2 0 0 7 A B C D E Annual World Bank Conference on Development Economics Regional Beyond Transition Edited by François Bourguignon and Boris Pleskovic Beyond Transition Annual World Bank Conference on Development Economics--Regional 2007 Beyond Transition Edited by François Bourguignon and Boris Pleskovic THE WORLD BANK Washington, D.C. © 2007 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 10 09 08 07 This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN-10: 0-8213-6843-5 ISBN-13: 978-0-8213-6843-5 eISBN-10: 0-8213-6844-3 eISBN-13: 978-0-8213-6844-2 DOI: 10.1596/978-0-8213-6843-5 ISSN: 1020-4407 Contents ABOUT THIS BOOK vii INTRODUCTION 1 François Bourguignon and Boris Pleskovic KEYNOTE ADDRESS Dynamics of Institutions, Development, and Elites 11 François Bourguignon KEYNOTE ADDRESS The Eurasian Growth Paradox 25 Anders Åslund and Nazgul Jenish KEYNOTE ADDRESS The Transition of Economics: The Cases of Israel and Russia 41 Gur Ofer KEYNOTE ADDRESS Interaction of Political and Economic Transition 51 Yegor Gaidar Growth After Transition: Is Rising Inequality Inevitable? Increasing Inequality in Transition Economies: Is There More to Come? 59 Pradeep Mitra and RuslanYemtsov Trade Liberalization, Inequality, and Poverty Reduction in Latin America 103 Guillermo Perry and Marcelo Olarreaga COMMENTS Jan Svejnar 141 Andrés Solimano 147 VI | CONTENTS Economic Space Comparing the Evolution of Spatial Inequality in China and India: A Fifty-Year Perspective 155 Kiran Gajwani, Ravi Kanbur, and Xiaobo Zhang COMMENTS Alan Gelb 179 Thierry Mayer 183 Governance The Institutional Determinants of State Capabilities in Latin America 193 Ernesto Stein and Mariano Tommasi Lowering the Cost of Capital in Emerging Market Economies 227 Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya Privatization: What Have We Learned? 249 Sergei Guriev and William Megginson COMMENTS Irena Grosfeld 297 Chong-En Bai 303 Judicial Foundations of a Market System Judicial Reform in Developing Economies: Constraints and Opportunities 311 Matthew C. Stephenson Transforming Judicial Systems in Europe and Central Asia 329 James H. Anderson and Cheryl W. Gray COMMENTS Ugo Panizza 357 Stefan Voigt 367 About This Book The Annual World Bank Conference on Development Economics is a forum for discussion and debate of important policy issues facing developing countries. The conferences emphasize the contribution that empirical economic research can make to understanding development processes and to formulating sound development policies. Conference papers are written by researchers in and outside the World Bank. The conference series was started in 1989. Conference papers are reviewed by the editors and are also subject to internal and external peer review. Some papers were revised after the conference, to reflect the comments made by discussants or from the floor, while most discussants' comments were not revised. As a result, dis- cussants' comments may refer to elements of the paper that no longer exist in their original form. Participants' affiliations identified in this volume are as of the time of the conference, January 18­19, 2006. François Bourguignon and Boris Pleskovic edited this volume. We thank Aehyung Kim for valuable suggestions and Leita Jones, the conference organizer, whose excel- lent organizational skills helped ensure a successful conference. Editing services were provided by Nancy Morrison, the book was typeset by Datapage Publishing Services, and book production and dissemination were coordinated by the World Bank Office of the Publisher. Introduction FRANÇOIS BOURGUIGNON AND BORIS PLESKOVIC The papers in this volume were presented at the Annual World Bank Conference on Development Economics (ABCDE), held January 18­19, 2006, in St. Petersburg, Russia. The conference series seeks to enhance the flow of ideas among development policy researchers and practitioners around the world and to open the Bank to the views of outside experts who can challenge or expand our knowledge of the theories and empirical evidence of development. Each year the topics selected for the confer- ence represent either new areas of concern or areas that we believe will benefit from a review of existing knowledge, as well as identifying areas for future research. The topic of the 2006 conference was "Beyond Transition," which encompassed four themes: growth after transition, economic space, governance, and judicial foun- dations of a market system. Keynote Addresses In his keynote address, François Bourguignon revisits the evidence on the role of institutions in growth, analyzes how elites influence whether institutions change or persist, and draws conclusions as to how international organizations can support Pareto-improving change. Growth modeling over the past decades has evolved by successively deeming as endogenous those factors that had been considered the ultimate sources of growth in previous models. In the latest permutation, growth depends on exogenous factors as well as accumulation of physical and human capital. Accumu- lation is determined by policies, which in turn are shaped by institutions. Efforts to define the impact of institutions have commonly been hampered by the dearth of objective indicators of institutional quality, as well as the difficulty of separating the influence of institutions from the influence of policy. Conversely, those studies that do use strong instruments to identify institutional quality do not allow for them to be endogenous to the political economy. What drives the evolution of institutions? François Bourguignon is chief economist and senior vice president, Development Economics, at the World Bank. Boris Pleskovic is research manager, Development Economics, at the World Bank. