The Political Economy of Fiscal Policy: Public Deficits, Volatility, and Growth

Lecture Notes in Economics and Mathematical Systems Founding Editors: M. Beckmann H. P. Kiinzi Managing Editors: Prof...

Author: Jaejoon Woo

48 downloads 724 Views 9MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Lecture Notes in Economics and Mathematical Systems Founding Editors: M. Beckmann H. P. Kiinzi

Managing Editors: Prof. Dr. G. Fandel Fachbereich Wirtschaftswissenschaften Fernuniversitat Hagen Feithstr. 140/AVZ 11, 58084 Hagen, Germany Prof. Dr. W. Trockel Institut fur Mathematische Wirtschaftsforschung (IMW) Universitat Bielefeld Universitatsstr. 25, 336 15 Bielefeld, Germany Editorial Board: A. Basile, A. Drexl, H. Dawid, K. Inderfurth, W. Kursten, U. Schittko

Jaejoon Woo

The Political Economv of Fiscal Policy Public Deficits, Volatility, and Growth

Q - Springer

Author Professor Dr. Jaejoon Woo Department of Economics Kellstadt Graduate School of Business DePaul University 1 East Jackson Boulevard Chicago, IL 60604 USA e-mail: jwool @ depaul.edu

ISSN 0075-8442 ISBN-10 3-540-29640-9 Springer Berlin Heidelberg New York ISBN-13 978-3-540-29640-9 Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer-Verlag. Violations are liable for prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media

O Springer-Verlag Berlin Heidelberg 2006 Printed in Germany The use of general descriptive names, registered names, trademarks, 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. Typesetting: Camera ready by author Cover design: Erich Kirchner, Heidelberg Printed on acid-free paper

4213153DK

5432 10

To Hyejung

Preface

Political economy is now one of the most active research fields in macroeconomics. My interest in the political economy and many of the ideas that are presented in this monograph were initiated during my PhD economics study at Harvard University. Alberto Alesina introduced me to the political economy literature in his graduate course, "Advanced Macroeconomic Policy." I am indebted to Alberto Alesina, Dani Rodrik, and Aaron Tornell for their insightful discussions and comments. They greatly helped my thinking on the subject. At various stages of writing the chapters (papers), I have also benefited from discussions with or comments from Daron Acemoglu, Gadi Barlevy, John Berdell, Lans Bovenberg, Daniel Cohen, Jorge Braga de Marcedo, Jason Furman, Steinar Holden, Philip Lane, Jongwha Lee, N. Greg Mankiw, Jordan Rappaport, Jeffrey Sachs, Jose Tavares, and Jürgen von Hagen. I also thank seminar/conference participants at Harvard University, State University of New York at Buffalo, ZEI at University of Bonn, the OECD Development Center, DELTA-ENS at Paris, DePaul University, University of Wisconsin at Milwaukee, Loyola University of Chicago, Western Michigan University, University of Calgary, University of Illinois at Chicago, University of Hawaii at Manoa, the European Economic Association Meeting in 2001, the North American Econometric Society Summer Meeting in 2002, the European Econometric Society Meeting in 2003, the International conference of Korea Economic Association and Korea-America Economic Association in 2004, the Southern Economic Association Meeting in 2004, and the Workshop on Fiscal Policy Issues at University of Oslo in 2005 for helpful discussions. A part of this book was written and revised while I was an economist at the Organisation for Economic Cooperation and Development (OECD) in Paris. At the OECD, I also benefited from discussions with colleagues, and learned enormously about the practice in "political economy" of policy debates. I also thank chairman Michael Miller and colleagues at the department for their encouragement and support. I am grateful to editor Katharina Wetzel-Vandai at Springer-Verlag for her encouragement and patience throughout this book

VIII

Preface

project. Mike Carlton and Rebecca Azaola provided excellent research assistance. Sean Moffat provided superb assistance with editing the manuscript. I gratefully acknowledge financial support from the Korea Foundation for Advanced Studies and the Kellstadt Business School at DePaul University for some of the work included here, and from the University Research Council at DePaul University (via Competitive Research Grant, No. 600201) for this book project. I always thank my parents for what I am today. I owe an enormous debt to my wife, Hyejung, who has been a good advisor and critic, and has created a suitable working environment. And to Elizabeth and Alexandra, who always make me laugh.

Chicago October 2005

Jaejoon Woo

Contents

1

Introduction

2

Fiscal Deficit, Fiscal Volatility and Growth: Social Polarization 2.1 Introduction 2.2 Endogenous Growth Model with Optimizing Interest Groups . . 2.3 Endogenous Fiscal Deficit and Volatility: Polarization 2.3.1 Non-cooperative Feedback Nash Equilibrium 2.3.2 Fiscal PoUcy with a Social Planner 2.3.3 Policymakers Faced with Political Uncertainty 2.4 Endogenous Fiscal Deficit and Growth 2.5 Concluding Remarks 2.6 Appendix

3

4

Inflation, Composition of Deflcit Finance, and Social Polarization 3.1 Introduction 3.2 The Economy 3.3 Inflation, Composition of Deficit Finance, and Social Polarization 3.4 Concluding Remarks Social Polarization, Industrialization, and Fiscal Instability 4.1 Introduction 4.2 The Economy 4.2.1 The Pre-Industrialization Regime 4.2.2 The Post-Industrialization Regime 4.2.3 Human Capital Accumulation, Threshold Externality, and Income Distribution 4.2.4 Income Inequality and Polarization 4.3 The Fiscal Policy

1 7 7 13 17 18 23 24 27 31 33 35 35 39 40 45 47 47 51 51 52 54 57 58

X

Contents

4.4 4.5 4.6 4.7 5

6

4.3.1 Polarization and Endogenous Fiscal Deficit 4.3.2 Polarization and Volatility of Fiscal Outcomes 4.3.3 Country Experiences Econometric Evidence Concluding Remarks Appendix A. The Social Planner's Solution Appendix B. Data

Economic, Political, and Institutional Determinants of Public Deficits 5.1 Introduction 5.2 Economic Variables on Public Deficits 5.2.1 The Benchmark Framework 5.3 Fiscal Politics 5.3.1 Political Instability 5.3.2 Government Fragmentation 5.3.3 Political Regime and Electoral Law 5.4 Social Polarization: Income Inequality and Ethnic Divisions . . . 5.5 Institutions 5.6 Robustness and Sensitivity Analysis 5.7 Concluding Remarks 5.8 Data Appendix Growth, Income Inequality, and Fiscal Volatility: Empirical Evidence 6.1 Introduction 6.2 Inequality and Growth: Cross-Country Regression 6.3 Income Inequality and Fiscal Volatility 6.4 Fiscal Volatility and Growth 6.5 Comparison with Sociopolitical Instability and Fertility Decision 6.6 Concluding Remarks 6.7 Data Appendix

58 60 61 64 73 74 75 77 77 80 80 84 84 87 89 92 96 100 103 107 117 117 119 129 135 140 144 146

References

151

Index

161

Introduction

In discussing issues of macroeconomic policy, the traditional approach used to assume a benevolent social planner who maximizes some social welfare function and to derive optimal policies. However, recurrent macroeconomic problems, which often manifested themselves in unsustainable macroeconomic imbalances (domestic and external) observed in many countries in the past decades, are at odds with the traditional optimal policy prescription. Indeed, it is hard to believe that the government is a monolithic entity that seeks to implement policies to achieve a socially optimal outcome. It has been increasingly recognized that the policymakers have their own objective function and agenda, and strategically interact with other current and/or future policymakers and with the pubUc. They may care about remaining in office or represent the interests of particular groups. At the same time, socio-political factors and the associated incentive constraints facing policymakers greatly influence policy decisions. This monograph presents new, theoretical and empirical studies on fiscal and macroeconomic issues, while highlighting social polarization as an essential organizing principle in a political economy approach. One of the most striking macroeconomic developments during the last three decades is the rise and persistence of large fiscal deficits in a number of countries. The rising public debts have been the subject of great concern to both policymakers and researchers since it has often been an important source of macroeconomic instability or even economic crises such as external debt crisis and hyperinflation. Accordingly, fiscal discipline and a strong fiscal position have been recognized as a necessary condition for sustaining economic growth. Many developing countries have embarked on major fiscal reforms in the wake of external debt crises. Also, some industrialized countries-for example, Denmark, Ireland, New Zealand, and Sweden-have adopted institutional changes that strengthened fiscal rules and procedures in order to achieve greater fiscal discipline in the 1980s and 1990s. In euro area, fiscal consolidation efforts culminated in the run up to monetary union as countries needed to comply with the convergence criteria stipulated in the Maastricht Treaty.

2

1 Introduction

Yet many countries suffer from recurrent large fiscal imbalances that often reflect lack of fiscal discipline. Recent financial crises in Brazil in 1999 and Argentina in 2001 were closely related to high budget deficits or unsustainability of fiscal outcomes.^ With the advent of the euro, the deficit bias has reemerged in euro area, though not in all countries. The 3% ceiling for the deficit-to-GDP ratio has been breached by a number of euro area countries (OECD 2005a).^ Some recent study even suggests that such institutional arrangements have induced cosmetic measures such as shifting expenditures to off-budget accounts or engaging in creative accounting, rather than genuine efforts to restrain fiscal spending (Easterly 1999). While it remains an active research topic how stringent fiscal rules and procedures will affect the fiscal performance in a longer time horizon, these developments vividly illustrate the difficulties in achieving fiscal discipline, and raise more fundamental questions such as why some countries have recurrent fiscal problems, while others do not. This question equally applies to developing countries of today. An important puzzle arising from the contrasting macroeconomic experiences across developing regions in the past decades is why certain countries, notably in Latin America and sub-Saharan Africa, have repeatedly adopted unsustainable fiscal policies, while others, notably in East Asia, have maintained sound fiscal policies.^ The unsustainable fiscal policies are often characterized by chronically large deficits and/or high volatility of fiscal outcomes, which are believed to have aggravated macroeconomic instability and contributed to poor growth or even growth collapses. Our research presented in this monograph is motivated by these fiscal developments and related macroeconomic issues. In particular, we advance social polarization thesis that social polarization and degree of social polarization are key factors behind the fiscal policy making process and the adoption of unsustainable policies that have caused macroeconomic problems of fiscal The Real Plan of Brazil was successful in reducing inflation between 1994 and 1998, but not in containing the fiscal deficit. The large fiscal deficit, close to 8% of GDP in 1998, also contributed to a widening of the current account deficit to 4.5% of GDP in the same year (IMF 1999), Argentine crisis in 2001 was largely driven by unsustainable fiscal debt dynamics compounded by the high initial debt level and the rigid exchange rate system. Argentine failed to strengthen its public finance sufficiently when it should have run a surplus during the years of strong growth. Thus, deficits were run throughout the 1990s (Hausmann and Velasco 2002; Allen 2003). ^ Thus, the excessive deficit procedure has been invoked for these countries, but its enforcement ended in a stalemate in November 2003. Since then, the rules have been amended, allowing under certain conditions more time to correct an excessive deficit. The European Council has made a decision in March 2005 that will allow more flexibility in applying the rules. ^ This is true even in the aftermath of the 1997-98 Asian crisis. Also, it is widely accepted that it was not caused by irresponsible fiscal policy but by heavy, private short-term borrowing, which was compounded by weakness of the financial system (see Radelet and Sachs, 1998).

1 Introduction

3

deficits, volatile fiscal outcomes, poor growth and inflation. We contribute to the literature by developing dynamic models of fiscal policy in which social polarization (conflicts of interest among socio-economic groups) plays a central role in generating aforementioned undesirable economic problems, and by presenting supporting evidence as well as comprehensive evaluation of relative importance between our explanation and other existing studies in the literature. (There is now a large political economy literature. See Drazen 2000; Persson and Tabellini 1999a for a literature survey.) Intuitively, a high degree of social polarization of preferences may make it hard for policymakers to agree on ideal government policies because of potential conflicts of interests, and hence may cause a coordination failure among the policymakers. In the presence of polarization of social preferences over public choices, heterogeneous policymakers may have greater incentives to insist on their preferred policies and may end up choosing individually rational but collectively inefficient policies, especially when institutional restraints on policymakers are lacking. We make this intuition more concrete and rigorous in theoretical models and successfully test the implications using cross-country data. Social polarization is one of the oldest ideas found in the political economy literature. Political scientists often distinguish between two major types of social polarization or social conflict: economic cleavage and cultural cleavage (Powell 1982). An important source of economic cleavage is unequal income distribution. The idea that inequality deepens factionalism and dissension in a society is really an old one, perhaps dating back to an ancient Greek philosopher Aristotle. James Madison also noted in 1787 that "the most common and durable source of factions has been the various and unequal distribution of property." Social polarization arising from struggles over the income distribution in turn can be a major impediment to successful economic performance.^ For example, high income inequality or ethnic fragmentation have long been mentioned as an important explanation for recurrent populist fiscal policies and macroeconomic crises in Latin America and sub-Saharan Africa. Conversely, initial low income inequality in East Asia seems conducive to economic growth by promoting stable macroeconomic environment and encouraging human capital accumulation. (See Sachs 1989; Kauffman and Stallings 1991; Birdsall et al. 1995; Rodrik 1996; Engerman and Sokoloff 1997 among others). Recently, empirical growth literature has found that income inequality and ethnic divisions are detrimental to growth (see Perotti 1996; Easterly and Levine 1997; Easterly 2002; Woo 2004 among others). In a celebrated book, The Elusive Quest for Growth, Easterly (2001) bluntly says "One way to summarize the conditions favorable for growth is that progrowth policies are more likely when the two most common forms of social polarization, class confiict and ethnic tensions, are absent." Some of the earlier work on this topic in modern time can be found in Tarantelli (1986) (see Sachs 1989).

4

1 Introduction

Surprisingly, however, there are very few systematic theoretical and/or empirical studies on the role of social polarization (of preference among socioeconomic groups) in explaining fiscal policy decisions and its fiscal outcomes. We fill this void in the literature. Importantly, the issue of social polarization is not confined to some developing countries inherited with a history of unequal income distribution or ethnic division. It is also relevant to industrialized countries. Some economists have warned that the on-going globalization process and technological advance can be a potential source of social polarization, creating a new set of class divisions, unless domestic policies and institutions are in place to cushion the negative effects associated with these trends. In an influential study, Rodrik (1997) writes "The on-going globalization process is exposing social fissures between those with the education, skills, and mobility to flourish in an unfettered world market—the apparent "winners"— and those without. These apparent "losers" are increasingly anxious about their standards of living and their precarious place in an integrated world economy. The result is severe tension between the market and broad sectors of society, with governments caught in the middle." He argues that this tension explains the growth of the size of governments in the international integration process as the governments try to preserve social cohesion by fiscal policy tools including social welfare systems.^ Social polarization is becoming a policy concern even in a so-called "miracle economy" like South Korea of today, as greater integration with the world economy and liberalization are generating a new set of class/sector divisions (Lee et al. 2004; Woo, S. 2005). Also, there is historical evidence that international trade and migration had significant consequences for income distribution, which is a source of social polarization. Williamson (1998) documents that globalization accounted for more than half of the rising inequality in rich, labor-scare countries (like United States, Argentina, and Australia) and for a little more than a quarter of the falling inequality in poor, laborabundant countries (like Sweden, Denmark, and Ireland) in the pre-World War I period. Therefore, we believe that the macroeconomic implications of social polarization are an important research subject relevant to industrialized and developing countries alike. ^ Anxieties about globalization are widely spread, although they are often exaggerated and are not always backed up by data. For example, OECD (2005b) recently reports that rising imports, outflows of foreign direct investment (sometimes tied directly to the relocation of production) and inflows of immigrants all contribute to rising job insecurity in OECD countries. The rapid integration into the world trading system of China and India, with their huge pools of low-wage labor, and the recent enlargement of the European Union have fuelled fears of job losses and wage cuts. In the United States, job outsourcing was one of the hotly debated issues during the 2004 presidential election campaigns. In a similar vein, Agell (1999) argues that the globalization of economic activity tended to lead to increased demand for various labor market rigidities.

