 r .In(2.35), we make use of the fact that Ec Em when τy ¼ τe > ¼ τe c 0, following from (2.28). An equilibrium with w 0 c exists if and only if w c Tw Tw. Because (2.35) is quadratic, subtracting (2.35) from (2.34) after substitut- c w ing Tw for Tw in (2.34), we can rewrite the necessary and sufﬁcient condition for w c Tw Tw into :¼ Àð À Þð À c Þ: ð : Þ Ew EH Em 1 k E Tw 2 36

τc τc Similarly, let w and c be the consumption tax rates at W2 and C1 in Fig. 2.4. τw :¼ τ jτ∗ðτ Þ¼ τc :¼ τ∗ð Þ Their formal deﬁnitions are c min c w c, n 0 and c c 0, n . Let us w :¼ ∕ð À τwÞ c :¼ ∕ð À τcÞ denote Tc 1 1 c and Tc 1 1 c for simplicity. Then the equilibrium conditions, (2.14), (2.26), (2.28), and (2.30) yield

À þð À Þ w þ wð À wÞ¼ ð : Þ E Ew k 1 ETc Tc 1 kTc 0 2 37 and

À þ cð À cÞ¼ : ð : Þ E Ec Tc 1 Tc 0 2 38

τe > ¼ τe The existence of an equilibrium with c 0 w is guaranteed if and only if c w c w Tc Tc . Subtracting (2.38) from (2.37) after substitutingTc forTc in (2.37) reduces the condition to

:¼ Àð À Þ cð À cÞ: ð : Þ Ew EL Ec 1 k Tc E Tc 2 39

Rewriting the equilibrium conditions, (2.14), (2.26), and (2.30), we ﬁnd that if τe > τe > τ τ an equilibrium with w 0 and c 0 exists, then the tax rates, w and c, and ∗ the ability of the median worker–voter in voting on consumption tax, Ec , are determined through the following system of equations:

E À Ew þ Eðk À 1ÞTc þ TcTwð1 À kTcTwÞ¼0 ð2:40Þ

À ∗ þ 2 ð À Þþ ð À Þ¼ ð : Þ E Ec kTc Tw 1 Tw TcTw 1 Tc 0 2 41 and 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 61

2 þ n FðEoÞþð1 þ nÞFðE∗Þ¼ , ð2:42Þ c 2

o where Tw :¼ 1∕ð1 À τwÞ and Tc :¼ 1∕ð1 À τcÞ. The deﬁnition of E is given by o ¼ þ 2 ð À Þ E E kTc Tw 1 Tw , which we obtain from (2.26) and (2.28) in the case of τy > c 0. Then, subtracting (2.41) from (2.40) yields

ð À Þ ð À Þ¼ ∗ À : 1 k Tc E TcTw Ec Ew

À > ∗ > Given that E TcTw 0 owing to a positive labor supply and Ec Ew, it follows < τe > τe > that k 1 is necessary for the existence of an equilibrium with w 0 and c 0. We next show that as illustrated in Fig. 2.5, the consumption tax reaction curve is steeper than the wage tax reaction curve at their intersection. Differentiating (2.40), we have the slope of the wage tax reaction curve,

∂T T λ c ¼À c < 0, ∂Tw Twλ þ Eðk À 1Þ

where λ :¼ 1 À 2kTcTw < 0 owing to the second-order condition spelled out in (2.16). Similarly, differentiating (2.41) and (2.42), and rearranging the terms, we obtain the slope of the consumption tax reaction curve,

∂T T λ þ α c ¼À c < 0, ∂Tw Twλ þ β where

ð oÞ α :¼ 2ð À Þþ f E 2ð À Þ < Tc k 1 ð þ Þ ð ∗Þ kTc 1 2Tw 0 1 n f Ec and

ð oÞ β :¼ ð À Þþ 2f E ð À Þ < : 2TwTc k 1 ð þ Þ ð ∗Þ kTwTc 1 Tw 0 1 n f Ec

Simple calculation then demonstrates that the consumption tax reaction curve is steeper, because 62 H. Konishi

T λ þ α T λ Δ c À c ¼ > 0, Twλ þ β Twλ þ Eðk À 1Þ ðTwλ þ βÞðTwλ þ Eðk À 1ÞÞ where

Δ :¼ Tcðk À 1ÞλðE À TwTcÞþETcðk À 1Þ ð oÞ 2 þ kf E Tc À λ þ ð À Þð À ÞÞ > : ð þ Þ ð ∗Þ 2 Tw E k 1 1 2Tw 0 1 n f Ec

Finally, given that the consumption tax reaction curve is steeper, it follows that τe > τe > the equilibrium with w 0 and c 0 is unique and that it exists if and only if w > c c > w < < jj Tw Tw and Tc Tc . This condition is reduced to EL Ew EH.

Appendix 3: Subgame Perfection of the Equilibrium Outcomes

Following Conde-Ruiz and Galasso (2005), we show that every structure-induced equilibrium obtained in the main text under the assumption of policy commitment is established as a subgame-perfect equilibrium outcome in an infinitely repeated voting game without policy commitment. Suppose that a voting game takes place in each period, where each worker and retiree announces a pair of wage and consumption tax rates. Let τwt and τct be the voting outcomes in period t 1, which are defined respectively as the medians of wage and consumption tax rates announced by voters. Let h1 be the history at the start of the game, and let ht be one at the start of period t. The latter is a combination of h1 and the outcomes having been realized until period t. The set H collects all c τ ¼ τe τ ¼ τe possible histories, and H contains only h1 and ht such that ws w and cs c for all s t À 1. We will denote by Ht the set of possible histories until period t. Each voter’s strategy in period t is a mapping from Ht to the set of the pairs of the σoð Þ two tax rates. Let i ht be the strategy of a retiree with ability Ei when voting in period t. Following (2.7), her payoff function is defined as

Vit ¼ð1 þ rÞ½Ai À τctð1 À τwtÀ1ÞðEi À EÞ þ ð1 þ nÞτwtNðτwt, τctþ1Þ,

where A and Ai are constant, satisfying Ai < A if and only if Ei < E. Similarly, let σyð Þ i ht be the strategy of a worker with ability Ei when voting in period t and, following (2.10), deﬁne her payoff function as

Btþ1 U ¼ð1 À τ þ Þð1 À τ ÞE À lnð1 À τ Þð1 À τ þ Þþ , it ct 1 wt i wt ct 1 1 þ r where 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 63

Btþ1 ¼ð1 þ rÞτctþ1ð1 À τwtÞNðτwt, τctþ1Þþð1 þ nÞτwtþ1Nðτwtþ1, τctþ2Þ:

ðτe τeÞ Now we will show that every structure-induced equilibrium w, c presented in Proposition 3 is established as a stationary subgame-perfect equilibrium outcome of the inﬁnitely repeated voting game by the combination of strategies, σoð Þ¼ðτo ð Þ τo ð ÞÞ σyð Þ¼ðτy ð Þ τy ð ÞÞ i ht wi ht , ci ht and i ht wi ht , ci ht for t 1, such that

τo ð Þ¼τo ðτeÞ τo ð Þ¼τe ð : Þ wi ht w c , ci ht c 2 43 for ht 2 Ht and Ei 2½E, E; τe if h 2 Hc τe if h 2 Hc τy ðh Þ¼ w t τy ðh Þ¼ c t ð2:44Þ wi t 0 otherwise; ci t 0 otherwise

2½ e for Ei E, Ec ; and τy ðE , τe, nÞ if h 2 Hc τyðE , τe , nÞ if h 2 Hc τy ðh Þ¼ w i c t τy ðh Þ¼ c i w t wi t 0 otherwise; ci t 0 otherwise: ð2:45Þ

2ð e e for Ei Ec, E , where Ec is the ability level that the equilibrium median worker– voter has in voting on consumption tax rates, implicitly deﬁned by τyð e τe Þ¼τe c Ec, w, n c. These strategies have the following properties. First, as (2.43) shows, concerning voting on wage tax rates, the equilibrium strategy of a retiree stipulates the same behavior as she chooses in the structure-induced equilibrium with policy commit- τe ment. As regards consumption tax rates, every retiree votes for c, whatever e happens in the past. Second, as (2.44) shows, the votes of workers with Ei Ec cluster at the pair of tax rates realized in the structure-induced equilibrium with policy commitment, as long as it has been repeatedly realized in the past, and otherwise they all vote for abolishing the social security system. Third, as (2.45) > e shows, workers with Ei Ec vote as described in the text whenever the outcome ðτe τeÞ w, c has been repeatedly realized, but otherwise they will vote for abolishing the ðτ τ Þ¼ðτe τeÞ 2 c social security system. Under these strategies, wt, ct w, c if ht H , and otherwise ðτwt, τctÞ¼ð0, 0Þ. Let us check whether these strategies constitute a subgame-perfect equilibrium, assuming that even a single vote can affect the voting outcome. τe To begin with, consider the strategy of retirees. If the consumption tax rate is c, τe τo ðτeÞ > τe they all want to increase the wage tax rate above w because w c w. To do this, τe they have to increase the votes for the tax rates higher than w. However, their votes are already higher than the level, and thus there is no room for them to change o the voting outcome. With respect to the consumption tax rate, retirees with Ei < E 64 H. Konishi

ð e τe Þ τe Ec, w want to increase it above c. However, they cannot manipulate the voting τe outcome in their desired direction because they already vote for c. A similar > oð e τe Þ reasoning applies to the voting behavior of retirees with Ei E Ec, w . c Now turn to the strategy for workers. First, in the case of ht 2 H , a similar reasoning applies. There is no room for each worker to manipulate the voting outcome in period t in her desired direction because she already votes in that e way. Next, suppose that workers with Ei Ew strategically voted for a wage tax τe τ rate below w and successfully reduced wt in period t. Then, the voting behavior stipulated in (2.44) will yieldτwtþ1 ¼ τctþ1 ¼ 0in period t + 1. This means that these workers receive no social security beneﬁts. If so, their best outcome in period t is τ ¼ ðτe τeÞ ð Þ wt 0. However, even when this happens, Ui w, c Ui 0, 0 holds for the ðτe τeÞ ð τeÞ e following reason. First, Ui w, c Ui 0, c for Ei Ew because of the single- ð τeÞ ð Þ < e crossing property of the utility function. Second, Ui 0, c Ui 0, 0 for Ei Ec τe < τyð Þ ∂2 ∕∂τ2 < because c c Ei,0,n and Ui c 0. Accordingly, they have no incentive to > e deviate from (2.44). Regarding workers with Ei Ew, because the majority of votes τe cluster at w, they cannot manipulate the voting outcome even if they change their votes on the wage tax rates. A similar reasoning applies to the voting on the consumption tax rates, and the above strategies result to form a subgame-perfect Nash equilibrium. jj

Comment Paper to Chapter 2

Naomi Miyazato Nihon University, 1-3-2 Misaki-cho, Chiyoda-ku, Tokyo, Japan e-mail: [email protected] This chapter presents an analysis of the political economy of financing social security by wage tax and consumption tax by using a model of majority voting within an overlapping-generations framework. The chapter focuses on the effects of wage tax and consumption tax on intra-generational redistribution as well as on the cost of intergenerational transfer and examines how population ageing affects the political equilibrium. Analytical results show that a society with low population growth is likely to have multiple equilibria with social security financed by wage tax only, consumption tax only, or both, whereas a society with rapid population growth has a unique equilibrium with social security financed by wage tax only. The results are very interesting and seem to elucidate the reasons behind the recent popularity of financing social security with consumption tax in some countries and the diversified composition of social protection receipts across countries. Here, I pose two questions that may appear trivial. First, people could transfer the burdens of financing social security onto the future. For example, Miyazato (2012) shows that policies implemented in the 1990s and the first half of the 2000s in Japan led to the transfer of burden onto future generations. Thus, people seem to have the policy option of transferring the social security burden onto 2 The Political Economy of Social Security Funding: Why Social VAT Reform? 65 future generations in addition to financing social security by consumption tax and wage tax. My question here is whether this transfer of burden onto future generations affects equilibrium. The next point regards the population growth rate in simulation analysis. According to the prediction of the model and simulations results, social security could be financed by taxing consumption only if the population growth rate is below 1.5 %. However, no country seems to be financing social security by levying only consumption taxes, even though many developed countries already have a population growth rate below 1.5 %.20 Therefore, my second question is what factors lead to this difference between the model prediction and the real world? Finally, I would like to consider the possibility of an empirical study of the theory presented in this chapter. The model predicts that the ratio of financing social security with consumption tax increases with population aging. Therefore, we could regress the ratio of financing social security with consumption tax on an aging indicator, as proposed by Razin et al. (2002) and Disney (2007), by using panel data from the Eurostat and OECD revenue statistics to check the main implications of the theory in this chapter.

References

Adema W, From P, Ladaique M (2011) Is the European welfare state really more expensive? Indicators on social spending, 1980–2012; and a manual to the OECD Social Expenditure Database (SOCX), OECD social, employment and migration working papers, No. 124. OECD Publishing, Paris Atkinson AB, Stiglitz JE (1980) Lectures on public economics. McGraw-Hill, New York Boadway RW, Wildasin DE (1989) A median voter model of social security. Int Econ Rev 30:307–328 Browning EK (1975) Why the social insurance budget is too large in a democracy. Econ Inq 22:373–388 Conde-Ruiz JI, Galasso V (2005) Positive arithmetic of the welfare state. J Publ Econ 89:933–955 Cooley TF, Soares J (1999) A positive theory of social security based on reputation. J Polit Econ 107:135–1160 Disney R (2007) Population ageing and the size of the welfare state: Is there a puzzle to explain? Eur J Polit Econ 23:542–553 Galasso V, Profeta P (2007) How does ageing affect the welfare state? Eur J Polit Econ 23:554–563 Hu SC (1982) Social security, majority-voting and dynamic efﬁciency. Int Econ Rev 23:269–287 Meltzer AH, Richard SF (1981) A rational theory of the size of government. J Polit Econ 89:914–927 Miyazato N (2012) Intergenerational redistribution policies of the 1990s and 2000s in Japan: An analysis using generational accounting. Mimeo Persson T, Tabellini G (2000) Political economics. MIT Press, Cambridge

20 The population growth rates of many developed countries can be found in Statistical Bureau, Ministry of International Affairs and Communications (2013). 66 H. Konishi

Razin A, Sadka E, Swagel P (2002) The aging population and the size of the welfare state. J Polit Econ 110:900–918 Shepsle KA (1979) Institutional arrangements and equilibrium in multidimensional voting models. Am J Polit Sci 23:23–57 Simonovits A (2007) Can population ageing imply a smaller welfare state? Eur J Polit Econ 23:534–541 Statistical Bureau, Ministry of International Affairs and Communications (2013) Sekai no Tokei 2013 (International Statistical Compendium 2013). Japan Statistical Association Tabellini G (2000) A positive theory of social security. Scand J Econ 102:523–645 Part II Fiscal Problems in Japan Chapter 3 Female Labor Supply, Social Security, and Fiscal Consolidation

Ryuta Ray Kato and Masumi Kawade

Abstract This paper numerically examines the impact of expanding female labor supply on economic growth and the government revenue in an aging Japan within a dynamic general equilibrium framework with multi-period overlapping generations. The difference in full-time and non full-time workers of both male and female labor force is explicitly considered. Simulation results indicate that even if all potential female labor force who cannot work currently due to child care becomes full-time workers and thus labor force in efficiency is most expanded, the impact of such an increase on the Japanese economy and fiscal consolidation through an increase in tax revenue as well as contributions to the public pension scheme is very much limited. Even in the most expanding case, production only increases by 1.50 % and the improvement in the government budget would be 1.34–1.46 % in year 2050. The most crucial reason of such limited impacts is found in the large gap in the wage profile between male and female workers, and if the gap vanishes, the impact drastically becomes quite large.

Keywords Computable general equilibrium (CGE) model • Economic growth • Female labor supply • Fiscal consolidation • Public pension • Simulation

1 Introduction

This paper numerically examines the impact of expanding female labor supply on economic growth as well as the government revenue in an aging Japan within a dynamic general equilibrium framework with multi-period overlapping generations. The future population forecast of Japan indicates that a drastic decrease not only in the labor force, but also in the total population is unavoidable in the future.

R.R. Kato (*) International University of Japan, 777 Kokusai-cho, Minami-Uonuma, Niigata 949-7277, Japan e-mail: [email protected] M. Kawade Nihon University, 1-3-2, Misaki-cho, Chiyoda-ku, Tokyo 101-8306, Japan e-mail: [email protected]

© Springer Japan 2015 69 T. Ihori, K. Terai (eds.), The Political Economy of Fiscal Consolidation in Japan, Advances in Japanese Business and Economics 8, DOI 10.1007/978-4-431-55127-0_3 70 R.R. Kato and M. Kawade

This prediction with rapid population aging in the near future implies that the increasing trend of national medical expenditure, further burdens on the future generations through the current pay-as-you-go public pension scheme, and even the decline of future economic growth are also unavoidable, if the current demographic structure does not change. As our research results (Ihori et al. 2006, 2011) suggest, two options could be considered to weaken the negative impact of the future demographic change on sustainable economic growth; stimulation of more female labor supply and/or opening a door to immigrants, in order to have the labor force enough for sustainable economic growth. Then, this paper particularly focuses on the impact of expanding female labor supply on economic growth, by examining to the extent how much an expansion of female labor supply improves future economic growth as well as the government revenue such as tax revenue and contributions to the public pension scheme. When female labor supply is explicitly considered, the pattern of labor supply by females over time should be taken into account. While the degree is different among different countries, a similar pattern of female labor supply over time is observed, which is the so-called ‘M shaped’ curve. Female labor supply first increases over time, and then due to child care it drops. After finishing child care they tend to get back to the labor market. Thus, in order to stimulate more female labor supply, it could be a key issue to the extent how much any policy is effective to reduce the opportunity cost of child care. In this paper, the impact of weakening the M-shaped curve is considered. Another key issue, when female labor supply is examined, is that labor force has recently been observed to be divided into two different categories in Japan; full- time labor force and non full-time labor force. The former is called ‘Seiki’ labor force, and the latter is called ‘Hi-Seiki’ labor force. The latter labor force includes part-time, dispatched, and fixed term workers. The recent division of labor force can be observed not only in female but also in male labor force. This paper also takes into account this difference in labor force explicitly, and assumes that there are four different types of labor force; male full-time, female full-time, male non full-time, and female non full-time labor force, since labor condition and wage differences can clearly been observed in data among these four categories of labor force. Several simulations will be conducted based on different scenarios on this aspect. Simulation results are as follows. First of all, even in the most stimulative case in terms of an expansion of female labor force measured in efficiency, production only increases by 1.50 % in year 2050. In this case, all potential female labor force who cannot work currently due to child care is assumed to become full-time workers. The impact of an increase in such female labor force on the Japanese economy is very limited. Another result shows that such a most stimulative expansion of female labor supply has little improvement in the government tax revenue and the total contributions to the public pension scheme as well. An increase in tax revenue through a consumption tax is at most 1.46 % in year 2050, and an increase in public pension contributions is 1.34 % in year 2050. Since the above results are obtained under the most stimulative assumption regarding the potential female labor force 3 Female Labor Supply, Social Security, and Fiscal Consolidation 71 who cannot work currently due to child care, the actual impact of stimulating such female labor force on the Japanese economy as well as fiscal consolidation would be smaller. Secondly, if the government tries to expand the Japanese economy and also expects fiscal consolidation by stimulating female labor force, then it should consider the wage profile gap in gender, otherwise the impact of a further inflow of female labor force into the labor market is quite limited. If the wage profile gap between currently working male and female labor force completely vanishes, then an inflow of female labor force as full-time workers who cannot work due to child care results in a drastic increase in production by 15.45 %. The tax revenue and the amount of contributions to the public pension scheme also increase by 13.98 % and 13.67 % in year 2050, respectively. Even though all of them enter as non full-time workers into the labor market, production, the tax revenue, and the amount of contributions will increase by 13.61, 12.26, and 12.07 %, respectively, if the wage profile gap vanishes. Thirdly, the impact of the change in the contract type of currently working female workers from non full-time to full-time on the economy as well as fiscal consolidation is quite limited. Even though the ratio of current full-time and non full-time female workers becomes the same as that of current male workers, production, the tax revenue, and the amount of contributions only increase by 2.68, 2.49, and 2.37 %, respectively, in year 2050. Finally, the impact of the so-called M-shaped pattern is also small. If the M-shaped pattern vanishes, increases in production, the tax revenue, and the amount of contributions are 4.06, 3.79, and 3.57 %, respectively, in year 2050. The paper is organized as follows. The next two sections explain the background of the discussion and the related literature. Section 4 introduces the model, and Sects. 5 and 6 explain the conducted simulation analysis in detail. Section 7 concludes the paper.

2 Background

Figure 3.1 shows the time trend of the female labor force ratio. The ratio is defined as the relative size of the actual female labor force to the total female labor force. Note that the number of females who choose not to work is not included in the definition of the total female labor force. As the figure shows, the so called M-shaped pattern is vanishing over time. Figure 3.2 gives comparison of the ratio among different countries. While the ratio in age 35–45 of Japan is still low in comparison with Sweden and France, the gap between Japan and the US is relatively small. The largest gap of the ratio between Japan and the US is observed in age 35–39, which is only 6.7 % points. Thus, these two figures suggest that as long as the total female labor force, which does not include females who choose not to work, is concerned, the impact of a further increase in female labor force on economic growth would be limited. Japan has already changed in the sense that the 72 R.R. Kato and M. Kawade

Fig. 3.1 The female labor force ratio

Fig. 3.2 International comparison of the female labor force ratio (year 2011) cost of child care is not so high, as long as the number of females in the labor market is concerned. However, the type of jobs, particularly the contract condition for female workers should also be considered. The labor force in Japan has recently been observed to be divided into two different categories; full-time labor force and non full-time labor force. The former is called ‘Seiki’ labor force, and the latter is called ‘Hi-Seiki’ labor force. The latter labor force includes part-time, dispatched, and fixed term workers. This phenomenon has been observed in both male and female labor force. Non full-time job is generally more flexible and thus it would provide females with more freedom. Thus, if females spend time on caring children, it would sometimes be more beneficial. However, it is argued that they choose non full-time jobs as if they are forced, and also that the wage profile is relatively lower in comparison with 3 Female Labor Supply, Social Security, and Fiscal Consolidation 73

Fig. 3.3 Wage proﬁles (year 2012: male non full-time 20-24¼1)

Fig. 3.4 Female labor force in year 2012 the one of full-time jobs. Furthermore, if the number of females who cannot work due to child care is not negligible, then stimulation of female labor force could still have a certain positive impact on economic growth. If such females are encouraged to work as full-time workers, then the magnitude of the impact would be larger, since labor supply measured in the efficiency unit of full-time workers is observed to be higher than that of non full-time workers. Figures 3.3 and 3.4 show the wage profiles and the detailed information on female labor force, respectively. Figure 3.3 74 R.R. Kato and M. Kawade shows the wage profiles based on the assumption that efficiency of the non full time male worker in age 20–24 is used to normalize efficiency of other workers in different age. In Fig. 3.4, the number of females who cannot work due to child care is shown. As the figure shows, the number starts increasing from the age group of 25–29 to 40–44. It reaches the highest in age 30–34, which is 6.6 %. This paper focuses on the potential female workers in this category, and the impact of several scenarios on economic growth and the government budget will be simulated.

3 Literature

Within the Auerbach and Kotlikoff framework (1987), many studies have been conducted in order to discuss the effect of tax and social security reforms of Japan. Homma et al. (1987) first applied Auerbach et al. (1983) to the Japanese context to examine the impact of the tax reform. Then, the conventional Auerbach and Kotlikoff model has been extended by incorporating actual future demographic forecast of Japan, in order to re-produce the future demographic structure within the model (Kato 1998, 2002), where several new features have also been taken into account such as government deficits and public capital. Kawade (2007; 2009) introduced an income difference in households generated by heterogeneity in labor efficiency in order to discuss the impact of fiscal reforms actually ongoing in Japan. Ihori et al. (2011) particularly paid attention to the national health services in an aging Japan to discuss the effects of several policy changes in the drastic reform of the national health services of Japan. One of our results shows that the increasing trend of national medical expenditure is mainly caused by population aging, and several policy changes in the reform such as an increase in the co-payment rate for the services has little effect on the reduction of the increasing trend in the future. Then the common wisdom obtained in the literature is that, as long as the current forecast of the future population is given, further burdens on the future generations are unavoidable, provided that the drastic reduction of pension benefits and the drastic increase in co-payments in the national health services are not conducted. The key issue is obviously population aging. Based on Projection of Future Population in Japan by the Institute of Population and Social Security Research, the existing literature has been discussing the impact of several reforms by taking the future population structure as given. Then next concerns are with the future demographic structure, and two options could be considered to change the given structure of the future generations; immigration and an expansion of female labor supply, particularly to stimulate future economic growth. Then, this paper tries to incorporate the endogenous labor supply behavior of females into our Auerbach and Kotlikoff model, in order to explore the impact of an expansion of female labor supply on economic growth as well as the government revenue. Regarding studies on labor supply in Japan, there has been several studies, particularly empirical studies on the estimation of several elasticities. In particular, on female labor supply in Japan, Abe (2009) and Kuroda and Yamamoto (2007) 3 Female Labor Supply, Social Security, and Fiscal Consolidation 75 should be mentioned. Kuroda and Yamamoto (2007) estimated several labor supply elasticities by using two different data sets. They distinguished the difference in estimation between the extensive margin and the intensive margin, and they also estimated the Frisch elasticities over time. Abe (2009) empirically studied the impact of a means-tested transfer program, so called ‘1.03 million yen ceiling’, on the female labor supply based on a theoretical framework.1 On the estimation of labor efficiency in age, Jinno (2010) used the long-run panel data to conclude that the age group of 40s achieves the highest efficiency in Japan, and also he considered the impact of population aging on labor efficiency. The estimated values of several key parameters particularly related to female labor supply and age-specific labor efficiency are referred in this paper. In the literature of the Auerbach and Kotlikoff model, a dynamic general equilibrium framework with multi-period overlapping generations, an explicit incorporation of female labor supply into the model has not been considered yet. In this paper, endogenous labor supply decision is considered within the Auerbach and Kotlikoff model, and not only the realistic future demographic structure but also the wage profiles of males and females are taken into account. Another difference from the existing literature is that the type of the work contract is also explicitly considered; full-time or non full-time working conditions. In the context of the Japanese labor market, the effect of such a difference in the contract condition has been argued recently. The difference in the wage profile among full-time and non full-time workers can be observed in Japan, and such a difference is also considered in this paper. Thus, four different groups of workers explicitly exist in our model; male full-time, female full-time, male non full-time, and female non full-time workers, in order to capture more realistic aspect of the Japanese labor market in the model.

4 Model

The computable general equilibrium model of this paper employs the dynamic multi-period overlapping generations model initially developed by Auerbach et al. (1983). This paper expands Kawade (2007; 2009) by explicitly considering the gender difference as well as the type of workers; Seiki (full-time workers) and Hi-seiki (part-time, dispatched, or ﬁxed term workers). Thus, there are four different types of workers; seiki male, seiki female, hi-seiki male, and hi-seiki female workers.

1 Iwamoto et al. (2001) estimated the deadweight loss of households who look after long term elderly patients. While the bottom of the M-shaped female labor supply over time is basically explained by child care, elderly care by females at home could also be considered as a reason why female labor supply is ceiled. 76 R.R. Kato and M. Kawade

Each household consists of these four different workers, and it is assumed that each household optimizes its intertemporal consumption through its lifetime, taking the wage rate, the interest rate, and its own survival rates as given. The tax system, and the public pension scheme are also assumed to be taken as given by each household. The household is assumed to obtain its wage income by supplying labor elastically until it retires, and once it retires it never returns to the labor market. There are no altruistic bequest motives and Ricardian equivalence does not hold. The ﬁrm is assumed to maximize its proﬁt by taking the wage rate and the interest rate as given. The wage rate and the interest rate are both determined in their fully competitive factor market in equilibrium. The government sector is assumed to collect taxes from households. For simplicity, the budget constraint of the government is assumed to be balanced in each period, so that the only balanced budget case is considered in order to concentrate on the impact of an expansion of female labor supply on the Japanese economy. The government sector has its public pension account as well as its general account. In order to capture the realistic aspect of its accounts, the government is assumed to have transfers from the general account to the public pension asocial account. The public pension account is assumed to be run under the pay-as-you-go scheme. It is assumed that there is no private life insurance, and thus there is no mechanism for the household to hedge the risk of dying in each period. Since the household is assumed to have no bequest motives, this assumption implies that the household leaves an accidental bequest when it dies. However, it is also assumed that there is no uncertainty in the whole economy in terms of the size of each generation, and thus there is no uncertainty in the total (aggregate) amount of bequests inherited in each period.