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / The World Bank 1 2 | FRANÇOIS BOURGUIGNON AND BORIS PLESKOVIC The political economy of institutions, and the influence of elites upon them, can be described in a simple two-sector model, in which an elite group inequitably holds more influence than the rest of the people. Both actors benefit from growth. They actively shape policy in accordance with their self-interest, but institutions set the rules of the game that both actors must follow. Institutions themselves, however, can also be influenced. Elites, while sometimes altruistic and almost never monolithic, will in many cases shape institutions to generate maximum rents, despite suboptimal outcomes for social welfare. When do institutions change, and when do they persist despite an adverse economic equilibrium? Historical evidence suggests that economic structure matters: plantation economies in Latin America and the southern United States long empowered rent-seeking elites and held back the development of equitable institutions, while small-holder farming promoted it early on in the northern United States. Furthermore, whether elites believe that they stand to win or to lose from a reform is important. Initial gradual steps toward liberalization in the Soviet Union met with resistance by the elite, which could be overcome only through deeper political reform. By way of contrast, elites did not resist economic reform in China, and no political reforms ensued. Recent strengthening of democracy in Africa has similarly produced mixed growth results. Institutional reforms are a sovereign function of each polity, and the influence of the international development community is limited, out of necessity. Yet international organizations can assist the process by analyzing prospective gains and losses, suggesting Pareto-dominant bundles, or devising com- pensation schemes and enforcement mechanisms. Well-negotiated aid conditionality can be a tool for consensus building and a commitment device--if the donor com- munity overcomes its own coordination and time-inconsistency challenges. Anders Åslund and Nazgul Jenish argues that in the first decade of postcommu- nist transition, the more radical and comprehensive the market economic reform was, the earlier a country returned to economic growth and the more vigorous its growth. Central Europe took the lead. However, since 2000 annual growth among the Commonwealth of Independent State (CIS) countries has been more than 4 per- centage points higher than the Central European countries. A regression analysis for 20 postcommunist countries shows with strong significance that a reduction of pub- lic expenditures has most effectively stimulated economic growth, while oil exports are also as positive and significant as expected. The distance from the European Union (EU) is also positive and significant: that is, the further from the EU, the higher the economic growth. The effect of corruption is negative for growth but only mar- ginally significant. Neither the laggard effect nor investment reveals any significant effect. Depressingly, the CIS countries that have generated impressive economic growth are largely authoritarian. In effect, the CIS countries have adopted the highly successful East Asian growth model lock, stock, and barrel, while the less dynamic Central European countries have adopted the EU model, which has not been con- ducive to high economic growth. Åslund concludes that, at least among postcommu- nist countries, high public expenditures and taxes are not conducive to economic growth. Not surprisingly, liberal economic policy or greater economic freedom does not seem to promote economic growth. Thus, more emphasis should be given to the need to reduce public expenditures to boost economic growth. INTRODUCTION | 3 Gur Ofer discusses the introduction of modern economics to Russia and Israel. After briefly describing the successful record of the New Economics School (NES) in Moscow, Ofer mentions that to a large extent, NES was modeled after the experience of introducing modern economics in Israel in the late 1940s. This is his reason for reviewing the Israeli experience in greater detail. The author describes how Don Patinkin single-handedly established modern economics in Israel, building on three pillars: (1) sending a group of students for PhD studies in the West; (2) developing curricula and lecture notes for new courses; and (3) establishing infrastructure for the study and research in the Israeli economy. The essence of Patinkin's economics, follow- ing the University of Chicago tradition, was the analytical positivistic version of neoclassical economics, which replaced both the old continental version of institu- tional economics as well as Marxist political economics. The method of teaching was to solve problem sets ahead of the discussion in class, which was a contradiction of the continental method of reading lectures. The U.S. pattern of combining teaching and research was also transferred to Israel, and Patinkin initiated the establishment of a research center devoted to the Israeli economy. Ofer further describes the evolution of the success story of economics in Israel in terms of achievements, quality, open- ness to the world, and a near optimal balance between theory and policy. There are many parallels--but also differences--between Israel's experience and that of Russia; indeed Israel served as a guide and inspiration for NES. Like Israel in the past, Russia needs modern economics and can gain from it even more than established market economies. Ofer further argues that, like Israel, Russia is in a potential position to compete successfully and move ahead of some of the countries in Western Europe that are moving rather slowly to advance modern economics. Russia, with the help of NES and other new institutions of higher education in economics, can provide important contributions to Russia's emerging market economy and also the advance of economics theory, in general and especially