1 Introduction

5

The plan of the book is as follows. In Chap. 2, we present a model of endogenous fiscal policy in a simple growth framework where social polarization (of preferences among different socio-economic groups) over the ideal composition of government spending plays a central in the fiscal dynamics. In a highly polarized society, a deficit occurs endogenously, fiscal spending path becomes more volatile, output collapses, and economic growth rate is reduced along the transition path to a new lower level of output. One novel feature is that the size of fiscal deficit, the magnitude of fiscal volatility, and the size of reduction in output and growth rate are explicitly shown to be increasing functions of the degree of social polarization. Thereby, we offer a fiscal instability channel that negatively links social polarization and growth, which is an alternative yet distinct explanation for the empirical finding that social polarization is harmful to growth. Moreover, we fully distinguish the incentive to engage in such short-term policies under political uncertainty from that under polarization. Polarization and political uncertainty are shown to be distinct forces that can drive the aforementioned macroeconomic problems. In the next chapter, we extend the framework that is developed in Chap. 2 to study infiationary consequence of the fiscal policy set by polarized policymakers by introducing money and infiation tax into the economy. We derive a positive relationship between inflation and social polarization, which is consistent with empirical findings in the literature. Interestingly, we show that inflation dynamics depend on the money-bond ratio in the deflcit finance and an increase in that ratio is less infiationary for a given degree of polarization. Chapter 4 is motivated by an important puzzle arising from the contrasting macroeconomic experience across developing regions, that is, in sharp contrast to East Asia, much of Latin America and sub-Saharan Africa has often engaged in unsustainable fiscal policies, leading to huge fiscal deficits, external debt crises, or hyperinfiation. We study the theoretical relation among income inequality, industrialization, and social polarization on the one hand, and the relation between the resulting social polarization and fiscal outcomes on the other hand. In a two-sector economy with a manufacturing sector and a traditional sector, the more unequal the initial income distribution, the larger the sectoral income gap during industrialization and the more likely the polarization of sector preferences for different types of government spending. In a highly polarized society, a fiscal deficit results and fiscal outcomes exhibit greater fluctuations over time. We successfully test these two predictions with data on consolidated public sector fiscal balance, and present the first econometric evidence on the positive link between income inequality and undesirable fiscal outcomes in a panel of 90 countries in the period of 1970-90. Countries that have suffered from the greatest fiscal instability (deficits and volatility) are those with highly polarized societies as measured by indicators of income inequality. In Chap. 5, we provide a comprehensive empirical test on a large pool of potential explanatory variables of public deficits in a panel of 57 countries including OECD countries and derive robust conclusions about which of

6

1 Introduction

these variables (or which theories) are important in explaining cross-country variation in fiscal balance. Financial depth, income inequality, assassinations, cabinet size, and centralization of authority in budgetary decisions are found to be significant and robust determinants of public deficits. We also propose a working hypothesis and provide supporting econometric evidence: social polarization is important in explaining differences in fiscal outcomes across countries, yet its effects may be even more pronounced or suppressed, depending on the political and institutional structures through which social polarization is linked to the fiscal policy-making process. Indeed, effects on public deficits of the sociopolitical variables tend to be smaller in countries with better institutional arrangements. Conversely, the sociopolitical polarization has very strong effects on deficits in the presence of poor institutions. In general, there are rich interactions among socio-political and institutional factors. The results are confirmed by extensive robustness tests such as the sensitivity analysis and the robust estimation method. Finally, we empirically investigate the fiscal instability channel that negatively links income inequality to growth in Chap. 6. Although we have derived the theoretical result on the fiscal instability channel in Chap. 2, we have not tested it yet. We present supporting new evidence in a cross section of 93 countries for the period of 1970-2000. There are numerous empirical papers on the relationship between income distribution and growth, yet this is the first study to present such an evidence on this specific fiscal instability channel. As a matter of fact, most studies have focused on the reduced-form growth regression that includes an income distributional measure as an explanatory variable, rather than investigating specific channels through which income distribution affects growth. When tested against other competing theories, the data continue to support our fiscal instability channel. We pay careful attention to robustness and consistency of our results with respect to various issues in running a regression, such as outlier and endogeneity problems.

Fiscal Deficit, Fiscal Volatility and Growth: Social Polarization

2.1 Introduction "The fundamental difference between redistributionist and developmentalist governments is social polarization. Societies divided into factions fight over division of the spoils; societies united by a common culture and a strong middle class creates a consensus for growth—growth that includes the poor." from Easterly (2001, p. 256) In recent years, empirical studies on long-term growth have found that social polarization that arises from ethnic divisions or struggles over the income distribution is detrimental to growth. (See Rodrik 1999; Easterly and Levine 1997; Alesina and Rodrik 1994 among others.) Countries with polarized societies, as measured by ethnic fractionalization or income inequality, seem to be more prone to adopt growth-retarding policies—for example, unsustainable fiscal policies that lead to large budget deficits, volatile fiscal outcomes, and growth collapses.^ (See Figs. 2.1 and 2.2.) In such countries, socio-economic groups may have sharp disagreements on ideal government policies, which may cause a coordination failure among policymakers that leads to an adoption of individually rational but collectively inefficient policies. Perhaps social polarization is one of the oldest ideas found in the political economy literature. Yet there are very few (or no) systematic theoretical studies on the role of social polarization (of preference) in collective decisionmaking process and in the development of aforementioned macroeconomic problems. In general, the heterogeneity of preferences is one factor that has Reprinted from European Economic Review, Vol. 49, August, 2005, pp 1451-1477, Woo J, "Social Polarization, Fiscal Instability and Growth", with permission from Elsevier. In a comprehensive study of public sector deficit, Easterly et al. (1994) conclude that large fiscal deficits are largely explained by conscious fiscal policy choices and not by external or by domestic macroeconomic shocks.

8

2 Fiscal Deficit, Fiscal Volatility and Growth: Social Polarization

not been well-recognized as critical to the coordination failure in collective action. This chapter provides a systematic analysis of the role of polarization among socio-economic groups in the evolution of fiscal instabilities (large deficits and fiscal volatilities) and their negative effects on the capital accumulation process.^ We build a dynamic game model of fiscal policy in a simple growth framework in which social polarization (more precisely, polarization of preference for types of government spending between socio-economic groups) plays a central role in both generating fiscal instability and growth collapse. Thereby, we emphasize that society's polarization and degree of polarization are key factors underlying policy decisions that are responsible for such undesirable macroeconomic outcomes. On the other hand, social polarization may not only be responsible for a coordination failure but is often thought to be associated with socio-political instability. Empirical growth literature also finds that socio-political instability is harmful to growth. (See Perotti 1996; Alesina et al. 1996.) High levels of socio-political unrest may not only make the downfall of the present government more likely but may dramatically shorten the horizons of politicians. With a shortened expected tenure in office, the government would be more likely to engage in short-term policies at the expense of macroeconomic stability. In this chapter, we fully distinguish the incentive for policymakers to engage in individually rational but socially inferior policies under political uncertainty from that under polarization of preferences. We make a contribution to the literature by clearly bringing out the different roles of polarization and political uncertainty (refiected in the discount factor) in generating fiscal deficits, volatile fiscal outcomes and output collapses, and in reducing economic growth along a transition path to a steady state in a unified framework. Interestingly, social polarization and political uncertainty are shown to compound to produce even worse outcomes in terms of fiscal instability and poor growth. Here we consider an economy in which two heterogeneous policymakers jointly control fiscal policy, but have different objective functions. Each policymaker maximizes her own utility from her public good provision that benefits a specific group (or sector) more than the other. Each represents a different group that may have a different preference for the public goods. The two groups can be thought of as either capitalists and labor workers, manufacturing (formal) and traditional (informal) sectors, right-wing and left-wing parties, urban and rural sectors, or two powerful ethnic groups. When the preferences for types of government spending differ substantially among policymakers (or equivalently among the groups they represent), a fiscal deficit occurs endogenously. This is due to strategic behaviors of the policymakers who have different preferences yet share the government budget. Each ^ Throughout this chapter, the term "fiscal instability" means both a large fiscal deficit and fiscal volatility.

2.1 Introduction

1

J

• Taiwan

•Kore%SfK&apore

• Hong Kong

• Malaysia • Thailand^ Botswana • Indonesia

• Ireland Portugal • Tunisia

I i

•Sweden ^ New Zealand • Switzeriand

^tGree«rf»«BWf^gn^. """^ •Mexico • Morocco # Nigeria #(IMIi|iiHifl »IJIHtfirt'iM \ •Ecuador • Costa Rica ^^^^^-••..„„,^^^^ • Argentina

^

• Venfeiteiaaica

'Wl^^^&A^^ß^^^^^ ^ ™ V B f e n ^ a l ^ p^^ • South Africa ^ cote d'lvoire • Liljeria

1

• Zambia • Sierra Leone 0

|

0.5

SOCPOLA

Fig. 2.1. A scatter plot of average growth rate of real per capita GDP for the period of 1970-98 against a composite index of social polarization (SOCPOLA) that is based on Gini coefficients, ethno-linguistic fractionalization, and institutional quality around 1970. Higher values of SOCPOLA indicate higher social polarization. Data Source: World Bank (2000), Deininger and Squire (1996), and Easterly and Levine (1997). policymaker is aware that whatever government resources she does not exploit may or may not be available for the future provision of her preferred public good, depending on the spending decision of the other policymaker. When policymakers disagree on the composition of government spending, each of them has a greater incentive to overexploit the common government resources and consequently exerts a net negative externality on the other. This prevents them from achieving a socially optimal fiscal outcome. The size of deficit rises with the degree of preference polarization because of the positive relationship between the preference polarization and the incentive to exploit the common resources (polarization effect). Political uncertainty facing policymakers, as reflected in a high discount rate, has an effect similar to that of polarization. Importantly, an economy with a higher degree of polarization will also exhibit greater fluctuations in fiscal spending in response to shocks to the government revenue. The higher the degree of polarization, the more volatile the fiscal path. Relatively heavy discounting of the future events by the policymakers will cause a volatile fiscal spending path too. In the presence of preference polarization (or impatience) among the policymakers, a shock to tax revenue is translated into a more than proportional change in spending. That is, if the degree of polarization is positive or if the policymakers' subjective discount rate is substantially high, then government spending rises (falls)

10


• Uganda

• Brazil

• Sierra Leone

• Argentina

•NiaP2imbia

^ Mala>^ ^ ^^~ • Malays«SSlÄtf^^^

I °]

• Italy

C O

i

* ^TßläH

• Belgluoi^

1 J

^ — « C ^ a •Spain tJi^SlI^^ j^KinaÔmited States rk • France • Sweden

• Sri ferteBP^Jca^^^,. '"'•TMorocco

• Egypt^

• C ^ s f i f e ' " ' ^ " *^®P; • South Africa ^ PhiiiDDinl&'^^^ • Costa Riq^Tunisia • Indonesia * Kniiippin^ • Vtnf tateid • B Salvactoeenegal • Colombia • Guatemala

• Norway

0

0.5

SOCPOLA

Fig. 2.2. A scatter plot of a measure of macroeconomic instability over the period of 1970-98 against the social polarization index (SOCPOLA). The index of macroeconomic instability is based on CPI inflation rates, central government deficits, and volatility of real GDP growth (measured by standard deviation), whose higher values indicate greater macroeconomic instabilities. Data Source: The same as in Figure 2.1.

more than proportionally in response to a positive (negative) shock to tax revenue—^this can shed light on the procyclicality of fiscal policies in Latin America that is extensively documented by Gavin and Perotti (1997) (see Woo 2005b for empirical evidence on this prediction). The intuition is similar to the polarization effect. This is due to an interaction between the shock to the government revenue and the policymakers' incentives to exploit it in the presence of the dynamic negative externality operated by the preference polarization or the political uncertainty. The output level and the transition dynamics of economic growth also depend crucially on polarization and political uncertainty. A fiscal deficit arising from polarization or political uncertainty leads to ineflficient capital accumulation in the private sector, and permanently lowers the levels of capital stock and output in the economy. This is because policymakers waste valuable government resources to maximize their utility from producing public goods; hence, they overspend beyond the socially optimal level for a given tax revenue. In the presence of polarization (and/or political uncertainty), the economic growth rate is reduced along a transition path to a new steady state of a lower level of output. The higher the degree of polarization or the subjective discount rate is, the sharper the decline in the growth rate is. In our model the growth collapse is caused by fiscal instability that is ultimately

2.1 Introduction

11

attributed to social polarization (and/or political uncertainty). We characterize the transitional dynamics of growth as a function of polarization and political uncertainty. Our fiscal mechanism is related to a growing literature on fiscal politics (see Alesina and Perotti 1995; Persson and Tabellini 1999a for a literature survey), and particularly to the common pool or pork barrel problem approach. The related papers are Weingast et al. (1981), Chari and Cole (1993a), Tornell and Lane (1998), Hallerberg and von Hagen (1999), and Velasco (1999).^ Under this approach, an excessive spending or deficit (i.e., an overexploitation of a common property) can arise because interest groups that have access to the government resource fail to internalize the full cost of their own appropriation. Each interest group enjoys the full benefit of a specific public spending, while it pays only a fraction 1/n of the cost (i.e., the cost spread through generalized taxation). These studies typically consider n-player symmetric games (although in different contexts), addressing issues such as whether an increase in the number of groups n leads to a worse economic outcome, the possibility of delayed fiscal reforms, or the effect of budgetary institutions on the size of deficits.^ Here we address different issues, however. In general, the heterogeneity of preferences among the players is one factor that has not been well-understood as critical to the coordination failure (i.e., dynamic negative externality) in collective action in a common pool setting. So is the subjective discount factor of the players. In this chapter, a more general model is developed that introduces two new dimensions of preference polarization and political uncertainty into a class of two-player common pool games.^ It is shown that preference polarization and discount factor are critical conditions for the dynamic negative externality to become operative in the common pool context. Not only would this help us to theoretically identify critical conditions leading to overexploitation of the common resources, but also would yield richer, yet distinct, theoretical and empirical implications for fiscal dynamics and the capital accumulation process as shown in the chapter. For example, our result suggests that the common pool problem would be more likely to occur and be more severe in societies with higher degrees of polarization (and/or political The common pool problem refers to a situation in which a productive asset is exploited jointly by economic agents whose noncooperative behavior results in an overexploitation of the asset that is not Pareto optimal. The existence of Pareto inefficient Nash equilibrium in this context was first shown by Levhari and Mirman (1980). See Fudenberg and Tirole (1992) for more details. ^ The relationship between the number of groups n and the common pool problem is addressed by Weingast et al. (1981), Tornell and Lane (1998), and Velasco (1999); the possibility of delayed fiscal reforms by Velasco (1999); and the effect of budgetary institutions on the size of deficits by Hallerberg and von Hagen (1999). ^ Thus, a typical two-player common pool model in the existing literature can be viewed as a special case in our framework.

12


instability). By sharp contrast, the existing literature tends to associate the severity of the common pool problem with the number of participants in the collective decision-making process. However, the theoretical relationship between the number of groups and the common pool problem is fragile because it turns out to depend crucially on the assumptions about the shape of utility function (see Kontopoulos and Perotti 1999). Moreover, we go well beyond the issue of budget deficit by analyzing the theoretical linkage of social polarization (and political uncertainty) to fiscal volatility, procyclicality of fiscal spending, and economic growth process. Again, in contrast, the existing models of fiscal deficits do not address the volatility or procyclicality of fiscal outcomes, let alone economic growth. The innovation of this chapter is that the size of fiscal deficit, the magnitude of fiscal volatility, and the decline in output and economic growth rate are explicitly shown to be increasing functions of the degree of social polarization and the subjective discount factor. Social polarization has long been mentioned as an important explanation for populist fiscal policies and poor macroeconomic performance in many developing countries.^ Yet there have been very few systematic theoretical studies on the role of social polarization per se in these problems.^ We fill this void in the literature by demonstrating how social polarization can cause the aforementioned fiscal and growth problems.

As a matter of fact, most of the existing studies tend to focus on either income inequality or ethnic fractionalization (sources of social polarization), rather than social polarization of preferences per se. For example, many economists have argued that unequal income distribution provides an important answer to the questions of why populist fiscal policies appear more often in Latin American countries than other regions. See Rodrik (1996) and Kaufman and Stallings (1991) among others. Even in this literature, however, there are very few theories that explain why unequal income distribution can lead to large deficits and volatile fiscal outcomes! A partial exception is Alesina and Tabellini (1990) who briefly discuss the linkage between polarization and budget deficit in a different mechanism. They show that faced with re-election uncertainty, the incumbent may fail to internalize the costs of additional debt and hence tend to have a deficit bias. Then they extend their argument into the case where two political parties have different preferences. Let alone the different mechanism we employ in our model (i.e., the common pool setting), however, in this type of model it is not the polarization of preference per se but the re-election uncertainty that is the critical condition for an endogenous deficit to arise. (See also Chari and Cole 1993b on this point.) If the government is not faced with the re-election uncertainty, deficit bias would not occur regardless of the opponent's preference. On the other hand, it is surprising that social polarization has largely been ignored in the empirical studies of fiscal deficits. Woo (2003b) presents the first econometric evidence that countries that have suffered greatest fiscal deficits also tend to be those with highly polarized societies as meeisured by indicators of income inequality in a panel of 90 countries over the period of 1970-90.

2.2 Endogenous Growth Model with Optimizing Interest Groups

13

Last but not the least important, we offer an alternative explanation for an important empirical finding that social polarization is harmful to economic growth: a fiscal instability channel. This can be viewed as an alternative fiscal policy channel, but distinct from the redistributive policy channel proposed by Alesina and Rodrik (1994) and Persson and Tabellini (1994). In their models, distributive conflicts within a society lead the government to engage in redistributive policies that may be harmful to economic growth. However, there seems to be a lack of empirical support. What matters for growth in their models is the distortion caused by income tax that accompanies redistributive spending. Perotti (1996) does not find any negative relationship between tax variables and growth. By contrast, our theoretical explanation is highly plausible since fiscal deficits are found to be harmful to growth in numerous empirical growth studies (Fischer 1993, to begin with). And recently Woo (2003a, 2003b) finds strong evidence that social polarization, as measured by income inequality, is robustly and positively associated with fiscal deficits in a comprehensive empirical investigation (see Chaps. 4 and 5). In Chap. 6, we also present evidence both for the positive relation between income inequality and fiscal volatility and for the negative relation between fiscal volatility and growth in a cross section of countries in the 1970-2000 period. The plan of the chapter is as follows. Sect. 2.2 presents a simple endogenous growth model with optimizing interest groups. Sect. 2.3 derives an endogenous government fiscal policy and establishes our main results on the linkage between social polarization and fiscal instability. The social planner's solution is then computed and compared with the non-cooperative feedback Nash equilibrium determined by two heterogeneous policymakers. This is followed by an extension of the model into the case of policymakers facing political uncertainty. Sect. 2.4 analyzes the effect of this endogenous fiscal policy on the capital accumulation and growth in the presence of polarization and political uncertainty. Our conclusions are in Sect. 2.5.