4.1 The Household

The representative household consists of four different types of workers. Workers differ, depending on the gender as well as the type of the work contract. Workers are differentiated by the difference in their work contract condition with their employer; Full-time work contract condition, or not. The former workers are called Seiki workers, and the latter workers are called Hi-seiki workers. The latter workers include part-time, dispatched, or ﬁxed term workers. Thus, there are four different workers in each household: Male seiki, female seiki, male hi-seiki, and female hi-seiki workers, respectively. Each household appears in the economy at age 20 as a decision maker. Although the household faces uncertainty regarding its death in each period, it dies with certainty at the end of its age of 99 if it is alive until age 99. It is assumed that there is no uncertainty regarding the size of the total population in each period. The household is assumed to maximize its expected lifetime utility with respect to its own consumption.ÂÃ The household’s expected lifetime utility of generation g, denoted by EVg , is given 3 Female Labor Supply, Social Security, and Fiscal Consolidation 77

ÂÃ X79 ðÞ1Àρ ¼ ðÞþ δ Às ucs,t, ls, t EVg Ps 1 À ρ , s¼0 1 where ρ is a reciprocal of the elasticity of substitution between consumption at the different time. δ is the time preference. Ps is a probability weight of the survival rate Ys ¼ deﬁned by Ps qi, where qj+1 is the conditional survival rate of a j years old i¼1 household survives to j + 1 years old. cs, t and ls, t are consumption and leisure of a s years old household at time t, respectively. Note that there is a relationship of t ¼ g þ s. The felicity function of u is given by:

ξ ξÀ1 ξÀ1 ξÀ1 ðÞ¼ξ þ κ ξ ucs,t, ls,t cs,t ls,t , where ξ denotes the elasticity of substitution between consumption and leisure, and κ denotes the weight parameter for leisure. The budget constraint of each household is: ÀÁ asþ1,tþ1 ¼ ½1 þ ðÞ1 À τr,t rt as, t þ 1 À τw,t À τp, t es, tðÞ1 À ls,t wt

þbt þ ðÞ1 À τh ht À ðÞ1 þ τc,t cs,t where as, t is the amount of assets held by a s years old household at the beginning of time t. es,t is the measure of efficiency of labor of the household, and es is the weighted average of efficiency of four different workers. Note that efficiency is different in age among all four different workers as well. Labor efficiency of four different workers can be obtained from the data. In this paper, Labor Force Survey 2012 (Rohdotyoku Chosa 2012) and Basic Survey on Wage Structure 2011 (Chinginkouzou Kihontoukei Chousa 2011) have been used to specify the efficiency profile of each worker over time. The weight for efficiency of the household in age s, es, has been calculated from these two data sets. In the simulation section, efficiency of the non full-time male worker in age 20–24 is used to normalize efficiency of other workers in different age.2 Note also that the wage rate the household faces is also the weighted average of wages of four different workers. The wage profiles of four different workers have been obtained from the above two data sets. Thus, the total labor supply by the household in age s at time t is such that:

2 While the household makes its decision each year in the theoretical section, it is assumed in the simulation section that the household makes its decision every 5 years, since the above two data sets only provide data for every 5 year period. 78 R.R. Kato and M. Kawade "# νs,t s, t þ νs,t s,t þ νs,t s,t þ νs,t s,t m,ftem, ft fe,ftefe,ft m,nf em,nf fe, nf efe, nt e ðÞ¼1 À l ðÞð1 À l 3:1Þ s,t s, t þνs s þ νs s s, t m,0em,0 fe,0efe, 0 where m, fe, ft, and nt denote male, female, full-time contract, and non full-time s contract, respectively. Thus, ek,n denotes labor efficiency of gender k of contract s type of n in age s. Labor efficiency is measured by the wage. ek, 0 denotes labor efficiency of gender k who does not work at all, and it is assumed to be zero for all s s and both males and females. vk,n denotes the weight of gender k of contract type of s n in age s in efficiency of the household. vk, 0 denotes the weight of gender k who 3 s s s does not work at all. Note that ek, n, vk,n and vk, 0 have all been calculated from Labor Force Survey (2012) and Basic Survey on Wage Structure (2011). τr,t, τw, t, τp,t, and τc, t are the interest income tax rate, the wage income tax rate, the public pension contribution rate, and the consumption tax rate, respectively. ht is the amount of bequests inherited at time t, and τh is the inheritance tax rate. bt is the amount of public pension benefits. wt and rt are the wage rate per the efficiency unit and the interest rate, respectively. Public pension benefits are given by E H ; s RH b ¼ t t , t 0; s, < RH where RH is the retirement age. εt and Ht denote the replacement rate, and the average annual amount, respectively. Ht is given by:

À 1 RHX1 Ht ¼ wtes,tðÞ1 À ls, t : RH s¼0

The ﬁrst order conditions then give us the following optimal equations such that:

½þ ðÞÀ τ 0 qsþ1,g 1 1 r,tþ1 rtþ1 1 þ τc,t 0 u ðÞ¼cs, t, ls, t u ðÞcsþ1, tþ1, lsþ1,tþ1 , 1 þ δ 1 þ τc, tþ1 ξ κðÞ1þτc,t ls,t ¼ cs,t ðÞ1Àτw,tÀτp,t wtes

3 Note that XX X ¼ s þ s 1 vk,n vk,0 k n k for all s. 3 Female Labor Supply, Social Security, and Fiscal Consolidation 79

4.2 The Firm

The ﬁrm is assumed to maximize its proﬁts, taking the wage rate and the interest rate as given. The wage rate and the interest rate are determined in perfectly competitive factor markets in equilibrium. The aggregate private production function is assumed to be Cobb-Douglas such that

¼ Ω α 1Àα Yt tLt Kt , where Yt, Kt denote aggregate output, and capital at time t. Lt is total labor demand measured in the efﬁciency unit. Ωt is technology of production of the private sector. The fully competitive assumption of factor markets yields:

Yt wt ¼ α , Lt Yt rt ¼ ðÞ1 À α À φ, Kt where φ is the depreciation rate.

4.3 The Government

The government sector consists of a general account and a public pension account. Expenditure in the general account includes the general government expenditure and transfers to the public pension account. Expenditure of the general account is ﬁnanced by taxes. For simplicity, the budget constraint of the general account is assumed to be balanced in each period. The general government expenditure includes government consumption, and government investments. The amount of transfers to the public pension account from the general account is characterized by η t, which is the ratio of the amount of transfers to the total amount of public pension beneﬁts at time t. The government sector is assumed to have no particular objective function which it maximizes. The general government expenditure, GEt, at time t is

GEt ¼ CGt þ IGt, where CGt, and IGt denote government consumption, and government investments, respectively. Then, the budget constraint of the general account is given such that:

þ η ¼ τ þ τ þ τ þ τ ð : Þ GEt tBt c, tACt r, tTAt w,twtLt hAHt, 3 2

η where Bt is the total public pension beneﬁts, and tBt is the amount of transfers from the general account to the public pension account. ACt, TAt, and AHt denote the 80 R.R. Kato and M. Kawade aggregated consumption, the aggregated assets, and the aggregated bequests, respectively. In this paper, the consumption tax rate is endogenously determined to satisfy the above the budget constraint of the general account. On the public pension account, the contribution rate, τp, t, is determined according to:

ðÞÀ η 1 t Bt τp,t ¼ , ð3:3Þ wtLt and the budget constraint of the public pension account is assumed to be balanced at η 4 each time with the transfers, tBt.

4.4 Equilibrium

All markets are assumed to be fully competitive. The equilibrium condition of the capital market is given by:

AtÀ1 ¼ Kt þ τhAHt, where AtÀ1 denotes the aggregated total private savings. The labor market equilibrium condition is given by:

RHXÀ1 Lt ¼ es, tðÞ1 À ls,t POPs,t s¼0 RHXÀ1hi ¼ νs,t s,t þ νs, t s, t þ νs,t s, t þ νs,t s,t ðÞÀ m,ftem,,ft fe,ftefe, ft m,nf em,nf fe,nf efe,nt 1 ls,t POPs,t, s¼0

where POPs, t denotes the total population of age s at time t, and thus the right hand side is the total labor supply at time t. On the goods market, the equilibrium condition is given such that:

Yt ¼ ACt þ Ktþ1 À ðÞ1 À φ Kt þ GEt

5 Simulation Analysis 5.1 Benchmark and Calibration

In the benchmark and scenario cases, the actual data has been used until year 2010. In order to highlight the impact of an increase in female labor force, changes in government activities are minimized in the benchmark and all scenario cases. In the

4 For simplicity, it is assumed that no public pension fund is accumulated. 3 Female Labor Supply, Social Security, and Fiscal Consolidation 81

Table 3.1 Sensitivity analysis Intratemporal Production Elasticity (normalized between by the baseline Technological Intertemporal Consumption value of year Growth Discount Rate Elasticity and Labor 1995) (Baseline: 5 %) (Baseline: À35 %) (Baseline: 0.4) (Baseline: 0.6) Baseline 10 % 1 % À30 % À40 % 0.5 0.3 0.7 0.5 1995 1.000 1.029 0.975 0.929 1.071 1.083 0.893 1.083 1.054 2000 1.108 1.184 1.044 1.028 1.188 1.203 0.986 1.203 1.168 2005 1.178 1.314 1.069 1.093 1.264 1.281 1.046 1.281 1.242 2010 1.176 1.373 1.025 1.090 1.263 1.282 1.041 1.282 1.242 2015 1.221 1.498 1.020 1.130 1.313 1.334 1.078 1.334 1.289 2020 1.274 1.645 1.018 1.178 1.372 1.395 1.121 1.395 1.345 2025 1.322 1.802 1.009 1.221 1.425 1.451 1.160 1.451 1.396 2030 1.355 1.952 0.986 1.250 1.462 1.489 1.185 1.489 1.431 2035 1.369 2.088 0.949 1.262 1.479 1.508 1.194 1.508 1.448 2040 1.373 2.220 0.906 1.264 1.485 1.516 1.193 1.516 1.452 2045 1.384 2.373 0.868 1.271 1.498 1.530 1.198 1.530 1.463 2050 1.399 2.549 0.834 1.284 1.517 1.551 1.208 1.551 1.480 Intratemporal Life Time Elasticity Welfare (devi- between Con- ation rate from Technological Intertemporal sumption and the baseline Growth Discount Rate Elasticity Labor value) (Baseline: 5 %) (Baseline: À35 %) (Baseline: 0.4) (Baseline: 0.6) Baseline 10 % 1 % À30 % À40 % 0.5 0.3 0.7 0.5 1995 0.000 0.485 À0.776 0.520 À1.367 0.868 À33.873 0.868 0.339 2000 0.059 0.547 À0.783 0.548 À1.222 0.873 À30.908 0.873 0.384 2005 0.117 0.602 À0.787 0.575 À1.082 0.879 À28.058 0.879 0.426 2010 0.175 0.651 À0.782 0.602 À0.945 0.884 À25.243 0.884 0.468 2015 0.228 0.693 À0.779 0.628 À0.818 0.889 À22.739 0.889 0.507 2020 0.278 0.730 À0.776 0.651 À0.699 0.894 À20.448 0.894 0.542 2025 0.324 0.761 À0.772 0.673 À0.593 0.899 À18.427 0.899 0.575 2030 0.364 0.787 À0.770 0.693 À0.500 0.903 À16.665 0.903 0.603 2035 0.400 0.808 À0.767 0.710 À0.417 0.906 À15.128 0.906 0.628 2040 0.432 0.826 À0.762 0.726 À0.343 0.909 À13.765 0.909 0.651 2045 0.461 0.841 À0.754 0.741 À0.276 0.913 À12.542 0.913 0.671 2050 0.488 0.853 À0.741 0.754 À0.216 0.915 À11.450 0.915 0.690 benchmark and all scenario cases, the technological progress is assumed to be 1 % every year from 2011 over time.5 The ratio of government expenditure in the general account to GDP is assumed to be constant at 30 % over time in all cases. On the public pension account, the replacement rate (εt), and the transfer rate from

5 The annual 1 % level corresponds to 5 % in 5 years, as shown in Table 3.1. 82 R.R. Kato and M. Kawade

the general to the public pension accounts (ηt) are both assumed to be constant at 35 % and 20 % in all cases, respectively. While the consumption tax rate (τc, t) and the public pension contribution rate (τp, t) are endogenously determined according to (3.2) and (3.3), the wage income tax rate (τw, t) and the interest income tax rate (τr, t) are both constant over time at 30 % and 20 % in all cases, respectively. In terms of the future demographic structure, the latest version of Projection of Future Population in Japan by the Institute of Population and Social Security Research is used to reproduce the future population structure within the model, and the model can perfectly trace the actual future structure of all different cohorts of future generations. On efficiency of four different labor force groups, Labor Force Survey (2012) and Basic Survey on Wage Structure (2011) have been used. For year 1990, 1995, 2000, 2005, and 2010, the actual calculated values of efficiency have been used in this section.6 The actual relative size of four different labor force groups has been used as the weight in (3.1). The difference between the benchmark case and several scenario cases appears in female labor force from year 2011.7 In the benchmark model, it is assumed that the situation of year 2010 continues for the next 50 years. This implies that the number of potential female workers who cannot work due to child care in year 2010 does still not change in the future, and they do not enter the labor market, so that the relative size of female full-time as well as female non full-time labor force do not change over time even after 2010. This benchmark case is presented by Fig. 3.5, which are compared with the scenario cases presented in the successive sections. The sensitivity analysis for the benchmark case is given in Appendix. Several points should be noted: Due to the assumption of a slight annual increase in technological progress by 1 %, the future production increases. Since female labor force is assumed to keep unchanged even after year 2010 in the benchmark, the ratio of efficiency labor to gross product keeps decreasing, thus, resulting in an increasing trend of the future wage rate. In the following simulation scenarios, the impact of an increase in the number of potential female workers who cannot work due to child care after year 2010 is simulated. Thus, the following effects would be expected; an expansion of production as well as a decrease in the wage rate by an increase in labor supply. The decrease in the wage rate also affects the optimal total labor supply of the household.

6 Figure 3.3 only shows the value of year 2010. 7 The actual data is given every 5 years only, so that the benchmark and scenario cases all calculate values every 5 years as well. 3 Female Labor Supply, Social Security, and Fiscal Consolidation 83

Fig. 3.5 Benchmark case

5.2 Simulations

The purpose of this paper is to investigate the impact of an increase in female labor supply on the Japanese economy. In particular, potential female labor force, which could enter the labor market as new labor force if the child care cost is reduced, is concerned. Such potential female labor force corresponds to ‘not working (child care)’ in Fig. 3.4, and it appears from the age group of 25–29 to 40–44. Further- more, the difference in the contract condition, namely full-time or non full-time, should be considered when they enter the labor market, since labor supply in efﬁciency differs between full-time and non full-time workers. Then, the following two scenarios are simulated.

5.2.1 Scenarios

The following two scenarios are considered and simulated in this section.

Scenario I: In year 2011, all of potential female workers who cannot work due to child care enters the labor market as full-time female workers. The simulated pattern of female labor force starts in year 2011, and it is exactly the same as the benchmark case until year 2010. This scenario corresponds to the highest case in the sense that it most stimulates the Japanese labor force measured in efﬁciency. Scenario II: Scenario I seems relatively too unrealistic, since it might be difﬁcult to assume that all of such potential female workers can obtain full-time jobs. Then, this scenario considers the lowest case that all of them will get non full-time jobs. 84 R.R. Kato and M. Kawade

The difference between Scenario I and II appears only in the contract condition; full-time or non full-time.

Note that in both scenarios potential female workers who cannot work due to child care start supplying their labor in all ages. Thus, both scenarios correspond to the case where the M shaped curve shifts upward just by the size of ‘not working (child care)’ in Fig. 3.4. Note also that regarding the difference in the contract condition, full-time or non full-time, the actual situation would be between Sce- nario I and II. Thus, the above two scenarios show the lower and upper bounds of the impact of an increase in the number of female workers who can currently not work due to child care.

5.2.2 Results

Table 3.2 shows the magnitude of the impact on several key values. The figures in the table show the relative changes from the benchmark value. In Scenario I, it is assumed that all potential female labor force who cannot work due to child care becomes full-time workers in all ages. Thus, it corresponds to the most stimulative case in terms of an expansion of labor force measured in efficiency. As table shows, even in this case, production only increases by 1.50 % in year 2050. This implies that the impact of an increase in female labor force who cannot work due to child care on the Japanese economy is very limited. Due to an increase in labor supply, the wage rate decreases, and the interest rate increases. The change in the wage rate also induces endogenous response of total labor supply of the household. On the impact on the government accounts, both tax revenue and contributions to the public pension scheme increase. However, the magnitude is also very small. Even in Scenario I, an increase in the tax revenue through a consumption tax is at most 1.46 % in year 2050, and an increase in public pension contributions is 1.34 % in year 2050. In the last column of the table, the welfare change is also shown. Note that in the case of the impact on welfare the year shown in the table corresponds to the year when the generation was born, so that the relative change in welfare is measured in comparison with lifetime utility of each generation in the benchmark case. As the table shows, the impact on lifetime utility is also very small, and the positive impact is less than 2 % even in Scenario I. In Scenario II, the impact is nearly a half of that of Scenario I. This implies that the actual size of the impact of an inflow of female labor force who cannot work due to child care into the Japanese labor market is between these two scenarios. Thus, in terms of the impact on the Japanese economy and the fiscal consolidation, it can be concluded that it would have little effect, even if all female labor force who cannot work due to child care can start working by the reduction of the child care cost. eaeLbrSpl,Sca euiy n iclCnoiain85 Consolidation Fiscal and Security, Social Supply, Labor Female 3

Table 3.2 Relative change from the benchmark case (Scenario I and II) Scenario I All potential female labor force becomes fullÀtime workers Production (%) Capital (%) Labor (%) Interest Rate (%) Wage Rate (%) Tax Revenue (%) Contributions (%) Welfare (%) 2010 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.90 2015 1.17 0.00 1.77 2.38 À0.60 0.58 0.83 1.92 2020 1.19 0.29 1.65 1.82 À0.46 0.73 1.02 1.92 2025 1.21 0.53 1.56 1.38 À0.34 0.87 1.15 1.92 2030 1.26 0.74 1.53 1.07 À0.27 0.99 1.22 1.91 2035 1.33 0.91 1.54 0.87 À0.21 1.12 1.23 1.90 2040 1.40 1.09 1.57 0.67 À0.16 1.24 1.23 1.89 2045 1.46 1.26 1.57 0.43 À0.10 1.36 1.27 1.89 2050 1.50 1.42 1.55 0.17 À0.04 1.46 1.34 1.88 Scenario II All potential female labor force becomes non fullÀtime workers 2015 0.64 0.00 0.97 1.31 À0.33 0.32 0.47 1.07 2020 0.65 0.16 0.90 1.00 À0.25 0.40 0.57 1.07 2025 0.66 0.29 0.86 0.76 À0.19 0.48 0.64 1.07 2030 0.69 0.40 0.84 0.60 À0.15 0.55 0.67 1.06 2035 0.73 0.50 0.85 0.48 À0.12 0.62 0.68 1.06 2040 0.77 0.60 0.87 0.37 À0.09 0.68 0.68 1.05 2045 0.81 0.69 0.86 0.24 À0.06 0.75 0.70 1.05 2050 0.83 0.79 0.85 0.09 À0.02 0.81 0.74 1.05 86 R.R. Kato and M. Kawade

6 Discussion

As the two scenario simulations have shown, the impact of an inﬂow of female labor force who cannot work due to child care on the Japanese economy as well as the ﬁscal consolidation is very limited. In this section, in order to seek other possibilities to more stimulate the Japanese economy, the impact of several more scenarios is simulated.

6.1 More Scenarios

As Fig. 3.3 shows, there is a large gap in the wage profile between male and female workers. Thus, even when a lot of female labor supply is expanded, its impact on production and the economy is limited. This observation infers that a wage profile gap in gender might result in a limited impact of an increase in female labor supply, and the following scenario simulates the impact of reduction of such a wage profile gap between males and females8:

Scenario III (The impact of the wage profile difference in gender): In this scenario, the wage profiles of both female full-time and female non full-time workers becomes the same as those of males, and the wage profile gap in gender completely vanishes. In order to highlight the impact of reduction of the wage profile gap in gender, it is assumed that the size of female labor force does not change, so that female labor force who cannot work due to child care does still not come into the labor market, and remains outside the labor market. This scenario investigates the pure effect of the gap in the wage profile in gender on the Japanese economy and fiscal consolidation. There is an argument that an opportunity of full-time jobs for females is rather limited so that female workers are forced to be non full-time workers, even if they are willing to work as full-time workers. Then the following scenario investigates the impact of the change in the type of job contract for female workers from the non full-time to the full-time contract: Scenario IV (The impact of the change in the contract type from non full-time to full-time): The female labor force who cannot work due to child care also enters the labor market, and further the ratio between full-time and non full-time workers of females becomes the same as that of males. Thus, in this scenario there is no difference in the ratio of full-time and non full-time workers in gender. Note that in this scenario the wage profile difference in gender still remains, but a certain ratio of currently working female non full-time workers becomes full-time so that the

8 While there are several arguments regarding the reason why there is a large gap in the wage profile between male and female workers, Scenario III simulates the impact without discussing the reason. 3 Female Labor Supply, Social Security, and Fiscal Consolidation 87 difference in the ratio of full-time and non full-time between male and female workers vanishes. While the M-shaped pattern is weakened over time, the shape implies that female workers leave the labor market for a certain periods after they first enter the labor market due to child care. In the next scenario, the impact of their leave from the labor market due to child care is simulated. Scenario V (The impact of the M-shaped curve): It is assumed that all female workers of age 25–29, irrespective of their contract type of full-time or non full- time, who enter the labor market, remain in the labor market and they do not quit their job, so that the M-shaped pattern of labor supply over time completely vanishes. In this scenario, the ratio of full-time and non full-time female workers is assumed to remain the same as that of age 25–29 female workers over time until they become age 40–44. After they become age 40–44, then female labor supply become the same as the benchmark case. Furthermore, two more scenarios are simulated as mixed cases. Scenario VI (Mixture of Scenario II and Scenario III): All female labor force who cannot work due to child care enters the labor market and all of them become non full-time workers. Then the wage profiles of both full-time and non full-time female workers become the same as those of males workers, so that there is no difference in wage profiles between male and female workers. In addition to the impact of vanishing the wage profile difference in gender in Scenario III, in Scenario VI the impact of an inflow of female labor force who cannot work due to child care as non full-time is added. Scenario VII (Mixture of Scenario I and Scenario III): While in Scenario VI all female labor force who cannot work due to child care enters the labor market as non full-time workers, Scenario VII simulates the case where all of them enter the labor market as full-time workers. Otherwise, all other assumptions remain the same as Scenario VI. Thus, in this scenario, there is no difference in wage profiles between male and female workers, irrespective of contract type; full-time, or non full-time.

Impacts in Scenarios Scenario III wage proﬁle difference in gender Scenario IV full-time or non full-time in female workers Scenario V M-shaped pattern

6.2 Results

Table 3.3 shows the simulation results of all scenarios. The table shows the impacts on production, welfare, the tax revenue, and the amount of contributions to the public pension scheme. The ﬁgures in the table show relative changes from the benchmark case. Note that Scenario VI and VII are both mixed effects of Scenario III with Scenario I and II, respectively, so that Scenario VI and VII are both extended 8RR aoadM Kawade M. and Kato R.R. 88

Table 3.3 Relative change from the benchmark case (Scenario III to VI) Production Welfare Scenario III Scenario IV Scenario V Scenario VI Scenario VII Scenario IV Scenario V Scenario VI Scenario VII (%) (%) (%) (%) (%) (%) (%) (%) (%) 2010 0.00 0.00 0.00 0.00 0.00 9.97 3.26 4.74 14.36 15.94 2015 6.57 2.02 3.07 9.86 11.22 10.03 3.26 4.87 14.41 16.09 2020 7.19 2.13 3.21 10.75 12.17 10.05 3.26 4.87 14.44 16.12 2025 7.69 2.22 3.33 11.46 12.93 10.07 3.26 4.86 14.46 16.14 2030 8.05 2.32 3.48 12.00 13.55 10.09 3.26 4.85 14.47 16.14 2035 8.31 2.41 3.63 12.39 14.02 10.09 3.25 4.85 14.47 16.14 2040 8.57 2.51 3.79 12.83 14.54 10.08 3.25 4.84 14.46 16.13 2045 8.83 2.61 3.94 13.24 15.02 10.06 3.24 4.84 14.43 16.09 2050 9.06 2.68 4.06 13.61 15.45 10.03 3.23 4.82 14.38 16.04 Tax Revenue Amount of Contributions 2010 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2015 3.25 1.00 1.52 4.88 5.55 5.05 1.46 2.19 7.47 8.42 2020 4.31 1.30 1.96 6.46 7.32 5.36 1.66 2.54 8.04 9.18 2025 5.24 1.55 2.34 7.84 8.87 5.76 1.85 2.85 8.73 10.02 2030 6.02 1.78 2.68 9.01 10.20 6.29 1.98 3.04 9.51 10.89 2035 6.66 1.98 2.99 9.99 11.32 6.88 2.10 3.19 10.36 11.80 2040 7.22 2.17 3.28 10.85 12.33 7.35 2.17 3.29 11.00 12.48 2045 7.70 2.34 3.55 11.62 13.22 7.74 2.26 3.42 11.57 13.10 2050 8.11 2.49 3.79 12.26 13.98 8.06 2.37 3.57 12.07 13.67 3 Female Labor Supply, Social Security, and Fiscal Consolidation 89 scenarios of Scenario III. In comparison among Scenario III, IV, and V, the impact of Scenario III is the largest. This implies that the reduction of the wage profile gap in gender induces the most effective result in stimulation of the Japanese economy as well as the fiscal consolidation. Scenario III shows the pure effect of the reduction of the wage profile gap in gender on the currently working female workers of both full- time and non full-time contracts, so that it does not include any effect of an inflow of female labor force who cannot work due to child care. If the impact of an inflow of such female labor force is also added in, then its effect is further expanded. Scenarios VI and VII show the case when an inflow of such female labor force becomes non full-time, and full-time, respectively. Scenarios VI and VII correspond to II + III and I + III, respectively. While the impact of Scenario I was at most 1.5 % in year 2050, it becomes more than 15 % if the wage profile gap vanishes for all female workers, when female labor force who cannot work due to child care enters the labor market as full-time workers. Even if such female labor force enters the labor market as non full-time workers, its impact is more than 13 % in year 2050. As Scenario IV shows, the impact of the change in the contract type from non full-time to full-time of all female workers is small. Even if all of currently working female workers are employed as full-time workers as males worker are, its impact on production is less than 2.68 % in year 2050. The impact on the fiscal consolidation is also limited, and the tax revenue and the amount of contributions to the public pension scheme only increase by 2.49 %, and 2.37 % in year 2050, respectively. The impact of the weakened M-shaped pattern is also limited. Scenario V shows the impact. Produc- tion, the tax revenue, and the amount of contribution increase by 4.06, 3.79, and 3.57 % in year 2050, respectively. If the government tries to expand the Japanese economy by stimulating female labor force, and it also expects the fiscal consolidation by the stimulation, then it should consider the wage profile gap in gender, otherwise the impact of a further inflow of female labor force into the labor market is quite limited. On the other hand, if the wage profile gap between male and female workers becomes smaller, then impact of an inflow of female labor force who cannot work due to child care is expected to be drastically large.

Concluding Remarks This paper has presented a computable general equilibrium (CGE) framework to numerically examine the effect of expanding female labor supply on economic growth and the government revenue in aging Japan within a dynamic general equilibrium framework with multi-period overlapping generations. In particular, the potential female labor force who cannot work currently due to child care is considered. The difference in the contract condition such as full-time (Seiki) and non full-time (Hi Seiki) has also been taken into account explicitly in the simulations.

(continued) 90 R.R. Kato and M. Kawade

(continued) Several simulations have been conducted in comparison with a benchmark model, and the obtained results are as follows. First of all, even in the most stimulative case in terms of an expansion of female labor force measured in efficiency, production only increases by 1.50 % in year 2050. In this case, all potential female labor force who cannot work currently due to child care is assumed to become full-time workers. The impact of an increase in such female labor force on the Japanese economy is very limited. Another result shows that such a most stimulative expansion of female labor supply has little improvement in the government tax revenue and the total contributions to the public pension scheme as well. An increase in tax revenue through a consumption tax is at most 1.46 % in year 2050, and an increase in public pension contributions is 1.34 % in year 2050. Since the above results are obtained under the most stimulative assumption, the actual impact of stimulating such female labor force on the Japanese economy as well as fiscal consolidation is much more small. Secondly, if the government tries to expand the Japanese economy and also expects the fiscal consolidation by stimulating female labor force, then it should consider the wage profile gap in gender, otherwise the impact of a further inflow of female labor force into the labor market is quite limited. If the wage profile gap between currently working male and female labor force completely vanishes, then an inflow of female labor force as full-time workers who cannot work due to child care results in a drastic increase in production by 15.45 %. The tax revenue and the amount of contributions to the public pension scheme also increase by 13.98 % and 13.67 % in year 2050, respectively. Even though all of them enter as non full-time workers into the labor market, production, the tax revenue, and the amount of contributions will increase by 13.61, 12.26, and 12.07 %, respectively, if the wage profile gap vanishes. Thirdly, the impact of the change in the contract type of currently working female workers from non full-time to full-time on the economy as well as the fiscal consolidation is quite limited. Even though the ratio of current full-time and non full-time female workers becomes the same as that of current male workers, production, the tax revenue, and the amount of contributions only increase by 2.68, 2.49, and 2.37 %, respectively, in year 2050. Finally, the impact of the so-called M-shaped pattern is also small. If the M-shaped pattern vanishes, increases in production, the tax revenue, and the amount of contributions are 4.06, 3.79, and 3.57 %, respectively, in year 2050. The drawbacks of the paper should also be mentioned. While the impact of an increase in potential female labor force who cannot work currently due to

(continued) 3 Female Labor Supply, Social Security, and Fiscal Consolidation 91

(continued) child care has been discussed in the paper, any explicit government policy to reduce the child care cost has not been incorporated into the model. If such potential female labor force enters the labor market, the child care cost to obstruct such female labor force should be reduced. The improvement in the environments for child care such as the development of public child care facilities could reduce the child care cost, and the optimal labor supply decision by females should be affected by the government policy instruments. In the paper, such policy instruments have not been incorporated yet, and further extensions should be considered. Another drawback is that this paper did not argue reasons why there is a large wage profile gap in gender. As long as the wage profile gap remains, the impact of an inflow of female labor force is quite limited. Thus, based on the empirical evidence, the government policy to reduce such a gap should be explicitly taken into account.

Acknowledgements We thank all participants at the tenth Irvine-Japan Conference on Public Policy held at UC Irvine on 7th February 2014. In particular, we thank Jinno for his helpful comments and suggestions.

Comment Paper to Chapter 3

Masatoshi Jinno Toyo University, 5-28-20, Hakusan, Bunkyo-ku, Tokyo, Japan e-mail: [email protected] Japan’s population is both aging and shrinking in size, thus requiring compensation for the reduced size of the working population. Using the computable general equilibrium framework, this study numerically analyzes the effects of expanding the female labor force by employing women who provide childcare and therefore do not currently work outside home. This study also focuses on differences between the Seiki and Hi-Seiki labor forces. This study considers two scenarios: all potential female workers who cannot work because they are providing childcare enter the labor market as full-time (Seiki) female workers (Scenario I), or as non-full-time (Hi-Seiki) female workers (Scenario II). Even in the more stimulative case (Scenario I), production increases by only 1.55 %; the impact of Scenario II is nearly one-half that of Scenario I. Thus, it becomes apparent that the impact of an increase in female labor force on the Japanese economy would be very limited. 92 R.R. Kato and M. Kawade

Scenario I focuses on women not working outside the home because they are providing childcare, who are few in number. In focusing on the M-shaped pattern of the female labor force, this study ﬁnds that it is vanishing over time. If we were to explore the possibly positive effects of an expansion in the female labor force, what would be the value of doing a simulation, wherein women work as if they were men? (This expansion, too, is feasible in the future Japan.) Even while examining the impact of expanding the female labor force in Japan, it is unclear how an increase in the female labor force should be treated. Additional explanation should be provided for the model. The pension system in Japan consists of two tiers: (1) National Pension and (2) Employees’ Pension Insurance. It is interesting to incorporate these pension ﬁnance systems into this model to a greater extent when undertaking quantitative analysis. We would like to know to what degree an increased expansion in the female labor force would alleviate the pension burden among working generations and hence increase their utility.