in the fields of institutional economics, political economy, and transition. Yegor Gaidar addresses the interaction of political and economic transition in Russia. He notes that serious, positive economic reforms, such as property rights and reduced taxation, were carried out in Russia, but by the year 2000 a reversal on the democratic front had occurred. It is often argued in the literature that first economic ground should be established for democratic development, and only then should the movement toward democracy be followed. Gaidar claims that this explanation is too simplistic and that the process should be analyzed from a historical perspective. He then discusses the main features of a socialist economic and political system. His main argument about the disintegration of socialism and of the Soviet Union is that socialism was an unstable economic system built on the basis of the authorities' willing- ness to employ violence. In the final decades of the Soviet Union, the ability to use violence was eroded and undermined by the evolution and development of a society that became more literate and educated. Gaidar also notes that the inefficient socialist model of industrialization and agriculture made the country dependent on the prices of raw materials, such as oil and natural gas, which fluctuated uncontrollably on a long-term basis. Gaidar concludes that the basis for the collapse of the Soviet Union was a political and economic structure that was not stable internally and that was 4 | FRANÇOIS BOURGUIGNON AND BORIS PLESKOVIC based on violence and later on importation of agricultural and other industrial products. As the level of development increased, it undermined the ability of the authorities to use unlimited violence. This was coupled with a crisis in agriculture, noncompetitive industry, and falling oil prices. According to Gaidar, this is the real reason for the catastrophic development of the Soviet Union. Growth After Transition: Is Rising Inequality Inevitable? Pradeep Mitra and Ruslan Yemtsov examine inequality in the transition economies of Eastern Europe and the former Soviet Union. Their paper decomposes changes in inequality, which has generally been increasing in the transition economies of Eastern Europe and the former Soviet Union, both by income source and socioeconomic group, with a view to understanding the determinants of inequality and assessing how it might evolve in the future. The paper's empirical analysis relies on a set of inequality statistics which, unlike "official" data, are consistent and comparable across countries and are based on primary records from household surveys recently put together for the World Bank 2005 study "Growth, Poverty, and Inequality in Eastern Europe and the Former Soviet Union: 1998­2003." The increase in inequality in transition, as predicted by a number of theoretical models, in practice differed sub- stantially across countries, with the size and speed of its evolution depending on the relative importance of its key determinants: that is, changes in the wage distribution, employment, entrepreneurial incomes, and social safety nets. Its evolution was also influenced by policy. This diversity of outcomes is exemplified on the one hand for Central Europe by Poland, where the increase in inequality has been steady but gradual and reflects, among other things, larger changes in employment and compensating adjustments in social safety nets, and on the other, for the Commonwealth of Inde- pendent States by Russia, where an explosive overshooting of inequality peaked in the mid-1990s before being moderated through the extinguishing of wage arrears during its recovery after 1998. Mitra and Yemtsov argue that the process of transition to a market economy is not complete and that further evolution of inequality will depend on (1) transition-related factors--such as the evolution of the education pre- mium, and a bias in the investment climate against new private sector firms, which are important vehicles of job creation--and (2) the regional impediments to mobility of goods and labor, as well as other factors, such as technological change and global- ization. The paper also contrasts key features of inequality in Russia in the context of other transition economies with trends in inequality observed in China, where rapid economic growth has been accompanied by a steep increase in inequality. The authors argue that China's experience is to a large extent a developmental phenome- non rather than a transition-related one, deriving from the rural-urban divide, and thus is of limited relevance for predicting changes in inequality in Russia. Guillermo Perry and Marcelo Olarreaga examine Latin America's trade liberalization in the late 1980s and early 1990s. They argue that the liberalization was accompanied in some countries by increases in skill premiums, wage and income inequality, and even INTRODUCTION | 5 in poverty: results unexpected by many. These phenomena were mainly the result of four factors. First, most Latin American countries are rich in natural resources (which in general are complementary with capital and skills), and are more capital abundant than other developing countries with large pools of unskilled labor, such as China and India, that were integrating into the world economy while Latin America liberalized. Second, dynamic effects of trade led to new goods being produced in the region through outsourcing, and to an acceleration of skill-biased technical change and Schumpeterian creative destruction, resulting in an increase in demand for skills in most industries. Third, initial conditions and contemporary events make predic- tions based on a simple factor abundance model difficult to generalize. Fourth, the impact that trade reform had on imperfectly functioning labor markets--such as potential transitions in and out of unemployment, informality, and