2.2 Endogenous Growth Model with Optimizing Interest Groups We consider an endogenous growth model with no population growth. The economy is populated by a government and a private sector composed of two groups, indexed by i, i = 1,2. These two groups may represent either capitalists and labor workers, manufacturing (formal) and traditional (informal) sectors, urban and rural sectors, two powerful vested interest (ethnic) groups, or right-wing and left-wing parties. Each group consists of a large number of atomistic individuals. The government and the private sector have perfect foresight. The infinitely-lived representative agent in group i seeks to maximize her lifetime utility, which is additively separable:

14


J'=

f

[log{ci)-{-Xilogigi) + {l-Xi)log(92)]e-^'dt,

(2.1)

where Ci is private consumption; gi and p2 are two different public goods provided by the government; and p is a subjective discount rate, p > 0. Being small, each member of group i takes gi as given and has the same preference for the two public goods within the group. But these two groups differ in their preferences for the public goods, which is reflected by Aj. We assume that 0 < Aj < 1, for i = l , 2 and A2 < | < Ai. This implies that group 1 prefers gi to ^2 and group 2 prefers p2 to pi. Even though the agent in group i may not like the public good gj, j ^ i as much as gi, it is included in her utility function because of non-exclusiveness of public goods. We also assume that she derives positive utility from the consumption of public good which is not her most favorite one.^ We define ^ = Ai — A2 and interpret it as the degree of difference in their preferences for two public goods. We can think of Ö as a degree of polarization between the two groups. We note that 0 < ^ < 1. While 9 = 1 implies the complete disagreement on the composition of two public goods between two groups, 0 = 0 implies the total agreement in their preferences. We will see the important role played by 6 in the evolution of fiscal deficit and fiscal volatility. Also, capital accumulation and growth process depend crucially on 0, as we will show later. In the economy there are two kinds of real assets: capital, denoted by k, and government bonds, denoted by b. The bonds are assumed to be a perfect substitute for capital and therefore to pay the same rate of interest, r. The dynamic budget constraint of the representative agent in group i is then for Vt > 0 and ao > 0 given, äit = rau - Cit - n,

(2.2)

where an is the asset held by an agent and hence an = ku + ba^ and TI is a lump-sum tax collected by the government from group i.^ We also impose the No-Ponzi-Game (NPG) condition: lim ttite"^* > 0.

(2.3)

As long as marginal utility is positive, the agent will not want to have increasing wealth forever at the rate of r, and that condition will hold as an equality. (See Barro and Sala-i-Martin 1995.) The representative agent in group i maximizes the lifetime utility function (2.1) with respect to c^, subject to equations (2.2) and (2.3). It follows from the first-order conditions for this maximization problem that we get ^ For example, even though an agent may care about the public education expenditures much more than the national defense expenditures, she will also benefit from the national defense. ^ We assume n = r2. The assumption of lump-sum taxation is mainly for simplicity of algebra and can be relaxed without affecting the qualitative implications in the chapter. In what follows, we mean ^ by x.

2.2 Endogenous Growth Model with Optimizing Interest Groups ^ = r - p .

15 (2.4)

Cit

Thus, the optimal consumption path of the agent i is at = coe^^-"'*,

(2.5)

where CQ remains to be determined. (See footnote 12 for the solution.) The budget constraint for the whole private sector is at = rat -ct-T,

Vt > 0,

(2.6)

where at = ait+0L2t, h = kit-{-k2t, h = 6it+&2t5 Q = Cit+C2t, and r = r i + r 2 , for alH > 0 (we normalize as if there is one agent in each group). Now, following Barro (1990), we introduce firms that have the linear production function: (2.7) y = f{k) = Ak, where A > 0 is the constant marginal product of capital. We can think of capital as encompassing human and nonhuman capital.^^ In a competitive equilibrium, the marginal product of capital is equal to the rental price for a unit of capital services. This is the first-order condition for maximization of profit. Therefore, in competitive equilibrium A = r-\-ö,

(2.8)

where 6 is a. constant depreciation rate and r -h (^ is the rental price for a unit of capital services. Thus, consumption at time t is given by^^ at = coe(^-^-^)*.

(2.9)

From the dynamic budget constraint equation (2.6) and the profit maximization condition equation (2.8), we get the following (We suppress the time index, t, when there is no confusion.): k = (A-6)k-c-gi-g2.

(2.10)

Thus, the equilibrium for the private sector is completely described by equations (2.9), (2.10), and (2.3), given government debt, 6, and the government budget constraint. ^° Also see Barro and Sala-i-Martin (1995). ^^ If the marginal productivity of capital is sufficiently large so that A - 5 - p > 0, then consumption grows over time. However, this does not yield unbounded utility because for given gi and p2, f=

/•oo

^0

[log{co) + (A-S-p)t

+ Xilog{gi) + (1 - Xi)log{g2)]e-'''dt < oo.

16


Suppose t h a t t h e government budget is balanced at each point in time. T h e n 6 = 0 and bt = bo,yt > 0. Under t h e balanced budget assumption, t h e equilibrium capital stock kt is given by^^

fc^ÎZl^

+ ê^^-P)*

(2.11)

T h e capital accumulation equation specified in equation (2.11) will be useful in computing t h e impact of polarization on capital stock and growth in Sect. 2.4. It is straightforward t o see t h a t t h e asymptotic growth r a t e of capital {k /k) is A — 6 — p. In fact, t h e growth rates of consumption, capital stock, and o u t p u t all asymptotically approach A — 6 — p}^ In other words, lim - = lim - = lim t—ôo k

t—»-oo c

'^=r-p

= A-S-p.

(2.12)

t-*oo y

For now, we assume t h a t t h e agents' subjective discount r a t e is equal t o t h e interest rate, i.e., p = r, so t h a t capital stock and o u t p u t stay constant under a balanced budget {b = 0). Also, note t h a t consumption is constant if p = r. This assumption is only made to serve as a benchmark and t o highlight ^^ Using equation (2.10) and the NFG condition (2.3), we get kt. Under the assumption of a balanced budget, equation (2.10) is k — rk = rbo — r — 2coe^^~^^*. To solve this first-order differential equation, multiply both sides of the equation by the integration factor, e~^*, and integrate it from t to oo. We then have /»oo

oo

/

e-"*(fc - rk)dt = / e-"*(röo - r - 2cQe^''-^^')dt. By applying the NPG condition limt-ôo kte~'^* = 0 to the integration, we can derive equation (2.11). Since y = Ak^ we easily find the output path by using equation (2.11): y = Ak. The initial level of consumption, co, is determined by the following condition. From equation (2.11), T-rbo 2co fco =

h

,

r p and Co is thus determined by the initial level of the capital stock, the lump-sum tax, and initial bond holdings. ^^ Note that in the standard AK growth model, there is no transitional dynamics; Ct, kt, and yt grow at the constant rate of A — S — p. However, in our model where we introduce tax and the government bond, the growth rates of kt and yt are not constant at A — S — p, but only asymptotically approach that rate. This can be checked in equations (2.11) and (2.12). This also implies that changes in polarization affect the growth rate of the economy only along the transition path to the steady state; the same is true for political uncertainty. For more about this, see Sect. 2.4.

2.3 Endogenous Fiscal Deficit and Volatility: Polarization

17

the main points of the chapter. Later, we will discuss the consequences of relaxing this assumption (see Sects. 2.3 and 2.4). Before we move to government fiscal policy, we make an additional assumption on the timing structure. Policymakers are assumed to simultaneously move before the private sector moves; therefore, when the private sector makes a decision, it has information about government fiscal policy that was determined by policymakers and takes the government policy as given.

2.3 Endogenous Fiscal Deficit and Volatility: Polarization In the previous section, we took the government budget as given and assumed a balanced budget, 6 = 0. Now we consider the endogenous fiscal policy controlled by two policymakers (interchangeably called ministers) who jointly represent the Fiscal Authority (FA) of the government. We assume that the government can transform the consumption goods produced by the private sector into two non-storable public goods, gi and p2. Two ministers indexed by i, i = 1,2, represent the corresponding group, i = 1,2, in the private sector. Minister i provides the public good gi to the private sector, which is financed by government revenues. Each minister, i, derives greater utility from the provision of her favorite public good gi than from the other gj. Since they have different preferences for the two public goods and seek to maximize their own utility, two ministers faced by the common government budget constraint behave strategically in determining the amount of public goods they provide. To describe this endogenous fiscal policy determination process, we consider a diflPerential game between two ministers. Specifically, we explore the set of feedback Nash equilibria, which are subgame perfect and time consistent, in a game-theoretic model in which two ministers can jointly exploit the government net revenues to maximize their utility from providing their favorite public goods.^^ Each minister, i, has the following objective function: V' = / [Xilog{gu) + (1 - Xi)log{g2t)]e-'^'dt, Jo

(2.13)

where 0 < A 2 < | < A i < l and the minister's discount rate is assumed to be equal to the interest rate.^^ Minister i prefers gi to gj, j ^ i and i, j = 1,2,

15

It is well known that the subgame perfect equilibrium is time consistent. For more about the feedback Nash Equilibrium and time-inconsistency problem, see Cohen and Michel (1988). Each minister's utility is assumed not to depend on her consumption, which might bias the policies towards an oversupply of public goods. Yet the assumption that the discount factor is equal to the interest rate delivers a constant consumption path as one can see from equation (2.5). Without loss of generality, we can nor-

18


which implies that she puts more weight on her favorite public good, gi, in her utility function. Minister i shares the same weight Xi with her favorite group, so that 0 = Xi — X2 is also the degree of preference polarization between the two ministers. Now we turn to the budget constraint of the government which faces the ministers. The government collects the lump-sum tax of r from the private sector (with normalization of the number of agents in each group to one). Government expenditures can also be financed by issuing bonds at a constant real rate of r. The government budget constraint at each instant is then b = rb-\-gi-\-g2-T,

(2.14)

where b is the stock of national debt.^^ 2.3.1 Non-cooperative Feedback Nash Equilibrium Each minister i chooses her control variable, pj, so as to maximize her utility function (2.13) subject to the government budget constraint equation (2.14) and the NPG condition for every possible choice of the other minister's control variable gj, j 7^ i. Here we employ the feedback Nash equilibrium concept, which allows players to revise their actions through time as the game evolves.-^^ To facilitate the computation of equilibrium in this game, we define government net revenue, Rt, as Rt = r - rbt.

(2.15)

In general, the feedback strategy is a function of time and state; however, very few diflFerential games can be solved in closed form because the firstorder condition for this feedback Nash equilibrium involves a system of partial differential equations. In order to get a closed-form solution, we restrict the strategy set to linear Markov strategies that depend on the current state. We

16

malize the constant consumption to 1. Then log(c) = 0 justifies the specification of the utility form. Note that the No-Ponzi-Game condition relevant for the government budget constraint is lim 6te-"* = 0. t—>oo

17

In differential games, open-loop and feedback Nash equilibria are among the most commonly employed equilibrium concepts. Open-loop strategies are ones for which each player chooses all the values of his control variable for each point in time at the outset of the game. This is relatively easy to solve for, but in general is time inconsistent. On the other hand, feedback strategy consists of a contingency plan that indicates what the best thing to do is for each value of the state variable at each point in time. That is, it allows the player to revise her action at each instant on the basis of the state at that point in time. Hence, the feedback strategy has the property of being subgame perfect.


19

then conduct transformation of the variables so that we can construct a game structure in which an open-loop strategy calls for the same rate of public good provision at every point in time as that of the feedback strategy. This is known as "synthesizing the feedback control." ^^ Let us now consider the following linear strategies: 9it=XiRu (2.16) where xi will be endogenously determined as a part of the solution. We assume that the set of strategies is xi ^ [0, oo). Let t/^t = logRt- Then, we can rewrite the minister's objective function (2.13) as /•oo

V^'(Xi,X2)= / [A^/op(xl) + (l-A,)/op(x2) + V't]e-^*rft, Jo

(2.17)

and the budget constraint becomes i^t = r-rxi

-rx2'

(2.18)

Now we solve for the feedback Nash equilibrium, which is also the Markov perfect equilibrium, by maximizing the objective function of minister z, equation (2.17), with respect to Xi subject to equation (2.18). Minister i's Hamiltonian is given by = l>îlo9{xi) + (1 - K)log{x2) + V'tje"^* + fiit[r - rxi - rx2], (2.19) where Xi is the control variable, fiu the costate variable, and tpt the state variable. The feedback Nash equilibrium to this game is as follows (See Appendix A for the derivation of the following feedback Nash equilibrium.): H'{xuX2,ipt)

Xi = Al, X2 = (1 - A2); and hence p*, = XiRt, git = (1 - \2)Rt- (2.20) Substituting p*^ = x?-^* and Rt = r — rbt into equation (2.14) yields bt = (Al - A2)(r - rb) = 0{T - rb) > 0.

(2.21)

Recall that the parameter 0 £ [0,1] is the degree of the polarization between two ministers. Whenever there are differences in the ministers' preferences for two public goods (i.e., Ö > 0), there occurs an endogenous fiscal deficit, 6 > 0.^^ This result is due to the strategic behaviors of these ministers who ^^ In the differential game literature, it is very common to consider linear strategies due to the aforementioned technical complications. See Fudenberg and Tirole (1992) for more about synthesizing the feedback control. ^^ However, the growth of debt is not explosive. If we solve the differential equation (2.21) for öt, assuming 60 = 0 for simplicity, we obtain ht = ^ — ê~^^*. Thus, the NPG (No-Ponzi-Game) condition is satisfied: limt-.oo btß-''* = {^-ê-^''*)e-''^ =

20


have different preferences, but share the government budget. Each minister is aware that whatever government resources she does not exploit may or may not be available for future provision of her preferred public good, depending on the spending decision of the other minister.^^ Thus, when they disagree on the ideal composition of government spending, each has an incentive to overexploit the common resource today. The polarization of preference leads each policymaker to insist on a higher spending for her favorite sector and to exert (net) negative externality on the other, contributing to bigger overall spending and a larger deficit than the social optimum. Whenever two ministers value public goods with different weights, the negative externality of minister j ' s one-unit provision of QJ on minister i's utility through the state variable h always dominates the positive effect that directly enters minister i's utility function. Moreover, the incentive for each minister to overexploit the government revenues increases with the amount of disagreement between the two ministers (polarization effect). This is because the positive effect of minister j ' s one-unit provision of QJ that directly enters minister i's utility function gets smaller, whereas its negative externality operating through the state variable h gets bigger, as the degree of polarization gets bigger. From the point of view of minister i, therefore, one unit of resource devoted to her opponent's favorite type of spending brings a greater net negative externality, inducing her to spend even more for her favorite item ahead of her opponent. This implies that the size of the current budget deficit is a positive function of the degree of polarization (see Fig. 2.3). Along with this result, the intertemporal budget constraint implies that polarization is positively associated with greater changes in fiscal outcomes over time, such as spending and fiscal balance for a given path of tax revenue. The greater the polarization is, the larger the fiscal spending and current fiscal deficit are. But this only raises the debt level more quickly and reduces available government resources, which forces policymakers to cut tomorrow's spending by more.^^ The intertemporal budget constraint means that larger 0. As t —^ oo, 6 approaches J . This is because the game and the strategies are constructed such that it is in each minister's best interest to spend less so as to satisfy the NPG condition. Note that each minister's spending gi depends only on the state variable, net tax revenue {r — rh). As debt (6) is accumulated, the net tax revenue shrinks, which forces each minister to spend less. As a result, total government spending shrinks asymptotically to zero when there is polarization (^ > 0). This is merely due to the lump-sum tax assumption. By contrast, when 0 = 0, the budget is then balanced, and total government spending equals the lump-sum tax for each time period. ^° It is not the level of tax that causes a deficit in our model. This result holds true for any given level of tax revenue. ^^ It should be noted that our model is not related to the Ricardian equivalence experiment. The Ricardian equivalence proposition implies that the timing of taxes does not matter as long as the present value of taxes is equal to the present


21

deficits today must be met by larger surpluses tomorrow, causing an even bigger swing of fiscal policy over time (see Fig. 2.3 again). Importantly, an economy with a higher degree of polarization will exhibit greater fluctuations in fiscal spending in response to shocks to government revenues. We can illustrate this point by using the solution for gi and p2Using equation (2.20), we can write the total government spending (g) at time t as 9t = gû + g^t = (1 + 0){T - rbt).