References

Auerbach A, Kotlikoff LJ (1987) Dynamic fiscal policy. Cambridge University Press, Cambridge Auerbach A, Kotlikoff LJ, Skinner J (1983) The efficiency gains from dynamic tax reform. Int Econ Rev 24:81–100 Abe Y (2009) The effects of the 1.03 million yen ceiling in a dynamic labor supply model. Contemp Econ Pol 27(2):147–163 Homma M, Atoda N, Iwamoto Y, Ohtake F (1987) Nenkin: Koureika shakai to nenkin seido (Pension: Aging society and public pension scheme). In: Hamada K, Horiuchi A, Kuroda M (eds) Nihon keizai no makuro bunseki (Macroeconomic analysis of the Japanese economy). University of Tokyo Press, Tokyo, pp 149–175 Ihori T, Kato RR, Kawade M, Bessho S (2006) Public debt and economic growth in an aging Japan. In: Kaizuka K, Krueger AO (eds) Tackling Japan’s fiscal challenges: Strategies to cope with high public debt and population aging. Palgrave Macmillan, New York Ihori T, Kato RR, Kawade M, Bessho S (2011) Health insurance reform and economic growth: Simulation analysis in Japan. Jpn World Econ 23(4):227–239 Iwamoto Y, Kohara M, Saito M (2001) Setai kouseiin no tyouki ryouyou ni kiin suru keizai kousei no sonshitsu ni tsuite (Deadweight loss of households by caring long term elderly patients). Kikan Shakai Hosyo Kenkyu 36(4):547–560 Jinno M (2010) Aging of the population and employment among the young: A reconsideration of the relationship. NIRA Report, pp 72–87 Kato RR (1998) Transition to an aging Japan: Public pension, savings and capital taxation. J Jpn Int Econ 12:204–231 Kato RR (2002) Government deficit, public investment, and public capital in the transition to an aging Japan. J Jpn Int Econ 16(4):462–491 Kawade M (2007) Welfare comparison among generations on public expenditure and pension. In: Tachibanaki T (ed) Scale of government and social security. University of Tokyo Press, Tokyo, Chapter 6 (in Japanese) 3 Female Labor Supply, Social Security, and Fiscal Consolidation 93

Kawade M (2009) Shotoku kakusa wo kouryo shita zaisei saiken no sai hyouka (The revaluation of the ﬁscal reconstruction in Japan: Incorporating intra-generational income gap). Niigata University Keizai Ronsyu 87:185–204 (in Japanese) Kuroda S, Yamamoto I (2007) Hitobito ha chingin no henka ni oujite roudou kyoukyu wo donoteido kaeru noka?: Roudou kyoukyu dansei chi no gainen seiri to wagakuni no deta wo mochiita suikei (How much do people change their labor supply by responding to the change in wage?: Survey on labor supply elasticities and estimation in Japan). Kinyu Kenkyu Chapter 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems Associated with Intergovernmental Transfers and Their Political Roles

Nobuo Akai

Abstract Japanese local public finances are supported by large intergovernmental transfers from the central government. Intergovernmental transfers have two roles. The first is to guarantee fiscal standards at local levels. The second is to reduce the degree of fiscal inequality among local governments. However, both roles induce incentive problems. This chapter focuses on the incentive problems associated with intergovernmental transfers in Japan and explores how the behavior of politicians in terms of designing the transfer system affects this problem.

Keywords Incentives • Intergovernmental transfers • Soft budget constraints

1 Introduction

Japan is now facing pressure to undertake fiscal consolidation or reconstruction. To improve Japan’s fiscal situation, it is essential to reform and redesign the fiscal system by removing or reducing inefficiencies. In this chapter, we focus on inefficiencies in local public finances and the intergovernmental relationship between the central and local governments. Japanese local public finances are supported by intergovernmental transfers from the central government. Intergovernmental transfers have two roles. The first is to guarantee fiscal standards at local levels. The second is to reduce the degree of fiscal inequality among local governments. However, both roles induce an incentive problem that distorts the efficiency of the fiscal management of local governments. For example, if the revenue increase or expenditure decrease from efficient fiscal management is offset by changes in intergovernmental transfers, the incentive for efficient fiscal management is reduced. We call this the incentive problem.

N. Akai (*) Osaka School of International Public Policy, 1-31 Machikaneyama-cho, Toyonaka 560-0043, Japan e-mail: [email protected]

This chapter focuses on this incentive problem associated with intergovernmental transfers in Japan and explores how the behavior of politicians in terms of designing the transfer system affects this problem. This chapter is organized as follows. In Sect. 2, we classify intergovernmental transfers and discuss their functions. In Sect. 3, we describe the intergovernmental transfers in Japan. Sect. 4 discusses the incentive problem associated with intergovernmental transfers and explores how this problem is affected by political distortions. Finally, we conclude the chapter with Sect. 5.

2 Classiﬁcation of Intergovernmental Transfers and Their Functions

In this section, we classify the types of intergovernmental transfers and discuss their functions.

2.1 Classiﬁcation of Intergovernmental Transfers

Intergovernmental transfers can reflect various objectives, such as the funding of services at the local level, subsidization of local services, or the equalization of fiscal disparities among local regions. Rules and conditions attached to intergovernmental transfers vary widely, ranging from transfers that grant full autonomy and come close to tax sharing or equalization transfers, to transfers where the higher level of government retains tight control over their use. Bergvall et al. (2006) and Blo¨chliger and King (2006) consider the taxonomy of transfers. The transfers are mainly divided into two categories: earmarked transfers, which the local governments have to use for a specific purpose, and nonearmarked transfers, which they may spend freely. Both types of transfers can be divided further into two subcategories: mandatory transfers and discretionary transfers. Earmarked mandatory grants may be further subdivided into two categories: matching and nonmatching grants. This taxonomy is shown in Fig. 4.1 below. For actual data, see Blo¨chliger (2006, Tables 9 and 10). If a transfer is nonearmarked, this implies an expansion of local public resources because it can be spent freely. Therefore, such a transfer only has an income effect and the incentives of the regional government are not affected.1 Therefore, as long as its purpose is justified from a social perspective, this type of transfer is desirable. On the other hand, earmarked transfers are only provided when the central government wants spending in a specific area at the regional level. Earmarked transfers include both flat-rate subsidies (the amount is fixed and independent of the

1 As described in Sect. 4, the incentives of the regional government are affected by the ex post transfer even if this transfer is nonearmarked. 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 97

Fig. 4.1 A taxonomy of transfers. Source: Blo¨chliger (2006, Fig. 3: A taxonomy of grants) total amount of expenditure at the regional level) and ﬁxed-rate subsidies (the amount varies with the total amount of expenditure at the regional level.) The transfer covers part of the cost of the spending in the speciﬁc area. Therefore, the behavior of the regional government may be affected because transfer affects the incentives of the government. Before discussing the system of transfers in Japan, we consider the functions of the various transfers.

2.2 Functions of Transfers

The transfers are designed to achieve the following three goals: (1) adjustment of fiscal disparity; (2) guarantee of fiscal resources; and (3) improving efficiency of resource allocation.

2.2.1 Function 1: Adjustment of Fiscal Disparity

The ﬁrst function is the adjustment of ﬁscal disparity. This adjustment also has two objectives, which are the adjustment of vertical disparity and the adjustment of horizontal disparity.

Adjustment of Vertical Disparity

The upper level of government has a comparative advantage in terms of the efﬁciency of taxation or public funding because of the interregional mobility of tax resources, heterogeneous allocation of tax resources, and economics of scale in taxation. Therefore, it is efﬁcient when the upper level of government levies the tax. Furthermore, the central government levies taxes, the amount of which exceeds its own expenditure. However, the regional government spends more than its own 98 N. Akai revenue. This gap between regional revenue and regional expenditure is the vertical disparity. To fund this gap, a transfer from the central government is made.

Adjustment of Horizontal Disparity

From the perspective of ﬁnancial horizontal equity, it is unreasonable that the geographic area of a local government determines: (1) the tax burden associated with a certain level of public services or (2) the level of public services for a certain tax burden. Horizontal equity holds when residents in a local government area are treated the same when there are no public services or tax burden as when there is. A horizontal disparity is created when tax revenues are distributed unevenly. To decrease this disparity, transfers are made from rich regions to poor regions. Theoretically, this should occur using a lump sum transfer, which has no effect on regional incentives associated with public management.

2.2.2 Function 2: Guarantee of Fiscal Resources

The second function of fiscal transfers is to guarantee fiscal resources. From the perspective of equity, the central government is required to guarantee a minimum standard of living (often called the national minimum) and has a responsibility to secure the fiscal resources to achieve that minimum level. To achieve this, the fiscal transfers must reflect the cost of providing public services in each region, which differs depending on the degree of cost efficiency. In this case, a nonearmarked block grant is desirable because the responsibility given to the central government is to guarantee fiscal resources and the regional government should decide how to allocate resources in order to maximize regional utility. Therefore, this function should be achieved by a lump sum transfer, which has no effect on regional incentives regarding public management.

2.2.3 Function 3: Improving Efﬁciency of Resource Allocation

The third function of fiscal transfers is to improve the efficiency of resource allocation. Public infrastructure such as airports or highways has positive spillover effects across regions. This externality creates inefficiency in the sense that the level of public services provided is too small. In this case, the fiscal transfers are designed to enhance public services, by reducing the cost of provision. An earmarked matching transfer is desirable for achieving this goal.2 As a result, this

2 As described in Sect. 4, when the transfer is designed ex post, the transfer has incentive effects. Therefore, nonearmarked transfers ex post may be useful for removing inefficiency if there exists inefficiency ex ante. See Caplan et al. (2000) and Ko¨thenbu¨rger (2002). 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 99 type of fiscal transfer improves resource allocation by internalizing the externality across regions. It is worth considering how the earmarked matching transfer is different from a lump sum transfer. After allocating a lump sum transfer, the incentives of the regional government are unbiased and this transfer only creates income effects, whereas the earmarked matching transfer affects the incentives of the regional government to increase the provision of the public services producing the positive externality.

3 Intergovernmental Transfers in Japan

3.1 Intergovernmental Transfers and the Present Functions

To realize these three functions, the Japanese system has two types of transfers: (1) national treasury disbursements and (2) local allocation tax grants.3 While national treasury disbursements, which involve earmarked grants, have incentive effects for the specific field of expenditure, local allocation tax grants are generally regarded as nonearmarked general-purpose grants. Theoretically, it is expected that while national treasury disbursements achieve the third function (improving efficiency of resource allocation), local allocation tax grants realize the first and second functions. However, in the real world, local allocation tax grants have similar effects to earmarked grants. In other words, the incentives of the regional government are affected dynamically by local allocation tax grants. This is because the calculation of the total amount of transfers depends on the past behavior of the regional government. In addition, national treasury disbursement is also used to realize the first and the second functions. Therefore, the three functions described can be achieved by various types of transfers. In fact, the second function (guarantee of fiscal resources) is realized by not only earmarked grants, but also nonearmarked general-purpose grants. If each transfer is designed for a specific function and the regional incentives for public management are adjusted appropriately, there is no need to discuss transfer reform in Japan. However, the current types of transfers in Japan, as described above, do not necessarily achieve these three functions and may harm equity and efficiency.

3 The total amount of local expenditure in ﬁscal year 2013 in Japan was 100 trillion yen, of which 16 % was national treasury disbursement and 18.7 % was local allocation tax grants. 100 N. Akai

3.2 Local Allocation Tax Grants

Before discussing inefficiency, it is useful for describing local allocation tax grants in Japan. The amount of local allocation tax grants allocated to each regional government is calculated according to the gap between (1) the standard fiscal needs and (2) the standard fiscal revenues. Standard fiscal needs are defined as the total expenditure level required to satisfy the fiscal needs that enable residents to achieve a minimum standard of living. Standard fiscal revenues are simply defined as the total revenue collected by the regional government. If standard fiscal revenues exceed standard fiscal needs, transfers are zero. Reverse transfer from the regional government to the central government never occurs even if the standard fiscal revenues exceed the standard fiscal needs. A region with zero transfer is called “Fu-koufu-dantai” (No-transfer region). The capital city, Tokyo, is one example of such a region.

3.2.1 Calculation of Standard Fiscal Needs

Standard Fiscal Needs for Each Area of the Public Service

i Standard ﬁscal needs for the public service in ﬁeld j in region i (Nj) are calculated as

i Nj ¼ ðÞA unitcost Â ðÞB unitof measurement Â ðÞC parameter for correction:

First, the unit cost is the unit price per measurement unit, which is the same nationwide. However, the unit cost is amended each year, depending on the standard payrolls of local government employees and the consumer price index. Second, the unit of measurement is the unit used to measure the ﬁnancial needs of each administrative item. For example, in the case of needs of a road, the area of the road is adopted as the unit of measurement. As for measuring the social welfare needs, the regional population is adopted as the unit of measurement. Technically its unit cost is calculated by establishing a standard hypothetical local government and then dividing the total expenditure of this hypothetical government by the unit of measurement. The parameter for correction is set in order to adjust for different circumstances in each region. For example, the climate and population of the region are taken into account during the adjustment. The standard ﬁscal needs in each region can be measured by setting the parameter for correction, which enables each local government to supply public services that guarantee a minimum acceptable standard of living. 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 101

Standard Fiscal Needs in Each Region

The standard financial needs in the region i (Ni) are calculated as the sum of the standard financial needs in each field. This is the amount guaranteed by the central government.

3.2.2 Calculation of Standard Fiscal Revenue

Standard ﬁscal revenue Ri is calculated as 75 % of the total regional tax revenue evaluated at the standard tax rate established by the central government. The reason why total revenue is not considered in the calculation and 25 % is removed from the revenue is to retain an incentive to increase revenue.4

3.2.3 Calculation of Local Allocation Tax Grants

Local allocation tax grants (Ei) to region i are calculated as ÂÃ Ei ¼ Max 0, Ni À ðÞ1 À 0:25 Ri :

If Ni À (1 À 0.25)Ri < 0, then local allocation tax grants equal zero. That is, a transfer from the rich region to the central government is not allowed in the present system; therefore, the power of the central government to equalize resources at the regional level is limited in the sense that the central government has no method of adjusting resources in the super-rich region.

4 Incentive Problem Associated with Intergovernmental Transfers in Japan

In the previous section, we presented an overview of Japan’s local allocation tax grants system, which involves contracts between the central and local governments for the transfer of public resources. In this section, we consider how such contracts affect the incentives of local governments. As described in the previous section, the local allocation tax grants are calculated as the gap between revenue and expenditure, given the parameter and the unit of measurement. The local government behaves optimally, following this grant system. If the amount of the transfer does not affect the behavior of the local

4 In addition, the incentive is incorporated in the calculation of the standard fiscal revenue. See Sect. 4.1.2. 102 N. Akai government, then the local government makes effort at the social optimal level; that is, its behavior is not biased. If the amount of the transfer changes with behavior, then the behavior is affected. In the actual system, the standard fiscal expenditure (including its components, the cost parameter, and the unit of measurement) and the standard fiscal revenue are adjusted according to local public finances and the regional economy. From the viewpoint of local fiscal needs, the standard fiscal expenditure and its cost parameter will be adjusted when the cost of providing public services in the region increases. From the viewpoint of the regional economy, the standard fiscal revenue will be adjusted when the revenue collected from the regional economy decreases. However, the expenditure or cost necessary for the supply of public services in each region may be able to be reduced by making effort and tax revenue amounts might also be increased by making effort. In this situation, the local government might decrease its efforts to reduce expenditure or increase revenue, because the shortage created by the expenditure increase or revenue decrease is covered by an adjustment of the local allocation tax grants. This means that, when the effort made ex ante affects the transfer amount ex post, the incentive for making the effort, which is desirable from the social viewpoint, is reduced. In the framework of contract theory, local allocation tax grants, which are transferred from the principal (the central government) to the agent (the local government), can be understood as a compensation arrangement for the agent in a principal–agent model. What is important in considering the incentive problem behind the allocation tax system is to distinguish between problems arising from what is included in the contract and problems arising from what is not factored into the contract. The former (problems under complete contracts) arises from the inadequacy of the design of the contract system; the latter (problems under incomplete contracts) arises from the limits of the contract. In the following, we discuss separately (1) the incentive problem under complete contracts based on the fact that the contract is not designed optimally in advance and (2) the incentive problem under incomplete contracts based on the fact that the contract cannot force commitment to a detailed agreement in advance.

4.1 Incentive Problem Under Complete Contracts: Moral Hazard

4.1.1 Incentive Effects of Intergovernmental Transfers

When the effort to decrease public expenditure and/or increase tax revenue is measured perfectly, the incentive problem is solved by setting the contract optimally. However, in the real world, it is difficult to measure regional effort and output excluding the risk factor because there exists the problem of asymmetric 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 103 information between the regional government and the central government. There- fore, the system of grants must be designed based on asymmetric information. Incentives cannot be controlled perfectly under information asymmetry, even if it is possible to set a complete contract. Local allocation tax grants are distributed according to the difference between standard financial revenues and standard financial needs. If the amount of local allocation tax grants is adjusted fully for changes in the tax revenue collected in the region, the incentive to collect tax is reduced.5 This corresponds to inefficiencies that have been brought by hidden action or private information created ex post; we call this “the moral hazard problem.” For the case of Japan, Tajika and Miyazaki (2008) empirically derive the result that the incentive for increasing tax revenue is distorted by the higher dependency on local allocation tax grants.

4.1.2 Incentive Design of the Intergovernmental Transfer System in Japan

The present system of local allocation tax grants has been designed toward elim- inating or reducing possible inefficiencies (incentive problems) resulting from asymmetric information, by removing the effect of the effort from the standard fiscal revenue. In other words, as described below, the system is designed such that the revenue created by the effort is not reflected directly. The first consideration for the incentive problem is to allow to reserve the part of the regional financial resources, tax revenues collected in the region. In the existing grants system, only 75 % of the actual tax revenue is used to determine the basic financial revenue. This percentage has been designed to maintain tax collection incentives and is called the “reserve funding ratio.” If the reserve funding ratio is set equal to 0, then the amount equal to tax revenue increase decreases from the grants, which reduces tax collection incentives. However, if the reserve funding ratio is set at 25 %, then 25 % of the tax revenue increase associated with the efforts of local governments is retained by the local government, and this institutional design partly removes disincentives for tax collection. The second consideration for the incentive problem is seen in the calculation of the tax revenue. The tax revenues, as the basis for calculating the standard fiscal revenue, are obtained by multiplying the collection rate and tax rate by the tax base. First, to calculate the tax revenue, we use the standard tax rate, not the actual tax rate, in order to remove the disincentive to collect additional tax revenue by setting the actual tax rate beyond the standard tax rate. In addition, the average collection rate across Japan is used. That is, the grants system is designed such as to maximize regional tax revenue in the sense of setting of local tax rate and local collection rate. Ishida (2014) carefully examines this behavior in Japan.

5 As described in Sect. 4.1.2, some factors addressing the disincentive effect of the grants have already been incorporated in Japan. 104 N. Akai

In addition, in the grants system the revenue from nonstatutory taxes (“hotel tax” and “recreational tax”) is not included in standard ﬁscal revenue. Tax collection effort has also been secured even in this aspect. This is also one way of maintaining tax collection incentives by separating the incentive effect from the tax collection effort for setting the additional taxes in the calculation of the standard ﬁscal revenue.

4.1.3 The Institutional Design of Grants and Political Aspects

Policy Implications

As described above, the inefficiency arises from the asymmetric information about the regional effort between the central government (principal) and the local government (agent). One of the methods of mitigating the problem is to reduce the asymmetric information between the central government (principal) and the local government (agent). For example, a transparent system for identifying the regional effort might be effective. Another method is to increase the reserve funding ratio. Then the revenue increase or expenditure saving by making efforts remains in the region, which increases the incentive for efficient fiscal management. However, the higher reserve funding ratio is not necessarily desirable when considering equity. As the ratio increases, the power to adjust interregional inequality among regions decreases. The institutional design of grants faces the problem of the efficiency– equity trade-off.

Political Aspects of the Incentive Problem

The political effect is not incorporated in the model described above. In other words, the inefficiency arises even if there do not exist any political distortions. Let us consider how the behavior of politicians affects this inefficiency. As described above, reducing the asymmetric information between the central government (principal) and the local government (agent) is effective in removing incentive problems and improving efficiency. If politicians behave benevolently from the social viewpoint, then rules regarding transparency and information disclosure will be introduced by politicians because social welfare increases under such a rule. However, if the politicians are captured in any regions, then they are reluctant to introduce this rule because the existence of information asymmetry is desirable for the regions and the politicians get reward from this region. It is worth noting that the latter political inefficiency relates to the number of regions. As mentioned before, when there is only one region, regional welfare does not increase by inefficient decision making ex ante because all additional costs created by inefficient decisions must be covered by that region; therefore, the residents do not capture the politicians. However, this means that as the number 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 105 of regions increases, the advantage to capture the politicians for the regional residents becomes large. Then, inefficiency becomes large through the political distortion.

4.2 Incentive Problem Under Incomplete Contract: Soft Budget (Bailout)

In the previous subsection, we examined how the grants system affects incentives under a contract set ex ante. However, there exists another incentive problem, namely incomplete contract, which arises from the fact that the central government cannot commit to a detailed agreement in the contract in advance. In the following, we consider this problem. When the subsidy policy is discretionary ex post, the central government will have an incentive to optimize the policy again at the time of the occurrence of the problem. In fact, the realized cost is regarded as a necessary cost and the grants guarantee the necessary cost ex post. However, the desirable policy ex post may not necessarily match the desirable policy ex ante. Therefore, there is a possibility that the desirable policy ex ante becomes inefﬁcient from the ex post point of view. This is called “time inconsistency.” As for the design of the standard ﬁscal needs, each component (unit cost, measurement unit, and correction parameter) is discretionary and set based on the cost actually realized, measurement unit the degree of importance of which is politically reported, correction parameter calculated by the cost actually realized. In this situation, the incentive problem might occur, as described below.

4.2.1 Soft Budget Constraint

A soft budget constraint is a concept that was originally used to characterize the financial relationship between the government and public enterprises in a socialist economy. This concept is pioneered by Kornai (1986). More specifically, it involves government bailouts of public companies that have fallen into financial difficulty. This concept is theoretically formulated by Dewatripont and Maskin (1995) and Qian and Roland (1998). In the context of local government finance, this is a situation where the central government bails out a local government for which expenditure exceeds revenue in its region ex post. Various reasons why the central government decides to bailout the local government are considered. Especially, the essential problem is the lack of commitment of the central government. When a local government faces fiscal problems, the central government has an incentive to bail out the local government and to make an intergovernmental transfer. The budget constraint of the local government becomes soft as a result 106 N. Akai of the pursuit of efficiency and fairness. This situation is referred to as a “soft budget constraint,” while the case where the budget does not change is referred to as a “hard budget constraint.” In this case, because a system to compensate the resources of the local government ex post is established, the incentive for the local government to make an effort is biased; thus, efficient fiscal management is not induced. This incentive problem due to the lack of commitment is referred to as the “soft budget constraint problem.” The criteria of policy desirability will vary from the ex ante stage to the ex post stage where the incentives of the local government are eliminated. Even though the central government acts normatively, efficiency (or fairness) is not matched ex ante and ex post. The fact that the government has the discretion of the subsidy policy makes it possible for the central government to bail out the local government through ex post optimization. The basic framework of the soft budget constraint problem can be explained using the following principal–agent model. The central government as the principal maximizes social welfare and the local government as the agent maximizes regional welfare. Assume the following fiscal game between the principal and the agent. The agent is the leader and the principal is the follower. The agent decides whether it makes an effort (to minimize the fiscal cost) or not ex ante. For simplicity, the effort level is selected from two extreme levels, namely 0 or 1. After that, the fiscal situation is realized ex post, depending on the effort level. If the local government makes an effort, the fiscal situation improves, whereas the fiscal situation deteriorates with no effort. If the central government does not respond to the deteriorated situation, the agent may go into bankruptcy. In this situation, ex post efficiency and ex post equity may force the central government to bail out the local government facing the deficit, namely softening the budget constraint by making the additional transfer become the ex post desirable policy. For example, for the local government in financial difficulty, ensuring the provision of the financial resources to guarantee the national minimum level may be justified from the viewpoint of interregional equity and the transfer will be socially acceptable ex post. Alternatively, bailouts of public enterprises may be desirable for employment. The incentive problem of the soft budget arises when the local government expects and incorporates this relief into their behavior. This can be described in Fig. 4.2 below. When making effort, e ¼ 1, the fiscal situation improves and thus the central government has higher welfare in the case without a bailout than with a bailout, because the central government can save social resources. That is, we have P1 < P1 + α1, where P1 is the benefit of the central government and α1 is the social benefit from saving social resources. Now the local government has regional welfare of A1. However, when making no effort, e ¼ 0, the fiscal situation deteriorates. In this situation, the social benefit may increase ex post after being bailed out and the central government has an incentive to bail out the local government ex post. That is, we have P0 + α0 > P0 where P0 is the benefit of the central government and α0 is the social benefit associated with a bailout. Now the local government has regional welfare, A0 + β0, where β0 is the social benefit of the local government associated with being bailed out. 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 107

Benefit (Principal, Agent)

Bailout P1 , A1 b 1

Principal e = 1 (Make effort) P1 + a1 A1 No bailout Agent Bailout 0 P0 + a 0 A 0 e = 0 (No effort) Principal

P0 A0 No bailout

Fig. 4.2 Ex post bailout and inefﬁciency

Given this situation, let us consider whether the local government makes an effort or not. Compare the regional welfare in two cases with and without making an effort. If the local government does not make an effort when A1 < A0 + β0, then the situation where the central government bails out the local government ex post becomes the subgame perfect equilibrium. Now A0 and A1 correspond to the benefits in the case with and without making effort. A1 > A0 might be assumed to hold in general. Therefore, when the additional social benefit for the local government bailed out ex post is high enough so as to satisfy 0 < A1 À A0 < β0, the soft budget constraint problem arises. The soft budget constraint means that the local government decides not to make any effort toward efficient management because of an expected bailout from the central government ex post. As the local government makes this decision ex ante, this decision is optimal. However, this decision may not be socially optimal for the following two reasons. The first is the higher opportunity cost of a bailout from an ex ante perspective and the second is the negative fiscal externality across multiple regions. The opportunity cost here equals the social benefit from saving the resources from the additional transfer in the case of making an effort. The negative fiscal externality means that the opportunity cost above is covered by all regions. If the bailout cost is self-financed by the region being bailed out, all costs are inter- nalized within the region and the region will make an effort as long as the opportunity cost is higher. As a result, the socially optimal level of effort will be made by the region. This means that the existence of multiple regions is one reason for this inefficiency. Needless to say, if the social benefit from bailing out ex post is higher than the opportunity cost, making no effort ex ante and bailing out ex post become socially optimal. Then, discretionary decision making becomes desirable, compared with the nondiscretionary rule. The trade-off between these two rules is analyzed by Besfamille and Lockwood (2008) and Sanguinetti and Tommasi (2004). 108 N. Akai

4.2.2 The Soft Budget Constraint Problem Behind the Calculation of Standard Fiscal Needs in Japan

As described above, the soft budget constraint problem means that the incentive to make an effort is biased by the ex post discretionary bailout policy. This problem is created by the lack of commitment of the central government for the policy. As shown in Sect. 3, the present local allocation tax grants to each region in Japan are objectively calculated by a fixed formula. However, the budget of the local government becomes soft as long as the central government ratifies the choice made by the region ex ante and designs the policy ex post. There exist many varieties of measurement units, and both the unit cost and parameter for correction are set in a discretionary manner corresponding to the actual needs and expenditures in each region. Although unit costs are uniform across regions, local allocation tax grants tend to be designed such that the regions facing a deficit are bailed out. First, let us consider the correction parameter. Local allocation tax grants have been allocated to local governments, where the costs of providing public services are higher, by adjusting the correction parameter to reflect regional characteristics. However, because regional characteristics are realized ex post, it is impossible to identify whether those relatively expensive costs are created by the special nature of the area or a reflection of inefficient fiscal management with less effort. The correction parameter is introduced in order to reflect expenditures based on characteristics such as social and natural conditions in each region in the standard fiscal needs, but the objective (scientific) criteria are not incorporated in the process of the determination of the objective facts. Second, let us consider the unit cost. The unit cost is modified depending on the economic conditions and the consumer price level, but its modification is not objective and policy makers have discretion in the calculation of the unit cost. It is not clear whether the increase in the unit cost is induced by inefficient fiscal management or exogenous economic factors such as an increase in consumer prices. Thus, we cannot deny the possibility that the central government ratifies the financial situation of local governments by adjusting the correction parameter or unit cost. Even if providing an additional transfer to the local government is ex post desirable from the national perspective at the stage of the financial deterioration of a local government, the local government, when expecting in advance such ex post assistance, no longer has an incentive to perform efficiently (at minimum cost). In addition to harming the fiscal discipline of local government, the ex post additional transfer will distort incentives for residents to monitor the management of the local government and characterize its discipline. If the additional costs created by inefficient fiscal management are paid for by residents (voters), the local residents will judge the local government by “voting with their feet” and as a result, the incentive of the local government toward cost minimization might increase. How- ever, if the transfer is discretionary, the incentives of residents will be distorted because they expect that the additional cost will be covered nationally. Then, an 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 109 inefficient higher cost of satisfying standard fiscal needs is realized ex post as a result of inefficient decisions ex ante. This leads to a larger required transfer. The soft budget constraint problem has been tested empirically by some studies. The pioneering work is Yamashita et al. (2002), which derives the result that a higher expenditure level is induced by a higher dependency on local allocation tax grants. Miyazaki (2010) clarifies how the budget is softened in the calculation of standard fiscal needs. The soft budget constraint problem then arises as an equilibrium in the dynamic game of the grant system between the central government and the local governments. Ex post discretion of the central government allows it to increase the supply of local allocation tax grants and the ex ante decision of the local government allows it to increase the demand for local allocation tax grants. The higher level of tax grants is realized as an equilibrium that matches supply with demand, which results in fiscal deterioration. The guarantee of fiscal resources by the central government means that the higher level of tax grants must be financed by the central government. However, there is a limit to the amount of money the central government can finance because it could face a financial crisis. As a result, the local governments rely on their own debt. This debt increases rapidly every year. The higher level of accumulated local bonds issued is realized as an equilibrium because of the moral hazard of the lenders such as bankers, based on the guarantee of local fiscal resources by the central government. Indeed, the small city of Yubari located in the Hokkaido region was fiscally bankrupted in 2007 because of excessive issuance of bonds. Banks lent large amounts of money to this small city. This seems to have been caused by moral hazard based on the guarantee of local fiscal resources by the central government.