income volatility-- is likely to affect and sometimes change the direction of the impact of trade reforms on income inequality and poverty. Finally, the paper concludes that the effect of trade on poverty (and income inequality) depends on complementary policies being implemented. The impact of trade on poverty reduction can be significantly enhanced (and the effects on inequality mitigated) by policies that increase the provision and access to skills and other productive assets to the poor. Economic Space Kiran Gajwani, Ravi Kanbur, and Xiaobo Zhang compare the evolution of spatial inequality in China and India. In the second half of the last century, both India and China underwent major transitions and moved to more liberalized economies. The paper relates the observed patterns in regional inequality to major events during this period. Because of China's institutional barriers to migration, regional inequality is much higher than in India. Also, China's decentralization and opening up to global- ization are closely related to the observed regional inequality--particularly the inland-coastal disparity--since the reform period. With openness, the rates of returns to labor--particularly skilled labor in the coastal areas--change, as well as for land. On the other hand, from the Green Revolution to the period of economic liberaliza- tion in India, the evolution of regional comparative advantage has shifted from the quality of land to the level of human capital as India integrates with the international market. Therefore, India's states have become clustered into two clubs: more educated and less educated ones. The empirical findings are also relevant to the ongoing debate on globalization's effects on regional inequality in developing countries. Convergence or divergence of a nation's economy is dependent upon not only its domestic polices but also on its openness. The authors' results show that openness has led to changes and increases in regional inequality by providing more favorable conditions for growth for coastal and better-educated regions. The implications for policy are a need to pay careful attention to those regions that are less able to take advantage of gains from openness. In China, this refers to inland regions, while in India this means less-educated regions. 6 | FRANÇOIS BOURGUIGNON AND BORIS PLESKOVIC Governance Ernesto Stein and Mariano Tommasi present the institutional determinants of state capabilities in Latin America. They argue that there are some qualities and charac- teristics of public policies that are (to some extent) independent of grand policy "titles" (such as "public" or "private") and that seem to lie behind their impact on behavior and outcomes. For instance, the performance of a given sector of the economy may be better or worse under either public ownership or private ownership cum public regulation, depending on some fundamental state capacities, such as the ability to commit to a policy course, the ability to adjust policies when circumstances change, the ability to enforce and implement policies, and the ability to focus on broad general welfare as opposed to narrow interests. In a nutshell, the main tenet of this research agenda has been to move the discussion away from "universal policy recipes" toward a focus on the determinants of policymaking capabilities, including the ability to reach reasonable degrees of societal consensus as a foundation for the credibility and effectiveness of public policies. In particular, the authors explore the politico- institutional determinants of good public policies by drawing from a framework that predicts that desirable policy characteristics (stability, adaptability, consistency, pub- lic regardedness) depend on the behavior of political actors in the policymaking process (PMP). The framework places particular emphasis on the ability of political actors to cooperate over time. The preliminary empirical work has uncovered no simple direct effects of some politico-institutional variables usually emphasized in the previous literature. These variables include characteristics of the electoral system, and legislative and partisan powers of the executive. Further empirical work on a broader data set is necessary in order to identify configurations that tend to produce better policies. One additional feature suggested by their work is that the "institu- tional blessings" behind high-quality policies and state capacities tend to develop slowly over time and tend to be the result of the ongoing behavior of many relevant political actors. In conclusion, studying the way in which different institutional char- acteristics are built over time would require theoretically structured comparative country studies that could pay special attention to the interaction between institutions and the specificities of political cleavages and socioeconomic structures behind the economic and social policies implemented in each country at each point in time. Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya discuss corporate governance and bankruptcy policy specific to emerging market economies (EMEs). They argue that lowering the cost of capital for firms in emerging market economies is one of the major tasks of economic development. The authors present key arguments in addressing various policy solutions for this task. First, the solution to the problem of lowering the cost of capital in EMEs is unlikely to resemble corporate governance and bankruptcy reforms in OECD countries. The emerging market economies are characterized by different ownership and capital structures and a different nature and depth of market and government failures. Although many EMEs have already adopted best-designed company and bankruptcy laws, these changes have not yet led to improvements in the cost of capital because of imperfect INTRODUCTION | 7 enforcement. In some