(2.22)

For any point in time t, government spending is proportional to the net tax revenue r — rb. Note that the policymaker revises her action at each instant on the basis of the state at that point in time, which should be r — rbt- • Thus, if we take the total differentiation on this equation at an instant t + , then (Qt-\= (1 + Ö) > 1 with equality when 0 = 0. (2.23) dr This yields a striking prediction that government spending rises more than proportionally in response to an increase in tax revenue if the degree of polarization is positive {9 > 0). Whenever there is a positive (negative) shock to the government net revenue, it is translated into a more than proportional increase (decrease) in government spending in the presence of polarization.^^ The absolute size of the change in 'g will be even greater with the size of 0. value of government spending plus the value of the initial government debt. This is because the government spending path is exogenously given, whereas taxes are endogenous. In sharp contrast, government spending is endogenous, while taxes are exogenous in our model. Therefore, government debt acts like net wealth for the private sector. Recall that the level of initial consumption depends on the initial levels of capital stock and government bond holdings, and the lump-sum tax (see Footnote 12). In the Ricardian world, higher bond holdings just mean a higher present value of future taxes, which makes government bonds irrelevant for consumption. In our model, however, higher initial bond holdings reduce the present value of future government spending. ^^ This result is reminiscent of the voracity effect in Tornell and Lane (1999) that interest groups' total appropriation of the economy capital stock rises more than proportionally to the windfall to the capital stock. Although the motive and the issues we address in our chapter are sharply different from theirs, the underlying mechanism for both polarization effect and voracity effect is the negative externality in common pool settings. Aside from other major differences from the existing common pool problem literature, the novel feature of our model is that we not only demonstrate that the polarization of preferences (and the political uncertainty, as will be shown in Sect. 2.3.3) cause the fiscal volatility to arise endogenously but also fully show that the size of fiscal volatility itself is an increasing function of the degree of polarization and the discount factor, which yields a new testable prediction. Moreover, the polarization of preference and the discount factor are shown to be the critical conditions for the dynamic negative externality to become operative in a more general two-player common pool games.


22

The intuition behind this result is quite similar to that behind the polarization effect. Recall that the equilibrium Markov strategy in equation (2.20) calls for minister I's spending to be equal to the multiproduct of Ai (or (1A2) for minister 2) and net tax revenue R. For a given shock to tax revenue Ar, minister 1 will claim Ai x A r , while minister 2 will want to increase her favorite spending by (1 — A2) x Ar. Unless Ai = A2 = 1/2, it will result in a more than proportional increase in total spending (Ap = {1 -\- 9) x A r ) . In the case of complete agreement (Ai = A2 = 1/2), the increase in tax revenue will be evenly split between two types of spending so that a balanced budget is maintained. 1

1

1

1

1

1

1

1

1

60

^^^-^"^^\^^,.^--^^^^"^^^^

40

e-

^^ ^^ 0), the growth rate of capital stock is also lower than that in the absence of polarization {0 = 0) for all finite periods of time. Respectively, the growth rates when ^ = 0 and 9 > 0 are given by

(r-p) (^)l.=o = ^ e - ( - p ) * + l = 0 and

(2.32)

30


i..n,

^ e - « « + 0 - . ) * + (,_^) r2co

^

•$rt

r2co

where 6o = 0 is assumed for simplicity, and the last term in each equation above is obtained under the assumption of p = r. It is clear from equations (2.32) and (2.33) that (klk)e^Q > (A:/A;)0>o, for all finite time t > 0. Over time, the growth rate of capital stock in the presence of polarization (9 > 0) converges to the rate of growth under a balanced budget, r — p, whereas the level of k itself is permanently lowered. That is, the asymptotic growth rate of capital is r - p: limt-.oo(|)^^ = r - p (see Fig. 2.8).'^^ Interestingly, however, the relationship between the growth rate and the degree of polarization for the economies with polarization (6 > 0) is not monotonic. A more polarized economy would experience a more dramatic change in its economic growth rate for a given period of time. For example, an economy with a higher degree of polarization 9 would see a sharper collapse of growth rate initially as it runs a larger deficit. However, this same economy would grow more rapidly later as its fiscal balance improves compared to an economy with a lower degree of polarization 6 (again Fig. 2.8). This reflects the fiscal deficit dynamics in relation to polarization as illustrated in Sect. 2.3.1. It is straightforward to see that the above results still hold even when policymakers are impatient enough to discount the future more heavily (i.e., p > r). So far, we implicitly assumed that the policymakers still share the same time preference rate p with the private sector, even when we allowed the case of p j^ r. We can further distinguish them by assuming that policymakers faced with political uncertainty discount the future more heavily than the private sector—^that is, p > p, where the policymakers' discount rate is p = p + p; p is the private agent's discount rate (which may be equal to r or not); and p is the constant probability of policymakers being removed from the office. Even in this case, the result still remains qualitatively the same. The transitional ^^ Since the marginal productivity of capital stock is a constant A (and hence the value of r—p is constant), it can produce perpetual growth without assuming some exogenous technological progress. However, the transitional dynamics of growth as a function of polarization 6 looks similar to that of neoclassical models such as the Solow model. For example, a change in saving rate in the neoclassical model can lead to a permanent effect on the level of output, whereas it does not change the steady-state growth rate. Similarly, a change in the degree of polarization 0 in our model leads to a permanent change in output and capital stock, but not to a permanent change in the steady-state growth rate. Thus, our model can explain why some nations are rich and others are poor, while the differences in growth rates across countries can be explained by appealing to the transition dynamics. From the empirical point of view, this interpretation that differences in growth rates across countries are due to the fact that countries are on different transition paths to their own steady states of output works reasonably well. For seminal empirical papers, see Mankiw et al. (1992), Barro (1991) and Barro and Sala-i-Martin (1992) among others.

2.5 Concluding Remarks

31

Fig. 2.8. The figure depicts the growth rate of capital stock, ( | ) ^ ^

=

'^''^———r < 0, for different degrees of polarization 6 (theta) under the assump+1] tion of r = 100, 6o = 0, 2co = 1, and r = p = 3.5%. dynamics of growth with respect to the probability of losing office p is similar to that with respect to polarization 6, whereas the steady-state growth rate is still r — p. To see this point, note that the government budget flow equation (2.27) becomes 6=(l + ^ - I ) ^ ( r - r 6 ) > 0 , if?>

j ^ ,

(2.34)

and that capital accumulation takes place according to kt =

^Jr-p)t

, jr-rbo)

+

j^_^^t

^

(öp+(i+ö)p)t

(2.35)

We can then clearly see that an increase in political uncertainty as reflected in p leads to a permanent reduction in output and capital stock, but not to a permanent decrease in the steady-state growth rate {r — p).

2.5 Concluding R e m a r k s This chapter has presented a dynamic model of fiscal policy in a simple growth framework where social polarization of preferences among socio-economic groups plays a central role in the evolution of fiscal instability and growth collapse. One key feature of the chapter is that the size of fiscal deficit, the

32


magnitude of fiscal volatility, and the size of reduction in output and growth rate are explicitly shown to be increasing functions of the degree of social polarization and the degree of political uncertainty (reflected in policymakers' subjective discount factor). Broadly consistent with recent empirical studies on fiscal instability and growth, our results can particularly contribute to explaining why many Latin American countries in the past decades have suflFered from chronic fiscal deficits, volatile fiscal outcomes, procyclicality of fiscal policies, and disappointingly poor growth. Our theory suggests that all of these problems can be ultimately attributed to the polarization within the government or among socio-economic groups in a society. Although social polarization might be among the oldest ideas in the political economy literature, we first theoretically demonstrate how social polarization can cause the aforementioned fiscal and growth problems. Furthermore, we fully distinguish the incentive for policymakers to engage in short-term policies under political uncertainty from that under social polarization of preferences. Not only do we show that these two are separate and distinct forces driving fiscal dynamics and growth process but also that these are critical to the coordination failure (that is, the negative dynamic externality) in a more general two-player common pool game. On the other hand, recent studies on budgetary institutions have presented evidence that stringent budgetary procedures and rules can directly influence fiscal outcomes (see Alesina and Perotti 1996 for a literature survey).^^ However, our theory suggests that social polarization may lie in depth behind the fiscal problems. So whether institutional arrangements (including budgetary institutions) can be made to mitigate the negative fiscal eflfects from social polarization remains to be an important question. In this regard. Woo (2003a) reports encouraging econometric evidence that the eflPects on public sector deficits of the social polarization tend to be smaller in countries with better (budgetary) institutional arrangements. Conversely, the social polarization has very strong effects on deficits in the presence of poor institutions. This also has important policy implications. For example, the fiscal decentralization that many countries have recently undergone may produce diflferent results depending on the underlying social polarization and institutional factors. In a highly polarized country, decentralizing power to local governments may only increase tension among central governments, regions and different groups, and can threaten macroeconomic stability unless it is checked by proper institutional restraints. This has been observed in some countries such as the Balkans, Indonesia, and Brazil in recent years.^^ ^° See von Hagen (1992), Hallerberg and von Hagen (1999), and Alesina et al. (1999) among others. ^^ Rodden et al. (2003) provide deatiled country case studies on various issues related to fiscal decentralization for eleven nations. In many countries, the fiscal decentralization often led to difficulty in maintaining fiscal discipline.

2.6 Appendix

33

2.6 A p p e n d i x A. Derivation of the decentralized feedback Nash equilibrium. Minister I's Hamiltonian is H\xuX2,i^t)

= [Ai/opxi + (l-Ai)/opx2+i/'t]e-"*+/iit[r-rxi-rx2]. (2.36)

The first-order conditions are Hl^ = —e-^* - fiir = 0; Hl = e"^* = - / i i ;

lim fi{t) = 0

(TVC).

(2.37) Using the transversality condition (TVC), we solve the first-order differential equation and we can get —rt flit) = -—. (2.38) Thus, the solution is Xi = Ai. Similarly, we can solve the optimization problem of minister 2, getting X2 = (1 — A2). Note that the first-order condition for the control variable, xii and the solution of fi{t) do not contain the state variable ip. This makes the open-loop and feedback strategies coincide. B. Derivation of the solution of social planner By using a social planner's objective function and the government budget constraint, we set up the following Hamiltonian function: = [(Ai+A2)/opxi+(2-Ai-A2)/opX2+2i/'t]e"''*+/xt[r-rxi-rx2]. (2.39) The first-order conditions are H{xi,X2,ipt)

Al + A2 _j.t ^ jj. 2 — Al — A2 _j.+ „ ,^ .^v e "^^ - /j,r = 0; (2.40) e '^'^ - fir = 0; H^^ = .-rt H^ = 2e-^' = - / i ; and lim fi{t) = 0 (TVC).

H^^ =

t—^00

Using the transversality condition (TVC), the first-order differential equation is solved to get ß{t) =

.

(2.41)

Thus, the solution is ^ Al -f- A2 , * 2 — Al — A2 Xi = —1^— and X2 = k -•

/r^ ^«s (2-42)

Inflation, Composition of Deficit Finance, and Social Polarization

3.1 Introduction "Milton Friedman's famous statement that inflation is always and everywhere a monetary phenomenon is correct. However, governments do not print money at a rapid rate out of a clear blue sky. They generally print money to cover their budget deficit. Rapid money growth is conceivable without an underlying fiscal imbalance, but it is unlikely. Thus, rapid inflation is almost always a fiscal phenomenon." from Fischer and Easterly (1990, pp. 138-39)

The previous chapter has developed a political economy model of fiscal policy in which social polarization plays a crucial role in driving macroeconomic outcomes such as fiscal deficits, fiscal volatility and poor growth. Now we extend the model to study inflationary consequence of the fiscal policy set by polarized policymakers by introducing money and inflation tax into the economy. The relation between fiscal deficits and inflation has long been a subject of economic research. As the above quote from Fischer and Easterly (1990) suggests, the fiscal consequences of the debt crisis seem to be the initiating disturbances in many countries that have suffered extreme inflation in the past decades (Dornbusch et al. 1990; Fisher et al. 2002). More recently, Catao and Terrones (2003) reaflfirm the positive association between deficits and inflation among high-inflation and developing country groups in a panel of 107 countries over the 1960-2001 period, although not among low-inflation advanced economies.^ The inflationary consequences of fiscal deficits are often more severe in developing countries because of the lack of central bank independence from political pressures to extend cheap credits or help to finance ^ Gosh et al. (2002) also report a positive and significant effect of fiscal deficit on inflation in a panel of 147 countries over the period 1970-99. Alfaro (2005) finds similar evidence in a cross-section of 130 for the period 1973-98, although the result becomes less significant in the annual panel data for the same period.

36

3 Inflation, Composition of Deficit Finance, and Social Polarization

budget deficits, and the lack of sufläciently developed domestic capital markets that can buy newly issued governments debt (see Chap. 5 and Caballero and Krishnamurthy 2004 for evidence on fiscal deficits and financial market development). In this chapter, however, we emphasize the role of social polarization as a fundamental force that may drive the relation between fiscal deficits and inflation. We derive a result that if there is a high degree of polarization and a large deficit arises as a result, there will be high steady-state inflation once the government turns to money financing. Moreover, the higher the degree of polarization, the higher the steady state inflation rate. Based on Israel's high inflation experience in the mid 1980s, Patinkin (1993) provides a supportive observation for what our model attempts to capture. He regards high inflation as resulting from the prisoners' dilemma in the coalition government. Out of fear that her position, relative to that of other minister may be worsened, each minister insists on her increased nominal budget. The overall budget, whose planned expenditures exceed its expected revenues, then generates inflation. As we saw in the previous chapter, when there is greater disagreements on the ideal composition of government spending and therefore high degree of polarization, each policymaker will have a greater incentive to insist on greater spending for her favorite type of spending program or public good, leading to overspending that causes inflation via monetary financing.^ Fig. 3.1 shows the scatter plot of inflation tax over the period of 1970-2000 against income inequality measured in 1970s (as a proxy for social polarization).^ A positive correlation is quite discernible. Indeed, several empirical studies have reported evidence of a positive correlation between income inequality and inflation over the period of 1965-1990 (Sachs 1989; Beetsma and Van Der Ploeg 1996; Al-Marhubi 1997). Desai et al. (2005) extend the sample period through 2000 and find a significant positive relation between the two variables in countries with democratic political systems in a panel of 120 countries. Also, there is a unconditional, positive and significant relation between inequality and inflation in the OECD country sample.^ Aizenman (1990) studies the consequences of inflationary finance when the inflation tax is determined by several policymakers, each of whom can print money via the central bank. He suggests that the failure of the central bank to impose the needed discipline with friction among the various policymakers may be responsible for the recent inflationary episode (Yugoslavia and Russia in the early 1990s). Cukierman et al. (1992) find that the level of inflation in a cross-section of countries is related to measures of political instability. Inflation tax is measured by 7r/(l4-7r), where TT is the inflation rate that is obtained from World Bank (2005). The income inequality indicator AGINI70 is the average of all available Gini coefficients in the 1970s in Deininger and Squire (1996) data set. Beetsma and Van Der Ploeg (1996) develop a theory of inflation that arises from distributional conflicts, which is similar to the reasoning in Alesina and Rodrik (1994). A democratic society in which the wealth distribution is unequal elects

3.1 Introduction

37

Argentin

.998283 IsraelUganda

Uruguay

funcey Venezuel

Costa Ri

ace

Dominica Kenya

Egypt Paldstan

South Af I

Sri Lanic

India United K Funiand

Norway Denmaric

Malawi

Jamaft^^'"^'^

Trinidad New Zeal

Barbaaos Fiji

Austraii traii Iwedei Sweden

Senegal

Thailand

Fr^niWglade

Hong Kon Belgium

.768751

Switzerl Germany

28.1

Singapor

AGINI70

65.38

Fig. 3.1. Inflation tax and income inequality. Interestingly, however, we show that inflation dynamics depend on the money-bond ratio in the deficit finance and an increase in that ratio is less inflationary for a given degree of polarization. That is, more reliance on moneyfinancing relative to bond-financing can be less inflationary. This paradoxical result is consistent with earlier studies by Turnovsky (1978) and Liviatan (1983). However, these studies derive the result for an exogenously given government spending. More important, our explanation is very diflFerent from theirs. The reason for our result is that as each policymaker tries to finance her expenditure through inflation tax, this increases the rate of inflation which, in turn, reduces the private agent's real money demand, tax base. Therefore, they would not be able to spend as much as they would with pure debt-finance. As a result, money-financing provides more discipline to these polarized ministers. Therefore, it follows that the money-financing can less inflationary if the government relies more on inflation tax relative to bond-financing from the beginning. In other words, bond-financing of fiscal deficit may not cause inflation initially. Yet as government debt accumulates as a result, it further

political parties that are likely to represent the interests of poor people. It is in the interests of their clientele of the resulting governments to attempt to levy inflation taxes in order to erode the real value of debt service and redistribute from the rich to the poor. Consequently, inequality and high levels of nominal government debt result in inflation in more democratic countries due to populist pressures for redistribution.

38


raises the potential inflation that would prevail when the government finally has to turn to seigniorage.^ More recently, Tornell and Velasco (2000) make a similar reasoning in a sharply diflFerent context in which they compare fixed exchange rate with flexible exchange rate regime for their fiscal implications. Conventional wisdom holds that fixed rates provide more fiscal discipline than do flexible rates. The claim that flxed rates induce more discipline stresses that sustained adoption of lax fiscal policies must eventually lead to an exhaustion of reserves and thus to a politically costly collapse of the peg. Hence, under fixed rates irresponsible fiscal behavior today leads to punishment tomorrow. Under flexible rates irresponsible fiscal behavior has costs as well. The diflPerence is in the intertemporal distribution of these costs: flexible rates allow the eflPects of unsound fiscal policies to manifest themselves immediately through movements in the exchange rate (and inflation). Hence, under flexible rates irresponsible fiscal behavior today leads to punishment today. They show that if fiscal authorities are impatient, flexible rates by forcing the costs to be paid up-front provide more fiscal discipline and higher welfare for the representative private agent. Despite the diflFerent mechanism at work, the implication of their result is similar to ours in that monetary financing (under the flexible exchange rate regime) may provide more fiscal discipline (than does fixed rate regime under which the budget deficit cannot be financed by monetary financing as long as the government is credibly committed to the fixed rate). The plan of the chapter is as follows. Sect. 3.2 describes the economy with money, and derives money demand of the private sector. Sect. 3.3 derives our main result on inflation, fiscal deficits, and social polarization. Sect. 3.4 concludes.