4.2.3 The Institutional Design of Grants and Political Aspects

Policy Implications

As described above, the inefﬁciency (incentive problem) arises from the lack of commitment in the contract between the central government (principal) and the local government (agent), and the ex ante incentive of the local government. One way of mitigating the problem is to limit discretion in the ex post policy by the central government and correct the distortion of the incentive in the ex ante decision of the local government. One way to do this is to set the rule clearly and objectively. In addition, improvement of governance by setting the rule clearly contributes to an improvement of the efﬁciency of the bond issuance by mitigating the moral hazard of the lenders such as banks. 110 N. Akai

Political Aspects of Incentive Problems

The political effect is not incorporated in the model described above. In other words, the inefficiency arises even if there never existed any political distortions. Let us consider how the behavior of politicians affects this inefficiency. As described above, setting clear and objective rules is effective in removing incentive problems. If politicians behave benevolently, then a clear rule to limit discretion and correct distortions should be introduced by politicians to increase social welfare. However, if politicians are captured in any regions, then the politicians will be reluctant to introduce any limitations because discretion ex post is desirable for the regions that have been bailed out and the politicians receive reward when a bailout occurs. It is worth noting that the latter political inefficiency relates to the number of regions. As mentioned before, when there is only one region, the regional welfare does not increase by inefficient decision making ex ante because all additional costs created by inefficient decisions must be covered by that region; therefore, the residents do not capture the politicians. However, on the other hand, this means that as the number of regions becomes large, the incentive for the regional residents to capture the politicians becomes high because most of additional costs must be covered by other regions. Then, the inefficiency becomes large through the political distortion.

Conclusion In this chapter, the incentive problems behind intergovernmental transfers in Japan, especially local allocation tax grants, were examined. This chapter clariﬁed that information asymmetry and the lack of commitment devices in the contract by the central government are key reasons for this incentive problem associated with intergovernmental transfers in Japan. As described above, these may be mitigated or not, depending on the purpose of the politicians, which are either benevolent or leviathan. While we focused on the transfer system between governments in this chapter, incentive problems have occurred in various public systems in Japan and these problems are closely related to the behavior of politicians. Future research should explore how these problems are affected by the behavior of politicians in terms of the design of transfer systems.

Comment Paper to Chapter 4

Takeshi Miyazaki Kyushu University, 6-10-1 Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan e-mail: [email protected] This study discusses Japanese local public finance, focusing on the intergovernmental transfer system known as “Local Allocation Tax” (LAT), and its incentive 4 Fiscal Consolidation and Local Public Finances in Japan: Incentive Problems... 111 problems and political roles. With respect to incentive problems, the study closely analyzes the calculation formula of LAT. In the formula, it is argued that the retention rate of “Standard Fiscal Revenues” (SFR) is 25 %, meaning it is equivalent to a very high 75 % marginal tax rate on taxable revenue. Moreover, problems such as discretionary adjustments in the components of “standard fiscal needs” (SFN) and SFR based on the economic situation, the existence of information asymmetry related to fiscal effort, and the effect of “soft budget constraints” (SBCs) are said to induce local governments to lower their fiscal efforts to raise tax revenue or to increase their tax base. The main comments on this study follow. Regarding the function of LAT, although the author did not discuss this in any depth, it would be better to point out that some recent theoretical works have found that fiscal equalizing transfer can contribute to attaining economic efficiency as well as equity, in the presence of competition with mobile tax base (e.g., Ko¨thenbu¨rger 2002; Smart 1998). More- over, to strengthen the argument of this study, another view to the one stated above concerning the effects of LAT on fiscal effort may also be discussed. That is, some researchers have stated that the LAT formula for the calculation of SFN is objective, and is measured on the basis of objective indices (e.g., Mochida 2006). Also, it is shown in this chapter that if the central government can commit to “no bailout”, then local government will choose to make fiscal efforts as its best course of action, and a socially optimal equilibrium can be achieved. As is well known, this form of equilibrium is reached under “hard budget constraints” (HBCs), so discussing HBCs will help readers understand the argument. Although there are several articles that have attempted to demonstrate the function of LAT, many of them can only explain its framework and basic statistics. In contrast, this study aims at revealing what outcome LAT will yield on the basis of up-to-date economic theory. As a result, it could become one of the more useful studies to which later researchers who intend to learn about the Japanese intergovernmental transfer system can refer.

References

Bergvall D, Charbit C, Kraan DJ, Merk O (2006) Intergovernmental transfers and decentralized public spending. OECD Journal on Budgeting 5:111–158 Besfamille M, Lockwood B (2008) Bailouts in federations: is a hard budget constraint always best? Int Econ Rev 49:577–593 Blo¨chliger H, King D (2006) Fiscal autonomy of sub-central governments. OECD Working Paper Caplan A, Cornes R, Silva E (2000) Pure public goods and income redistribution in a federation with decentralized leadership and imperfect labor mobility. J Public Econ 77:265–284 Dewatripont M, Maskin E (1995) Credit and efﬁciency in centralized and decentralized economics. Rev Econ Stud 62:541–555 Ishida M (2014) Chihou-koufu-zei ga tyousyuuritsu ni ataeru kouka no suitei – gyoukaku insenthibu santei no kouka to koufuzeiseido ni naizaisuru yugami no kensyou (Estimating the effect of local allocation tax grants on local tax collection rates: verifying the effect of the 112 N. Akai

administrative reform incentive assessment and the distortion underlying local allocation tax grants). Keizai Bunseki 188:22–43 Kornai J (1986) The soft budget constraint. Kyklos 39:3–30 Ko¨thenbu¨rger M (2002) Tax competition and fiscal equalization. Int Tax Public Finance 9:391–408 Miyazaki T (2010) Chihou-koufu-zei ni okeru sofutona yosanseiyaku no kensyou – keijyou keihi ni okeru hosei keisuu no kettei (Verification of soft budget constraints in the local allocation tax grants – determination of the parameter for correction in operating expenses). Keizai Bunseki 183 Mochida N (2006) Local government organization and finance: Japan. In: Shah A (ed). Public expenditure analysis, World Bank, Washington DC Qian Y, Roland G (1998) Federalism and the soft budget constraint. Am Econ Rev 88:1143–1162 Sanguinetti P, Tommasi T (2004) Intergovernmental transfers and fiscal behavior insurance versus aggregate discipline. J Int Econ 62:149–170 Smart M (1998) Taxation and deadweight loss in a system of intergovernmental transfers. Canad J Econ 31:189–206 Tajika E, Miyazaki T (2008) Chihou-koufu-zei to chihou jichitai no zaisei kaizen doryoku – zenkoku sichouson de-ta ni yoru bunseki (Financial improvement efforts of local government and local allocation tax grants – analysis by municipalities data). Kaikei Kensa Kenkyu 38 Yamashita K, Akai N, Sato M (2002) Chihou-koufu-zei ni hisomu insentyibu kouka (Incentive effect lurking in local allocation tax grants). Fyinansyaru Rebyu 61:120–145 Chapter 5 Public Policy and Economic Growth in the Integrating Japanese Economy

Hiroki Kondo

Abstract In an economy where goods and factors markets are progressively integrating and modern industries with scale economies and externalities are prevailing, economic activities tend to concentrate in a limited number of regions. This paper considers the outcomes of ﬁscal competition, competition in tax and subsidies and competition in public infrastructure provision among regions in this process. There is no pure strategy Nash equilibrium. A region’s choice of capital subsidy or public infrastructure provision is not deterministic. We therefore consider a mixed strategy Nash equilibrium. Fiscal competition tends to be ﬁercer in this equilibrium. Concentration of economic activities causes serious and persistent income gaps among regions. As a consequence, regions are impelled make more aggressive efforts to attract modern industries within their borders. Yet the industries the regions seek to introduce will settle in only a limited number of regions. Many regions fail in introducing an industry after making tremendous investments in public infrastructure. As a consequence, huge amounts of public infrastructure remain are left unused in many regions, incurring huge welfare losses.

Keywords Fiscal competition • Industrial agglomeration • New economic geography

1 Introduction

High economic growth has been accompanied by drastic restructuring in interregional specialization patterns and resource allocations. The integration of goods and factor markets has promoted economic growth and the restructuring of interregional specialization. In this process, modern industries tend to concentrate in limited areas due to characteristics such as plant-level scale economies,

H. Kondo (*) Department of Economics, Sophia University, 7-1 Kioicho Chiyoda-ku, Tokyo 102-8554, Japan e-mail: [email protected]

© Springer Japan 2015 113 T. Ihori, K. Terai (eds.), The Political Economy of Fiscal Consolidation in Japan, Advances in Japanese Business and Economics 8, DOI 10.1007/978-4-431-55127-0_5 114 H. Kondo local-in-scope technological and knowledge externalities, and significant vertical input-output linkages, as shown in Krugman (1991) and Ellison and Glaeser (1997).1, 2 This process causes serious and persistent income gaps among countries and regions. As a consequence, countries and regions are impelled make more aggressive efforts to attract modern industries within their borders. One means they use to do so is to drastically reduce tax rates on capital. Another means is to provide public infrastructures such as transportation infrastructures and industrial parks to attract modern industries. In postwar Japan, the integration of goods and factor markets proceeded rapidly. From the early or mid 1970s, rural regions of Japan rapidly increased investment in public infrastructure (Fig. 5.1). Yet modern industries and population have concentrated in a limited number of regions of the country. Some regions managed to introduce clusters of modern industry, whereas many others failed to introduce or develop industry in spite of tremendous investments in public infrastructures.3 This paper considers the outcomes of fiscal competition, competition in tax and subsidies and competition in public infrastructure provision among regions in an economy where the progressive integration of goods and factor markets is concen- trating modern industries into a single region or a limited number of regions. Our framework introduces the vertical linkage of modern industry in the manner proposed in the new economic geography models of Krugman and Venables (1995) and Venables (1996). We show that there is no pure strategy Nash equilibrium in an economy where regions are competing to attract modern industries. A region’s choice of capital subsidy or public infrastructure provision is not deterministic. We therefore consider a mixed strategy Nash equilibrium. Fiscal competition tends to be fiercer in

1 Krugman (1991) and Kim (1995) investigated regional agglomerations of manufacturing industries in the U.S. computing the Gini coefficient of spatial inequality. Ellison and Glaeser (1997) exploited an index which can measure spatial agglomeration distinguishing it from industrial concentration due to plant level scale economies and applied it to the case of the U.S. manufacturing industry. However, the results depend on the choice of the size and shape of the spatial unit. To overcome this drawback, Duranton and Overman (2005) measured the distances between all the pairs of the plants in the same industry and observed the distribution of the distances to investigate the case of the U.K. manufacturing industry. 2 Holl (2004a,b) and Chandra and Thompson (2000) show that the improvements in transportation facilities cause the concentrations of economic activities. 3 Public infrastructures such as roads, railways, and industrial parks play a significant role in the introduction and development of modern industries. Yoshino and Nakajima (1999) examine the productivities of disaggregated public infrastructures and show that production infrastructures such as roads have significant impacts on the productivity of the secondary sector of industry. Yoshino and Nakano (1994, 1996) show that public infrastructures have higher in urban areas where modern industries are based. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 115

Fig. 5.1 Rations of administrative investment in Metropolitan area and Rural area to the amount in total Note: Metropolitan area includes Ibaraki, Tochigi, Gunma, Yamanashi, Nagano, Saitama, Chiba, Tokyo, Kanagawa, Gifu, Shizuoka, Aichi, Mie, Shiga, Kyoto, Nara, Osaka, Hyogo, and Wakayama prefectures. Rural area includes Hokkaido, Aomori, Iwate, Miyagi, Akita, Yamagata, Fukushima, Niigata, Toyama, Ishikawa, Fukui, Tottori, Shimane, Okayama, Hiroshima, Yamaguchi, Tokushima, Kagawa, Ehime, Kochi, Fukuoka, Saga, Nagasaki, Oita, Kumamoto, Miyazaki, Kagoshima, and Okinawa prefectures. From “administrative investment” by Ministry of Internal Affairs and Communications this equilibrium. The amount of public infrastructure provision tends to far exceed the optimal level. Modern industries may even leave regions that are richly endowed with public infrastructures. When they do, extensive public infrastructures developed are left unused. According to traditional studies into tax competition (often referred to as ‘basic tax competition models’) such as those by Wilson (1986), Zodrow and Mieszkowski (1986), and Wildasin (1988), the tax rate on a mobile factor such as capital tends to be lower than optimal. Most studies have employed a single-sector model of a standard neoclassical, non-increasing-return-to-scale production function. In this framework, the allocation of a mobile factor changes elastically and continuously with the gap in the tax rate among countries. Countries or regions compete to attract larger amounts of capital by lowering their capital tax rates, which renders the tax rate in equilibrium very low. Recent studies into tax competition employ the frameworks of new economic geography and find tax competition to be less intensive. Kind et al. (2000), Ludema and Wooton (2000), Baldwin and Krugman (2004), Borck and Pfluger (2006), Kondo (2009) and Kondo (2013) employ a new economic geography model with 116 H. Kondo two countries and two industries, one with scale effects.4, 5 As integration proceeds, the industry with scale effects concentrates in one of the two countries. In contrast to the case seen in a basic tax competition model, the country hosting industrial agglomeration can raise its capital tax rate up to a level where the tax cost equals the agglomeration benefit without risking a capital outflow. In equilibrium, the industrialized country sets higher capital tax rates than the less industrialized country.6 Yet many studies focus on the case where an industry with scale effects is concentrated in one country and a one-sided pure strategy Nash equilibrium is reached.

2 The Model

Consider an economy which consists of two regions, 1 and 2, manufacturing industry sector and agricultural sector. In each region, unit measure of households inhabits which are immobile between regions. Also in each region, the government levies taxation on capital operating in that region and corrects lump sum tax from the inhabitants to provide public infrastructure there. The government can distribute the capital taxation revenue to the households as a lump sum transfer. The government can also opt to subsidize the capital ﬁnancing the subsidies by levying lump sum tax on the households. Households have the following identical utility function:

α X Y : lnCi þ Ci , i ¼ 1, 2, ð5 1Þ

X where Ci is a composite index of the consumption of differentiated manufacturing Y goods and Ci is the consumption of agricultural goods in region i. The composite X index Ci is a subutility function deﬁned by a constant elasticity of substitution (CES) form:

4 Baldwin and Krugman (2004) analyze a game of sequential move and its Stackelberg equilibrium. The other studies and this work consider a game of simultaneous moves. 5 Baldwin et al. (2003) summarize the results in fiscal competition in the new economic geography models. 6 The tax rates in equilibrium differ in the basic tax competition model as well, if countries with different size are introduced. In Bucovetsky (1991) and Wilson (1991), the larger country has the lower elasticity of capital outflow and chooses higher tax rates in equilibrium. Haufler and Wooton (1999) and Ottaviano and Ypersele (2005) introduce international trade and imperfect competition, and show that a larger country sets higher capital tax rates (or lower subsidy rates) yet it can attract more firms as its larger market is more attractive to them. In contrast to these studies, Wilson (1987) analyzed the endogenous determinations of the specialization pattern and the gap in capital tax rate between ex-ante identical countries, introducing international trade and sectors that differ in the intensity of capital. The country specializing in capital-intensive goods chooses the lower capital tax rate than the country specializing in labor-intensive goods. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 117

2 31 Z γ 6 7 X 4 ω γ ω5 : Ci ¼ ðCið ÞÞ d , i ¼ 1, 2, ð5 2Þ

ω2I1[I2 where Ci(ω) denotes differentiated manufacturing goods ω consumed in region i, Ii is the set of differentiated manufacturing goods produced in region i, and γ 2 (0, 1) is the constant rate of elasticity of substitution. Let Pi denote the minimum cost of X purchasing a unit of Ci . This is calculated as:

2 3 1 Z 1Àε 6 7 ω 1Àε ω : Pi ¼ 4 ðpDið ÞÞ d 5 , i ¼ 1, 2, ð5 3Þ

ω2I1[I2 where pDi(ω) is the consumer price (c.i.f. price or delivered price) of differentiated goods ω in region i, and ε ¼ 1=ð1 À γÞ > 1. Optimization of a household yields the demand function for a differentiated manufacturing good that has a constant price elasticity ε ¼ 1=ð1 À γÞ > 1. See Appendix 1 for the derivation of the demand function. One unit of agricultural goods is produced using one unit of labor as inputs and supplied in a perfect competitive market, which implies marginal cost pricing. Assuming that the preference for agricultural goods is sufficiently large for the sector to operate in both regions, the prices and then the wages wi are equalized between two regions. We set the price of agricultural goods and wages as the numeraire. Differentiated manufacturing goods are produced and supplied by a set of monopolistically competitive firms. Each good is produced by using manufacturing differentiated goods themselves as intermediate inputs, as well as labor force. Public infrastructure enhances the productivity of differentiated manufacturing goods firms within the region where that is provided. Specifically, the production of each good takes a form of Cobb-Douglas function of labor and CES composite of differentiated manufacturing goods with CES composite share a 2 (0, 1). We assume that CES rate in the CES composite in the production function is the X same as that in the CES composite Ci of (5.2) in the utility function (5.1). Public infrastructure in region i, Gi, enhances the productivity of the manufacturing industry sector there so that the unit cost to produce a differentiated manufacturing a η good is ðPiÞ =ðGiÞ . In addition to this, production requires a unit of capital equipment. Therefore, the total cost to produce qi amount of goods in region i is a= η ððPiÞ ðGiÞ Þqi þ ri, where ri is the return to capital. The demand function for a good has the property of a constant price elasticity ε ¼ 1=ð1 À γÞ > 1. Thus the choice for the mill (f.o.b.) price by a firm in region i so 118 H. Kondo as to maximize monopolistic competition profits is a constant (gross) markup 1/γ a η over the unit cost ðPiÞ =ðGiÞ :

a ðPiÞ : : pi ¼ η , i ¼ 1, 2 ð5 4Þ γðGiÞ

The lower the elasticity of substitution rate γ and thus the lower the price elasticity ε ¼ 1=ð1 À γÞ, the higher the (gross) markup ratio 1/γ. Manufacturing goods cannot be traded interregionally without transaction costs. As is common in new economic geography models, we assume an iceberg form of transaction costs such that one unit of differentiated good melts down to 1/τ (τ 1) units when it crosses the border. If a good is produced in region i, its consumption price (c.i.f. price) in region i, pDi, is equal to the mill price (f.o.b price) pi.In Dj contrast, its consumption price in the other region j, p , is equal to τ pi. From (5.4) and these facts, (5.3) is written as:

"# 1 a ð1ÀεÞ a ð1ÀεÞ 1Àε i 1 ðPiÞ ðPjÞ P ¼ ni η þ nj τ η , i ¼ 1, 2, i 6¼ j, ð5:5Þ γ ðGiÞ ðGjÞ

where ni is the number of ﬁrms in manufacturing in region i. As a ﬁrm employs a unit of capital equipment, ni is also the mass of capital operating in region i.

3 Location Patterns

In the economy there is a unit measure of capital which is equally shared by households and freely mobile. Equilibrium in the capital market requires:

n1 þ n2 ¼ 1: ð5:6Þ

The interregional allocation of capital, n1 and n2, determines operating profits (total revenue minus payments for labor and intermediate inputs) per firm in each region, and all the operating profits are absorbed by the payment for capital, ri. Put differently, firms bid up the payment for capital ri until it is equal to the operating profits. See Appendix 1 for the details about the derivation of ri. The government levies taxation on capital operating in that region at the rate ti. Therefore, net returns to the owner of the capital operating in region i is ð1 À tiÞri.If there still exists the gap in the net returns ð1 À tiÞri even after r1 is determined to absorb all the operating profits per firm in region i, the owners of capital shift the capital to the region in which its net return is highest. The capital allocation n1 and n2 changes satisfying (5.6), and accordingly operating profits and thus the returns to capital change. Figure 5.2a plots the ratio of returns to capitalr1=r2 as a function of n1. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 119

Fig. 5.2 (a) The ratio of returns to capital between two regions in the manufacturing industry as a function of n1. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74 and ε ¼ 3. 8. (b) The ratio of returns to capital between two regions in the manufacturing industry as a function of τ when n1 ¼ 1, G1 ¼ G2, and t1 ¼ t2. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74 and ε ¼ 3. 8

Firms stably disperse into two regions if ni which satisfies r1 ¼ r2 exists in [0, 1]. In contrast, If ri > rj is satisfied for ni ¼ 1, firms can stably agglomerate in region i. When ni ¼ 1, return to capital in region i is:

2ð1 À γÞα r ¼ , ð5:7Þ i 1 À aγ and that in region j is: γ α ηðεÀ1Þhi ð1 À Þ Gj ðaÀ1Þð1ÀεÞ ð1þaÞð1ÀεÞ rj ¼ ð1 À aγÞτ þð1 þ aγÞτ , ð5:8Þ 1 À aγ Gi

Note that rj will be attained by a ﬁrm that initially locates in region i with all the other ﬁrms but moves to region j alone, which is not actually observed. From (5.7) and (5.8), the ratio of after-tax returns to capital is written as:

η ε 1 À t G ð À1Þ 2 h τ, G , G , t , t i i : ð i j i jÞ¼ ðaÀ1Þð1ÀεÞ ð1þaÞð1ÀεÞ 1 À tj Gj ð1 À aγÞτ þð1 þ aγÞτ ð5:9Þ

Figure 5.2b plots hðτ, Gi, Gj, ti, tjÞwhenG1 ¼ G2 andt1 ¼ t2.Ifhðτ, Gi, Gj, ti, tjÞ1, the manufacturing industry can stably agglomerate. 120 H. Kondo

ab t G2 2

Q1 \ Q2 0.21 Q2 \ Q1

− 0.26 0.21 ∩ Q1 Q2 0 t1 ∩ Q1 Q2

− 0.26 Q2 \ Q1 Q1 \ Q2

0 G1

Fig. 5.3 (a) Areas of Q1 and Q2. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8 and τ ¼ 1. 4. (b) Areas of Q1 and Q2. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8 and τ ¼ 1. 4

Let Qi denote the set of ðτ, G1, G2, t1, t2Þ under which the manufacturing industry agglomerates in region i. That is,

τ τ : Qi ¼fð , G1, G2, t1, t2Þjhð , Gi, Gj, ti, tjÞg 1

Figure 5.3 shows the areas of Q1 and Q2. In the following proposition we summarize the properties of Qi and thus the manufacturing industry sector can stably agglomerate. And then discuss the intuitions. Proposition 1. (i) When interregional transaction costs τ are lower, the manufacturing industry agglomerates in one region. (ii) When interregional transaction costs τ are lower and the gap in the capital tax rate is smaller, the manufacturing industry can agglomerate in either region. When the gap is larger, the manufacturing industry agglomerates in the region with the lower capital tax rate. (iii) When interregional transaction costs τ are lower and the gap in the amount of public infrastructure is smaller, the manufacturing industry can agglomerate in either region. When the gap is larger, the manufacturing industry agglomerates in the region with the larger amount of public infrastructure. Proof. See Appendix 2. In the following two subsections we consider the intuitions behind Proposition 1. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 121

3.1 Transaction Costs and Agglomeration

In our model, the manufacturing industry has significant vertical linkages. That is, firms buy and sell their outputs as intermediate inputs with one another. Thus, in an economy with interregional transaction costs, the interregional location pattern of firms determines the ratio of net returns to capital, through affecting the gap in the costs to purchase intermediate inputs and the gap in the scale and fierceness of competition of the domestic market between the two regions. The relative sizes of three forces determine whether or not agglomeration takes place: the forward linkage effect, the backward linkage effect, and the local market competition effect. In a region with more input suppliers, a firm can pay lower production costs as it can purchase its intermediate inputs at lower transaction costs. Hence, a firm will prefer to stay in this region or be eager to move there. This propensity of location choice is called the forward linkage effect. A firm will also be eager to locate itself in a region with more customers who buy its products. This propensity is called the backward linkage effect. In our model, a manufacturing good is produced using the other goods in that manufacturing industry as intermediate inputs. Under this situation, the suppliers of a given firm’s inputs are also the customers of the firm’s outputs in the manufacturing industry. Hence, a region with more firms in the manufacturing industry will attract more firms in that industry through both the forward and backward linkage effects. Yet at the same time, this region will also face fiercer competition. Firms prefer to locate in less competitive market. This propensity of location choice, the so-called local market competition effect, makes it less attractive to be based in a region with more firms. We consider how international transaction efficiency τ affects the relative size of the three offsetting effects mentioned earlier (the forward linkage effect, backward linkage effect, and local market competition effect) and thereby determines the stability of agglomeration. We suppose that firms in the manufacturing industry agglomerate in region 1, and discuss the stability of the agglomeration by considering whether or not a firm that moves from region 1 to region 2 alone will earn lower profits than a firm that remains in region 1. If the firm does earn lower profits after relocating to region 2, the agglomeration of that manufacturing industry in region 1 is stable. Interregional transaction costs shelter domestic markets from imports. When transaction costs are higher, the profit per firm will be more sensitive to the number of rival firms in that industry in the domestic market. If transaction costs are extremely high and no firms are based in region 2, a firm can almost monopolize the domestic market in that industry in region 2 by relocating there alone. That is, when a firm moves from region 1 to in region 2 alone, the increase in profits from in region 2’s market will be large enough to counterbalance the decrease in profits from region 1’s market. This means that the local market competition effect dominates the forward and backward linkage effects at work in region 1 and de-stabilizes the agglomeration. 122 H. Kondo

When transaction costs fall, the domestic market is less protected and a shift to a region with fewer firms becomes disadvantageous. Specifically, when τ becomes lower than ~τ as in Fig. 5.2b, a firm in region 1 will choose to stay in the agglomeration of that industry in region 1 rather than move to region 2 alone. That is, the forward and backward linkage effects in region 1 dominate the local market competition effect and stabilize the agglomeration.7

3.2 Capital Tax, Public Infrastructure and Agglomeration

The gaps in the capital tax rate and amount of public infrastructure do not directly determine whether or not the manufacturing industry can stably agglomerate. The interregional transaction efficiency τ determines the stability of agglomeration. More specifically, the interregional transaction costs τ determine it by affecting the curvature of the graph in Fig. 5.2a: when the transaction costs are lower, the graph in Fig. 5.2a is more likely to have a positive slope. Hence, the entry of more firms in a region increases the profits per firm in that region via the mechanism discussed in the last subsection. While the gaps in tax rates ti and public infrastructure provision Gi exert upward and downward shifts of the graph in Fig. 5.2a, they have no affects on the curvature.8 When interregional transaction costs are lower, the manufacturing sector stably agglomerates, as we saw in the last subsection. In this case, the gaps in tax rates and public infrastructure provision determine the region where the manufacturing sector agglomerates. If gaps in tax rates and public infrastructure provision are small, the manufacturing industry can agglomerate in either of the two regions in the two-region model. Firms prefer a region with a lower capital tax rate and more extensive public infrastructure. And as we saw in the last subsection, firms are willing to locate in a region with larger numbers of firms when interregional transaction costs are low.

7 Note that when transaction costs fall, a firm moving from region 1 to 2 can expect less of a decrease in its profits from region 1’s larger market, as its access to the foreign market is improved. Lower transaction costs reduce the opportunity costs for a firm leaving the agglomeration region 1, and thus de-stabilize the agglomeration again. Yet a transaction cost in a medial range, such as τ∗ in Fig. 5.2b, is half-finished, in the following sense: for a firm moving to region 2 alone, τ∗ is too small to protect the domestic market in region 2 and ensure sufficiently large profits from that market, yet too large to ensure access to the larger foreign market in region 1 and sufficiently large ∗ profits from that market. That is, when τ ¼ τ , locating in region 2 outside of the agglomeration becomes the most disadvantageous, and the agglomeration force in region 1 becomes the most prominent. When τ becomes smaller than τ∗, the agglomeration force becomes less prominent but remains strong enough to stabilize agglomeration. 8 When region i hosts the manufacturing industry, a rise in ti increases the disposable income in the region and it in turn expands the market size. If we introduce a utility function in which demands for manufacturing goods have income effects, a change in ti therefore affects ri=rj through an additional channel. These effects are too small, however, to change the stability of agglomeration. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 123

Firms will agglomerate in that region, even if the capital tax rate is higher and public infrastructure is less extensive. In contrast, the manufacturing industry will only agglomerate in a region with a lower tax rate and more extensive public infrastructure if the interregional gaps in tax rates and public infrastructure provision are large. Gaps in these policies can overwhelm the agglomeration beneﬁts. Thus, the agglomeration in a region with a much higher tax or much lower infrastructure provision cannot be stable. In an economy with efﬁcient transaction technologies, manufacturing industries base themselves in a limited number of regions. Changes in tax rates and public infrastructure induce the locations of the manufacturing industries to changes not continuously and uniformly, but in a lumpy, sporadic fashion. This pattern may compel regions to adopt extreme policies, as we will discuss in Sect. 4.

3.3 Location Patterns in an Economy with More than Two Regions

An extension of our two-region model would allow us to analyze the pattern of interregional specialization pattern in an economy with more than two regions, but not without considerable complexity. To discuss how industries with scale economies locate in an economy with many regions, we will instead examine the results of the earlier research in combination with results from our two-region model.9 In an integrated economy where transaction costs among the regions are very low, the manufacturing industry will cluster in a limited number of regions. The number of regions hosting the manufacturing industry and the distance between any two of the regions depend on the importance of the industry and how prominently its scale economies work. Specifically, they depend on the share of differentiated manufacturing goods in the utility function a, the share of the CES composite index α in the production function, and the elasticity of the substitution rate γ. In the real world, however, transaction costs between any two of regions in an economy will not decrease uniformly. Transaction efficiencies start to improve between small numbers of regions and their neighbors. Meanwhile, transportation facilities outside this group of regions remain inefficient. In this case, a large number of firms in the manufacturing industry sector agglomerate in either one region, or a very limited number of regions in the group of regions with efficient transportation facilities. The rest of the sector disperses among regions outside this group. That is, differentiated goods firms are located in many regions where transportation is inefficient, but are small in number overall. These firms tend to base themselves nearby their final consumers in areas where transportation facilities are not well developed. We call the area constituting a group of regions with

9 Fujita et al. (2001) summarize the development of theoretical frameworks in the new economic geography models, including the settings of more than two regions and continuous space. 124 H. Kondo

Case 1. Location of the manufacturing sector when transaction costs are high The manufacturing share in the region

Region unit Transaction costs between two regions

Case 2. Location of the manufacturing sector when transaction costs are low among a small number of regions

Rural area Metropolitan area Rural area

Case 3. Location of the manufacturing sector when transaction costs between any two of regions are low

Fig. 5.4 Transaction costs and the location pattern of the manufacturing sector efﬁcient transaction facilities the “metropolitan area”, and all the other areas, collectively, the “rural area”. Figure 5.4 demonstrates how the location of the manufacturing sector changes when transaction costs decreases in a manner discussed above. When transaction costs start decreasing in the rural area as well, ﬁrms agglomerate in a smaller number of regions, as discussed above. Many regions in the rural area lose the manufacturing industry in this process.