circumstances, the transplantation of OECD laws to an EME may actually be detrimental to a country's financial development, rather than just being ineffective. Second, there is a substantial variation between the EMEs, which implies that there is no "one-size-fits-all" solution. Implementation of reforms will depend crucially on the distribution of political and economic power in each partic- ular country, as well as its cultural and social environments. Thus, instead of suggest- ing ready solutions, the authors identify key conceptual trade-offs in the areas of corporate governance and bankruptcy that can help inform policy debate about the costs and benefits of specific policy choices. The importance of these costs and ben- efits for each particular country would depend on its economic and political environ- ment. Third, since the difference in the environment between the EMEs and OECD countries in many instances is much greater than among the emerging market coun- tries, it is possible to draw a few general lessons. One clear message from the paper is that corporate governance, bankruptcy, judicial, and political reforms are highly complementary in EMEs. At the same time, one of the main obstacles to financial development--poor enforcement of law and contracts--arises from weaknesses in political institutions. Improving enforcement requires policy intervention at many dif- ferent levels, including deep political transformation with fundamental constitutional change, and administrative and regulatory reforms. Since the level of enforcement is necessarily an outcome of political economic game among interest groups, improving enforcement is an immensely difficult task. Under poor contractual and law enforce- ment, countries seem to be better off when they rely on private mechanisms of investor protection. As debt-financing plays the major role in EMEs and bank- ruptcy is the crucial mechanism for protecting creditor rights, one cannot consider corporate governance reform without bankruptcy reform. Therefore, the priority of corporate governance and bankruptcy reforms in EMEs should be on protecting property rights of majority claimants. Sergei Guriev and William Megginson examine how privatization has changed the economic landscape since the late 1970s, given that it is one of the major phe- nomena of recent economic history. Their paper summarizes empirical research on the effect of privatization on the performance of privatized firms and on society. The extant evidence in many developed and developing countries shows that pri- vatization usually results in increased productivity and has positive effects on society. Achieving a positive effect depends, however, on having critical economic institu- tions in place--in particular, the rule of law, competition, hard budget constraints, high-quality governance, and effective regulation. Guriev and Megginson pay special attention to the cases of Russia and China. In Russia (and some other CIS coun- tries), privatization seems to have produced few benefits for the privatized firms or for society, whereas China has managed to pursue a reform package that, so far, has not included mass privatization of state-owned enterprises and yet has pro- duced very impressive results. The authors argue that in both cases--as well as in other controversial privatization examples, such as Latin America--the outcomes can be explained within the conventional framework once one accounts for an appropriate counterfactual. 8 | FRANÇOIS BOURGUIGNON AND BORIS PLESKOVIC Judicial Foundations of a Market System Matthew C. Stephenson argues that there has been an extraordinary increase in the attention paid to the role that public institutions play in promoting economic devel- opment over the last decade. Indeed, the assertion that "institutions matter" has become commonplace, perhaps even cliché. This institutionalist revival in the devel- opment community has included a resurgence of interest in the role that legal and judicial institutions play, or ought to play, in promoting material improvements in the quality of life of the world's poor. Academics and policy analysts have sought to better understand the relationship between legal and judicial institutions and eco- nomic performance, while the development community has promoted legal and judicial reform projects that range from modest efforts to improve court adminis- tration to ambitious attempts to eliminate judicial corruption, promote judicial independence, and craft better, more equitable, and more market-friendly legal systems. The diversity and complexity of the debate about legal and judicial reform, and of the myriad reform projects that have already been undertaken, put compre- hensive overview of the field beyond reach. In his presentation, Stephenson first identifies what he sees as basic and recurring problems that bedevil efforts to design and implement effective legal and judicial reform projects. Second, he suggests some conceptual tools that can be used to address these difficulties by describing three particular problems. The first is a straightforward resource constraint problem. Improving the capacity and quality of a judicial system requires material and human resources that are often in short supply in developing economies. The second problem is what one might think of as an incentive compatibility problem. The judiciary's capacity to perform the economic and other functions assigned to it by law-and- development theorists depends in large part on the willingness of affected parties to use the courts to resolve disputes and to abide by judicial decisions, and on the will- ingness of judges and other legal officers to behave in a manner that is consistent with the requirements of a well-functioning judicial system. But aligning incentives in this way is often difficult. The third problem is an institutional version of the