This is reminiscent of the argument of Sargent and Wallace (1981) that bondfinancing may be more inflationary for a given deficit, although the reason is very different. They show that when the government deficit is held constant, a temporary tightening monetary policy may eventually lead to higher inflation. This is because tight current monetary policy implies greater bond issuance and growing debt service to finance the given deficit. Related to this, Blanchard (2004) makes an interesting argument regarding the inflationary effects of inflation targeting in an open economy setting. If high inflation triggers a tight monetary policy stance under an inflation targeting that results in an increase in the real interest, it may lead to a real depreciation and to a further increase in inflation. It is because the increase in the real interest rate increases the probability of default on the government debt, which may cause a real depreciation. In this case, fiscal policy, not monetary policy, is the right instrument to decrease inflation. He argues that this is the case of Brazil in 2002 and 2003.

3.2 The Economy

39

3.2 T h e Economy The economy is populated by a government and a private sector composed of two groups, indexed by i, i = 1,2, as in the previous chapter. Each group consists of a large number of atomistic individuals. The government and the private sector have perfect foresight. Now we incorporate money into the lifetime utility by considering a Sidrauski type utility function as follows: [log{ci) + ^logirrii) + Xilogigi) + (1 - Xi)log{g2)]e-'''dt,

(3.1)

^0

where ci is private consumption; P the price level; m = yr the real money balance; and 7 > 0. We assume that the private discount rate p is equal to the real interest rate r, and focus on the inflationary consequences of deficit finance solely from social polarization (that is, there is no political uncertainty either because p = r and hence p = 0).^ As previously, gi and p2 îre two different public goods provided by the government. Being small, each member of group i takes gi as given and has the same preference for the two public goods within the group. Yet these two groups may differ in their preferences for the public goods as reflected in A^ G [0,1]. We continue to assume that A2 < | < Ai, which implies that group 1 prefers gi to p2 and vice versa. An agent can hold her wealth in the form of capital k, government bonds 6, and real money balance m. The bonds are assumed to be a perfect substitute for capital and therefore to pay the same rate of interest r. The flow budget constraint of the agent of group i is now for Vt > 0 and ÜQ > 0 äit = rau - Cit - (r + 7rt)mit - n,

(3.2)

where an is the asset held by an agent; an = kn -h hn + rriit ; and r^ is a lump-sum tax collected by the government from group i. The utility maximization subject to the above budget constraint and NPG condition yields the solution Ct = c, and rrit = 7(r + 7rt)~\

(3.3)

where TTt is an inflation rate. Consumption is constant since the private discount rate is assumed to be equal to the real interest rate r. Before we move to government fiscal policy, we make an additional assumption on the timing structure. Following Cohen and Michel (1988), we assume that the fiscal authority moves before the individual at each point of ^ Since the discount rate is assumed to be equal to the interest rate, then consumption will be constant over time as one can recall from a private agent's life-time utility maximization problem in the previous chapter. This makes it easier to focus on the inflationary consequences of endogenous fiscal policies, which is the main purpose of this chapter. This separability assumption will render the behaviors of the financial sector completely independent of the production sector.

40


time and is able to precommit for an instant of time. Noting that the private consumption is independent of each minister's public good provision decisions, and taking into account the money demand of the private agent, each minister chooses her own public good provision level, g*, which is the Markov perfect equilibrium. The fiscal authority moves first, announcing its gi and p2 obtained by solving the optimization problem subject to the budget constraint and to the response of the private agent, Ct and rut. The agent takes the government spending as given and solves for Ct and rrit so that she can maximize her life-time utility subject to her budget constraint.^

3.3 Inflation, Composition of Deficit Finance, and Social Polarization Each minister i has the following objective function: /•CX)

V' = /

[Xilog{git) + (1 - Xi)log{g2t)]e-'^*dt,

(3.4)

where 0 < A 2 < | < A i < l ; and the minister's discount rate is assumed to be equal to the interest rate r. Minister i prefers gi to gj, j ^ i and i, j = 1,2, which puts more weight on her favorite public good, pi, in her objective function. As in the previous chapter, we define ^ = Ai ~ A2, and interpret it as a degree of polarization between ministers (equivalently between two groups). Each minister would maximize her objective function subject to the relevant government budget constraint and the private agent's real money demand function (3.3). We assume that money does not enter each minister's utility function because it is only used to finance her public good provision, gi. Thus, the money supply is indirectly determined by the ministers' spending levels, gi. Once the government finances its spending by seigniorage as well as debtfinancing, the relevant government budget constraint becomes b-\-m = gi-\-g2-hrb

— T — 7rm,

(3.5)

where irm is the inflation tax. Using the money demand equation (3.3), we can get irm = 7 — rm. Note that as inflation rate goes up, the inflation tax increases. But there is a maximum of inflation tax revenues that can be collected by printing money because money demand depends negatively on the nominal interest rate.^ ^ Here we do not consider the distinct role of the central bank explicitly by assuming that the fiscal authority controls the central bank for simplicity of analysis. There is no distinction between the central bank and the fiscal authority. ^ To see this, we compute the first and second derivatives of 7rm with respect to TT as follows:

3.3 Inflation, Composition of Deficit Finance, and Social Polarization

41

Since we are primarily interested in inflationary consequences of deficit finance and in the relationship among inflation, composition of deficit finance, and degree of polarization, we consider the following experiment. The fiscal authority begins by using both debt-finance and money-finance at a given ratio of money and bonds (^) and then turns to pure money financing at time T in the future, either because there is a maximum level of debt service or because the fiscal authority has no other way than inflation tax to continue to finance its expenditures.^ The maximum level of debt service is determined by the lump-sum tax r in our model.^^ Let us start with the simplest case of ^ = 0. That is, the government initially uses pure bond-financing before switching to pure money-financing. Also, let us assume, for simplicity, that the money supply is constant at the level 7r~^ before the switch in order to guarantee money market equilibrium. This switch takes place at time T when the fiscal authority changes its financing method so as to freeze the debt at its then current level bx, and hence 6 = 0 from time T. Thus for alH < T government debt is increasing, money balances are constant, and so is private consumption. After T the government debt is not increasing, and money balances are decreasing as the government begins to finance its expenditures by printing money and generating inflation. The Mt = 0 for t < T ; 6t = 0 for t > T ; and the government budget constraints are thus

dTTt

(r-f-7rt)2

but the second derivative is negative. That is, dTT?

(r + 7r03

'

^ There is another incentive to turn to inflationary finance. When the government has issued a large debt, the government might want to reduce the burden of debt by inflating away the debt. This incentive is studied by Missale and Blanchard (1994). They derive the maximum maturity of debt in which this incentive to inflate the debt burden is restrained by the reputational consideration. 10 Temporarily, the government might not be able to finance its spending by printing money, so the government relies on pure bond-financing. For example, the central bank often pursues an independent monetary policy by not accommodating the fiscal policy. Empirically, however, most independent monetary policy regimes are of short duration in practice. See Klein and Marion (1994). Dornbusch et al. (1990) and Drazen and Helpman (1990) also study the inflationary finance and the government fiscal deficits. Specifically, Drazen and Helpman (1990) discuss stabilization policies and inflation, and study how expectations of a policy switch affect economic dynamics before the switch, whose timing or mix among expenditure cuts, tax increase, or increases in money growth rates may be uncertain.

42

3 Inflation, Composition of Deficit Finance, and Social Polarization b = 9i-\-g2-^rb-r, for t < T

(3.6)

m = pi + p2 + rbr — r — Trm, for t>T. We can solve for the endogenous g* by maximizing the minister's utility subject to equation (3.6) and to the response of private agent, equation (3.3). The feedback Nash equilibrium gi for t > T is p^ = Ai(7rm+r—rör) and P2 = (1 — A2)(7rm + r — rbr)- Substituting this solution into the second equation of the government budget constraint equation (3.6) yields the following equation: m = 6>(7rm + r - rbr), for t > T.

(3.7)

We can see that the change in real money balance will be greater as the degree of polarization increases. Note that the dynamics of m around its steady state is unstable (i.e., the coefficient of m in equation (3.7) is On > 0). The unique convergent perfect foresight path includes an immediate jump to the new steady state level: TT^H- = TT*. Furthermore, the higher the degree of polarization 0, the higher the steady state inflation rate that is reached after the switch, which we show below. Since m = 0 in the steady state, by using equation (3.7) and m = 7(r + 7r)~^ in equation (3.3), we can get the steady sate inflation rate TT*: ^'"'^ =rbT-T. r-t-TT

(3.8)

From the above steady state inflation, dn* _ (r -f TT*)^ > 0. dbr 7

(3.9)

The steady state inflation rate is higher if the level of debt at T is higher. Also, an additional bond issuing at T before the switch accelerates the steady state inflation that would prevail after the switch. One can show that the government debt bx at time T should be equal to (assuming 6o = 0)^^ (3.10) br = - - -e-'^^. r r Combining equations (3.8) and (3.10) and then diflPerentiating TT with respect to 0 yields (3.11) - ^ = (r^ + TrrT) > 0. du

The steady state TT* will be higher if the degree of polarization within the government becomes larger. ^^ The government debt path under the non-cooperative Nash Feedback equilibrium is exactly identical to that in Chap. 2 since for t < T the fiscal authority uses the pure bond financing.

3.3 Inflation, Composition of Deficit Finance, and Social Polarization

43

Next, we consider a more general case that the fiscal authority uses both debt-finance and money-finance at a given ratio ^ of money and bonds (i.e. ^ = :^) at the beginning, and turns to pure money financing at time T. Each minister would determine her public good provision level gi, by solving the maximization problem subject to the government budget constraint and the money demand function of the private sector. Let d = b-\-m. Using the expression for d, we can rewrite the government budget constraint as follows: d = gi+g2 + rd — T — {r-^7r)m

(3.12)

= gi-^g2+rd-T-j. The feedback Nash equilibrium, p^ and pj? is Pi = Ai(7 -h r - rd), and P2 = (1 " ^2){7 + r - rd).

(3.13)

To facilitate the interpretation of the relationship among polarization, fiscal sending and inflation rate, note that d = {b -\- m) = {1 -\- Qb where ^ _ m rpj^^ feedback Nash equilibrium, p^ and pj? is equivalent to g^ = Al(7 4- r — r ( l + f)6) and g^ = (1 — A2)(7 + r — r ( l + ^)6). One can quickly notice that each minister's public good provision g* lessens as the moneybond ratio (f) increases. This can be explained intuitively. Since the private agent's money demand is inversely related to the inflation rate, there is a limit on (real) revenues that can be collected by inflation tax for any given lumpsum tax and money-bond ratio in the deficit finance. As ^ becomes larger, the portion of revenue orginated from inflation tax from which the polarized ministers can extract decreases and their spending declines as well. The steady state inflation after the switch takes place at T can be obtained as follows. Since 6 = 0 at T and m = 7(r-|-7rt)~^, the steady state inflation rate TT* is (recall that the government budget is c? = ^(7 + r — rd) from equations (3.12) and (3.13)) ^.^Jib-_rl_ (3.14) 7 -h r — ro Note that the inflation rate will be higher if the debt at T when the switch takes place is higher. Using ^ = y and the derived g*, the government budget constraint can be written as 6 = ^ ( 7 + T - r ( l + 0^>)-

(3.15)

Since d = (1 + 0^^ d={l-\- ^k We go further by computing the br using equation (3.15) and the NPG condition, U _ (7 + -r) ,^ .^r^ (1-e-^^^). (3.16)

r(l + 0

44


It is clear from equation (3.16) that a higher money-bond ratio in the deficit finance or a lower polarization leads to lower debt. Substituting equation (3.16) into equation (3.14) yields (7 + r ) ( l - e - ^ ' - ^ ) - r r ( l + 0 The steady state inflation rate depends on several variables such as polarization, money-bond ratio, and switch time T. First, we note that there is a monotonic positive relationship between the degree of polarization and inflation rate. To see this, we compute the diflPerentiation of TT* with respect to 0. As in the first case, either the higher the degree of polarization or the higher the resulting debt at time T, the higher the steady state inflation rate after switching to monetizing the debt service. If the government had a higher money-bond ratio at the beginning but turns to money finance later, however, then this economy will have a lower steady state inflation rate despite the polarization within the government. In other words, there is a monotonic relationship between ^ and inflation rate for any given 6, dTT* d^

-(7-hT-rr)(l-e-^^^) < 0, V e > 0, (7 + r)(e + e-^-^)2

(3.19)

where 7 + r — r r > 0. The above result suggests that money-financing can be less inflationary. This result is consistent with earlier studies by Turnovsky (1978) and Liviatan (1983), although they derive the result for an exogenously given government spending.^^ Thus, our explanation is very diflPerent. In our model, it is because as ^ increases, each minister's spending decreases. In a sense, money-financing provides discipline to these polarized ministers. When she tries to finance her expenditure by printing more money, this increases the inflation rate that in turn reduces the private agent's real money demand (i.e. tax base). With a money market equilibrium condition and the resulting limit on the inflation tax revenue, she cannot spend as much as she would with pure bond-financing. Therefore, it follows that the money finance is less inflationary if the government relies more on inflation tax relative to bondfinance from the beginning. In other words, bond-finance of fiscal deficit may not cause infiation initially. Yet as debt accumulates, this also increases the potential inflation that would prevail when the government finally turns to seigniorage. ^^ Calvo (1985) and Liviatan (1988) also study the implications of changing the money-bond ratio in deficit financing for an exogenously given government spending.


45

3.4 Concluding Remarks We have examined the inflationary consequence of fiscal deficit in relation to social polarization, and derived a positive relation between social polarization and inflation: the higher the degree of polarization, the higher the steady state inflation. This prediction is consistent with empirical evidence on inequality and inflation found in the existing studies, given that income inequality has long been mentioned to be an important source of social polarization in the political science literature. Interestingly enough, however, an increase in the money-bond ratio in the deficit financing turns out be less inflationary for any given degree of polarization because the money financing may provide more fiscal discipline to these polarized policymakers.

Social Polarization, Industrialization, and Fiscal Instability

4.1 Introduction This chapter is motivated by an important puzzle t h a t arises from t h e contrasting macroeconomic experience across developing regions in recent decades. In sharp contrast t o East Asia, much of Latin America and subSaharan Africa has often engaged in unsustainable fiscal policies, leading t o huge fiscal deficits, external debt crises, or hyperinflation (see Table 4.1).^ For instance, Korea's largest public deficit in t h e 1980s was 4.3 percent of G D P in 1982 against Mexico's 15.4 percent of G D P in 1982 or Zambia's 28.5 percent of G D P in 1986. Despite t h e substantial progress made on fiscal reform since t h e late 1980s, many developing countries still suffer from chronically recurrent large deficits. Brazil's recent financial crisis is closely related t o its high budget deficit, which is 8.4 percent of G D P as of January in 1999. (See Newsweek of J a n . 25, 1999.) Moreover, past fiscal policies in Latin America and sub-Saharan Africa are characterized by high volatility of fiscal outcomes such as fiscal balance and its components, compared t o East Asia (see Table 4.1).2 Reprinted from Journal of Development Economics, Vol. 72, October, 2003, pp 223-252, Woo J, "Social Polarization, Industrialization, Fiscal Instability: Theory and Evidence", with permission from Elsevier. Even in the wake of the recent Asian crisis, most economists agree that it was not caused by irresponsible macroeconomic policy but by heavy, private shortterm borrowing, which was compounded by weakness of the financial system (see Radelet and Sachs 1998, for example). East Asian countries hit by the financial crisis in 1997 showed somewhat declining fiscal surpluses or emerging deficits in 1997-98, yet the size of deficits is still modest. It ranges from -1.37 percent of GDP in Korea, -0.87 percent of GDP in Thailand, -0.67 percent of GDP in Indonesia, to 2.99 percent of GDP in Malaysia in 1997, which are the most recent figures available from World Development Indicators of World Bank (2000). Closely related to this volatility is the procyclicality of fiscal policy in Latin America, which has been extensively documented by Gavin and Perotti (1997).

48

4 Social Polarization, Industrialization, and Fiscal Instability

Table 4 . 1 . Comparison of Developing Regions: Fiscal Balance, Volatility of Fiscal Outcomes, and Income Distribution East Asia Latin America Sub-Saharan Africa Fiscal Balance (% of G D P ) : 1970 - 90 Central Government Balance" General Government Balance^ Public Sector Balance*^

-2.34 -2.12 -3.60

-3.80 -4.52 -4.89

-4.68 -5.21 -6.73

Volatility of Fiscal Outcomes: 1970 - 90' Central Government Balance General Government Balance Public Sector Balance Government Expenditures Tax Revenue Expenditure Components Wages and Salaries Purchases of Goods and Services Subsidies and Transfers

2.40 2.75 2.53 2.91 1.80

3.47 4.14 3.78 4.15 2.41

3.53 3.62 4.43 4.69 2.62

3,35 5.53 4.09

5.00 7.71 8.81

5.19 8.20 4.10

Income Distribution around 1970 Gini Coefficient for Income^ Gini Coefficient for Land Ownership-^

38.99 45.20

50.83 81.19

49.45 47.19

Sources: "" World Bank (1999). ^ IMF (various issues). ^ Easterly et al. (1994). '^ Averages of country-specific standard deviations are reported. All but expenditure components are measured as a percentage of GDP, and the components of expenditure are me8isured as a percentage of government expenditure. Data on the central government expenditure components are from World Bank (1999). ^ Deininger and Squire (1996). f Taylor and Hudson (1972).