4 Fiscal Competition

Suppose that transportation facilities in the rural area initially remained inefficient but are now improved. As we saw in Sect. 3, firms in the manufacturing sector that used to be dispersed among regions start agglomerating in a limited number of regions. We analyze the outcome of a game where regions recognize that the integration of the goods market will lead to an agglomeration of the manufacturing sector, and respond by providing public infrastructure and subsidizing capital in order to attract the agglomeration. We employ the two-region model. Transaction costs τ were high but now are falling. Differentiated manufacturing goods firms that heretofore have been dis- persing are now expected to agglomerate in one of the two regions. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 125

4.1 Setup

In this section, we consider the outcome of ﬁscal competition between governments aiming to maximize the utility of households in their regions. From (5.1), we can see that utility of a household in region i mainly depends on disposable income and price index of manufacturing differentiated goods Pi:

Ei À αlnPi, i ¼ 1, 2, ð5:10Þ where Ei is disposable income of a household in region i. This depends on the location pattern of manufacturing industry, which in turn depends on transaction costs τ, the capital tax rates and the public infrastructure of the two regions, t1, t2, G1 and G2, as we have seen in the last section. Let uðτ, Gi, Gj, ti, tj, niÞ denote the utility of a household in region i when it hosts ni measure of manufacturing differentiated goods ﬁrms. When the manufacturing industry is agglomerating in region i, we can calculate uðGi, Gj, ti, tj, niÞ as: ð1 À γÞα αη uðτ, Gi, Gj, ti, tj,1Þ¼1 þð1 þ tiÞ À Gi þ lnGi, ð5:11Þ 1 À aγ 1 À a ð1 À γÞα αη uðτ, Gj, Gi, tj, ti,0Þ¼1 þð1 À tiÞ À Gj þ lnGi À αlnτ: 1 À aγ 1 À a ð5:12Þ

See Appendix 3 for the details about the derivations of (5.11) and (5.12). From (5.11) and (5.12), we can see that the disposable income, the sum of the ﬁrst three terms in these equations, is much larger in region i where the manufacturing industry is agglomerating, as region i enjoys larger tax revenue from capital agglomerating in that region. In contrast, region j has no capital tax revenue as it does not have manufacturing industry. Moreover, by raising ti, region i can increase its disposable income. Unless ti becomes much higher than tj, the agglomerations in region i never collapse and no capital leaves the region. Thus, capital tax revenue in that region drastically increases. Thus, a region hosting the manufacturing industry sector has a large incentive to raise its capital tax rate. However, if region i sets a much higher capital tax rate, it loses the manufacturing industry. It faces a drastic decrease in disposable income, and higher prices of manufacturing goods. The higher prices reduce its utility by αlnτ, the third term in (5.12). 126 H. Kondo

The utility of a household in each region is: 8 > τ > uð , ti, tj, Gi, Gj,1Þ > τ ∖ , <> if ð , ti, tj, Gi, GjÞ2Q1 Q2 λ τ λ τ τ iuð , ti, tj, Gi, Gj,1Þþð1 À iÞuð , ti, tj, Gi, Gj,0Þ Uð , ti, tj, Gi, GjÞ¼> τ > if ð , ti, tj, Gi, GjÞ2Q1 \ Q2, > τ :> uð , ti, tj, Gi, Gj,0Þ τ ∖ : if ð , ti, tj, Gi, GjÞ2Q2 Q1 ð5:13Þ

As we saw in Proposition 1, when interregional transaction costs are low and the gaps in capital tax rate and the amount of public infrastructure are moderate so that τ ð , ti, tj, Gi, GjÞ2Q1 \ Q2, the manufacturing industry agglomerates in either region. We let λi 2½0, 1ðλi ¼ 1 À λjÞ denote the probability with which the manufacturing industry agglomerates in region i.

4.2 Public Infrastructure Provision

We consider the competition in providing public infrastructure between regions. Figure 5.5a shows the utility of a household in region i as a function of Gi and Gj. Figure 5.5b shows the reaction function of the two regions.

b G2 a Q \ Q G high 2 1 0.190

Region 1’s reaction ∩ Q1 Q2

Region 2’s reaction G 0.083 ~ 0.067 G Q1 \ Q2

G 0 0.067~ 0.083 0.190 1 G G G high

Fig. 5.5 (a) Utility of a household in region 1. (b) Reaction function. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8, λ ¼ 0. 5 and τ ¼ 1. 4 5 Public Policy and Economic Growth in the Integrating Japanese Economy 127

Let G~ denote the amount of public infrastructure that maximizes (5.11). This is calculated as: αη G~ ¼ : ð5:14Þ 1 À a

In order for region i to host the agglomeration with probability of unity, it has to choose Gi large enough to satisfy hðτ, Gj, Gi, tj, tiÞ < 1. Therefore, if the other region j’s provision of public infrastructure Gj is in a large amount, region i has ~ to choose Gi that is larger than G in order to host the agglomeration with probability of unity. By choosing such a large amount of public infrastructure Gi, region i can enjoy much lower c.i.f prices of manufacturing goods. On the other hand, however, disposable income of a household in region i decreases, since the lump sum tax burden on households in that region to ﬁnance the large amount of public infrastructure provision drastically increases. Therefore, if the other region j’s choice of Gj is extremely large, region i will give up to host the manufacturing industry but rather it will choose Gi ¼ 0 and free ride on the other region’s very large Gj. Region i must import all the manufacturing goods. However, since region j provides public infrastructure in a very large amount Gj, households in region i can enjoy very cheap prices of manufacturing goods which are reduced by a huge amount of Gj without paying for this. House- holds in region j can enjoy the lower prices of the goods, yet they bear a huge tax burden to ﬁnance Gj. We consider G and Ghigh that satisfy:

τ high τ uð , G , G, ti, tj,1Þ¼uð ,0,G, ti, tj,0Þ, : high ð5 15Þ hðτ, G, G , tj, tiÞ¼1:

The amount of G in (5.15) is the threshold in that region i will switch its strategy depending on whether or not the other region’s choice of Gj is larger than that amount.

If region j chooses Gj smaller than G, region i will choose larger Gi and attract the agglomeration of the manufacturing industry with probability of unity. The amount of Gi required to do so is larger than Gj yet not extremely so. Then, in region i, the beneﬁt of a decrease in manufacturing goods c.i.f. prices dominates the additional tax burden on households required to ﬁnance this relatively small Gi. In contrast, if region j chooses Gj larger than G, region i will not attract the manufacturing industry. Public infrastructure plays no role in a region without the manufacturing industry sector. Hence, once region i decides not to attract the manufacturing industry sector, it chooses Gi ¼ 0. It would rather enjoy the prices of goods drastically reduced by region j’s public infrastructure, without paying for it. 128 H. Kondo

b t2

Region 1’s reaction \ a Q1 Q2 0.21 thigh 0.155

−0.345 −0.26 0.155 0.21 t low t 0 thigh t1 t

Q1 ∩ Q2

Region 2’s reaction

0.26 Q2 \Q1

tlow 0.345

Fig. 5.6 (a) Utility of a household in region 1. (b) Reaction function. Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8, λ ¼ 0. 5 and τ ¼ 1. 4

Therefore, the choice of Gi by region i never equals to that by region j, Gj. Region ~ i chooses G or larger Gi if Gj G, and it chooses Gi ¼ 0ifGj > G. Hence, there is no symmetric pure strategy Nash equilibrium. The results are summarized by: Proposition 2. The competition in providing public infrastructure between regions has no pure strategy Nash equilibrium.

4.3 Tax Competition

Next we consider the competition in capital tax between regions. Figure 5.6a shows the utility of a household in region i as a function of ti and tj, and Fig. 5.6b summarizes the best reactions of the two regions. A region will set ti as high as possible, if it could continue hosting the agglomeration of manufacturing industry. As long as ti does not exceed the one in hðτ, Gi, Gj, ti, tjÞ¼1 and thus the gap in capital tax rate remains modest enough, the manufacturing industry agglomerates in region i with positive probability. Thus, a rise in ti brings larger expected capital tax revenue. If region i chooses much higher ti, it cannot attract the manufacturing industry. Region i must import all the manufacturing goods. In addition, it loses its capital tax revenue. However, if the other region j announces that it will subsidize capital at a very high subsidy rate (negative and very low tj), the disposable income of a household in region i can increase. Region i can beneﬁt from region j’s capital subsidy policy without paying the costs. Inversely, region i may be willing to choose ti that is much lower than tj. Setting i j i j ti low enough to satisfy hðτ, Gj, Gi, tj, tiÞ < 1 (and ðt , t Þ is in Q ∖Q ), it can host the 5 Public Policy and Economic Growth in the Integrating Japanese Economy 129 agglomeration of the manufacturing industry with probability of unity. Then, households in region i can enjoy much lower c.i.f. prices of manufacturing goods. At the same time, however, disposable income of a household in region i decreases. In order to attract the manufacturing industry, region i has to subsidize capital at very high capital subsidy rate (negative and very low ti). The success in attracting the manufacturing industry is accompanied by a massive increase in subsidy payments and a massive increase in lump sum tax burden on households in that region to ﬁnance the subsidy policy. We consider t, tlow and thigh that satisfy:

low high high uðτ, Gi, Gj, t , t,1Þ¼λuðτ, Gi, Gj, t , t,1Þþð1 À λÞuðτ, Gi, Gj, t , t,0Þ, low hðτ, Gj, Gi, t, t Þ¼1, high hðτ, Gi, Gj, t , tÞ¼1: ð5:16Þ

If region j chooses tj strictly higher than t, region i will choose much lower ti and attract the agglomeration of the manufacturing industry. As tj is high, ti required to attract the manufacturing industry that is a bit lower than tlow, is much lower than tj yet not extremely low. Then, in region i, the beneﬁt of a decrease in manufacturing goods c.i.f. prices dominates the additional tax burden on households required to ﬁnance this relatively small capital subsidy policy. In contrast, if region j chooses tj that is lower than t, region i will abandon its hopes of attracting the manufacturing industry with probability of unity. It would rather maximize its expected capital tax revenue (or minimize its capital subsidy payment) by raising ti up to thigh. Therefore, a region never chooses ti that is equal to tj. It chooses higher ti iftj t, and it chooses lower ti if tj < t. That is, a region’s reaction function is discontinuous at t and never crosses the 45 degree line. Hence, there is no symmetric pure strategy Nash equilibrium. The results are summarized by: Proposition 3. Tax competition between regions has no pure strategy Nash equilibrium.

4.4 A Nash Equilibrium of Mixed Strategies

We consider the game of mixed extension. However, it is formidably complicated to apply the game of mixed extension to our model assuming an infinite action space. Hence, we set up a finite game and give its numerical example. We let Θi denote a finite set of Gi that is a large but finite set of equally spaced ψ points. And we let iðGiÞ denote the function that shows the probability for region ’ ’ ψ ’ i to choose Gi. Given region i s belief about region j s strategy jðGjÞ, the region i s expected utility when it chooses Gi is: 130 H. Kondo X τ ψ : ViðGiÞ¼ Uð , Gi, Gj, ti, tjÞ jðGjÞ Gj2Θj

ψ Then, region i determines iðGiÞ (how to mix its strategies Gi) so as to maximize: X ψ : Wi ¼ ViðGiÞ iðGiÞ Gi2Θi

ψ ψ If the 1ðG1Þ and 2ðG2Þ actually chosen correspond to their initial beliefs about the other region’s behavior, then such a combination of mixed strategies is a Nash equilibrium. ψ # ψ # Ξ Ξ The combination of the probability functions i ðG1Þ and 2ðG2Þ on 1 and 2, which are the subsets of Θ1 and Θ2 is a Nash equilibrium of mixed strategy if X ψ # ψ # i ðGiÞ2ð0, 1Þ and i ðGiÞ¼1, Ξ X Gi2 i X τ ψ # j τ ψ # Uð , Gi, Gj, ti, tjÞ j ðG Þ¼ Uð , Gi0 , Gj, ti, tjÞ j ðGjÞ Gj2Ξj Gj2Ξj i 0 Ξ 5:17 for any two Gi and Gi on , ð Þ andX X τ ψ # j > τ ψ # Uð , Gi, Gj, ti, tjÞ j ðG Þ Uð , Gi0 , Gj, ti, tjÞ j ðGjÞ Gj2Ξj Gj2Ξj Ξ Θ ∖Ξ : for any Gi 2 i and Gi0 2 i i

ψ # ψ # There is no incentive for either region to deviate from 1ðG1Þ and 2ðG2Þ. If region ψ ψ # Ξ i changes iðGiÞ from i ðGiÞ but limits its choice of Gi on the set of i, there is no ψ # change in the expected utility Wi under the given j ðGjÞ. Yet if region i includes the points Gi out of the set of Ξi into its own strategy, the expected utility does necessarily decrease. This symmetric game has a symmetric mixed strategy Nash ψ # ψ # equilibrium in which 1ðG1Þ¼ 2ðG2Þ when G1 ¼ G2. By the similar treatment we can obtain a symmetric mixed strategy Nash ϕ# equilibrium in the tax competition. We let i ðtiÞ the probability of region i to choose the tax rate ti in the mixed strategy Nash equilibrium. ψ # ϕ# Figures 5.7 and 5.8 show examples of i ðGiÞ and i ðtiÞ, respectively. If the region where the manufacturing industry sector agglomerates is predetermined, that region will choose public infrastructure in the amount of G~ and the other region will choose to provide no public infrastructure. In reality, however, the region where the manufacturing sector agglomerates is not predetermined. There is no pure strategy Nash equilibrium. Competition in providing public infrastructure rises to a more intensive level. The amount of public infrastructure Gi tends to be very large. The amount on average is larger than G on average. And Gi included in the mixed strategy in an equilibrium has very large variance. 5 Public Policy and Economic Growth in the Integrating Japanese Economy 131

Fig. 5.7 ψ ðGiÞ in a mixed strategy Nash equilibrium Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8, λ ¼ 0. 5 and τ ¼ 1. 4

Fig. 5.8 ϕðGiÞ in a mixed strategy Nash equilibrium Note: α ¼ 0. 4, a ¼ 0. 4, γ ¼ 0. 74, ε ¼ 3. 8, λ ¼ 0. 5 and τ ¼ 1. 4 132 H. Kondo

~ As we saw in Sect. 4.2, ðGi, GjÞ¼ðG ,0Þ cannot be a Nash equilibrium. If region i chooses public infrastructure in the modest amount of G~ , region j will choose not Gj ¼ 0 but Gj, an amount large enough to attract the manufacturing industry. Given ~ the strategy of region j in this case, region i will choose a Gi larger than G to prevent the manufacturing industry sector leaving that region.

As we saw, by raising Gi up to G, region i can discourage region j from taking steps to attract the manufacturing industry. Region i, however, will not always choose public infrastructure in an amount as large as G in a Nash equilibrium. Assume that a Nash equilibrium has region i to choose G. The best reaction of region j is to give up hosting the agglomeration of the manufacturing sector, and to provide no public infrastructure Gj ¼ 0. Given this choice by region j, region i will not choose G, but smaller G~ . This contradicts our ﬁrst assumption. We may guess that the range of Gi region i includes in its strategy in a mixed strategy Nash equilibrium fall within the range of G~ < Gi < G. If region j decides to adopt the same strategy, a symmetric mixed strategy Nash equilibrium is reached. Region j may, however, respond to region i’s strategy by including a much larger Gj. Because region i includes Gi smaller than G in its strategy, region j will attract the manufacturing industry agglomeration with probability of unity by choosing Gj that is higher than G yet close to or lower than Ghigh. In a mixed strategy Nash equilibrium, the range of Gi each region includes in its ~ strategy is much wider than the range of G < Gi < G. Assume that region i chooses

Gi within this wide range. Region j will include Gj that is much larger than G and high close to G , that is, Gj that will attract the whole of the manufacturing industry with a probability of unity when region i chooses Gi smaller than G. Region i, however, may also choose Gi that is larger than G as well. In this case, region j will also include much smaller Gi, the preferable choice when region i chooses very large Gi. That is, given the strategy of region i assumed above, region j will adopt a similar strategy. Therefore, this strategy is in equilibrium in a mixed extension of the game. Next, we consider the tax competition. If the region where the manufacturing industry agglomerates is predetermined, the region will set ti as high as possible. Again, in reality, however, the where the manufacturing sector agglomerates is not predetermined. Tax competition becomes much more ﬁercer. The capital tax rate tends to be low. The rate on average may be negative (capital subsidy). Tax rate included in the mixed strategy in an equilibrium has very large variance. A Nash equilibrium does not have one region to choose only high capital tax rates. Assume that in a Nash equilibrium region i chooses only very high ti. Then region j will set tj that is low enough to host the manufacturing industry’s agglomeration with probability of unity. Given the strategy of region j like this, region i will lower ti to prevent the manufacturing industry sector leaving that region. This contradicts our ﬁrst assumption. Neither a Nash equilibrium has region j to choose capital tax in the rate as low as t. Assume that in a Nash equilibrium region i always chooses t. The best reaction of 5 Public Policy and Economic Growth in the Integrating Japanese Economy 133 region j is to choose thigh. As we saw in Sect. 4.3, region j will not choose tlow that is low enough to attract the manufacturing industry with probability of unity. Region j would rather choose higher tj that maximizes the expected capital tax income. As long as region j chooses tj that is higher but not extremely so, it can still expect that the manufacturing industry will agglomerate in that region with positive probability < λ < high 0 j 1. By choosing t , the highest tj included in Q1 \ Q2, region j maximizes its expected capital tax income. Given the strategy of region j, region i will not choose t. It can attract the agglomeration of the manufacturing sector with probability of unity by choosing ti that is a bit lower than t and thus it is willing to do so. This contradicts our assumption. The range of ti in a mixed strategy Nash equilibrium is very wide including t and lower ti. Assume that region i will choose ti within this wide range. Region i chooses ti that is higher than t with positive provability. Then region j will include tj that is as low as or closer to tlow in its strategy, with which it can attract the whole of the manufacturing industry with a probability of unity in the case that region i actually chooses ti higher than t. Region i, however, may also choose ti that is lower than t.It is preferable for region j to choose tj that is much higher than t yet closer or lower high than t , when region i chooses ti that is lower than t. The probability for region j to attract the manufacturing industry with such tj is lower than unity but positive. When region j raises capital tax rate up to thigh, it can keep the positive probability to attract the manufacturing industry, and maximize its expected capital income tax revenue. Therefore, region j will also include tax rates higher than t and closer to thigh in its strategy. Therefore, region j will adopt a strategy similar to that of region i.

Conclusion and Discussion In an economy where goods and factors markets are progressively integrating and industries with scale effects and externalities are prevailing, economic activities tend to concentrate in a limited number of regions. When local governments in many regions try to introduce industries with scale effects independently, they invest huge amounts in public infrastructure and announce signiﬁcant subsidies on capital. Yet the industries the regions seek to introduce will settle in only a limited number of regions. Some regions thus succeed in introducing an industry by making huge or even moderate investments in public infrastructure, while many others fail after making tremendous investments in public infrastructure, incurring huge welfare losses. As a consequence, huge amounts of public infrastructure remain are left unused in many regions. Many studies have pointed out that the dependence of local government on the central government induces the former to provide excessive public infrastructure. Speciﬁcally, the central government provides subsidies to induce local governments to carry out some public projects. The central government

(continued) 134 H. Kondo

(continued) may also be prepared to bail out projects when local governments fail, further incentivizing local government to invest in public projects to excess.10 No central government is involved in the model examined in this paper. The local governments decide the policies in their regions freely and independently, and bear full responsibility for the results. Our analysis shows even in the absence of a central government, local governments still tend to invest excessively in public infrastructure in an economy where market integration is proceeding and modern industries are agglomerating in only a limited number of regions. In the real-world setting of Japan, public investment in infrastructure increased rapidly from the early or mid 1970s, especially in rural areas. Many regions failed to host the agglomeration of modern industries after making massive investments in public infrastructure. Though huge in scale, many of these public infrastructures have had only negligible effects on the productivity in the regions. To reiterate, no central government is involved in the model examined in this paper. We can, however, discuss the role of a central government in the conclusion of this paper. Can we introduce a central government or a government on a tier above the local governments where in can enhance economic welfare by coordinating the local governments and mitigating ﬁscal competition? If a central government could choose the regions where industries with prominent scale effects and externalities were based in advance, it could preempt the excess investment in public infrastructure by many regions. In this case, however, public infrastructures provided by a very limited number of selected regions would tend to be smaller than optimal. Hence, the central government would induce the regions to provide public infrastructure through subsidies. In Japan, the central government has planned out regional infrastructure provision to a considerable extent. Especially, for the public infrastructures that affect the locations of modern industries in the future, such as transportation facilities and industrial parks, the central government has planned out the regions where the infrastructure should be concentrated and has guided and induced the local governments in the targeted regions to follow its plans. In reality, however, the central government has not created these plans wholly on its own. Local governments and the related interest groups positively have taken active part in the planning process. In an economy where industries agglomerate in a limited number of regions, the welfare of a region signiﬁcantly depends on whether or not the region can host the agglomerations. Hence, regions will act aggressively, independently and

(continued)

10 See Akai (2014). 5 Public Policy and Economic Growth in the Integrating Japanese Economy 135

(continued) non-cooperatively in approaching the central government to sell themselves as the regions where the industries should be based. As a consequence, the central government may select more regions than optimal.11 When local governments take account of such a process in the formation of a “centralized” plan, independent local governments assumed in our model are still valid and. Fiscal competition will be ﬁerce. In addition, if the central government subsidizes the investment in public infrastructure, or more precisely, if regions let the central government subsidize their investment in public infrastructure, the subsidies will induce more excessive investment, exacting more serious losses to economic welfare.

Acknowledgements I am indebted to comments made by the participants at the Japan ﬁscal research group seminars in University of California, Irvine, Max Planck Institute in Munich, and National Graduate Institute for Policy Studies (GRIPS) in Tokyo, and to the discussant Takashi Fukushima, whose suggestions much improved the paper. I am grateful for the support of the Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Scientiﬁc research (C) number 22530241.

Appendix 1: Derivations of (5.7) and (5.8)

Households maximize the identical utility function (5.1) subject to the budget constraint in which the prices of agricultural goods and the wages are normalized as unity. The maximization yields:

ε α p ω À ω ð ið ÞÞ : Cið Þ¼ 1Àε , i ¼ 1, 2, ð5 18Þ ðPiÞ Y α : Ci ¼ Ei À , i ¼ 1, 2, ð5 19Þ where Ei denotes the disposable income per capita in region i. a= η ω The unit costs in producing a good in region i are Pi Gi . Hence, when a good ω a= η ω is produced in the amount of q( ), the total variable costs are ðPi Gi Þqð Þ. Applying the Shepard’s Lemma to it yields demands for labor and intermediate goods by a good ω ﬁrm in region i: a Pi ð1 À aÞ η qðωÞ, i ¼ 1, 2, ð5:20Þ Gi

11 See Nagamine and Katayama (2001). 136 H. Kondo a ω0 Àε Pi ðpDið ÞÞ ω : : a η 1Àε qð Þ i ¼ 1, 2 ð5 21Þ Gi ðPiÞ

ω a= η ω Àε= 1Àε A differentiated good firm faces the demand of aðPi Gi ÞððpDið ÞÞ ðPiÞ Þq ðω0 Þ by each firm and of (5.18) per consumer. Since in each region all firms face the same factor prices and demand function, all choose the same mill price pi and production amount qi. Thus, from (5.4), (5.20) and (5.21), the demands for labor, intermediate goods in that region and that in the other region by a firm in that industry in region i are:

γ : ð1 À aÞpiqi, i ¼ 1, 2, ð5 22Þ ε p À γ ð iÞ : apiqi 1Àε , i ¼ 1, 2, ð5 23Þ ðPiÞ ε ðτp ÞÀ γ j : : apiqi 1Àε , i, j ¼ 1, 2, i 6¼ j ð5 24Þ ðPiÞ

From (5.23), (5.24) and (5.18), the condition that a differentiated good’s market is cleared is written as follows:

Àε ÀÁτ1Àε Àε α γ ðpiÞ α γ ðpiÞ : qi ¼ ðÞþ anipiqi 1Àε þ þ anjpjqj 1Àε , i, j ¼ 1, 2, i 6¼ j ðPiÞ ðPjÞ

The total revenue of the manufacturing sector in each region can be written as: ÀÁ α γ Ω α γ Ω : nipiqi ¼ ðÞþ anipiqi ii þ þ anjpjqj ij, i, j ¼ 1, 2 i 6¼ j, ð5 25Þ

Ω = 1Àε Ω τ = 1Àε where ii niðpi PiÞ and ij nið pi PjÞ are the share of the expenditure by a household in region j for the differentiated goods produced in region i. Solv- ing (5.25) for nipiqi (i ¼ 1, 2) yields: "# Ω þ Ω þ aγðΩ Ω À Ω Ω Þ n p q ¼ α ii ij 12 21 11 22 , i i i 2 ð5:26Þ 1 À aγðΩ11 þ Ω22ÞþðaγÞ ðΩ11Ω22 À Ω12Ω21Þ i, j ¼ 1, 2, i 6¼ j :

Inserting (5.26) into (5.22), we can derive the labor force employed by a ﬁrm in region i as follows: "# ð1 À aÞγα Ω þ Ω þ aγðΩ Ω À Ω Ω Þ L ¼ ii ij 12 21 11 22 , i 2 ð5:27Þ ni 1 À aγðΩ11 þ Ω22ÞþðaγÞ ðΩ11Ω22 À Ω12Ω21Þ i, j ¼ 1, 2, i 6¼ j: 5 Public Policy and Economic Growth in the Integrating Japanese Economy 137

The shares of (1 À a)γ and aγ of total revenue (5.26) are paid for labor as in (5.27) and for intermediate inputs, respectively, and the rest 1 Àγ is retained as operating proﬁts which are absorbed by ri as: "# ð1 À γÞα Ω þ Ω þ aγðΩ Ω À Ω Ω Þ r ¼ ii ij 12 21 11 22 , i 2 ð5:28Þ ni 1 À aγðΩ11 þ Ω22ÞþðaγÞ ðΩ11Ω22 À Ω12Ω21Þ i, j ¼ 1, 2, i 6¼ j:

If the manufacturing sector agglomerate in region i, P1, P2, and Ωij ði, j ¼ 1, 2Þ are calculated as follows:

hiη 1 1 1 1Àa 5:29 Pi ¼ ð Þ γ Gi

hiη 1 1 1 1Àa ð5:30Þ Pj ¼ τ γ Gi Ω Ω ii ¼ ij ¼ ni, ηðεÀ1Þ Gj ð1þaÞð1ÀεÞ Ωji ¼ nj τ : Gi ηðεÀ1Þ Gj ðaÀ1Þð1ÀεÞ Ωjj ¼ nj τ : Gi

With these Eq. (5.28) is rewritten as (5.7) and (5.8).

Appendix 2: Proof of Proposition 1

Proof of Proposition 1(i)

We show that when τ is small, hðτ, Gi, Gj, ti, tjÞ is larger than unity. The derivative of the denominator of (5.9) with respect to τ is

ε ε ð1 À εÞτð1þaÞð1À ÞÀ1½ð1 þ aÞð1 þ aγÞþðÀ1 þ aÞð1 À aγÞτÀ2ð1À Þ:

We let τ∗ denote τ which makes it zero:

1 ε ∗ 1 þ a 1 þ aγ 2ð À1Þ τ ¼ > 1: 1 À a 1 À aγ

When τ ¼ 1 (no transaction costs), Gi ¼ Gj and ti ¼ tj, hðτ, Gi, Gj, ti, tjÞ¼1. For ∗ τ 2½1, τ , hðτ, Gi, Gj, ti, tjÞ is increasing function and thus it is necessarily larger ∗ than unity. For τ > τ , hðτ, Gi, Gj, ti, tjÞ is decreasing function. However, unless τ is 138 H. Kondo extremely large, hðτ, Gi, Gj, ti, tjÞ is still larger than unity. Therefore, when τ is small, hðτ, Gi, Gj, ti, tjÞ is necessarily larger than unity.

Proof of Proposition 1(ii) and (iii)

In the ratio of after-tax returns to capital ½ð1 À tiÞ=ð1 À tjÞðri=rjÞ, the part ðri=rjÞ depends on ni and τ, yet it is independent of ti. Thus, the gap in capital tax rate between two regions affects the size of ½ð1 À tiÞ=ð1 À tjÞðri=rjÞ only through the part ½ð1 À tiÞ=ð1 À tjÞ.Ifτ is low enough and if 1 ½ð1 À t1Þ=ð1 À t2Þ,1 ½ð1 À t1Þ=ð1 À t2Þðr1=r2Þ with n1 ¼ 1 necessarily holds. Moreover, if 1 ½ð1 À t1 Þ=ð1 À t2Þ ðr1=r2Þ with n1 ¼ 1, 1 ½ð1 À t2Þ=ð1 À t1Þðr2=r1Þ with n2 ¼ 1 also holds, and agglomeration can take place in region 2 either. In contrast, if ðr1=r2Þ < ½ð1 À t1Þ=ð1 À t2Þ, agglomeration takes place only in region 1.

Appendix 3: Derivations of (5.11) and (5.12)

The disposable income of a household in each region consists with the wage income, dividend from differentiated manufacturing goods ﬁrms, and the lump sum tax or transfer by the government. As a unit measure of capital is equally shared by the measure of two households in the economy, the measure of capital owned by a household is 1/2. When the manufacturing industry agglomerates in region i, = ð1ÀγÞα from (5.17) the dividend for a household is ð1 2Þð1 À tiÞri ¼ð1 À tiÞ 1Àaγ . Hence, disposable income of a household in each region is: ð1 À γÞα E ¼ 1 þð1 À t Þ À T , ð5:31Þ i i 1 À aγ i ð1 À γÞα E ¼ 1 þð1 À t Þ À T , ð5:32Þ j i 1 À aγ j where Ti and Tj are the lump sum tax or transfers in regions i and j, respectively. In region i where the manufacturing industry agglomerates the government has capital ð1ÀγÞα ’ tax revenue in the amount tiri ¼ 2ti 1Àaγ . Therefore, the government s budget constraint is: ð1 À γÞα T ¼ 2t þ G : ð5:33Þ i i 1 À aγ i

In region j the government has no capital tax revenue. Thus the government’s budget constraint is: 5 Public Policy and Economic Growth in the Integrating Japanese Economy 139

Tj ¼ Gj: ð5:34Þ

With (5.33) and (5.34), (5.31) and (5.32) are rewritten as: ð1 À γÞα E ¼ 1 þð1 þ t Þ À G , ð5:35Þ i i 1 À aγ i ð1 À γÞα E ¼ 1 þð1 À t Þ À G : ð5:36Þ j i 1 À aγ j

With (5.18), (5.19), (5.29), (5.30), (5.4), (5.35) and (5.36), utility in each region (5.10) is written as (5.11) and (5.12).

Comment Paper to Chapter 5

Takashi Fukushima National Graduate Institute for Policy Studies, 7-22-1 Roppongi, Minato-ku, Tokyo 106-8677, Japan e-mail: [email protected] This chapter by Prof. Kondo deals with the problem of overspending, or excessive infrastructure investment by local government. Traditionally, the problem was examined as a moral hazard of local government facing soft budget, that is, the local government has an incentive to overspend on the local public goods and infrastructures because the overspending is compensated by the central government in the form of local subsidy payments. Prof. Kondo offers another channel to explain the wasteful government spending by employing a general equilibrium model of industry agglomeration. To this end, a theoretical model of two identical regions is presented, where the utility functions, the production functions, and factor endowments are all identical between the two regions. The existence of transportation cost, increasing returns to scale in manufacturing sector, and the infrastructure provided by local government are the driving forces to inter-regional trades and agglomeration of the manufacturing industry. For a manufacturing industry, there have to be a sufficient profit incentive to find a favorable region to locate. Therefore, the transportation cost and the infrastructure provided by local government are the keys to create the profit incentive. When the transportation cost is high, locating and producing in both regions becomes more profitable than agglomerating in one region because importing and exporting the manufactured goods are more expensive with higher transportation cost. When the transportation cost is low, agglomeration in one region is more profitable with increasing returns to scale in producing manufactured goods. 140 H. Kondo

The regional government’s overspending occurs, in the Kondo model, when the two regional governments compete to attract manufacturing sector. When a region succeed to do so, she becomes better off. Therefore, Kondo argues, the regional governments tend to overspend. The spending is particularly wasteful when one of the regions invests heavily in local infrastructure and fails to attract the manufacturing industry. The model is interesting in its own light to describe the process of agglomeration, but it has some drawbacks as well. First, the model, as many other theoretical models, lacks a clear link to actual Japanese local economy. Some of the assumptions made in the model, such as two identical regions, no migration from one regions to the other, no explicit central government, etc., must be critically examined. Second, the model is very complex. I would like to see a little richer result if the model is this much complicated, or a little simpler model to show the same. A simpler partial equilibrium model may be sufﬁcient for the purpose.

References

Akai N (2014) Fiscal consolidation and local public finance in Japan: approach from the incentive design of Inter-governmental transfers. In: Ihori T, Terai K (eds) The political economy of fiscal consolidation in Japan. Springer, New York Baldwin RE, Forslid R, Martin P, Ottaviano GIP, Nicoud FR (2003) Economic geography and public policy. Princeton University Press, Princeton Baldwin RE, Krugman P (2004) Agglomeration, integration and tax harmonization. Eur Econ Rev 48:1–23 Borck R, Pfluger M (2006) Agglomeration and tax competition. Eur Econ Rev 50:647–668 Bucovetsky S (1991) Asymmetric tax competition. J Urban Econ 30:167–181 Chandra A, Thompson E (2000) Does public infrastructure affect economic activity? Evidence from the rural interstate highway system. Reg Sci Urban Econ 30:457–490 Duranton G, Overman HG (2005) Testing for localization using micro-geographic data. Rev Econ Stud 72:1077–1106 Ellison G, Glaeser E (1997) Geographical concentration in U.S. manufacturing industries: a dartboard approach J Polit Econ 105:889–927 Fujita M, Krugman P, Venables AJ (2001) The spatial economy: cities, regions, and international trade. MIT Press, Cambridge Haufler A, Wooton I (1999) Country size and tax competition for foreign direct investment. J Publ Econ 71:121–139 Holl A (2004a) Manufacturing location and impacts of road transport infrastructure: empirical evidence from Spain. Reg Sci Urban Econ 34:341–363 Holl A (2004b) Transport infrastructure, agglomeration economies and firm birth: empirical evidence from Portugal. J Reg Sci 44:693–712 Kim S (1995) Expansion of market and the geographic distribution of economic activities: the trends in U.S. regional manufacturing structure, 1860–1987. Q J Econ 110:881–908 Kind HJ, Knarvik KHM, Schjelderup G (2000) Competing for capital in a ‘lumpy’ world. J Publ Econ 78:253–274 Kondo H (2009) Providing public infrastructure competition and new economic geography. Sophia Econ Rev 54:1–28 Kondo H (2013) International R&D subsidy competition, industrial agglomeration and growth. J Int Econ 89:233–251 5 Public Policy and Economic Growth in the Integrating Japanese Economy 141

Krugman P (1991) Geography and trade. MIT Press, Cambridge Krugman P, Venables AJ (1995) Globalization and the inequality of nations. Q J Econ 110:857– 880 Ludema RD, Wooton I (2000) Economic geography and the ﬁscal effects of regional integration. J Int Econ 52:331–357 Nagamine J, Katayama T (2001) kokyo-toushi to douro-seisaku (Public Investment and Road Policy). Keiso Shobo Ottaviano GIP, van Ypersele T (2005) Market size and tax competition. J Int Econ 67:25–46 Venables AJ (1996) Equilibrium locations of vertically linked industries. Int Econ Rev 37:341– 359 Wildasin D (1988) Nash equilibria in models of ﬁscal competition. J Publ Econ 35:229–240 Wilson J (1986) A theory of interregional tax competition. J Urban Econ 19:296–315 Wilson J (1987) Trade, capital mobility, and tax competition. J Polit Econ 95:835–856 Wilson J (1991) Tax competition with interregional differences in factor endowments. Reg Sci Urban Econ 21:423–452 Yoshino N, Nakano H (1994) Shutoken he no koukyou-toushi. Hatta T ens. Tokyo ikkyoku-shuchu no keizai-bunseki (Public investment in metropolitan area. In: Hatta T (ed) Economic analysis of Tokyo monocentralizatio). The Nikkei Yoshino N, Nakano H (1996) koukyo-toushi no chiiki-haibun to seisan-kouka (Interregional distribution and productivity effects of public investment). Financ Rev 41:16–26 Yoshino N, Nakajima T (1999) koukyo-toushi no keizai-kouka (Economic effects of public investment). Nippon Hyoron Sha Zodrow R, Mieszkowski P (1986) Pigou, Tiebout, property taxation, and the underprovision of local public goods. J Urban Econ 19:356–370 Part III Institutional Reforms Necessary for Fiscal Consolidation Chapter 6 Tax Policy Under the “Generational Election System”

Takero Doi

Abstract This chapter investigates the effects of introducing the “generational election system” proposed by Ihori and Doi (Nihon-seiji no keizai-bunseki (Economic analysis of Japanese politics), Bokutaku-Sha, Tokyo, 1998). The generational election system (or the election district by generation) consists of election districts divided by not only region but also generation. In industrial countries, intergenerational conflicts of interest are large at present. In particular, the older generation has more political power because of aging and fewer children. In an electoral system that consists of election districts divided only by region, conflicts of interest among regions can be dealt with in the Congress, but intergenerational conflicts are buried in each district because the opinions of older people dominate those of younger people. Therefore, this chapter analyzes the effects of introducing the generational election system using an overlapping generations model. The results of the voting equilibrium show that the preferred policy of the younger generation can be better represented in the generational election system compared with the current majoritarian system. Furthermore, the selected policy does not depend on the turnout rate of the younger generation. These results suggest that introducing the generational election system benefits both the younger and future generations.

Keywords Generational Election System • Intergenerational equity • Tax policy • Turnout rate • Voting equilibrium

I would like to thank Prof. Haruki Kondo and other participants of the tenth Irvine-Japan Conference on Public Policy at University of California, Irvine and the 15th Annual Conference of the Association for Public Economic Theory at the University of Washington for helpful comments. I have received ﬁnancial support from the Grant-in-Aid for Specially Promoted Research from Japan’s Ministry of Education, Culture, Sports, Science and Technology: “Economic Analysis of Intergenerational Issues: Searching for Further Development” (Grant Number 22000001). All remaining errors are my own. T. Doi (*) Keio University, 2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan e-mail: [email protected]

1 The “Generational Election System”

This chapter investigates the effects of introducing the “generational election system” proposed by Ihori and Doi (1998). The generational election system (or the election district by generation) consists of election districts divided by not only region but also generation. In Japan, as well as industrial countries, intergenerational conflicts of interest are large at present. In particular, the older generation has more political power because of aging and fewer children. In the current electoral system that consists of election districts divided by only region, conflicts of interest among regions can be dealt with in the Congress, but intergenerational conflicts are buried in each district because the opinions of older people dominate those of younger people. In particular, the turnout rate of the younger generation in national elections is notably lower than that of the older generation in Japan. Figure 6.1 shows turnout rates by age group in national elections in Japan. It suggests that the opinions of older people dominate those of younger people in each district. If election districts were divided by generation, such different turnout rates could block to reflect in election results. Moreover, there is malapportionment in Japanese national elections. Electoral districts in urban areas, where more young people live, are more populous, while those in rural areas, where more old people live, are less populous. Though the Supreme Court rendered a judgment that the degree of malapportionment of allocated seats in the national election was “state of unconstitutional,” it is

% 90

0 1989 1990 1992 1993 1995 1996 1998 2000 2001 2003 2004 2005 2007 2009 2010 2012 2013

20s 30s 40s 50s 60s 70s and over

Fig. 6.1 Turnout rates of national elections in Japan. Source: Association for Promoting Fair Elections website 6 Tax Policy Under the “Generational Election System” 147 fundamentally unchanged. The ratio of the most populous electoral district to the least populous was 2.4 in the 2012 general House of Representatives election, and 4.77 in the 2013 regular House of Councilors election. This malapportionment reduces intergenerational equity. Therefore, the generational election system may be a solution in such situations. This chapter analyzes the effects of introducing the generational election system using an overlapping generations model.

2 Model

The model established in this chapter is based on the overlapping generations model in Ihori (1987). In this model, each household lives for two periods. Nt denotes the population of generation t and n denotes the rate of population growth. This means that Nt ¼ (1 + n)Nt À 1. There are Z election districts in this economy. Households do not migrate across regions over time. The utility function of a household of generation t can be written as

i ¼ i þ β i þ γ i þ δ i ð : Þ u lnct lnxtþ1 t lngt t lngtþ1 6 1

γi δi where t and t vary among individuals. The budget constraint of the household i of generation t is

i ¼ i À i À τ ct wt st t i ¼ ðÞþ i À ρ xtþ1 1 rtþ1 st tþ1

Hence, the lifetime budget constraint of the household can be rewritten as

x i ρ i þ tþ1 ¼ i À τ À tþ1 : ct wt t 1 þ rtþ1 1 þ rtþ1

The budget constraint of the government can be written as

τ þ ρ ¼ Nt t NtÀ1 t Gt ρ τ þ t ¼ Gt : t gt 1 þ n Nt

τ ¼ ρ 2þn τ ¼ Assume t t. Then 1þn t gt. The household i of generation t maximizes its lifetime utility with respect to st, τt, and ρt+1 given the tax rates or amounts of public goods in both periods. The optimal amount of consumption in each period is determined by the ﬁrst- order conditions 148 T. Doi

τ c i ¼ t t γ i t βρ þ x i ¼ t 1 tþ1 δ i t i xtþ1 ¼ βðÞþ þ : i 1 rt 1 ct

Under these conditions, tax rates which the household i of generation t prefers are as follows.

γ i τ i ¼ t w i ð6:2Þ t þ β þ γ i þ δ i t 1 t t δ i i t i ρ ¼ ðÞ1 þ r þ w ð6:3Þ tþ1 þ β þ γ i þ δ i t 1 t 1 t t

A household, that is, a voter, casts a vote sincerely in an election. Furthermore, I assume that malapportionment in the electoral system is not allowed in this economy.

3 Properties of Voting Equilibria with Homogenous Districts

In this section, I investigate the properties of voting equilibria if all election districts are homogenous. Before considering a voting equilibrium, some assumptions γi γ are made. For simplicity, the parameter t is the same ( t) for all members of δ ¼ βγ i generation t. I assume t t. wt is predetermined for the household and different i for each household. The distribution of wt is presumed to follow a uniform τi ρi distribution. Hence, the distributions of t and t follow uniform distributions because of Eqs. (6.2) and (6.3). Without loss of generality, I assume there is no population growth (n ¼ 0: and NtÀ1 ¼ Nt N), and that there are six election districts (Z ¼ 6) in this economy, for simplicity. Therefore, there are N/3 electorates per election district. In this section, the income distribution is assumed to be the same in each district. In this sense, all i districtsÂÃ are homogenous. Let w t denote the income of household i in a district, I set i ; wt U wt wt , where wt denotes the lower bound of the uniform distribution, and wt denotes the upper bound of the uniform distribution. The household whose income is wt prefers γ τ t w : t ðÞþ β ðÞþ γ t 1 1 t 6 Tax Policy Under the “Generational Election System” 149

Furthermore, the household whose income is wt prefers γ τ t : t ðÞþ β ðÞþ γ wt 1 1 t

ÂÃ τ iÀτ τ i τ ; τ τi t t Therefore, t U t t . Hence, the cumulative distribution function of t is τ Àτ ÂÃ t t τ i ∈ τ ; τ for t t t . À In an analogous way, the household of generation t 1 whose income is wtÀ1 prefers βγ t ρ ðÞ1 þ rt w À : t ðÞþ β ðÞþ γ t 1 1 1 t

Furthermore, the household whose income is wtÀ1 prefers βγ ρ t ðÞþ : t ðÞþ β ðÞþ γ 1 rt wtÀ1 1 1 t hi ρ iÀρ t ρ i ρ ; ρ ρi t Therefore, U t . Hence, the cumulative distribution function of is ρ Àρ t t t t hi t for ρ i ∈ ρ ; ρ . t t t This economy implements an indirect democracy, in which the election for the Congress representatives occurs in each election district in the ﬁrst stage, and the representatives select a lump-sum tax rate by ballot in the Congress in the ﬁnal stage. Furthermore, I can consider a voting equilibrium in a setting where the median voter theorem holds. In comparing both electoral systems, the majoritarian system and the genera- β > τi < ρi i ¼ i tional election system, I assume (1 + rt) 1. This means that t t,ifwt wt À 1 and γt ¼ γtÀ1 from Eqs. (6.2) and (6.3). Moreover, electorates of the younger generation may abstain in the election. I assume that the turnout rate of the younger generation is α(Â100) %, where 0 α 1, and electorates of the older generation are certain to vote. The turnout rate is presumed to be the same across all election districts and to be independent of income levels.

3.1 Voting Equilibrium of the Majoritarian System

First, the following majoritarian system is introduced in this model. This means that there are N/6 electorates for each generation. In each election district in period t, the tax rate preferred by the pivotal voter, τt, is as follows. 150 T. Doi

τ À ρ τ À τ t 1 þ α α t t þ t ¼ , τt À τ ρ À ρ 2 t t t because of the cumulative distribution functions and the median voter theorem. Therefore, in each election district,

τtρ À ðÞ1 À α τ ρ À ατ ρ t t t t t τt ¼ : ð6:4Þ τt À τ þ α ρ À ρ t t t In each election district, a representative who has the preferred tax rate Eq. (6.4)is elected. Next, the representatives from each election district convene in the Congress to decide the nationwide tax rate. However, all districts are homogenous, such that all representatives prefer the same tax rate τt. Thus, the preferred tax rate in the Congress is τt, in Eq. (6.4). Equation (6.4) implies that the tax rate determined in the Congress depends on the turnout rate α. The tax rate depends strongly on the values of the older generation, ρ and ρ , when the turnout rate α decreases. t t

3.2 Voting Equilibrium of the Generational Election System

Ihori and Doi (1998) proposes the “generational election system” that consists of election districts divided by not only region but also generation. Therefore, in the model below, the election districts are divided by not only region but also generation. Thus, there are three election districts of the younger generation. Furthermore, there are three election districts of the older generation. For the younger generation in period t, the pivotal voter is the voter who has the median income, because both the income distribution and the distribution of the preferred tax rate are uniform distributions and the turnout rate, α, is independent of the income level. Thus, the tax rate preferred by the median voter of the younger generation, τt, in each election district is as follows: τ þ τ τ ¼ t t , ð6:5Þ t 2 because of the cumulative distribution functions and the median voter theorem. Similarly, for the older generation in period t, the pivotal voter is the voter who has the median income. Thus, the tax rate preferred by the median voter of the older generation, ρt, in each election district is as follows: ρ þ ρ t ρ ¼ t , ð6:6Þ t 2 because of the cumulative distribution functions and the median voter theorem. 6 Tax Policy Under the “Generational Election System” 151

Finally, I investigate the voting equilibrium of the generational election system in the Congress. The tax rate is determined based on the median voter theorem. This result depends on the values of τt and ρt. However, from Eqs. (6.5) and (6.6), the tax rate determined in the Congress does not depend on the turnout rate, α. This is an important property of the voting equilibrium.

4 Properties of Voting Equilibria with Heterogeneous Districts

In this section, I investigate the properties of voting equilibria if all electoral districts are heterogeneous. This case is more general and complicated than that in the previous section. Before considering a voting equilibrium with heterogeneous districts, I make the γi γ same assumptions as in the previous section; t is the same ( t) for all persons of δ ¼ βγ β > i generation t, t t, (1 + rt) 1, and the distribution of wt is presumed to follow a uniform distribution. Furthermore, n ¼ 0(NtÀ1 ¼ Nt N) and there are six election districts (Z ¼ 6) in this economy. I assume that there are three types of people in each generation; types A, B, and C. There is an equal number of each type (N/3). The distribution of each type is ij assumed to be different. In this sense, all districtshi are heterogeneous. Let wt denote ij j; j ¼ the income of household i in type j, I set wt U wt wt for j A, B, and C, where j j wt denotes the lower bound of the uniform distribution of type j, and wt denotes the A < B < C upper bound of the uniform distribution of type j. I assume that wt wt wt and A < B < C wt wt wt in each generation. j The household whose income is wt prefers γ τ j t w j j ¼ A, B, and C: t ðÞþ β ðÞþ γ t 1 1 t

j Furthermore, the household whose income is wt prefers γ τ j t j ¼ : t ðÞþ β ðÞþ γ wt j A, B, and C 1 1 t hi τi τ ij τ j; τ j As the distributions of t are uniform because of Eq. (6.2), t U t t for τ ijÀτ j ¼ τij t t j A, B, and C. Hence, the cumulative distribution function of t is τ jÀτ j for hi t t τ ij ∈ τ j; τ j τ A < τ B < τ C τ A < τ B < τ C t t t . In this situation, t t t and t t t . À j In an analogous way, the household of generation t 1 whose income is wtÀ1 prefers 152 T. Doi

βγ j t j ρ ðÞ1 þ rt w À j ¼ A, B, and C: t ðÞþ β ðÞþ γ t 1 1 1 t

j Furthermore, the household whose income is wtÀ1 prefers βγ ρ j t ðÞþ j ¼ : t ðÞþ β ðÞþ γ 1 rt wtÀ1 j A, B, and C 1 1 t hi ρij ρ ij ρ j; ρ j ¼ As t follows uniform distribution because of Eq. (6.3), t U t for j A, B, t hi ρ ijÀρ j t ρij t ρ ij ∈ ρ j; ρ j and C. Hence, the cumulative distribution function of t is ρ jÀρ j for t t .In t t t this situation, ρ A < ρ B < ρ C and ρ A < ρ B < ρ C. t t t t t t This economy implements an indirect democracy, in which a representative is elected to the Congress from each election district in the ﬁrst stage, and the representatives then select a lump-sum tax rate by ballot in the Congress in the ﬁnal stage. Furthermore, I can consider a voting equilibrium in which the median voter theorem holds. As there are six election districts in this economy, there are N/3 electorates per election district. Moreover, as in the previous section, electorates from the younger generation may abstain in the election. The turnout rate of the younger generation is α, where 0 α 1, and electorates from the older generation are sure to vote. The turnout rate is presumed to be the same in all election districts and to be independent of the income level.

4.1 Voting Equilibrium of the Majoritarian System

First, the following majoritarian system is introduced in this model. I assume that election districts are divided by type, which means that there are N/6 electorates of type j of generation t and N/6 electorates of type j from generation tÀ1 in each district, for j ¼ A, B, and C. Furthermore, there are two election districts that are comprised of type j electorates. In the type j election district in period t, the following is true for the tax rate τj preferred by the pivotal voter, t.

j j τ j À ρ j τ À τ t 1 þ α α t t þ t ¼ j ¼ A, B, and C, τ j À τ j ρ j À ρ j 2 t t t t because of the cumulative distribution functions and the median voter theorem. Figure 6.2 describes this situation from the viewpoint of the voter distribution. Therefore, in an election district of type j, 6 Tax Policy Under the “Generational Election System” 153

ij tt j j tt tt

r ij t r j r j t

1 + a 2 ij ij t t r t

Fig. 6.2 Distribution in each election district under the majoritarian system

τ jρ j À ðÞ1 À α τ jρ j À ατ jρ j τ j ¼ t t t t t t ¼ : ð : Þ t j A, B, and C 6 7 τ j À τ j þ α ρ j À ρ j t t t t

From each election district, a representative who has the preferred tax rate Eq. (6.7) is elected. Next, representatives from every election district are convened to decide the tax rate nationwide in the Congress. There are two representatives whose preferred tax τA τB rate is t , two representatives whose preferred tax rate is t , and two representatives τC whose preferred tax rate is t . Under certain conditions, the preferred tax rate of the median representative in the Congress is 154 T. Doi

τ Bρ B À ðÞ1 À α τ Bρ B À ατ Bρ B τ B ¼ t t t t t t : ð : Þ t 6 8 τ B À τ B þ α ρ B À ρ B t t t t

Equation (6.8) implies that the tax rate determined in the Congress depends on the turnout rate α. The tax rate depends mainly on the values of the older generation, ρ j and ρ j, when the turnout rate α decreases. t t

4.2 Voting Equilibrium of the Generational Election System

In this model, under the generational election system, the election districts are divided by not only region but also generation. Therefore, there is one election district of the younger generation of each type because the population of the younger generation of each type is N/3. Furthermore, there is one election district of the older generation of each type. For the younger generation in period t, the pivotal voter is the voter who has the median income, because both the income distribution and distribution of the preferred tax rate have uniform distributions and the turnout rate, α, is independent of income level. Therefore, the tax rate preferred by the median voter of the τ j younger generation, t , for type j election district is as follows.

τ j þ τ j τ j ¼ t t j ¼ A, B, and C, ð6:9Þ t 2 because of the cumulative distribution functions and the median voter theorem. Figure 6.3 shows this situation from the viewpoint of the voter distribution. Similarly, for the older generation in period t, the pivotal voter is the voter who has the median income. Therefore, the tax rate preferred by the median voter of the ρj older generation, t, for type j election district is as follows.

ρ j þ ρ j t ρ j ¼ t j ¼ A, B, and C, ð6:10Þ t 2 because of the cumulative distribution functions and the median voter theorem (see Fig. 6.3). Finally, I investigate the voting equilibrium of the generational election system in the Congress. The tax rate is determined based on the median voter theorem. This τj ρj result depends on the values of t and t. However, from Eqs. (6.9) and (6.10), the tax rate determined in the Congress does not depend on the turnout rate, α. This is an important property of the voting equilibrium. 6 Tax Policy Under the “Generational Election System” 155

• Structure in each election district of the younger generation

a generation t ij t t j j tt tt Median Voter

• Structure in each election district of the older generation

generation t-1

ij r t j j rt rt Median Voter

Fig. 6.3 Distribution in each election district under the generational election system

5 Numerical Analysis of Voting Equilibria

Now, I set the values of the parameters in this model. Furthermore, I set β ¼ 0.95, γ ¼ 0.3, α ¼ 0.5, and the interest rate at 5 %. I assume that type A implies iA iB 1 wt 4, as shown in Table 6.1. Similarly, type B implies 2 wt 5, and iC type C implies 3 wt 6 as shown in Table 6.1. The income distribution of generation tÀ1 is presumed to be the same as for generation t.

5.1 Voting Equilibrium of the Majoritarian System

In each election district, there exist households of both the young generation and the old generation. I can calculate the upper bound and lower bound of the uniform distribution of the preferred tax rate in each type of generation, as shown in Table 6.1. In this situation, the tax rate preferred by the median voter in the Congress is 0.4759. 156 T. Doi

Table 6.1 Baseline case b = 0.99 r = 0.05 g = 0.3 a = 0.5

Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 4 5 6 lower bound 1 2 3

Results Type ij τ t A BC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ij ρ t ABC upper bound 0.4822 0.6027 0.7233 lower bound 0.1205 0.2411 0.3616

under the majoritarian system ABC j τ t 0.3569 0.4759 0.5949 median tax rate (M) 0.4759

under the generational election system ABC τ jt 0.2899 0.4059 0.5218 ρ jt 0.3014 0.4219 0.5425 median tax rate (G) 0.4139

(M)/(G) 1.1499 6 Tax Policy Under the “Generational Election System” 157

5.2 Voting Equilibrium of the Generational Election System

Under the generational election system, election districts are divided by not only region but also by generation. First, I consider the election districts of the younger generation. The preferred tax rate of the median voter of the younger generation in each district is shown in Table 6.1. Second, I examine the election districts of the older generation. The preferred tax rate of the median voter of the older generation in each district is shown in Table 6.1. In the Congress, the tax rate is determined by six representatives. In this situation, the median voter theorem holds. Therefore, the tax rate determined in the Congress is 0.4139. The tax rate determined in the generational election system is lower than that in the majoritarian system. The ratio of the tax rate determined in the majoritarian system to that determined in the generational election system is 1.1499, as shown in Table 6.1.

5.3 A Higher Turnout Rate

Table 6.2 shows the results when the turnout rate of the younger generation is higher than that of the baseline case. I set α ¼ 0.75. The voting equilibrium of the generational election system is the same as that in Sect. 6.5.2. However, the voting equilibrium of the majoritarian system is different from that in Sect. 6.5.1. The tax rate under in the majoritarian system for α ¼ 0.75 is closer than that in the generational election system. The ratio of the tax rate determined in the majoritarian system to that determined in the generational election system is 1.0638, as shown in Table 6.2.

5.4 A Lower Turnout Rate

Table 6.3 shows the results when the turnout rate of the younger generation is lower than that of the baseline case. I set α ¼ 0.25. The voting equilibrium of the generational election system is the same as Sect. 6.5.2. However, the voting equilibrium of the majoritarian system is different from that in Sect. 6.5.1. The tax rate under the majoritarian system for α ¼ 0.25 is larger than that in the generational election system. The ratio of the tax rate determined in the majoritarian system to that determined in the generational election system is 1.2714, as shown in Table 6.3. 158 T. Doi

Table 6.2 In case of α ¼ 0.75 b = 0.99 r = 0.05 g = 0.3 a = 0.75

Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 4 5 6 lower bound 1 2 3

Type ij τ t ABC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ρij t ABC upper bound 0.4822 0.6027 0.7233 lower bound 0.1205 0.2411 0.3616

under the majoritarian system ABC j τ t 0.3217 0.4403 0.5588 median tax rate (M)

under the generational election system ABC j τ t 0.2899 0.4059 0.5218 j ρ t 0.3014 0.4219 0.5425 median tax rate (G)

(M)/(G) 1.0638 6 Tax Policy Under the “Generational Election System” 159

Table 6.3 In case of α ¼ 0.25 b = 0.99 r = 0.05 g = 0.3 a = 0.25 Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 4 5 6 lower bound 1 2 3

Results Type ij τ t ABC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ij ρ t ABC upper bound 0.4822 0.6027 0.7233 lower bound 0.1205 0.2411 0.3616

under the majoritarian system ABC j τ t 0.4066 0.5262 0.6458 median tax rate (M) 0.5262

under the generational election system ABC j τ t 0.2899 0.4059 0.5218 j ρ t 0.3014 0.4219 0.5425 median tax rate (G) 0.4139

(M)/(G) 1.2714 160 T. Doi

These results show that the preferred tax rate of the younger generation, which is smaller than that of the older generation, can be achieved in the Congress by introducing the generational election system.

5.5 Voting Equilibrium in the Case Where the Older Generation Is Richer

I also assume that the mean and median incomes of generation tÀ1 are higher than iA those of generation t. I assume that type A implies 2 wt À 1 8, as shown in iB Table 6.1. Similarly, type B implies 4 wt À 1 10, and type C implies iC 6 wt À 1 12, as shown in Table 6.4. I can calculate the upper bound and lower bound of the uniform distribution of the preferred tax rate in each type of generation, as shown in Table 6.4. In the majoritarian system, the tax rate preferred by the median voter in the Congress is 0.7093, as shown in Table 6.4. In contrast, under the generational election system, the tax rate determined in the Congress is 0.5623. The ratio of the tax rate determined in the majoritarian system to that in the generational election system is 1.2614, as shown in Table 6.4.This ratio is higher than the one in the case where the income distribution of generation tÀ1 is presumed to be the same as that of generation t, as shown in Table 6.1. Furthermore, I analyze the case where the turnout rate of the younger generation is higher. I set α ¼ 0.75. In this case, the tax rate preferred by the median voter in the Congress under the majoritarian system is 0.6123, as shown in Table 6.5. The voting equilibrium of the generational election system is the same as that in Table 6.4. The ratio of the tax rate determined in the majoritarian system to that determined in the generational election system is 1.0890, as shown in Table 6.5. This ratio is higher than the one in the case where the income distribution of generation tÀ1 is presumed to be the same as for generation t, as shown in Table 6.2. Moreover, I investigate the case where the turnout rate of the younger generation is lower. I set α ¼ 0.25. In this case, the tax rate preferred by the median voter in the Congress under the majoritarian system is 0.8725, as shown in Table 6.6. The voting equilibrium of the generational election system is the same as that in Table 6.4. The ratio of the tax rate determined in the majoritarian system to that determined in the generational election system is 1.5517, as shown in Table 6.6. This ratio is higher than the one in the case where the income distribution of generation tÀ1 is presumed to be the same as for generation t, as shown in Table 6.3. 6 Tax Policy Under the “Generational Election System” 161

Table 6.4 In case that the older generation is richer (α ¼ 0.5) b = 0.99 R = 0.05 g = 0.3 a = 0.5

Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 8 10 12 lower bound 2 4 6

Results Type ij τ t ABC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ij ρ t ABC upper bound 0.9644 1.2055 1.4465 lower bound 0.2411 0.4822 0.7233

under the majoritarian system ABC j τ t 0.5319 0.7093 0.8866 median tax rate (M) 0.7093

under the generational election system ABC j τ t 0.2899 0.4059 0.5218 j ρ t 0.6027 0.8438 1.0849 median tax rate (G) 0.5623

(M)/(G) 1.2614 162 T. Doi

Table 6.5 In case that the older generation is richer (α ¼ 0.75) b = 0.99 r = 0.05 g = 0.3 a = 0.75

Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 8 10 12 lower bound 2 4 6

Results Type ij τ t ABC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ij ρ t ABC upper bound 0.9644 1.2055 1.4465 lower bound 0.2411 0.4822 0.7233

under the majoritarian system ABC j τ t 0.4475 0.6123 0.7772 median tax rate (M) 0.6123

under the generational election system ABC j τ t 0.2899 0.4059 0.5218 j ρ t 0.6027 0.8438 1.0849 median tax rate (G) 0.5623

(M)/(G) 1.0890 6 Tax Policy Under the “Generational Election System” 163

Table 6.6 In case that the older generation is richer (α ¼ 0.25) b = 0.99 r = 0.05 g = 0.3 a = 0.25

Type

wt ABC upper bound 4 5 6 lower bound 1 2 3

wt-1 ABC upper bound 8 10 12 lower bound 2 4 6

Results Type ij τ t ABC upper bound 0.4639 0.5798 0.6958 lower bound 0.1160 0.2319 0.3479

ρij t ABC upper bound 0.9644 1.2055 1.4465 lower bound 0.2411 0.4822 0.7233

under the majoritarian system ABC j τ t 0.6742 0.8725 1.0708 median tax rate (M) 0.8725

under the generational election system ABC j τ t 0.2899 0.4059 0.5218 j ρ t 0.6027 0.8438 1.0849 median tax rate (G) 0.5623

(M)/(G) 1.5517 164 T. Doi

6 Conclusion

The generational election system, proposed by Ihori and Doi (1998) consists of election districts divided by not only region but also generation. This chapter examined the effects of introducing the generational election system using an overlapping generations model. These results show that the preferred tax rate of the younger generation, which is smaller than the older generation, can be achieved in the Congress by introducing the generational election system. Furthermore, the tax rate determined in the Congress does not depend on the turnout rate of the younger generation. This is an important property of the generational election system. This suggests that introducing the generational election system beneﬁts both the younger and future generations.