Why have certain countries, notably in Latin America and sub-Saharan Africa, repeatedly adopted unsustainable fiscal policies in recent decades, while East Asian countries have maintained stable fiscal policies? We argue that social polarization and degree of polarization are key to understanding diflferences in fiscal outcomes across countries. The unsustainable fiscal paths can best be viewed as politico-economic equilibria that arise from social conAlthough we focus on volatility rather than procyclicality, we can shed some light on this issue.

4.1 Introduction

49

flicts of interest over government spending, created by interaction between initial income distribution and industrialization. In an economy with a manufacturing sector and a traditional sector, the more unequal the initial income distribution, the larger the sectoral income gap during industrialization and the more likely the polarization of sector preferences for different types of government spending. In a highly polarized society, each policymaker has a greater incentive to insist on higher spending for her preferred sector.^ A large fiscal deficit results, and the fiscal spending path becomes more volatile. Income inequality has long been mentioned as a major source of social polarization. Many economists have argued that class and sectoral divisions due to unequal income distribution provide an important answer to the question of why populist fiscal policies appear more often in Latin American countries than other regions. (See Rodrik 1996; Kauffman and Stallings 1991; Sachs 1989; Berg and Sachs 1988.) Yet there are very few theories that explain why unequal income distribution can lead to large deficits and volatile fiscal outcomes. On the other hand, a growing literature on the political economy of fiscal deficits emphasizes the importance of government fragmentation and political instability in understanding fiscal deficits.'* While this chapter shares the same spirit in emphasizing the political economy aspects of fiscal policies, it is also related to the tragedy of the commons literature in that the endogenous fiscal policy is jointly determined by heterogeneous policymakers who share common government resources.^ This chapter, however, explores the theoretical linkage between initial income distribution and social polarization along the industrialization process. It then highlights the role of polarization in the development of fiscal instability, deriving new directly testable implications. Moreover, we address both fiscal deficits and volatilities in a single framework, whereas the existing models of fiscal deficits do not deal with the volatility of fiscal outcomes. In our model, industrialization is viewed as the adoption of modern manufacturing technology that combines skill and capital stock (Goldin and Katz 1998). Industrialization provides an incentive to accumulate human capital and enjoy a higher wage. But not everyone can invest in human capital because education is costly. The agents who mainly benefit from the industrialization are those who are already rich enough to invest in human capital. Assuming the human capital externality with a threshold property, the tranOne of the important findings of a comprehensive study of public sector deficits by Easterly et al. (1994) is that largefiscaldeficits are largely explained by conscious fiscal policy choices and not by external or by domestic macroeconomic shocks. For example, see Alesina and Drazen (1991) and Alesina and Tabellini (1990). Also, papers somewhat related in the growth literature are Alesina and Rodrik (1994) and Persson and Tabellini (1994), who suggest that there may be a tendency of the majority voting to favor large redistributive spending in a democratic country with an unequal income distribution. See Velasco (1999), Hallerberg and von Hagen (1999), Tornell and Lane (1998), and Weingast et al. (1981) among others.

50


sition dynamics can be dramatically different depending on the initial income distribution.^ If industrialization starts with a large fraction of the population participating, it will experience a narrowing of wage/income inequality as observed in East Asia, contrary to historical experience and contemporary evidence of the Kuznets curve in other regions J By contrast, when only a small fraction of the population is able to participate in industrialization, it leads to a permanent segmentation between the manufacturing and traditional sectors that yields a large sectoral wage gap.^ If a wide wage gap between the two sectors results, the polarization of preference—conflicts of interest over government spending—is sharper. The polarization of preference leads a policymaker representing each sector to spend more for her favorite sector, collectively contributing to bigger overall spending. Given that two policymakers share the government budget, whatever government resources a policymaker does not exploit may or may not be available depending on the other's behavior. When policymakers disagree on the desirable composition of government spending, each of them has a greater incentive to overexploit the common resources, exerting negative externality on the other, which prevents them from achieving the social optimum. Interestingly, a more polarized society also suffers from greater fluctuations in its fiscal outcomes. The higher the polarization, the bigger the fiscal spending and the larger the current fiscal deficit (polarization effect). But this raises the debt level more quickly and reduces government resources, forcing policymakers to cut tomorrow's spending by more. This polarization effect implies that a shock to tax revenue is translated into a more than proportional change in spending. Moreover, the higher is the degree of polarization, the greater is the change in spending in response to a shock to the tax revenue. We test these predictions on fiscal deficits and volatility by running panel regressions, based on a panel data set covering 90 countries over the period of 1970-90. Here we utilize recently assembled quality data on public sector deficits (Easterly et al. 1994) and income inequality (Deininger and Squire 1996). As our theory suggests, income inequality as a proxy of polarization is found to be statistically significant and robust in the regressions of fiscal deficit and volatility. Countries that have suffered from the greatest fiscal instability ^ Human capital externality as a "trickle down" process has been extensively studied in growth literature. See Azariadis and Drazen (1990) and Perotti (1993) among others. ^ Kuznets's inverted-U hypothesis has been questioned by recent empirical studies such as Birdsall et al. (1995) and Fields and Jakubson (1994). The East Asian experience has contributed to this argument (see World Bank 1993). ^ Large wage differentials found in Latin America and sub-Saharan African countries reflect unequal distribution of schooling, which is in turn related to income distribution. Consequently, much of these countries' inequality is associated with large wage differentials, and education is a key channel through which income inequality is perpetuated. See Inter-American Development Bank (1998) for a comprehensive study on income inequality in Latin America.

4.2 The Economy

51

are those with highly polarized economic societies as measured by indicators of income inequality.^ To our best knowledge, this is the first econometric test that reports the significance of income inequality in explaining different fiscal outcomes across countries.^^ Sect. 4.2 describes an economy under a post-industrialization regime and discusses the dynamics of income distribution in the process of industrialization. Sect. 4.3 derives an endogenous fiscal policy and establishes the main results, followed by country experiences that support our conclusions. Sect. 4.4 presents econometric evidence, and Sect. 4.5 concludes. Appendix derives the social planner's optimal fiscal policy. A list of our sample countries and a table of pairwise correlation of consolidated public sector balance with selected variables are provided in the data appendices.

4.2 T h e Economy We consider a two-sector small open economy with a government {fiscal authority) and a continuum of atomistic individuals endowed with perfect foresight. The economy is populated by overlapping generations of individuals who live two periods: young and old. The agents are equally able, but they are endowed with different amounts of initial wealth. For simplicity, we assume that the economy's two sectors produce essentially the same product (i.e., perfect substitutes). The price is normalized to one, p = 1. The population is constant and its size is normalized to one, L = 1. 4.2.1 The Pre-Industrialization Regime The pre-industrialization regime is characterized by linear production technology that uses only labor in both the manufacturing and traditional sectors. Production is described by Xf = X f = J L *

(4.1)

where Xf and X{^ are outputs in the manufacturing and traditional sectors at time t, respectively. The marginal productivity of labor is J > 0 in both sectors. With this technology, the unskilled labor is assumed to be as productive as the skilled labor. The skilled labor is more productive than the unskilled only when combined with capital stock. Thus, there is no incentive for an agent to get an education and become skilled because there is no wage differential {w^ = w^ = 6). ^ It is interesting to note that Malaysia, which had run the largest fiscal deficit and exhibited the greatest volatility of fiscal outcomes among East Asian countries in the period of 1970-90, also had the most unequal income distribution in this group. ^° A notable partial exception is Berg and Sachs (1988), who found that countries with high income inequality had a significantly greater likelihood of having rescheduled the external debts than countries with low income inequality.

52


4.2.2 The Post-Industrialization Regime The post-industrialization economy consists of the modern manufacturing sector using skilled labor and capital and the traditional sector using unskilled labor only. Respectively, the production technologies are: X f = F ( L f , i ^ , ) , and

(4.2)

X f =:((5 + 0(Lf))Lf,

(4.3)

where F(-) is a constant return-to-scale (CRS) function that exhibits diminishing marginal productivity; L^ and L^ are skilled and unskilled labor, respectively (of course, L^ + L^ — L); and K is capital stock and depreciates completely in one period. To capture the benefit of industrialization for the rest of the economy, we assume that there is a positive human capital externality, 0(L'^), which spills over to the traditional sector, enhancing its labor productivity. The externality term (t>{L^) is a monotonically increasing function of L*^ such that (pLS > 0 and 0(0) = 0. It is assumed that FLS{L^, Kt) > S + (t){L^) for 0 < L^ < L so that the skilled worker is always more productive than the unskilled. It is also assumed that investment in human capital is made one period in advance and that both labor markets are in perfect competition. The government has free access to international capital markets and can issue bonds at the constant world interest rate, r € (0,1). It is assumed that there is no default risk of government debt. While individuals can lend any amount at this rate, they cannot borrow against future labor earnings due to a severe moral hazard problem. Since the number of skilled workers is known one period in advance, the capital stock installed each period is set to ensure that FK{Lf,Kt) = r + l. (4.4) Thus, the skilled worker earns the wage w^ = FL{L^,Kt) in the absence of government subsidy to the manufacturing sector, and w^ is a constant for given r and technology because F(-) is a CRS function. Note that the unskilled worker is paid w^ = MP^j = 5 -\- (j){L^). Each individual can work as unskilled labor in both periods or invest in human capital when young and work as skilled when old. Each supplies one unit of labor inelastically in each working period. Following Galor and Zeira (1993), we assume the indivisibility of human capital. To become a skilled worker, one has to invest /i > 0. Since no individual can have access to the capital market, a poor individual with e* < /i is unable to invest in human capital. The distribution function H{e) of the endowment is defined over the support e* G [e,e] and L = J^ dH{e). Let fit = Jh^î^t) ^^ ^^^ fraction of the population that is rich and 1 — fit = f^ dH{et) the poor. The young do not consume. Each individual cares about her offspring and leaves a nonnegative bequest if she can afford to do so. The utility function of an individual i is given by

4.2 The Economy

53 (4.5)

U' = u{cudt),

where u{') is a concave function, satisfying the Inada condition; Ct and dt are the second period consumption and the bequest. There are two different types of government spending that affect the individual's budget constraint: the subsidy to the manufacturing sector, gt, and the government spending for the traditional sector, ft. As illustrated later, the government spending programs that may disproportionately benefit certain groups can take many different forms. For simplicity, we call any government spending that benefits the skilled workers the subsidy to the manufacturing sector, g}^ By using g in this way, we try to capture the fact that in many developing countries fiscal transfers to rent-seeking groups (such as industrial powers or public enterprises) take the form of government inputs, such as electricity, gas, soft credit, and raw materials that are provided at subsidized prices. Similarly, the spending / represents any kind of government program that favors workers in the traditional sector such as subsidized food, gas, fertilizer, rural electrification, public-works programs or provision of public employment (for example, see Gillis et al. 1996). These gt and ft are endogenously determined by two policymakers representing different sectors, and individuals take them as given. Since the policymakers share the government budget but seek to maximize their objective functions, they behave strategically in determining the spending level of each item. In a noncooperative dynamic game, policymaker S chooses {gt,gt+i} to maximize her own objective function subject to the government budget constraint for any possible {ft, ft+i} policymaker N chooses. Similarly, we define policymaker AT's constrained optimization problem: 1

Max

J' = Y,ßno^H9t+j)

+ (1 - cx')v{ft+j)], i = S,N.

(4.6)

j=o

subject to h+j - bt+j-i = rbt+j-i + gt+j + ft+j -T,

j = 0,1,

(4.7)

where 0 < ^5 < 1 is the subjective discount rate of the policymakers; b is the government debt; T is the lump-sum tax revenue; and v(-) is a concave function satisfying the Inada condition. We assume that 0 < a^ < | < o^*^ < 1 so that policymaker 5, who represents the manufacturing sector, prefers g to / and the other policymaker N, who represents the traditional sector, prefers / to g. We then define the degree of polarization, 9 = a^ — a^ e [0,1], where ^ = 0 means complete agreement on the composition of spending, and 9 = 1 complete disagreement. ^^ The profit of manufacturing sector is given by7r = X — (1 + r)K — FL{L^ , K) • L^ -\- g > 0 with equality if ^ = 0. The manufacturing sector profit (essentially subsidized by the government) is assumed to be equally distributed to the skilled workers in the sector so that they receive w^ = FL{L^,K) + g/L^.

54


4.2.3 Human Capital Accumulation, Threshold Externality, and Income Distribution Now consider an optimization problem of an agent with initial endowment e. Only an agent with e > h can invest in human capital and become a skilled worker because no individual has access to the capital market. The optimization problem of an agent with e is, for given g and / : max U{e) = (1 - 7) log Ct + 7 log dt

(4.8)

{ct, dt}

s.t.

ct-hdt 7(1 + r). ^'=,_^l

(4.12)

+ r)^^'-T-il+r)h],

The wealth distribution in the steady state becomes more equitable as the initial fraction of the skilled (ßi) is higher.-''* The income inequality (measured as a ratio of y^ and y^) also declines as î increases: / y^

[{l + r){e^-h) + w^-T] [(1 + r)e'^(Mi) + (2 + r)(h, if a > h, = 7{(1 + r){et + (5 + (t>{ßt-i)) + ((^ + (t>M) -\-f-T}

< h, \i Ct < h.

Note that if ei < /i, then 62 < h and hence /i2 = Mi- That is, given that ßi < n and Lf = Oj ei < h implies that 62 = 7{(1 + r){ei +Ö) + (S + (t>{yii)) -¥ f-T} < 7{(1 + r)h + (2 + r){5 + (t>{ß)) + f-T} = h. Similarly, if 62 < h, then es = 7{(1 + r){e2 + (5 + (t>(fii)) -\-{Ö + (l>{f^2)) + / - T} < 7{(1 + r)h + (2 + r){S + (/£)) + / - T} = /i, and hence /xs = M2 = A*i < /£• Repeating the same procedure establishes that if /xi < /x, then /xi = /it+i < M, Vt > 1. ^'^ Note that e^ = e^(fii) and e^'ißi) = i z : ^ ^ [ ( 2 + r)(t>'{fii)] > 0. Also e^ declines in fii for given r, T, g, and technology because w^ = FL{L^ , K)-\-g/L^ = FL{fii,K)+g/fii,

56


larger than h. Since Lf = 0 and (t>{Lf) = 0, there can exist an unskilled worker endowed with ei < h who leaves a bequest bigger than h in period 2 only if (/>(/ii) > (2 + r)(p{jjL). Otherwise, there will be none. In either case, however, /X2 cannot be smaller than ßi. Since 0(^*2) > 0(/î) > ^(/f)? unskilled workers who inherit 62 € [62, h) now leave bequests greater than /i, and their descendants in all future generations become skilled, where 62 is a level of endowment such that 63 = 7{(1 -f-r)(e2 -\-S-\-(/xi)) -\-o-\- 0(//2) -\-f — T} = h. Thus, /is = j^dH{e^) > /i2 = J^dH{e2). Similarly, unskilled workers who inherit 63 € [es,h) leave bequests larger than h in period 4, where 63 is defined to satisfy 64 = 7{(1 + 0(^3 + ^ + (M2)) 4- (^ + (/)(/i3) + / - T} = /i, and so on.^^ Therefore, if //i > //, then /i2 > /î and //t+i > A*t, for Vt > 2. Once industrialization begins, the inequalities of income and wealth continue to decrease, converging to unique steady-state levels of income and wealth (see Fig. 4.1).

et+i

e^(/.f)

e\îf)

e^(/if) = e^(/.f)

Fig. 4 . 1 . Countries A and B fall into Case 1 (i.e., ßi < ßt < ß)^ whereas country C belongs to Case 2 (i.e., /£ < /if). For simplicity of the graph, we assume that these countries are identical except for /ii and that f = g — 0.

^^ There is a finite maximum level on / that can be sustained as an endogenous equilibrium as shown in Sect 4.3. Thus, we can rule out the trivial case that everyone becomes educated after the transfer regardless of her initial endowment by choosing parameters h, T, r, 5, and (j){') so that the threshold is binding for a given income distribution.