Comment Paper to Chapter 6

Haruo Kondoh Seinan Gakuin University, 6-2-92 Nishijin, Sawara-ku, Fukuoka 814-8511, Japan e-mail: [email protected] This paper analyzes the consequences of introducing the “electoral system by generation” theoretically and numerically. First, I will overview the paper. In the first step, the household’s optimal consumption and “preferred tax rate” is characterized by using a standard overlapping generation model. In the second step, based on the first-stage results, the voting equilibria are examined. In the third step, by using numerical analysis, voting equilibria of a majoritarian system are compared with those of a system by generation. The main results obtained in this paper are as follows. (1) Under some assumptions, the tax rate preferred by the older generation (denoted by ρ) is higher than the tax rate preferred by the younger generation (denoted by τ). (2) Numerical examples show that the younger generation’s preference is reflected more in the electoral system by generation, in particular the turnover rate of the young is low. Secondly, I will present my overall comments about this paper. This work is meaningful in the sense that it is the very first attempt to analyze the effects of the electoral system by generation. It is also meaningful because numerical analysis shows the electoral system is beneficial in making policy reflective of a younger generation’s preference. However, some assumptions seem to be too restrictive or at least worth examining as to whether or not changes in the conclusion depend on varying assumptions. To be concrete, in numerical simulation, no population growth (Nt ¼ Nt + 1) and the same income distribution between generations are assumed. In this setting, the conclusion of this paper is quite plausible. However, if the “no population growth” and “same income distribution” assumptions are relaxed, the conclusion may change. For example, it may be important to check 6 Tax Policy Under the “Generational Election System” 165 the robustness of the conclusion under the following alternative assumptions. What if we assume that distribution of tax preference is different by generation? What if we assume negative population growth (Nt > Nt + 1)? What if we assume the ratio of old to young is different between regions? This paper’s contribution will be more significant if the original conclusion of the paper holds under these alternative assumptions. Finally, I will give some suggestions. It may be important to discuss the validity of assumptions such as no population growth or income distribution of young and old. In addition, it may be useful to disaggregate public goods into multiple types of public goods (e.g., public goods for young and for old) or introduce assumptions about political behavior in order to emphasize conflict between generations.

References

Ihori T (1987) Tax reform and generational incidence. J Public Econ 33:377–387 Ihori T, Doi T (1998) Nihon-seiji no keizai-bunseki (Economic analysis of Japanese politics). Bokutaku-Sha, Tokyo Chapter 7 Budgets Under Delegation

Kimiko Terai and Amihai Glazer

Abstract Consider a principal who sets a budget that the agent allocates among different services. Because the preferences of the agent may differ from those of the principal, the budget the principal sets can be lower or higher than in the ﬁrst-best solution. When the principal is uncertain about the agent’s preferences, the agent may choose an allocation that signals his type, thereby affecting the size of the budget the principal will set in the following period. The equilibrium may have separation or pooling. In a pooling equilibrium, the agent may mis-represent his preferences, aiming to get a large budget in the future period.

Keywords Budget process • Delegation • Signaling

Notation

pi Marginal cost of service i v(Á ) Utility from consumption of a service X Budget set by the principal xi Quantity of service i provided k xti Choice of service i in period t by a type-k agent in the absence of signaling considerations k∗ xti Optimal choice of service i in period t by an agent of type k when signaling considerations are present P xi ﬁrst-best allocation for the principal of service i ∗ X ﬁrst-best level of the budget which the principal would give the agent when their preferences are the same Y Principal’s endowment in each period αA Parameter describing preferences of the agent

K. Terai (*) Keio University, 2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan e-mail: [email protected] A. Glazer University of California, Irvine, Irvine, CA 92697, USA e-mail: [email protected]

© Springer Japan 2015 167 T. Ihori, K. Terai (eds.), The Political Economy of Fiscal Consolidation in Japan, Advances in Japanese Business and Economics 8, DOI 10.1007/978-4-431-55127-0_7 168 K. Terai and A. Glazer

αP Parameter describing preferences of the principal πL Prior probability that αA ¼ αL π~ L Posterior probability that agent is of type-L, or has αA ¼ αL ~ ¼ q L Posterior probability that p2 pL

1 Introduction

Government often has one person or group set a budget, and another group or person decide how to allocate that budget across different services. We can think of Congress allocating a budget to the Federal Aviation Administration, with the Administration deciding where to allocate air traffic personnel, how many hours each facility should be open, and so on. A state legislature may give a budget to a state university, with the university deciding how many faculty to hire in the humanities, how many in the social sciences, and so on. Or, a central government may transfer funds to a local government, with the local government deciding how to spend the budget. Such situations are examples of principal-agent situations, where the principal sets the budget, and the agent chooses how to spend the budget. Problems with such delegation arise when the principal’s preferences differ from the agent’s. The first issue we address is whether the principal will give a budget larger or smaller than he would if the agent’s preferences were the same as the principal’s. The answer is not obvious. The principal may want to give a small budget because he fears that the agent will spend the money in ways the principal dislikes. But the opposite effect may arise if the principal worries that the agent will spend little on the service the principal favors; to assure provision of that service the principal may have to give a large budget. The second issue this chapter addresses is strategic behavior by the agent. The agent, who allocates any budget given him across the different services, prefers a large budget. If the principal is unsure of the agent’s type, then the agent in one period may allocate the budget in a way that would induce the principal to give a large budget in the next period. We also examine the Japanese fiscal problem, focusing on public works. The Ministry of Finance can be viewed as the principal, and the spending ministry as the agent. The spending ministry may be pressured by local governments or by local interests, and may have underestimated the costs of local public works with the aim of increasing the budget the Ministry of Finance provides. Consequently, some public projects including roads, bridges, and airports may be inefficient.

2 Literature

2.1 Reputation

One issue we shall address is how an agent may take action which affects the principal’s beliefs about the agent’s type. A large literature examines behavior 7 Budgets Under Delegation 169 intended to affect reputation, with the principal often viewed as voters, and the agent as an elected official. Reputational concerns may lead a politician to terminate a policy that he, but not the voters, knows has failed (Beniers and Dur 2007). And reputational concerns can give rise to political correctness: an adviser who wishes to avoid a reputation for bias may not truthfully reveal his information (Morris 2001). A career-concern model where the incumbent attempts to signal ability is analyzed by Canes-Wrone et al. (2001). Our assumptions resemble those in Fox (2007), who shows that an agent who cares about his reputation may adopt policies commonly associated with a high-quality agent, though the state of nature would call for a different policy. He further shows that if an agent can hide his actions from the public, this distortion can be reduced. Prat (2005) shows that a career-driven agent who knows that his action is observed has an incentive to conform. The principal is damaged by such behavior, and may want to commit to keep the agent’s action secret. Carpenter (2004) uses a career- concerns model to argue that the U.S. Food and Drug Administration may delay approving some drugs because it wants to safeguard its reputation for protecting the public’s health. An incumbent may increase his chances of winning election by pandering to the public, taking actions the public may incorrectly believe are best (Maskin and Tirole 2004; Smart and Sturm 2013). If a project will likely fail even under a skilled leader, a leader (whether skilled or not) may prefer projects likely to fail over projects likely to succeed (Majumdar and Mukand 2004). Indeed, a politician with a bad reputation may favor a highly risky policy—if the policy fails, he would have lost the next election anyway, but if the policy succeeds, his reputation and so his chances of re-election improve. This idea is applied by Hess and Orphanides (1995) to claim that a president with a bad reputation who goes to war gets an opportunity to improve his reputation. Relatedly, if voters learn about a politician’s ability from the performance of a new project he undertakes, then an incumbent ignorant of his own ability will adopt too many projects if he is at risk of losing re-election; he will adopt too few projects if he is likely to win re-election (Biglaiser and Mezzetti, 1997). Kessler (2010) considers a model in which local government officials with private information about the benefits of a policy have an incentive, when engaging in cheap talk, to hide that information from the central government. The imperfect information held by the central government can generate overspending, universalism, and uniformity. Most of that literature has a bad type want to mimic a good type. In our model, a good type may want to mimic a bad type, because that can increase the budget. The strategic behavior of agents we examine relates to the ratchet effect, which considers a worker who may exert little effort today: he anticipates that the employer may infer that high effort signals a low cost of effort, inducing the employer to offer a lower wage in the future. For example, in Lazear (1986) and Gibbons (1987) the worker has private information about the firm (such as the job’s difficulty), which he is reluctant to reveal. In Aron (1987) and in Kanemoto and MacLeod (1992) the worker has private information about a worker-specific attri- bute, such as ability. 170 K. Terai and A. Glazer

2.2 Preferences of Agents

The agent’s preferences may differ from the principal’s because the agent is corrupt, or influenced by special interest groups. The differences can also appear when the agent is intrinsically motivated, caring about policy or outcomes, rather than only about the income he earns. Work in the public administration literature provides evidence of intrinsic motivation among public-sector employees (Guyot 1962; Crewson 1997). Other work investigates whether individuals with higher levels of intrinsic motivation are more often found in the public sector. For example, Gregg et al. (2011) use British survey data to investigate whether prosocial behavior (as measured by the probability of working extra, unpaid, hours) is more prevalent in the nonprofit than in the for-profit sector. These authors find that individuals in the nonprofit sector are significantly more likely to work such extra hours. Survey data studied by Georgellis et al. (2011) also support the hypothesis that individuals are attracted to the public sector more by intrinsic than extrinsic rewards.1

2.3 Competition Among Agents

Our consideration of a principal allocating money among agents relates to work on the Good Samaritan Dilemma, where an altruist donor gives more money to a recipient the poorer is that recipient (Buchanan 1977). The Rotten Kid Theorem states that if all potential recipients get transfers from an altruist, then under some (but not all) conditions the potential recipients, even if selﬁsh, gain from maximizing the joint income of donors and recipients (Becker 1974).

2.4 The Budget Process in Japan

Earlier work has considered the budgetary process in Japan and has proposed reforms that would improve ﬁscal discipline. Sato (2002) describes negotiations between the Ministry of Finance and the ministry in charge of providing grants to local governments; he shows that the equilibrium has the Ministry of Finance give a large budget to the spending ministry, anticipating its subsequent behavior. Tanaka (2011) presents international comparative studies on ﬁscal reforms. None of these analyses, however, deal with asymmetric information between the principal and the agent.

1 For a selective review of research on the existence and the effects of prosocial behavior among individuals working in public organizations, see Polidori and Teobaldelli (2013). 7 Budgets Under Delegation 171

Akai (2014) and Ihori (2014) emphasize that the government should commit not to accommodate ex post fiscal needs. On the other hand, Besfamille and Lockwood (2008) show that a hard budget constraint may not be always optimal. This chapter studies difficulties government faces when it cannot commit, and shows that strict fiscal restraints may give rise to manipulation of information by the agent who expects a large budget in the future.

3 Agent Provides Only One of the Two Services

We begin with a simple model with an agent who provides only one of two services. The agent’s utility in any period is

αAvAðx1Þþð1 À αAÞvAðx2Þ, ð7:1Þ where αA captures the agent’s preferences for service 1 relative to service 2; x1( 0) 0 00 and x2( 0) are the quantities of services 1 and 2. Assume that vi > 0, vi < 0, and that vi(0) ¼ 0. Let αA ¼ 1 or else αA ¼ 0. Thus, the agent would want to spend all his budget on either service 1 alone, or else on service 2 alone. The principal gives the agent a budget X 0. The principal’s utility in any period is

ð : Þ αPvPðx1Þþð1 À αPÞvPðx2ÞÀX, 7 2 where again αP ¼ 1 or else αP ¼ 0. We assume in this section that αP ¼ 1; that is, the principal prefers service 1 to service 2. Consider ﬁrst outcomes under perfect information. If the principal knew that the P agent also had αA ¼ 1, then the principal would give the agent a budget X satisfying 0 ð PÞ¼ vP X 1, and the agent would spend all that budget on service 1. If the principal knew that the agent had αA ¼ 0, then the principal would choose X ¼ 0, fearing that the agent would spend all the budget given to him on policy 2. Even if this stage game in each period is repeated for multiple periods, the principal and the agent will behave the same way in each period. But now suppose that the principal does not know whether αA ¼ 1 (type 1) or αA ¼ 0 (type 2). The principal sets a budget Xt in period t. The timing of the game is as follows. In period 1:

Two cases must be distinguished. In a separating equilibrium, in period 1 the agent allocates the budget in a way that reveals his type. Then in period 2 the P principal sets X2 ¼ 0 if the agent values service 2, and the principal sets X2 ¼ X to 0 ð Þ¼ satisfy vP X2 1 if the agent values service 1. In a pooling equilibrium the agent in period 1 spends all his budget on service 1, even if he values service 2 (we shall suppose that if the agent spends anything at all on service 2, even if not all the budget, then the principal believes that the agent has αA ¼ 0, or that the agent values only service 2). The principal in period 2 then does not know the agent’s type. Let the prior probability that the agent has αA ¼ 1be π1, with 0 < π1 < 1. Then the principal’s expected utility is

ÀX2 þ π1vPðX2Þ: ð7:3Þ

π 0 1 In period 2 the principal will set X2 ¼ X to satisfy v ðX2Þ¼ . Thus, in the pooling P π1 equilibrium the principal is uncertain about the agent’s type; in a separating equilibrium, the principal is certain about the agent’s type, and in period 2 he gives the P π P budget such that X2 ¼ X for type 1 and X2 ¼ 0 for type 2. Note that 0 < X < X . We can determine the value X1 that will induce a separating rather than a pooling equilibrium. A type-2 agent (that is an agent who values only service 2) will reveal his own type if

P vAð0ÞþvAðX ÞvAðX1ÞþvAð0Þ, ð7:4Þ

P or if vAðX1ÞvAðX Þ. That is, inducing truthful revelation (instead of pooling) requires that the budget allocation in period 1 be large. The budget allocation is P costly to the principal, so that he will choose X1 ¼ X . On the other hand, a type-2 agent will want to mimic a type-1 agent if

π vAð0ÞþvAðX ÞvAðX1ÞþvAð0Þ, ð7:5Þ

π or if vAðX1ÞvAðX Þ. Thus, the principal does not need to set a large budget to π induce a pooling equilibrium, so that he chooses X1 ¼ X . Will the principal prefer full revelation to a pooling equilibrium? If he gives a π small budget in period 1, X , inducing a pooling equilibrium, then the principal’s expected utility over two periods is π π π π vPðX ÞÀX þ π1vPðX ÞÀX : ð7:6Þ

P If in period 1 the principal allocates X , inducing a separating equilibrium, his expected utility is

P P P P π1vPðX ÞÀX þ π1½vPðX ÞÀX : ð7:7Þ 7 Budgets Under Delegation 173

Note that

π π π π P P ð : Þ π1vPðX ÞÀX < π1½vPðX ÞÀX < π1½vPðX ÞÀX : 7 8

Because the principal can then set the optimal budget for each type in period 2, his utility in period 2 is greater in the separating equilibrium than in the pooling equilibrium. We cannot, however, determine whether utility in period 1 is higher in (7.6) than in (7.7). Another possibility is that the principal hides his type. If the agent thinks that the principal is equally likely to value the two services, then the agent will spend his budget in period 1 on the good the agent values, thereby revealing his type. Note that with many agents there is a higher probability that at least one of the agents has the same preferences as the principal does. That higher probability reduces the incentive to mimic the principal’s type.

4 Agent Provides Two Services

This section extends the model to have the principal and the agent value both services rather than only one or the other. The principal sets the budget, and the agent allocates it between two services. The utility function of the agent or of the principal is

αjvðx1Þþð1 À αjÞvðx2Þþmj, ð7:9Þ where 0 < αj < 1, j ¼ P, A, and mj is private consumption. The principal and the agent are assumed to have the same function v. The principal is endowed with Y. When he delegates the allocation of the budget between services to the agent, his private consumption is

ð : Þ mP ¼ Y À X, 7 10 where X is the total budget set for the agent. This private consumption may include spending on private goods, or spending on services other than those provided by the agent. The agent’s private consumption is

¼ À À ð : Þ mA X p1x1 p2x2, 7 11 where p1 is the marginal cost of providing service 1, and p2 is deﬁned analogously. 174 K. Terai and A. Glazer

4.1 First-Best Allocation

ð P PÞ The ﬁrst-best allocation for the principal, denoted by x1 , x2 , is obtained by ’ þ maximizing the principal s utility (7.9) subject to p1x1 p2x2 Y. The associated ﬁrst-order conditions are

α 0 1 À α 0 P ð PÞ¼ P ð PÞ¼ P þ P < ; ð : Þ v x1 v x2 1, p1x1 p2x2 Y 7 12 p1 p2 α 0 1 À α 0 P ð PÞ¼ P ð PÞ P þ P ¼ : ð : Þ v x1 v x2 1, p1x1 p2x2 Y 7 13 p1 p2

4.2 Delegation

The principal sets the total budget X, delegating to the agent the allocation of it between two services. Delegation can appear for several reasons, including lack of time or skill by the principal to provide the services. Or the agent may be better informed than the principal about local ﬁscal needs.

4.2.1 Perfect Information

Consider first behavior under perfect information: the principal knows the agent’s preferences (captured by αA) and knows the costs of providing different services, ð Þ given by p1, p2 . The timing of the game is as follows. 1. The principal sets the budget, X. 2. The agent allocates the budget for private consumption and for the two services, as ðx1, x2Þ. Note that the principal’s choice works as a fiscal cap on the agent’s decision. We will explore the subgame perfect equilibrium by examining the game backwards. In stage 2, the agent allocates the fixed amount X among service 1, service 2, and his own consumption to maximize his utility (7.9). Given X, his best response is α À α A 0 ð Þ¼1 A 0 ð Þ¼ þ < ; ð : Þ v x1 v x2 1, p1x1 p2x2 X 7 14 p1 p2 α 0 1 À α 0 A ð Þ¼ A ð Þ þ ¼ : ð : Þ v x1 v x2 1, p1x1 p2x2 X 7 15 p1 p2

In stage 1, the principal sets a budget, anticipating the agent’s responses given by (7.14) and (7.15). The principal’s objective is to 7 Budgets Under Delegation 175

α ð Þþð À α Þ ðÞþ À max pv x1 1 P vx2 Y X, ð7:16Þ X subject to

ð Þ¼ α ð Þþð À α Þ ðÞþ À À ; ð : Þ x1, x2 argmaxðz1, z2Þ Av z1 1 A vz2 X p1z1 p2z2 7 17 Y X; ð7:18Þ þ : ð : Þ X p1x1 p2x2 7 19

Note that (7.17) is the participation constraint for the agent. The principal’s budget constraint is (7.18); the agent’s budget constraint is (7.19). Anticipating the agent’s response, the principal chooses X. If he gives a sufﬁ- ciently large budget for the agent to choose the quantities of the two services without binding the agent’s budget constraint (7.19), the principal’s utility declines with the size of the budget—a smaller budget allows the principal to increase spending, mP, on services other than those the agent provides. The principal should therefore make the agent’s budget constraint (7.19) bind. Then the agent’s response in the neighborhood of the equilibrium strategy is derived by using (7.15)as

∂ð Þ ∂ð Þ 2ð À α Þ 00 ðÞ p1x1 p2x2 p1 1 A v x2 < ¼ 1 À ¼ 00 00 0 ∂ 2α ð Þþ 2ð À α Þ ðÞ ∂X X p2 Av x1 p1 1 A v x2 ¼ 1 < : ð : Þ 00 1 7 20 2 αA v ðx1Þ p2 þ p 00 1 1 1 À αA v ðx2Þ

000 In (7.20), when v > 0 and the agent with a high αA spends more on service 1, the 00 00 absolute value of v (x1) is smaller than the absolute value of v (x2), and hence some of an increased budget will still be spent on service 1. Alternatively, when αA is large and v000 < 0, little of an increased budget will be spent on service 1. Therefore, with (7.20), the associated ﬁrst-order condition for the principal is

∂ð Þ α ∂ð Þ À α 0 p1x1 P 0 p2x2 1 P v ðx1Þþ v ðÞx2 ∂X p ∂X p 1 2 À α ∂ð Þ α À α 1 P 0 p1x1 P 0 1 P 0 ¼ v ðÞþx2 v ðx1ÞÀ v ðÞx2 1: ð7:21Þ p2 ∂X p1 p2

Thus, if the principal has a large endowment Y, the marginal benefit from the agent’s increased provision of the two services induced by a larger budget equals the marginal cost of reduced consumption of services not provided by the agent. If αP ¼ αA, that is if the agent’s preferences for services are the same as the principal’s, the first-best allocation is realized in equilibrium. Otherwise, the principal would suffer an efficiency loss arising from delegation. 176 K. Terai and A. Glazer

Fig. 7.1 Budget allocation αP p v in the ﬁrst-best solution and 1 α > 1 1−αP v by delegation ( P 2, p 2 αA < αP, but the divergence between αP and αA is not sufﬁciently large)

μ αP p v 1

1−αP p v 2 P P x x x x x x 0 2 2 1 1 1, 2

Without loss of generality, let αP > 1=2, so that the principal prefers service ¼ ¼ 1 over service 2. Also let p1 p2 1. For a given budget, the principal would want ∗ ¼ P þ P ð P PÞ P > P more spending on service 1. Define X x1 x2 where x1 , x2 , x1 x2 , is given ∗ by (7.12) and (7.13). The budget X corresponds to the budget the principal would give the agent when their preferences were the same. We are interested in whether ∗ the equilibrium budget is larger or smaller than X when their preferences differ. As 000 shown later, it depends on the utility functions. For instance, suppose v > 0. Then, as already mentioned, an increase in the budget is expected to be allocated between the two services in a balanced way. Suppose that αA < αP, or that the agent’s preferences for service 1 are weaker than the principal’s. Consider first outcomes when their preferences do not greatly α Àα differ. Define μ ¼ P vðxPÞ¼1 P vðxPÞ in (7.12) and (7.13). Represent it by p1 1 p2 2 ∗ dotted horizontal lines in Fig. 7.1. For a given budget X , the agent will allocate ð Þ P < < < P ’ it as x1, x2 such that x2 x2 x1 x1 , as illustrated in Fig. 7.1. The principal s benefit from marginally increasing the budget is given by the weighted sum of 0 0 αP 1ÀαP v ðx1Þ and v ðÞx2 in (7.21) that are given by two bold vertical lines in p1 p2 Fig. 7.1. That benefit from the marginalÀÁ increase in the budget can be smaller than α 0 Àα 0 0 the weighted sum of P v ðxPÞ and 1 P v xP , which equals μ, since the function v p1 1 p2 2 is strictly convex under the assumption that v000 > 0. Comparing this marginal benefit to the marginal cost of increasing the budget (that is, reducing consump- ∗ tion of other services), the principal would choose a budget smaller than X . When the divergence in their preferences is sufficiently large, the principal may ∗ anticipate the allocation of X by the agent, as shown in Fig. 7.2. The agent’s parameter value αA is sufficiently small so that the agent spends more on service 2 than on service 1. With a sufficiently large endowment Y, the principal will then 7 Budgets Under Delegation 177

Fig. 7.2 Budget allocation αP p v 1 in the ﬁrst-best solution −α 1 1 P v and by delegation (αP > , p 2 2 αA < αP, and the divergence between αP and αA is sufﬁciently large)

αP p v 1

1−αP p v 2

P P x x x x x x 0 1 2 1 2 1, 2

Fig. 7.3 Budget allocation αP p v in the ﬁrst-best solution 1 −α 1 1 P v and by delegation (αP > , p 2 2 αA > αP)

αP p v 1

1−αP p v 2

P P x x x x x x 0 2 2 1 1 1, 2

∗ 0 αP give the agent a budget larger than X , since the weighted sum of v ðx1Þ and p1 0 1ÀαP v ðÞx2 given by two bold vertical lines may be greater than μ. p2 000 Figure 7.3 shows outcomes when αA > αP. With v > 0, the weighted sum of the marginal beneﬁt from increased spending on service 1 and service 2 exceeds its ∗ marginal cost, and the principal may choose a budget larger than X . Intuitively, a divergence in preferences between the principal and the agent has two, opposing, effects. First, the agent will spend some of the budget on the service the principal little values. That effect induces the principal to give a small budget. Second, the agent will spend little of the budget on the service the principal highly values. That effect would induce the principal to increase the budget he gives the agent, to induce the agent to provide more of the service the principal values. 178 K. Terai and A. Glazer

The results are summarized as follows: 1. When the preferences of the principal and the agent differ, but not greatly, the principal may choose a budget that is smaller than in the first-best solution. 2. When the preferences of the principal and the agent differ, and the divergence is sufficiently large, or the agent’s preferences are extreme, the principal may choose a budget that is larger than in the first-best solution. If v000 < 0, the results may be reversed. The principal may give the agent a large budget when the agent’s preferences are close to his. If the agent has extreme preferences, the principal may give the agent a small budget.