4.2 The Economy

57

4.2.4 Income Inequality and Polarization Industrialization can engender not only earning differentials but also potential conflicts of interest between income groups over the composition of government spending. Suppose the economy ends up with a permanent segmentation and a wide sectoral income gap (the case of /xi < jj). The resulting sectoral income gap may induce workers in the traditional sector to demand higher redistributive spending /.^^ On the other hand, the permanent segmentation is likely to encourage the rich to strongly influence the fiscal policy in their interest, leading to a greater amount of transfer and subsidy g. Thus, the polarization of preferences for the composition of government spending would be sharper in this economy. To illustrate this point, we use the indirect utility function, V^(y^ : •) = lny% i = 5, N}'^ The welfare gap between sectors is then: V ^ ( / : e>h)-

V'iy''

: e < /i) = In

( ^ )

V[(l + r-)e^(w) + (2 + r)(5 + ,^(Mi)) + / - T ] ; -

^'

'

Note that either a higher / or a higher initial /xi reduces this welfare gap at the steady state for any given level of g and T. For any given level of g, the incentives for unskilled workers to demand higher welfare spending / (as measured by their marginal utility of / ) become much stronger when the initial income distribution is more unequal: dV^/df = l/y^ is greater if //I is lower (i.e., d'^V^/dfdfii < 0). On the other hand, the skilled worker's marginal utility of g is dV^/dg = l/{y^ * Mi)- Again, the more unequal the initial income inequality and hence the lower the //i, the higher the marginal utility of the manufacturing subsidy (i.e., d'^V^/dgdfii < 0). On the contrary, the economy with fii > /i will converge to a unique distribution of income and wealth, causing the welfare gap to decline towards zero. Unskilled workers would be less likely to demand higher redistributive spending as their marginal utility with respect to / becomes smaller due to declining income inequality. By assumption, policymaker S prefers p to / , whereas policymaker N prefers / to g. Since the marginal utility of g is negatively related to //i, policymaker S, representing the manufacturing sector, will put a higher value of a^ on the utility she derives from the transfer g when //i is smaller. Similarly, the policymaker N will put a greater value of (1 — a^) on her utility derived ^^ Related studies on redistributive politics are Meltzer and Richards (1981) and Benabou and Ok (2001). In these models, income inequality or lack of upward social mobility favorably influences the preference of the majority for redistributive policies, leading to a larger amount of redistribution. ^^ Since we assume a Cobb-Douglas utility function and the price is normalized to one, the indirect utility function is in a simple logarithmic form of ln(y).

58


from / when //i is smaller. The degree of polarization 0 = a^ — a^ G [0,1] will then be most likely higher when //i is lower since both a^ and (1 — a^) are inversely associated with /ii. Therefore, people in a more unequal society would have much more divergent preferences for the composition of fiscal spending because there is a sharp conflict of interests among them. As Birdsall et al. (1995) note, the likelihood that fiscal policies are polarized between the extreme of serving the narrowly defined myopic of the rich and an equally myopic populist extreme will be higher where higher levels of income inequality widen the distance between the interests of the rich and the poor. Now we look at the role of this polarization in the evolution of fiscal spending and deficit.

4.3 The Fiscal Policy The fiscal policy consists of two different types of government spending {gt, ft}u=i and taxation T. While the two different types of policymakers play a non-cooperative dynamic game, each policymaker of a generation also cares about the fiscal spending levels of the next generation, which cares about the government spending of the following generation and so on. Altruism is limited in our model in the sense that each generation derives utility only from its own spending and that of its immediate successor. ^^ This game-theoretic setup introduces two critical driving forces for the fiscal outcome: polarization of sector preference and limited altruism towards the next generation. For simplicity, we assume a lump-sum tax, T. However, the rise of an endogenous fiscal deficit does not rely on the level of tax revenue, as becomes clear later. 4.3.1 Polarization and Endogenous Fiscal Deficit Assuming a simple Cobb-Douglas utility function for v{'), we can write the optimization problem for the policymaker i of generation t as follows: 1

Max J' = Y1 ^'[^' l^SPt+i + (1 - a') log /,+,], i = S,N

(4.15)

3=0

subject to h+j - h+j-i

= rbt+j-i + gt+j + ft+j -T,

j = 0,1.

For a concrete result, we focus on the linear stationary Markov perfect equilibrium that depends only on the current state variable, bt.^^ Let us ^^ Limited altruism models have been explored in a variety of contexts. See Bernheim and Ray (1989), Leininger (1986) and the references therein. ^^ In a Markov strategy set, past decisions influence current play only through their effect on a state variable that summarizes the direct effect of the past on the

4.3 The Fiscal Policy

59

write the government net revenue as Rt = T — rbt-i and consider the linear strategies gt = X^^t and ft = X^^t, where the set of strategies is X* € [0,00), i = S,N. By focusing on stationary equilibria, we describe the evolution of fiscal policy decisions when all generations of each type of policymaker select the same Markov strategy x*Policymaker S solves max{(l+/?)[a^logx^ + (l-c^^)logx'^ + logi^t] + / ? [ l o g ( l - r ( x ^ + x ' ^ - l ) ) ] } . (4.16) The first-order condition for her optimization problem is ßrx^ = (1 + ß)a^[l - rix^ + x ^ - !)]•

(4.17)

Similarly, the first-order condition for policymaker N is ßrx"" = (1 + /3)(1 - a ^ ) [ l - r(x^ + x'^ - 1)1-

(4.18)

The Markov perfect equilibrium is then: _

(l+/?)(l+r)a^

_ (l + / ? ) ( l + r ) ( l - a ^ )

Substituting this solution and Rt = Tt — rbt-i into the government budget constraint (4.7) yields bt-bt-i = k{-){T-rht-i), where k{') = ri"'^^ît^^

(4.20)

> 0. We establish the following proposition.

Proposition 4.1. (i) The higher the degree of polarization, the larger the fiscal deficit, (ii) The less patient the policymakers, the bigger the deficit. (Hi) When there is polarization (i.e. 6 > 0), the spending level and the size of deficit are always greater than the social optimum. Proof, (i) dk{-)/dO > 0. (ii) dk{')/dß

< 0. (iii) See Appendix A.

Given that two policymakers share the government budget, whatever resources one does not exploit may or may not be left, depending on the other's behavior. When they disagree on the ideal composition of government spending, each has an incentive to overexploit the common resource today. That is, the polarization of preference leads each policymaker to insist on a higher current environment. A Markov perfect equilibrium (MPE) is a profile of Markov strategies that yields a Nash equilibrium in every proper subgame. Thus, the MPE is subgame perfect and time consistent. See Leininger (1986) and Bernheim and Ray (1989) for the existence of Markov perfect equilibria in a model of growth with altruism between generations (so-called bequest games).

60


spending for her favorite sector and to exert negative externality on the other, contributing to bigger overall spending and a larger deficit than the social optimum. In Sect. 4.2.4, we saw that the more unequal the income distribution, the more likely the polarization of sector preference over different types of fiscal spending. Thus, the society with greater income inequality is more likely to run larger deficits. When policymakers care less about the next generation and discount the future more heavily, they also have a deficit bias.^^ 4.3.2 Polarization and Volatility of Fiscal Outcomes In this subsection, we establish that a country with a higher degree of polarization will experience greater fluctuations in its fiscal outcomes. We begin with a simple case of lump-sum taxation. For any given tax revenue, a rise in polarization 6 has two effects. The first, which we call polarization effect, is that the level of government spending (pt = 9t -^ ft) rises (see Proposition 1). The second effect is that government spending falls more quickly because the increase in debt lowers Rt, the pool of the government resources.^^ Thus, greater polarization means larger fiscal deficits today and more drastic spending cuts tomorrow. The country with higher 6 will exhibit a bigger swing of its fiscal spending path for a given level of tax revenue.^^ 20

If we view a high subjective discount rate as reflecting political uncertainty facing the policymakers, this result reaches the similar point made by Alesina and Tabellini (1990). They showed that the incumbent government, faced with the uncertainty over re-election, may have a debt bias because the costs of additional debt are borne only if re-elected. A country with high 6 will spend more until some finite time; thereafter, it will spend less than a country with low 0. Let us say 6i > 62- The ratio of gu to p2t is greater than one until for some finite time t, and thereafter becomes less than one, where the ratio is given by 9u ^ (l + î)[/3 + ( l + / 3 ) ( l + ^2)](l-fe(î)r)*-^ 92t (1 + 02)[ß + (1 + ß){l + Öi)](l k{02)ry-^'

22

In practice, it does not have to be government spending that must adjust. The government can use taxes or seigniorage to pay the debt. No matter which instruments are used, however, today's larger deficits still mean more drastic adjustments in fiscal policies in the future, making fiscal outcomes volatile over time. In most hyper inflationary episodes, large budget deficits were often an initial cause. In Chap. 3, we studied the inflationary consequence of budget deficits in relation to the polarization effect. On the other hand, it is a commonly shared view that developing countries in general have much narrower tax bases relative to those of industrial countries. Also, the collection of tax revenue in developing countries is often hindered by limited administrative capacity and political constraints (Agenor and Montiel 1999). This implies that developing countries might find it harder to raise taxes in order to cut budget deficits than cutting expenditures. Consistent with this line of reasoning, total expenditure and its


61

More importantly, a more polarized society would suffer from a greater fluctuation in its government spending in response to a shock to tax revenue of the same magnitude. Note that the fiscal spending is given by 9t = p^^q^^5ji^p^(^* - rbt-i) from equation (4.19). Hence, for any given

where /(•) = m]|!(i-^m(i-}-^)]r ^ ^' ^^^ amount of fiscal spending change in response to a shock to tax revenue is increasing in the degree of polarization (i.e., dl{')/d6 > 0). An increase in tax revenue is translated into a more than proportional increase in spending through the polarization effect. Moreover, the higher the polarization, the bigger the increase in p. But this leads to a sharper reduction in subsequent spending because the increase in tax revenue is dissipated more quickly. Similarly, the absolute size of spending change resulting from a shock to tax revenue is bigger when the policymakers are less patient (i.e., dl{')/dß < 0). Proposition 4.2. (i) The higher the polarization, the more volatile the fiscal spending and the fiscal balance, (ii) The less patient the policymakers, the more volatile the fiscal spending and the fiscal balance. (Hi) When there is polarization, the fluctuations in fiscal outcomes are always greater than the social optimum. Proof, (i) Prom equation (4.19), it is straightforward to see that for any given bt-u \{d{bt - bt-i))/dT\ = k{-) and dk{-)/de > 0. (ii) dk{-)/dß < 0. (iii) For any given bt-u \^/depolarization = KO, and |c^/^T|^^^.^;^^„^^^^ = ((1 H- ß){l -h r))/{[ß + (1 + ß)]r) (see Appendix A). One can easily show that \d9ldT\^^,^^,^^,,^^ > \d9/dTl^^,^,^,^^^^^ when ^ / 0. If Ö = 0, \dg/dT\^î^^^^^^^^^ = \dg/dTl^^.îpi^^^^^. Similarly, for any given 6t-i, \{d{bt - bt-i))/dT\ôiarization > Mbt " &t-i))/rfT|,,,,,,^^^^^,, with equality when ^ = 0. 4.3.3 Country Experiences It is a well-established fact that education is the most important determinant of wage differentials in developing countries (for instance, see Gillis et al. 1996). Moreover, the wage differentials by educational level are much greater in developing countries than in developed countries (Psacharopoulos 1985). On the other hand, school enrollment is found to be strongly and negatively correlated with income inequality (Birdsall et al. 1995; Clarke 1995). Indeed, a large wage gap has been reported in Latin America and sub-Saharan Africa components as well as fiscal balance have been more volatile than tax revenues across developing regions (see Table 4.1).

62


where initial income inequalities were greater. In Colombia and Jamaica, workers earn nonagricultural wages that exceed agricultural wages by 150% (Fields 1994). According to Berthelemy and Bourguignon (1996), in Cote d'lvoire the average wage of workers in the manufacturing sector was 364% higher than the average wage in the informal sector in 1984-85. In sharp contrast, this sectoral wage gap in East Asia is only about 20% (Fields 1994). The favorable initial conditions in East Asia helped to set up a virtuous circle: initial low inequality in income led to educational expansion, which reinforced narrowing wage gap and low inequality. Table 4.2. Social Welfare Spending and Subsidies and Transfers as a Percentage of GDP in Selected Countries and Years Social security and welfare*^ Subsidies and transfers 1975 1980 1985 1990 1995 1975 1980 1985 1990 1995 E a s t Asia Indonesia Korea Malaysia Singapore Thailand Average

N/A 0.83 0.71 0.33 0.65 0.63

N/A 1.09 1.14 0.27 0.51 0.75

N/A 0.92 1.04 0.43 0.62 0.75

N/A 1.46 1.08 0.44 0.51 0.87

0.91 1.54 1.42 0.79 0.55 1.04

3.42 5.04 7.29 1.37 2.48 3.92

5.33 5.88 5.43 1.11 2.69 4.09

5.11 6.23 3.78 2.65 1.60 3.87

3.78 7.44 4.78 2.44 1.22 3.93

Latin America Argentina Brazil Chile Colombia Mexico Panama Uruguay Venezuela Average

N/A N/A 8.20 N/A 3.39 3.04 10.7 1.33 5.33

5.21 6.46 9.04 N/A 3.15 3.17 10.6 1.44 5.58

5.76 5.94 11.8 2.73 2.23 3.78 12.0 1.16 5.68

4.84 8.81 7.24 0.76 2.34 5.22 13.0 N/A 6.03

7.79 N/A 6.71 N/A 3.19 5.73 18.5 N/A 8.38

N/A N/A 11.2 N/A 3.82 3.62 12.3 4.76 7.74

7.81 12.9 13.0 5.04 4.99 4.30 9.38 4.08 8.20

10.4 10.7 15.6 6.74 5.02 4.65 10.5 6.35 9.09

5.97 8.73 13.6 N/A 10.5 10.3 4.88 N/A 2.98 6.89 6.20 6.65 12.9 18.7 7.70 6.02 8.15 10.25

Sub-Saharan Africa N/A 0.97 7.31 Cote d'lvoire Zambia 0.19 0.70 0.54 N/A 2.02 1.16 Zimbabwe Average 0.19 1.23 3.00

0.23 0.30 N/A 0.27

N/A 0.70 N/A 0.70

N/A 7.03 N/A 7.03

4.26 6.35 9.37 N/A 9.06 10.57 7.56 8.46

7.34 N/A 4.73 4.37 4.80 N/A 5.62 4.37

Egypt

4.68 4.94 5.22 3.58 4.54 25.0 16.2

Source: IMF (various issues). ^ Does not include spending on Education and Health.

13.9

7.34

2.34 8.65 4.64 1.83 1.03 3.70

8.72


63

Large income gaps across sectors and skills in Latin American and subSaharan African countries may have contributed to political pressures for the government to raise redistributive spending in various forms such as social welfare spending, in-kind transfers, public employment, and public-works programs.^^ The segmentation between the rich and the poor due to income gap may also have helped the economic elite to strongly influence fiscal policy in their interest.^^ In contrast to East Asia, fiscal expenditures in Latin America and sub-Saharan Africa in the past decades are characterized by (i) higher share of social welfare spending in GDP, (ii) higher share of subsidy and transfer in GDP, and (iii) rapid expansion of the public sector such as public sector employment. Social welfare spending (as a percentage of GDP) in Latin American countries was more than six times greater than that of East Asia for the period of 1975-95 (see Table 4.2). During the same period, the subsidy and transfer (as a percentage of GDP) was often much larger in Latin American countries than in East Asia. On the other hand, the state-owned enterprise (SOE) deficit has been important for fiscal instability. According to World Bank (1995), it accounted for 35% of fiscal deficits in 34 developing countries for the period of 1978-92. It is widely accepted that a substantial portion of government subsidy and transfer has been devoted to covering large deficits of SOEs. For example, diverting SOE operating subsidies to basic education would increase central government education expenditures by 50% in Mexico, 74% in Tanzania, and 160% in Tunisia. This stands in stark contrast to 0.1% in Thailand (World Bank 1995). This warrants some special attention because the macroeconomic role of SOEs in East Asia was not smaller than in other developing regions where SOEs' expansion often led to greater fiscal deficits and inefficiency in the form of patronage, subsidies, and overstaffing. The provision of public employment to satisfy the excess demand for highwage employment is also an all-too-common example of populist policies. The public sector employment grew more rapidly than employment in the private sector in Latin America and sub-Saharan Africa. The median share of the public sector in the increase in total employment was 71-87% in Latin America and sub-Saharan Africa, compared to 10% in Taiwan and 33% in Thailand (Gelb et al. 1991). The macroeconomic costs of public employment were often very large, and cutting the wage bill in the public sector was a key ingredient of the fiscal reform package.^^ ^^ In Latin America and sub-Saharan Africa, however, income-transfer programs are still rudimentary when compared to industrialized countries. Consequently, their governments tend to rely on public employment, in-kind transfers, and publicworks programs as well. ^^ The allocation of limited fiscal resources for tertiary education, so common in Latin America, is another example (Birdsall et al. 1995). ^^ However, it has been difficult to restructure the civil service due to strong opposition and lobbies from the insiders. Mackenzie et al. (1997) found that Chile and Ghana were the only two countries to achieve a substantial restructuring of the

64


4.4 Econometric Evidence The basic implication of the model is that countries with greater polarization are more likely to run larger fiscal deficits and exhibit more volatile fiscal outcomes. We test these implications of social polarization for fiscal outcomes by running panel regressions of fiscal balance and its volatility. Many of the empirical studies on budget deficits use data on the central government's fiscal balance. However, using only the central government deficit data can be problematic because a large part of the fiscal deficits in developing countries is attributable to the rest of the public sector (see Easterly et al. 1994). Here we use an expanded panel data set for 90 developed and developing countries, which combines consolidated public balance data of 57 countries from Easterly et al. (1994) with general government balance data of an additional 33 countries from IMF's Government Finance Statistics (GFS). All other data are from World Bank's World Development Indicators CD-ROM unless specified. Our panel is comprised of decade averages of variables for 1970-79 and 1980-90. See Appendix B for a complete list of these 90 countries. Our basic regression equation for fiscal balance is as follows: SURPit = aiLRGDPit

+ a2GRGDPit + aÎNFLATu

+ ßXu

(4.22)

where i and t denote the country and decade (1970-79 and 1980-90), and e is an error term. SURP is the average of fiscal surplus (as a percentage of GDP). We follow the existing studies (for example, Edwards and Tabellini 1991; Roubini 1991) to include the initial income level, the real GDP growth rate, and the inflation rate in our basic regression. According to the tax-smoothing model of fiscal deficits, budget deficits will emerge when output is temporarily low relative to its permanent level. We include the decade average rate of real GDP growth (GRGDP) as a proxy for the cyclical effects on budget deficits (such as booms and recessions), which might proxy for the different degree to which the countries in question faced economic recessions during the sample period. The log of real per-capita GDP at the start of each decade, LRGDP, is introduced to control for potential effects of economic backwardness on the public deficits. Poor countries tend to have relatively inefficient tax and spending systems and may therefore be more prone to budget deficits. We also include the decade average of the rate of inflation of the consumer price index (INFLAT). Inflation can affect fiscal deficits through different channels. Rapidly rising inflation can raise flscal deficits through higher nominal interest payments. High inflation can also lead to lower real tax rev-

civil service through restraining wage and cutting employment in the period of 1978-93 among the eight countries in their study.