4.2.2 Imperfect Information About the Agent’s Preferences

We now turn to considering a principal who is unsure about the agent’s preferences. Consider two periods. We are interested in the agent’s strategic behavior in period 1 to affect the principal’s beliefs, and thus the budget in period 2. Suppose that αA can take one of two values, αH or αL,withαL < 1=2 < αH < αP. The prior probability that αA ¼ αL is πL. Building on the results derived in Sect. 4.2.1, suppose that αL is sufficiently small to induce the principal to choose L ∗ X > X under perfect information; αH is sufficiently close to αP to induce the H ∗ principal to choose X < X . A larger budget benefits the agent, so that the agent may be motivated to behave as if his preferences were given by αL when his preferences are given by αH. The timing is as follows. In period 1:

1. Nature determines αA. 2. The principal sets the budget X1. 3. The agent allocates the budget between the two services as ðx11, x12Þ. In period 2: 1. The principal updates his beliefs about the agent’s type. 2. The principal sets a new budget X2. 3. The agent allocates the budget between the two services as ðx21, x22Þ. Introducing asymmetric information between the principal and the agent, we explore perfect Bayesian equilibrium. Because any manipulation is useless at the ﬁnal stage of the game, in the last stage of period 2 the agent spends the budget X2 according to his preferences, that is, according to (7.15). Denote a type-k agent’s choice of ðxt1, xt2Þ that maximizes his ð k ð Þ k ð ÞÞ ¼ one-period utility (7.9) according to (7.15) given Xt by xt1 Xt , xt2 Xt , t 1, ¼ k ð Þ ’ 2, k H, L.Thus xti Xt corresponds to the agent s choice in the absence of signaling considerations. Figure 7.4 displays choices by the two types of agents, when each faces a budget Xt: the marginal rate of substitution equals the ratio of the 7 Budgets Under Delegation 179

x Fig. 7.4 Budget allocation t2 by the agent without signaling considerations Xt

xL(X ) t2 t

Type L’s indifference curve

xH(X ) t2 t Type H’s indifference curve

xL(X ) xH(X ) x 0 t1 t t1 t Xt t1

marginal costs for providing the two services. For any pair of services ðxt1, xt2Þ, each type’s marginal rate of substitution satisﬁes the following single-crossing property:

α 0 ð Þ α 0 ð Þ Lv xt1 Hv xt1 ð : Þ 0 < 0 : 7 22 ð1 À αLÞv ðxt2Þ ð1 À αHÞv ðxt2Þ

In the second penultimate stage in period 2, the principal sets the budget for period 2. When he has the posterior beliefs such thatπ~ L ¼ 1, he will give the agent L ∗ X , which exceeds X . On the other hand, when he infers that π~ L ¼ 0, he will give H ∗ the agent the budget X < X . If the principal is unsure of the agent’stype,with the posterior beliefs the same as the prior beliefs, he will give the agent the budget π H π L X L , such that X < X L < X . In period 1 the agent thus has an incentive to behave strategically, inﬂuencing the principal’s beliefs: given a larger budget, the agent can spend more on each service according to (7.20), and may increase his utility. We will then examine behavior by each type of agent in period 1, where the agent’schoiceofx11 and x12 affects the principal’s beliefs about the agent’s type. Let ∗ denote the equilibrium choice of the agent with his signaling considerations. We ﬁrst examine the agent’schoiceinperiod1inaseparatingequilibrium ð L∗ L∗Þ 6¼ð H∗ H∗Þ π~ ð L∗ L∗Þ¼ π~ ð H∗ H∗Þ¼ satisfying x11 , x12 x11 , x12 , L x11 , x12 1, and L x11 , x12 0, π~ ð H∗ H∗Þ¼ where the agent reveals his type. Given L x11 , x12 0, a type-H agent should ð H∗ H∗Þ¼ð H ð Þ H ð ÞÞ choose x11 , x12 x11 X1 , x12 X1 in equilibrium. We assume that the principal’s beliefs at the decision nodes in the information set off the equilibrium path 180 K. Terai and A. Glazer

π~ ð Þ¼ ð Þ 6¼ð L∗ L∗Þ are such that L x11, x12 0for x11, x12 x11 , x12 . Then the necessary conditions for a separating equilibrium are2 ÂÃ L∗ L∗ αLvðx Þþð1 À αLÞvðx Þ hi11 12 þ α ð L ð LÞÞþð À α Þ ð L ð LÞÞ Lv x21 X 1 L v x22 X ÂÃð7:23Þ L L αLvðx ðX1ÞÞ þ ð1 À αLÞvðx ðX1ÞÞ hi11 12 þ α ð L ð HÞÞþð À α Þ ð L ð HÞÞ ; Lv x21 X 1 L v x22 X

ÂÃ L∗ L∗ αHvðx Þþð1 À αHÞvðx Þ hi11 12 þ α ð H ð LÞÞþð À α Þ ð H ð LÞÞ Hv x21 X 1 H v x22 X ÂÃð7:24Þ H H αHvðx ðX1ÞÞ þ ð1 À αHÞvðx ðX1ÞÞ hi11 12 þ α ð H ð HÞÞþð À α Þ ð H ð HÞÞ : Hv x21 X 1 H v x22 X

In a separating equilibrium, a type-L agent allocates his budget in a way that a type- ð H ð Þ H ð ÞÞ H agent would not want to mimic. Then a type-H agent chooses x11 X1 , x12 X1 that maximizes his utility in period 1 but reveals his type; he cannot get a larger ð L∗ L∗Þ budget in period 2. Note that x11 , x12 does not always coincide with ð L ð Þ L ð ÞÞ x11 X1 , x12 X1 . A type-L agent separates himself from a type-H agent by choosing the allocation that makes a type-H agent worse off by mimicking even if he can get a large beneﬁt in period 2. Such an allocation may induce a type-L agent to spend more on service 2 than he would in the absence of signaling considerations. Anticipating the agent’s subsequent behavior, in the second penultimate stage in period 1, the principal sets a budget. His choice should induce a type-H agent to reveal his type, so that the following condition should hold: ÂÃ L∗ L∗ αHvðx Þþð1 À αHÞvðx Þ hi11 12 þ α ð H ð LÞÞþð À α Þ ð H ð LÞÞ Hv x21 X 1 H v x22 X ÂÃð7:25Þ H H ¼ αHvðx ðX1ÞÞ þ ð1 À αHÞvðx ðX1ÞÞ hi11 12 þ α ð H ð HÞÞþð À α Þ ð H ð HÞÞ : Hv x21 X 1 H v x22 X

2 It follows from the discussion in Sect. 4.2.1 that mA ¼ 0 should hold in equilibrium. 7 Budgets Under Delegation 181

The second term on the left-hand side is greater than that on the right-hand side in (7.25), because the agent gets a larger budget. For (7.25) to be satisfied, the first term on the right-hand side should be larger than that on the left-hand side. ðð L∗ L∗Þ ð H∗ H∗ÞÞ ð L∗ L∗Þ¼ð H∗ H∗Þ Moreover, x11 , x12 , x11 , x12 satisfying x11 , x12 x11 , x12 ,given π~ ð L∗ L∗Þ¼π π~ ð Þ¼ ð Þ 6¼ð L∗ L∗Þ L x11 , x12 L and L x11, x12 0 for any x11, x12 x11 , x12 , constitutes a pooling equilibrium if a type-H agent benefits most from spending in a way that makes him indistinguishable from a type-L agent. That condition is ÂÃ α ð L∗Þþð À α Þ ð L∗Þ Lv x11 1 L v x12 ÂÃπ π þ α ð L ð L ÞÞþð À α Þ ð L ð L ÞÞ Lv x21 X 1 L v x22 X ÂÃð : Þ L L 7 26 αLvðx ðX1ÞÞ þ ð1 À αLÞvðx ðX1ÞÞ hi11 12 þ α ð L ð HÞÞþð À α Þ ð L ð HÞÞ ; Lv x21 X 1 L v x22 X

ÂÃ α ð L∗Þþð À α Þ ð L∗Þ Hv x11 1 H v x12 ÂÃπ π þ α ð H ð L ÞÞþð À α Þ ð H ð L ÞÞ Hv x21 X 1 H v x22 X ÂÃð : Þ H H 7 27 αHvðx ðX1ÞÞ þ ð1 À αHÞvðx ðX1ÞÞ hi11 12 þ α ð H ð HÞÞþð À α Þ ð H ð HÞÞ : Hv x21 X 1 H v x22 X

For example,

ð L∗ L∗Þ¼ð H∗ H∗Þ¼ð L ð Þ L ð ÞÞ ð : Þ x11 , x12 x11 , x12 x11 X1 , x12 X1 , 7 28 may constitute a pooling equilibrium. Figures 7.5 and 7.6 describe intertemporal choices by a type-H agent in this pooling equilibrium. In period 1, a type-H agent

x12 Type H’s indifference curve X1

L x12(X1)

Type L’s indifference curve

Fig. 7.5 Budget allocation by the agent in period 1 with xL (X ) X x signaling considerations 0 11 1 1 11 182 K. Terai and A. Glazer

x22 X πL

H πL x22(X ) H H x22(X ) Type H’s indifference curve

H H H πL x 0 x21(X ) x21(X ) 21

Fig. 7.6 Budget allocation by a type-H agent in period 2 gets lower utility by mimicking than by not. In period 2, however, his loss by L H mimicking is compensated for by getting a larger budget X rather than X . In this pooling equilibrium, the principal’s choice in the second penultimate stage in period 1 must satisfy the incentive compatibility constraint for a type- H agent in (7.27).Thesecondtermontheleft-handsidein(7.27) exceeds that on the right-hand side because a larger budget is given to the agent. The principal’s optimal choice of X1 depends on the details of the model, but comparing (7.25)to(7.27) shows that it may be easier for the principal to induce a pooling equilibrium rather than a separating one, as already mentioned in Sect. 3. In a separating equilibrium, in period 2 (after the agent allocates his spending in period 1), a type-H agent whose preferences are close to the H principal’sgetsasmallerbudget X . In contrast, a type-L agent whose L preferences greatly differ from the principal’s gets a larger budget X . Alternatively, if 1=2 < αL < αH < αP, the results may be similar to the results derived in Sect. 3: under perfect information, an agent gets a larger budget the closer his preferences are to the principal’s preferences; an agent whose preferences differ greatly from the principal’s may gain a large budget in period 2 by manipulation in period 1.

4.3 Extensions

4.3.1 Imperfect Information About the Costs of Services

Next consider asymmetric information about the marginal cost of a service. Let ¼ < p1 1. Moreover let p2 take one of two values pL or pH such that pL pH, 7 Budgets Under Delegation 183

¼ with some exogenous prior probability qL that p2 pL. The agent knows which value has occurred, but the principal cannot observe it. Let us consider a two-period model, with asymmetric information about the marginal cost of service 2. The agent, by manipulating the information on the cost of service 2 in period 1, may get a large budget in period 2. Suppose that pL and pH satisfy the following relationship. α α α α pH P > pL P > pH A > = > pL A : À α þ α À α þ α À α þ α 1 2 À α þ α 1 P pH P 1 P pL P 1 A pH A 1 A pL A ð7:29Þ

This inequality implies that the principal wants to spend more on service 1 even if the marginal cost of service 2 is low; on the other hand, the agent values service 1 less than does the principal, so that when the marginal cost for providing service 2 is low, he wants to spend more on service 2. Under the supposition (7.29), the previous argument about the agent’s signaling behavior and the principal’s choice of a budget can be used. In the final stage in period 2, the agent sincerely spends a budget given by the principal, according to (7.15). The agent will be better off if he gets a larger budget in that period. In the second penultimate stage in period 2, the principal sets a ~ ¼ budget for the period given his posterior beliefs about p2. With q L 0, he expects that the agent spends more on service 1 than service 2, but less than the principal would prefer. Therefore, the principal will give a smaller budget than in the first- ~ ¼ best solution for the principal. If the principal infers that q L 1, he may set a larger budget than in the first-best solution, to induce the agent to spend enough on service 1. Such behavior by the principal may induce the agent to behave strategically in period 1, in order to influence the principal’s beliefs. We can examine the agent’s strategy in period 1 in the same manner as in Sect. 4.2.2. In a separating equilibrium, the agent truthfully reveals p2 to the ¼ ¼ principal. In a pooling equilibrium, the agent behaves as if p2 pL when p2 pH in fact. Thus, in a pooling equilibrium, the agent pretends to face a low marginal cost of the service that the principal prefers less, and induces the principal to give a larger budget. Such behavior produces inefficiency.

4.3.2 Imperfect Information About Each Other’s Preferences

Now suppose that the principal and the agent do not know the preferences of each other. Suppose further that the principal’s preference parameter αP takes one of two values αL and αH as for the agent, and that these two values do not greatly differ. If the agent knows the principal’s type, he can get a larger budget by pretending to have the same preferences as the principal, even when that is false. We are interested in the agent’s behavior when he is uncertain about the principal’s type. 184 K. Terai and A. Glazer

In the final stage in period 1, if the principal’s preference parameter αP ¼ αL was perfectly revealed by his choice of budget in that period, the subsequent choice by the agent in period 1 would constitute a separating or pooling equilibrium. In a pooling equilibrium, a type-H agent would pretend to be type L, to gain a larger budget. If the principal’s choice of budget revealed his preference parameter to be αP ¼ αH, a type-H agent may no longer pretend to be of type L, because he would get a smaller budget in period 2 by this manipulation than he would gain without manipulation. If the principal takes the inefficiency produced by the agent’s strategic manipulation seriously, in the first period, the principal may randomize his choice of budget. In equilibrium, with the principal’s type incompletely revealed, the agent may also use a mixed strategy or reveal his type. Intuitively, an agent who is unsure about the principal’s preferences finds it difficult to behave as if his preferences were identical to the principal’s, and induce the principal to give him a larger budget. We can apply the same discussion to an agent who can elicit a larger budget from the principal by pretending that his preferences differ from the principal’s.

5 Fiscal Problems in Japan

The models used in Sects. 3 and 4 suggest that an agent captured by special interests, taking advantage of his private information, can get a large discretionary budget that enables him to benefit the interest groups. Using the example of spending on public works, we apply our analysis to explain increased government spending and the resultant fiscal deficits in Japan.

5.1 Misreporting Costs

Spending on public works is discretionary. It is well known that initial annual budgets in each ﬁscal year for public works in Japan have been allocated among projects for forestry and water control, roads, ports, harbors, airports, water services, waste disposal facilities, and infrastructure for agriculture at a stable ratio; see Figs. 7.7 and 7.8.3 In Fig. 7.7, spending appears to have changed in a similar manner for almost all categories. Figure 7.8, which displays the annual growth rate of spending in each category in the initial annual budget, supports this idea. This phenomenon, which should be observed when the principal gives the agent discretion in spending across different public projects, has been said to occur because the

3 We used the data until 2000 because the definition of some categories changed after that fiscal year. We excluded two categories because: one was subject to changes of definition during the period from 1985 to 1999, and the other is miscellaneous. 7 Budgets Under Delegation 185

Fig. 7.7 Spending in each category in the original annual budget. Data source: Ministry of Finance Japan

Fig. 7.8 Annual growth rate of spending in each category in the original annual budget. Data source: Ministry of Finance Japan government departments in charge (the agents) were pressured by local interest groups, local governments, and politicians elected in the jurisdictions that will beneﬁt from those projects if implemented. See Fig. 7.9, which shows the principal-agent relationship in the budget process for spending on public works, and the inﬂuences on the agent from the vested interests. 186 K. Terai and A. Glazer

interest groups local government Diet members

Ministry of Finance spending ministry (principal) (agent)

central government

Fig. 7.9 Budget process for public works spending

Japan has a parliamentary system, with the political parties controlling a majority of the seats in the House of Representatives of the Diet forming the government. The central government’s budget, prepared by the Ministry of Finance and submit- ted to the Diet, has usually won legislative approval without major conflicts between the government and the Diet. Therefore, negotiations between the Ministry of Finance and each spending ministry, which is also a constituent of the government, prior to the submission to the Diet have been important in the budget process. Since 1955, except for a few recent years, the Liberal Democratic Party (the LDP) has formed the government of Japan. It has been frequently pointed out that factionalism in the LDP may have caused common-pool problems. Diet members belonging to a faction are connected to special interests and attempt to get large budgets for policies that benefit them. In combination with vested interests, they may misreport costs of public works. Moreover, corruption, which can influence decision-making in favor of those groups, may hide information about costs. Some important public projects were more expensive than initially planned, and were criticized as inefficient and wasteful after completion. One example is Kansai International Airport. To avoid noise pollution, the airport is built on an artificial island in the middle of Osaka Bay in Osaka Prefecture. Its construction was planned in the 1960s, anticipating increased demand for air services. Construction started in the 1980s. In the course of construction, mis-estimation of costs was revealed. The artificial island sunk much more than predicted, because of soft soils in Osaka Bay. Additional spending was needed for adjustments. Moreover, the airport was debt- financed. It had to expend further funds on unanticipated interest payments because completion was delayed. Compensation to local fisheries also exceeded the initial estimate. Thus the project became more expensive as it went on. The airport was opened in 1994, but as of FY2013, it still receives subsidies from the central government for stable management.4

4 We thank Nobuo Akai and Takero Doi for this example. 7 Budgets Under Delegation 187

To address the problems of misreporting by the agent, costs of public works should be estimated more precisely and inspected more carefully. Using the information on other projects or on other agents’ behavior as a yardstick may be helpful. The principal can allocate a budget to an agent depending not only on the agent’s choices but also on any other observation available. Indeed, the principal can use ~ these observations as a yardstick when updating his beliefs (i.e., calculating q L in Sect. 4.3.1). Recall that the agent manipulates information in order to get a large budget later. If the agent knows that the principal is less likely to be cheated, cheating may be less attractive for the agent. Also, to obtain information about the value of the project, the government can hold a public auction or collect private funds, for example by issuing revenue bonds. Thus, acquiring much information enables the principal to induce agents to compete. These measures can reduce inefﬁciency produced by delegation and asymmetric information.

5.2 Uncertainty About the Agents’ Preferences

In the budget process in Japan, the principal can be considered as the prime minister or as the Ministry of Finance, as described in Fig. 7.9. The agents may be considered as spending ministries or local governments. The principal may be unsure about the preferences of different ministers. Such uncertainty about preferences is especially likely to occur under a new prime minister, or under new ministers, or in a coalition government where ministers do not necessarily share the prime minister’s preferences. The coalition need not be a formal one, but can be an effective one, as with the LDP in Japan. Referring to our analysis in Sect. 4.2.2, which examined the effect of uncertainty about the agents’ preferences, the total budget may be large under a new administration. In 2009, after the victory in the House of Representatives election, the Demo- cratic Party of Japan formed the new government in place of the LDP. Figure 7.10 shows the annual growth rate of the total annual initial budget, as well as budgets for major policy ﬁelds.5 In the FY2010 budget, which was drawn up for the ﬁrst time by the new government, spending for redistributive purposes, including social security spending, and local allocation tax grants, greatly increased.6 Moreover, spending shifted from public works to education and science, as the Democratic Party of Japan promised in the campaign. Despite a large decline in spending on public works, total spending net of debt-servicing costs, which is plotted in the bold curve, rose, as our model predicts. The Democratic Party of

5 The budgets for FY2009 were large, to address the effects of the Lehman shock. 6 The new administration abolished ﬁscal caps in drawing up the FY2010 budget. It motivated each spending ministry to require increased budgets. Accordingly, ﬁscal caps were restored next year. 188 K. Terai and A. Glazer

Fig. 7.10 Annual growth rate of spending on major policy ﬁelds in the annual original budget. Data source: Ministry of Finance Japan

Japan was in charge of drawing up budgets for FY2010, FY2011, and FY2012. The budget gradually declined, perhaps reﬂecting the resolution of uncertainty.

5.3 Supplementary Budgets

The Japanese government, after adopting an initial annual budget, often later adopts a supplementary budget. In terms of our model, we can think of the initial budget as the budget in period 1, and the supplementary budget as the budget in period 2. Figure 7.11 shows the change in spending for public works included in the initial annual budget and the supplementary budget of each fiscal year.7 One reason for increased supplementary budgets in the 1990s is that the government used supplementary budgets for public works in economic stimulus packages to address the economic downturn in that period; inefficient spending may also have favored vested interests. The Japanese government has attempted to limit the size of the initial budget by setting a fiscal cap on discretionary spending included in it. But the supplementary budget has been free from fiscal restraint. Even if the government adopted a reasonably balanced budget at first, added budgets enlarged the deficit. The puzzle is why the supplementary budget allowed great spending. Figure 7.12 shows by category the ratio of spending on public works in the supplementary budget to such spending in the initial annual budget. Almost all time-series, except spending for maintenance and improvement of housing and

7 Spending for post-disaster recovery projects is excluded. 7 Budgets Under Delegation 189

Fig. 7.11 Spending for public works in the original annual budget and supplementary budget. Data source: Ministry of Finance Japan

Fig. 7.12 The ratio of spending in each category in the supplementary budget to spending in the original annual budget. Data source: Ministry of Finance Japan urban environment, move in a similar way. The data thus suggest that agencies have discretion in spending on public works, even in the emergency economic stimulus packages. The measures we examined in Sect. 5.1 to address the problems of misreporting by the agent can also be used when considering modifications of the 190 K. Terai and A. Glazer budget. Fiscal restraint on the supplementary budget using the results of inspection can reduce inefficient government spending. Those measures can be also applied to the medium-term fiscal plan. In Japan, the government creates a medium-term fiscal plan, but it cannot bind spending. The government may induce the agent to spend according to the plan by monitoring his annual spending and using results from monitoring in setting a budget for the next year. We can see, by returning to Fig. 7.10, that spending on public works rose in the initial FY2013 budget, drawn up by the revived LDP government. We have to be careful again in monitoring spending on public works.

Conclusion In many democracies, whether at the central or local level, decisions about the budget and spending are divided among multiple persons or departments. We can regard a person or a group setting a budget as a principal who delegates the allocation of the budget to the agent. This principal-agent relationship may distort the principal’s decision on the size of a budget when their preferences diverge, since the principal must set a budget anticipating the agent’s response. First we examined inefficiency caused by delegation under the assumption of perfect information. In equilibrium, the budget may be lower or higher than in the first-best solution, depending on how the marginal utility from each service declines as the quantity of a service increases. Then we developed a two-period model where the agent’s spending on each service in period 1 signals his own preferences or the marginal cost of providing a service to the principal, and affects the principal’s decision on a budget in period 2. A separating or pooling equilibrium may occur. In a pooling equilibrium, the agent is driven by strategic considerations to gain a large budget in the following period. In a separating equilibrium, the agent reports truthfully, allowing the principal to set an optimal budget for each type in the following period. In the separating equilibrium, however, the agent may have to be given a sufficiently large budget in period 1 so that he would not demand a large budget in period 2. If the principal adopts some measures to elicit information, the expected benefit from cheating is reduced; the agent may be induced to tell truth with a smaller budget in period 1. Thus inefficiency may be reduced by reforms of the budgetary process.

Acknowledgements The authors thank Nobuo Akai, Takero Doi, Keisuke Hattori, Toshihiro Ihori, Shintaro Nakagawa, Shuhei Shiozawa, Shinichi Suda, and the seminar participants at Keio University and at the University of California, Irvine, for helpful comments and suggestions. In particular, comments by Keisuke Hattori on our Figs. 7.1, 7.2, 7.3 were most helpful in revising our graphical presentation. Financial support by the Seimeikai Foundation and a Grant-in-Aid for Scientiﬁc Research from the Ministry of Education, Culture, Sports, Science and Technology are gratefully acknowledged. 7 Budgets Under Delegation 191

Comment Paper to Chapter 7

Keisuke Hattori Osaka University of Economics, 2-2-8, Osumi, Higashiyodogawa, Osaka, Japan email: [email protected] This chapter constructs a principal–agent model of delegation, in which a principal sets a budget that an agent then allocates between two different public services (“good” or “bad” services from the principal’s point of view). It investigates (i) how differences in preferences between the principal and agent affect the size of the budget that the principal sets, and (ii) how the agent behaves to obtain a larger budget when the principal is uncertain about the agent’s preferences. The following results are obtained using plausible assumptions regarding preferences: (i) the more divergent the preferences of the principal and agent, the larger the budget that the principal tends to give the agent, and (ii) the agent has strategic incentives to obtain a large budget. Regarding this latter, it is further determined that in a separating equilibrium, an agent who prefers bad services spends much more on bad services in a manner that an agent preferring good services would not want to mimic. In a pooling equilibrium, however, even an agent who prefers good services mimics the bad agent and inefficiently spends on bad services. The basic signaling model is extended by incorporating asymmetric information regarding the cost to provide each service. It is shown that, in a pooling equilibrium, an agent who has low (high) marginal costs of good (bad) services will behave as if he/she has high (low) marginal cost of good (bad) services to elicit a larger budget from the principal. The authors stress the importance of asymmetric information regarding agent’s cost for providing public services for explaining increased government expenditure in Japan, but I believe that it is equally possible that the asymmetric information regarding agent preferences has the same importance as that of agent’s cost. For example, a national government prefers welfare services to road maintenance services but a local government prefers the opposite, which may cause an inefficient allocation of government expenditure. Furthermore, differences in preferences for some public services between national and local governments may be due to positive or negative externalities of local public services (e.g., libraries, sewer systems, parks, and city streets). On this point, the authors can also connect their ideas to the “decentralization theorem” by Wallace Oates. In addition, I would like to know how the national subsidy to local governments for providing good services affects inefficiency in delegation. I found the analysis very interesting and thought provoking even in the perfect information setting. In addition, I believe that the model can be applied to budget allocation problems within firms, organizations, and families, all of which await future research. 192 K. Terai and A. Glazer

References

Akai N (2014) Fiscal consolidation and local public finance in Japan: Incentive problems behind Inter-governmental transfers and its political roles. In: Ihori T, Terai K (eds) The political economy of fiscal consolidation in Japan. Springer, Tokyo Aron DJ (1987) Worker reputation and productivity incentives. J Lab Econ 5:87–106 Becker GS (1974) A theory of social interactions. J Polit Econ 82:1063–1094 Beniers KJ, Dur R (2007) Politicians’ motivation, political culture, and electoral competition. Int Tax Publ Finance 14:29–54 Besfamille M, Lockwood B (2008) Bailouts in federations: is a hard budget constraint always best? Int Econ Rev 49:577–593 Biglaiser G, Mezzetti C (1997) Politicians’ decision making with re-election concerns. J Publ Econ 66:425–447 Buchanan J (1977) The Samaritan’s dilemma. In: Phelps ES (ed) Altruism, morality and economic theory. Russell Sage Foundation, New York Canes-Wrone B, Herron MC, Shotts KW (2001) Leadership and pandering: a theory of executive policymaking. Am J Polit Sci 45:532–550 Carpenter DP (2004) The political economy of FDA drug review: Processing, politics, and lessons for policy. Health Aff 23:52–63 Crewson PE (1997) Public service motivation: Building empirical evidence of incidence and effect. J Public Admin Res Theory 7:499–518 Fox J (2007) Government transparency and policymaking. Publ Choice 131:23–44 Georgellis Y, Iossa E, Tabvuma V (2011) Crowding out intrinsic motivation in the public sector. J Publ Admin Res Theory 21:473–493 Gibbons R (1987) Piece-rate incentive schedules. J Lab Econ 5:413–429 Gregg P, Grout PA, Ratcliffe A et al (2011) How important is pro-social behaviour in the delivery of public services? J Publ Econ 95:758–766 Guyot JF (1962) Government bureaucrats are different. Publ Admin Rev 22:195–202 Hess GD, Orphanides A (1995) War politics—An economic, rational-voter framework. Am Econ Rev 85:828–846 Ihori T (2014) Fiscal consolidation in a political economy of Japan. In: Ihori T, Terai K (eds) The political economy of fiscal consolidation in Japan. Springer, Tokyo Kanemoto Y, MacLeod WB (1992) The ratchet effect and the market for secondhand workers. J Lab Econ 10:85–98 Kessler AS (2010) Communication in federal politics: Universalism, policy uniformity, and the optimal allocation of fiscal authority. CEPR Discussion Paper No. DP7910 Lazear E (1986) Salaries and piece rates. J Bus 59:405–31 Majumdar S, Mukand SW (2004) Policy gambles. Am Econ Rev 94:1207–1222 Maskin E, Tirole J (2004) The politician and the judge: Accountability in government. Am Econ Rev 94:1034–1054 Morris S (2001) Political correctness. J Polit Econ 109:231–265 Polidori P, Teobaldelli D (2013) Prosocial behavior in the production of publicly provided goods and services: An overview. Int Rev Appl Econ 27:285–296 Prat A (2005) The wrong kind of transparency. Am Econ Rev 95:862–877 Sato M (2002) Intergovernmental transfers, governance structure and fiscal decentralization. Jpn Econ Rev 53:55–76 Smart M, Sturm DM (2013) Term limits and electoral accountability. J Publ Econ 107:93–102 Tanaka H (2011) Zaisei kiritsu to yosan seido kaikaku (Fiscal discipline and budget process reform). Nippon Hyoron Sha, Tokyo Index

A Dynamically efﬁcient economy, 38, 39, 57 Abenomics, 4 Dynamic general equilibrium framework, 69 Asymmetric information, 104, 178, 182 Auerbach and Kotlikoff framework, 74 Automatic built-in stabilizers, 10 E Automatic ﬁscal stabilizers, 28 Earmarked transfers, 96 Electoral system by generation, 164 Endogenous labor supply behavior, 74 B Endogenous labor supply decision, 75 Bailout, 105–110 Externalities, 114, 133, 134 Benevolent government, 23

F C Female labor force, 90, 91 Central government, 27 Female labor force ratio, 71 Change in government, 6 Female labor supply, 69 Change in the type of job contract for Females who cannot work due female workers, 86 to child care, 74 Child care, 84, 91 First-best solution, 19 Cobb-Douglas, 79 Fiscal cap, 174, 188 Complete contracts, 102–105 Fiscal competition, 114, 116, 124–135 Computable general equilibrium model, 75 Fiscal consolidation, 44, 71 Condorcet winner, 38, 40, 45–47, 50–53 Fiscal consolidation, stimulating Contract condition, 72 female labor force, 90 Contribution rate, 80 Fiscal disparity, 97–98 Corrupt governments, 26 Fiscal federalism, 27 Counter-cyclical ﬁscal policy, 15, 24 Fiscal management, 95, 104, 108 Credible commitments, 28 Fiscal multipliers, 7 Fiscal privileges, 17 Fiscal reconstruction, 33 D Fiscal reforms, 29 Deﬁcit ceiling, 15, 20 Fiscal situation, 13 Delegation, 168, 174 Fiscal structural reform, 55 Discretionary policy, 10 Fiscal Structural Reform Act, 16 DP government, 7 Full-time, 75

Full-time/non full-time in female workers, 87 Lack of commitment, 106, 109, 110 Future demographic structure, 82 Local Allocation Tax (LAT), 110 Local government, 27

G General account, 79 M Generational election system, 145–165 Majoritarian system, 149–150, 152–156 Great East Japan Earthquake, 11 Majority voting, 37, 38, 40, 42, Guarantee of fiscal resources, 98 45, 50, 57 Malapportionment, 146 Marginal utility of public good, 25 H Matching, 96 Hard budget constraint, 106 Microeconomic fiscal policy, 13 Hi-seiki labor force, 70 Microeconomic reforms, 30 Hi-seiki workers, 76 Ministry of Finance, 170, 186 Horizontal disparity, 98 Moral hazard, 102–105, 109 M shaped curve, 70 M-shaped pattern, 87 I Impact of M-shaped curve, 87 Impact on welfare, 84 N Improving efficiency, 98–99 New economic geography, 114–116, Incentive design, 103–104 118, 123 Incomplete contract, 105–110 Nonearmarked transfers, 96 Industry agglomeration, 139 Non full-time, 75 Information asymmetry, 12, 110 Non-Keynesian effect, 8 Information disclosure, 104 Nonmatching, 96 Initial annual budget, 184 Numerical analysis, 155–163 Institutional design, 104–105, 109–110 Integrated reform, social security and tax systems, 9 O Integration, 113, 114, 116, 124, 134 Overlapping-generations, 37, 40, 57 Interest groups, 14, 17, 170 Intergenerational equity, 147 Intergenerational redistribution, 44 P Intergenerational transfers, 38, 39, 46, 47, Parliamentary system, 186 49, 51, 54, 55, 57, 64 Perfect information, 171, 174 Intergovernmental financing, 14 Policy commitment, 43, 49, 58, 62, 63 Intrinsic motivation, 170 Political circumstance, 11 Political economy, 25 Political efforts, 13 J Politically quasi-strong, 29 Japanese government, 29 Pooling equilibrium, 172, 181, 183 Population aging, 36, 39, 40, 46, 47, 51, 52, 54–57 K Potential growth rate, 12 Kansai International Airport, 186 Principal-agent relationship, 185, 190 Koizumi administration, 6, 30 Pro-cyclical fiscal policy, 24, 26, 31 Public infrastructures, 114–117, 120, 122–128, 130, 132–135 L Public investment, 8 Labor efficiency, 77 Public pension account, 79 Labor market equilibrium, 80 Public pension benefits, 78 Index 195

R T Recession, 26 Tax competition, 115, 116, 128–130, 132 Reputation, 169 Tax revenue, 26 Reserve funding ratio, 103, 104 Technological progress, 81 Transparency, 104 Truthful revelation, 172 S Turnout rate, 146, 157, 160 Sales tax, 16 Scale economies, 113, 114, 123 Scenarios, 83–84 U Second best, 19 Uncertainty, 187 Seiki labor force, 70 Utilitarian government, 18 Seiki workers, 76 Separating equilibrium, 172, 179, 183 Signal/signaling, 169, 178, 179, 183 V Simulation analysis, 80–85 Vertical disparity, 97–98 Social security funding, 35, 37–39, Voting equilibrium, 149–150, 152–156, 52–58 160–163 “Social” valued-added tax (VAT), 35, 39, 57 Soft budget constraint, 105–111 W Standard fiscal needs, 100–101, Wage profile gap, 71 108–109 in gender, 87, 89 Standard fiscal revenues, 100, 101 between males and females, 86 Structure-induced equilibrium, 38, 39, Wasteful government spending, 139 52–55, 58, 62, 63 Weaker deficit ceiling, 22 Subgame-perfect, 43, 58, 62–64 Supplementary budget, 188 Y Sustainability of government debt, 8 Yardstick, 187