4.4 Econometric Evidence

65

enue.^^ We try to capture the effects of inflation on fiscal deficits by including INFLAT. Besides decade dummies, Dt, we include regional dummies, RDi, for East Asia, Latin America and the Caribbean, and sub-Saharan Africa. Regional dummies are intended to control for structural characteristics related to geographical location. We always include the aforementioned economic variables throughout our econometric exercise. Finally, Xu represents other variables, which we specify later. Table 4.3 presents the direct evidence supporting the model's prediction that the fiscal deficit is a positive function of degree of polarization multiplied by net tax revenue as indicated by equation (4.20). As suggested in this chapter, we proxy polarization using the Gini coeflScient, AGINI (see below for detailed explanation). The net tax revenue (a percentage of GDP), NTR, is central government tax revenue minus interest payment. The coefficients of AGINI*NTR have the correct sign (-) and are all significant at the 1-5% significance level. The coefficients of LRGDP, GRGDP and INFLAT all have the expected signs, but only that of LRGDP is consistently significant. In columns (4) and (5), we also include the multiproduct of growth rate of terms of trade and trade share in GDP (EXT) to control for the external shocks to the economy. External shocks can be an important source of fiscal instability, especially in developing countries. Changes in export and import prices can affect the public sector balance either through the profits of exporting public enterprises or through import tariffs and taxes on exports. The growth of terms of trade is expected to be associated with smaller budget deficits and to have a greater impact in economies that are more open to trade. Although the coefficients of EXT have the expected sign (-h), none of them are significant. In column (5), we include frequencies of cabinet changes (CABCHG) and military coups (COUPS) as proxies for political instability. Recent studies on fiscal politics emphasize political instability as an important cause of fiscal deficits (for example, see Alesina and Tabellini 1990). They suggest that if faced with uncertainty about the government's survival, the incumbent government may have a bias towards overspending and fiscal deficits. Edwards and Tabellini (1991) and Roubini (1991) found evidence in support of this political instability channel.^'^ Both the coefficients of CABCHG and COUPS are of the expected sign (-), but they are insignificant. We will have more to say about this later. Table 4.4 reports our main finding that countries with highly polarized societies tend to run larger fiscal deficits. As an indicator of polarization, we use GINI and AGINI. (i) GINI: high-quality data on income inequality measured ^^ If, however, income taxes are not indexed to inflation, the above effects of inflation on deficits can be at least partially offset by the positive effect of bracket creep on income tax revenue. ^^ Based on consolidated public sector balance data of 57 countries, Chap. 5 provides a comprehensive empirical evaluation of different theories of fiscal deficits.

66


Table 4.3. Pooled Regressions Based on the Model. Dependent variable: Consolidated Public Sector Surplus (% of GDP) (1) 2.336** East Asia (2.171) 0.314 Latin America (0.407) Sub-Saharan Africa 2.167*** (1.852) 2.392* LRGDP (5.324) -0.003** AGINI*NTR (-2.406) GRGDP

(2)

(3)

(4)

(5)

1.328 (1.081) 0.521 (0.679) 2.614** (2.278) 2.690* (5.029) -0.004* (-2.652) 0.330 (1.549)

1.576 (1.249) 0.954 (1.173) 2.773** (2.322) 2.674* (5.008) -0.004* (-2.817) 0.227 (0.979) -0.003* (-2.766)

1.541 (1.218) 1.121 (1.291) 2.791** (2.299) 2.658* (4.852) -0.004* (-2.642) 0.231 (0.985) -0.004* (-2.756) 0.085 (0.428)

1.522 (1.144) 1.155 (1.395) 3.233* (2.804) 2.938* (5.759) -0.004* (-2.624) 0.354 (1.487) -0.002 (-1.309) 0.114 (0.593) -0.773 (-1.098) -1.144 (-0.833)

115 0.63

113 0.63

109 0.62

107 0.64

INFLAT EXT CABCHG COUPS No. of Observarions Adj. R^

115 0.62

The decade dummies were included in the regressions (not shown to save space). White heteroskedasticity-consistent t-statistics are in parentheses. Levels of significance are indicated by asterisks: * 1%, ** 5%, *** 10%.

at the start of each decade from Deininger and Squire (1996); and (ii) ACINI: the decade average of all available data of Gini coefficients from Deininger and Squire (1996).^^ We expect the high degree of social polarization as proxied by 28

To avoid reverse causality from public deficits to income inequality, GINI was limited to countries for which we have high-quality data measured close either to the end of each previous decade, the 1960s and 1970s, or to the start of each decade, the 1970s and 1980s. However, income inequality measured by Gini coefficients is very persistent over time as evidenced by high correlation between GINI and AGINI. The pairwise correlation is 0.92. When GINI is used, the results are often stronger than when AGINI is used. Moreover, if one is concerned with measurement error in a specific year, using a ten-year average of Gini coefficients like

c»

o

(-1

OH

t+-t

Q O

CO

E

'>< O o CL.

CQ

Co

.2 *-+-' SH

3. ^

Co

H:»

;3 (1.

,^

p ^

o

0) CO

-1-3

CO

1 'o o o o

3 *o CO

CO

'? Ö

w

.2

o

j3

0) bO O

T3

Ö 0)

73

^ 3 3 .^ o *5 CL, g ^ ^ Ti5

-2

a o H Q

3Co

CO CO CO CO

(X5

00 to

^

o o o

o o

o o Ö Ö o^ o^ Ö1 o

Ö1 Ö1 Ö Ö Ö Ö O^ CO an r^ y ^ r^ 10 P c<j

Ö d Ö d 9 r^

" ^ 10 -Tt

t- 22 S t^ o b-

oocooiTiî-iootU';^ l O C O C O O O C O ^r HJ S CO

1-1

00

C5 10 C^ Ö

CO 00 o (M O O CD CO

^ '"J rH

rH iO ;ZJ $2

l^ 10 O Ö

CO 00

CO CO ^.

^.

. ö r-; P rH

CO

0

^ Z^^ C.^ S'ö d (M

T-H

^

CO CO CO ^ O ^ CO CO CO 9 r-i C CM

1

CO 1—1

*

CM CM rH

Ö

CJi 05 CM CM

co" 1—1

CM

CO

1

CM

^ CM" lO

CM CM

rH

r-1

'^

^

CO CO LO 0 0

O ffi

o m < o

t- o

CO b LO ^ CO CO

CO

Ü

O

13

OH

LO O O LO CM CO

Ö Ö

CO CO

73 ^

T3

^°

Is

^

CO flJ

Ji

I^ O 73

rH

Ö

LO CO

r-i

LO CO

Ö

^ ^
3

^

CO

CO CO CNI

CD T-H

^ ^

'^

w

Ü

q

Ü

r-i

Co

L. •

0

^ CO CO

Ö ^ ^
0 (or if TRANS> 0), and 0 otherwise, is significant at the 5% level in the basic regression. Higher values of INFLEX and TRANS represent greater inflexibility of budget execution and greater transparency, respectively. Both range from 0 to 4.

98

'S o

p0) T3 r^

cu

'o o ra

p

'•4^

T}* b -

*

Tf

CO (M

1—1

;0

^8 29

>^

»n

oo'

O JIJ 00 ^ ^ d

co" o co" #

CO

1-t

"--^ 9

<M 0 5

^-^ d Tt*

>^ d

csi 9 --^ o Ä c5

8 ^- S

CO tie

O

cî ^ --^ o CO iO iC O ; 0 O CO d d c4

c6 isS e o

S^3^9S§" cv5i«de°:iL§:i

ci

°Oc^^c4^qOc5 O PL,

1 ^

1

b ô '6 bO

n3 0

?0 lO P2

a Q

T3 Ö . ^

^ '^

5 Economic, Political, and Institutional Determinants of Public Deficits

O ^

It t^

IC U5 >0 * 3 »H ^ CO d

^ eo ^ T c^

^'T^

Sgo

t^ o S ^ !O> t- o ^ o CQ ö 9 Ö c4

^:

^- ^ o 2 c:^

t-H •?(•

t^ T-! ^ vJ^ O v i . P

S3 2 CO ^ o

^

o o

^% iS Ö ^ 9 ? 9

l5 ^ I S § I I f

3^8SgS

IC

q o 09 CO Ö 9 Ö

d ^ ci

IC

ö s o 9,0 2.

co^ 00 5q c4 Tf ^

q

^ ^

§5 S3dS§9§ä§!^ (M CO lO ^ 00 ^

g


103

bounds. According to the stringent EBA criterion, none of the variables are 'robust.' However, in view of potential multicollinearity among explanatory variables and the interaction between sociopolitical variables and institutional variables, we believe that our results are strong. We check the robustness of our results also in terms of the observations, not the conditioning variables, by using a robust regression method. OLS estimates tend to be sensitive to outliers, either observations with unusually large errors or influential observations with unusual values of explanatory variables (often called leverage points). One of the most common ways to deal with outliers is to drop observations one at a time. But this is often inadequate because it may miss a group of outliers due to the masking effect. Similarly, single-case diagnostics such as Cook's distance measure, the studentized residual, and DFITS that are often used to detect outliers may fail to identify a group of outliers. Robust estimation is intended to obtain estimates that are not sensitive to outliers, and hence allows us to characterize the most coherent part of the data set. We use the least median of squares (LMS) estimator due to Rousseeuw (1984), which is given by min median ef,

(5.3)

ß

where e^ is the residual of the 2th observation with respect to the LMS fit. This LMS estimator, typically computed by approximate algorithms, can resist the effect of nearly 50% of contamination in the data. A disadvantage of the LMS method is its lack of efficiency because of its unusually slow convergence, making it unsuitable for inference. To deal with this problem, we use the LMS estimates to classify some of the observations as outliers, and then carry out a simple reweighted least squares (RWLS) procedure by assigning zero weight to outliers and full weight to the rest of the observations, as recommended by Rousseeuw and Leroy (1987). Table 5.9 displays the robust estimates of the coefficients of the explanatory variables that were found to be highly significant in the panel regressions. Once again, they are very close to the results presented in the previous sections, and in fact, the results are much stronger. They are all significant at the 1% level.^^

5.7 Concluding R e m a r k s We tested different theories of public deficits, using more than 40 economic, sociopolitical and institutional variables, and could identify which of the vari^^ Among the most frequently identified outlying countries are Italy, India, Belgium, Zaire (now Congo), Japan, Ghana, Malaysia, Greece, Philippines, Argentina, Spain, Mexico, Indonesia, Malawi in decreasing order of frequency. All of these countries (except India, Mexico and Philippines) were among the 15 countries with the largest (or smallest) deficits either for the 1970s or the 1980s or both. See Data appendices.

Q

O

X5 OH

CO

O Ü

'S

I

Ö

-^

00

CO

3

104

°° p

^ fe

t>05 Oi C5

Cl bCJ5 <Ji

Tj< CO

05 CO

(M (M 00

CO 00 CO 05

Tf O T—I LO

ro (N 05

(M 00 iO

t^ lO CO 05

rt^

lO CO CS 05

CO CO CO

o^ T—1

(M

o

lO LO lO

q o

00

Oi

^ l>00

C75 CO 00

oo

T—1

CO (M

o

-^ ,_( ^

Tf l^ (M 03 (M LO b- CO CO c^ CO l^ t^ t- t-- t-

^ Tt^

05 00

rt< (M l>-

CO

o

Tf«

LO

00

(M

NCO 05

t^ C5

CQ CO 00

05

^ CO CO (M

CO CO 00 l>CO I:CO b-

CO

t~ CO LO

T—5

CO

LO LO

^ - t^ LO lO o

CO

o CO

c^

CO

CO ^ o0LO0

t--

lO CO CO rH

c> ö ö 05 CQ CO 05

T—1 o cq T-H co CO o

^ LO 05

1—1

CS

1> <Ji 00

c^ ^ o

CS

^

00 05

CS

C75

o

CS 00

o

TH

CO

^

lO CO CO 05

r-i

CO 05 05

05 05

t-

ö ö

^ !>•

CO 1>05

LO LO

CD

N-

T-H

t>- rf LO CO CO

ö

o

LO CO CO

CO

LO CO

t-

00 00

^ 05 CS

LO CO

^^

,-i 03

Q Q

a=! Jö ; >^ o3

3CO

o

o

o

05 LO

1—1

^ ö CS

00 CO 00 00 CS

LO CO LO CS

J

CO

O H:I > < W O

0)

ö ö

05 05 C75

t-

CO LO 00

o rH o o q '"!

CO

05 CS

^

$ J C S ° - : C S ? o LT^ i s j « ^ „„ , ^ , _. CS

The Political Economy of Fiscal Policy: Public Deficits, Volatility, and Growth

The Political Economy of Growth

Entrepreneurship, Growth, and Public Policy

Volatility and Growth

Volatility and growth

Rich Democracies: Political Economy, Public Policy, and Performance

Fiscal policies and growth in the world economy

The Political Economy of American Monetary Policy

Monetary and fiscal policy, - Credibility

Mayors and Money: Fiscal Policy in New York and Chicago (American Politics and Political Economy Series)

The Political Economy of Diet, Health and Food Policy (Routledge Frontiers of Political Economy, Number 24)

Monetary and fiscal policy, - Politics

The Political Economy of Public Administration: Institutional Choice in the Public Sector (Political Economy of Institutions and Decisions)

Growth Theory and Growth Policy

The Political Economy of Public Administration: Institutional Choice in the Public Sector (Political Economy of Institutions and Decisions)

Macroeconomic Policy: Demystifying Monetary and Fiscal Policy

Monetary Vs Fiscal Policy

Aid and the Political Economy of Policy Change

Demographic Change and Fiscal Policy

Political Economy of Rule Evasion and Policy Reform (Routledge Frontiers of Political Economy, 45)

Fiscal Policy Without a State in EMU?: Germany, the Stability and Growth Pact and Policy Coordination

Drugs, Crime and Public Health: The Political Economy of Drug Policy

The Limits of Stabilization: Infrastructure, Public Deficits and Growth in Latin America

The Political Economy of Work (Routledge Frontiers of Political Economy)

Seeking Growth under Financial Volatility

Handbook of Global International Policy (Public Administration and Public Policy)

An outline of political economy: Political economy and Soviet economics

Fiscal Deficits in the Pacific Region (Routledge Studies in the Modern World Economy, 44)

China and the Global Political Economy (International Political Economy)

Handbook of Conflict Management (Public Administration and Public Policy, Vol. 104) (Public Administration and Public Policy)

The Political Economy of Fiscal Policy: Public Deficits, Volatility, and Growth

The Political Economy of Growth

Entrepreneurship, Growth, and Public Policy

Volatility and Growth

Volatility and growth

Rich Democracies: Political Economy, Public Policy, and Performance

Fiscal policies and growth in the world economy

The Political Economy of American Monetary Policy

Monetary and fiscal policy, - Credibility

Mayors and Money: Fiscal Policy in New York and Chicago (American Politics and Political Economy Series)

The Political Economy of Diet, Health and Food Policy (Routledge Frontiers of Political Economy, Number 24)

Monetary and fiscal policy, - Politics

The Political Economy of Public Administration: Institutional Choice in the Public Sector (Political Economy of Institutions and Decisions)

Growth Theory and Growth Policy

The Political Economy of Public Administration: Institutional Choice in the Public Sector (Political Economy of Institutions and Decisions)

Macroeconomic Policy: Demystifying Monetary and Fiscal Policy

Monetary Vs Fiscal Policy

Aid and the Political Economy of Policy Change

Demographic Change and Fiscal Policy

Political Economy of Rule Evasion and Policy Reform (Routledge Frontiers of Political Economy, 45)

Fiscal Policy Without a State in EMU?: Germany, the Stability and Growth Pact and Policy Coordination

Drugs, Crime and Public Health: The Political Economy of Drug Policy

The Limits of Stabilization: Infrastructure, Public Deficits and Growth in Latin America

The Political Economy of Work (Routledge Frontiers of Political Economy)

Seeking Growth under Financial Volatility

Handbook of Global International Policy (Public Administration and Public Policy)

An outline of political economy: Political economy and Soviet economics

Fiscal Deficits in the Pacific Region (Routledge Studies in the Modern World Economy, 44)

China and the Global Political Economy (International Political Economy)

Handbook of Conflict Management (Public Administration and Public Policy, Vol. 104) (Public Administration and Public Policy)

Recommend Documents