General Theory of the Second Best: When a legal system is suboptimal in more than one respect, improving the law or the courts along one dimension may not improve overall institutional performance, and may even worsen it. Understanding this principle is important to understanding--and attempting to avoid--the pitfalls associated with the necessarily incremental and partial nature of virtually all efforts at legal and judicial reform. James H. Anderson and Cheryl W. Gray argue that the judicial systems in the transition countries of Central and Eastern Europe, the Balkans, and the former Soviet Union are under heightened scrutiny these days, 17 years after transition began. In Central and Eastern Europe, the European Union is exerting strong pres- sure on new members and candidate countries to root out corruption and improve the functioning of their judiciaries. Further east, judicial systems in Russia and other countries in the former Soviet Union have been increasingly in the spotlight due to high-profile roles in controversial cases, such as the Yukos case in Russia INTRODUCTION | 9 and the dispute surrounding the presidential elections in Ukraine. As economic reforms mature and these countries become increasingly interconnected with the outside world, the need for good governance and the constraints imposed by weak judicial systems are rising in visibility and importance. A recent World Bank report by Anderson, Bernstein, and Gray (2005), Judicial Systems in Transition Economies: Assessing the Past, Looking to the Future, reviewed the experience of transition countries with judicial reform since 1990 and drew on numerous data sources to compile a snapshot of the state of their judiciaries in the first few years of the twenty-first century. Anderson and Gray's paper updates that report by incorporating the findings of a large survey of enterprises throughout the region undertaken in spring 2005, the third EBRD­World Bank Business Environment and Enterprise Performance Survey, or BEEPS, and going into further detail on the judicial reform programs underway in transition countries. Anderson and Gray examine three broad questions: (1) the kinds of judicial reforms needed for successful transition from socialism to market-based economies and sequence of reforms; (2) progress made in this transition both by individual countries and by subregion and factors explaining the extent of progress to date; and (3) how firms' evaluations of judicial systems in transition countries--and by implication, the priorities and chal- lenges that these systems face--compare with those in more advanced countries, to the extent that transition countries share common concerns and priorities with countries in Western Europe. Keynote Address Dynamics of Institutions, Development, and Elites FRANÇOIS BOURGUIGNON I am very grateful to the Russian government and the city of St. Petersburg for hosting this conference. It is a harbinger of the events that will take place later this year related to the G8, which is chaired by Russia, and of the role of this great country in development--not only in the region, but in the world. Institutional change and the role that elites play in that change are integral to this address. It is increasingly realized today that economic development depends very much on the quality and nature of institutions. The experience of those countries that have "transitioned" from centrally planned economic systems to market economies is perhaps the best example of the major role that institutions can play in the process of economic development. Considerable progress has been made recently in our understanding of the relationship between institutions and development. Yet our knowledge is still very partial. In particular, a point that we need to understand better is that of the endogeneity of institutions. What explains the fact that institutions are modified in one society, thus improving economic performance, and persist in another, despite very unsatisfactory economic results? Without some answer to this question, recommendations for institutional change in a given country may be purely academic, with little chance of ever being embraced. A key actor in the evolution--or, on the contrary, the persistence--of institutions are the elites, or borrowing from Mancur Olson, the groups within a nation that share, give, or get part of political or economic power. In a perfect democracy, the political elite would be the whole population and the concept would not be very useful. But democracies are imperfect, and some groups end up playing a more important role than others in public decision making--whether because of their eco- nomic power or other reasons. In addition, there are still many countries that are run by autocratic regimes, which fit more closely the preceding definition of elites. This address is about the role of elites in determining the institutions in which the economy of a country has to work, and about what this role implies for third parties like the international development community. As with most of my work at the François Bourguignon is chief economist and senior vice president, Development Economics, at the World Bank. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / The World Bank 11 12 | FRANÇOIS BOURGUIGNON World Bank, there are two sides to this presentation: on one side drawing on what we do know, and on the other, dealing with what we don't. Regarding the first part, I will rely on the extensive work done by the Research Group of the World Bank, in close cooperation with the whole development research community. On the more dif- ficult part--discussing what we don't know--we have to take a more theoretical view. And we need to alert our colleagues in the development economics community to the unresolved issues that are important to policy makers and development practitioners. I will begin the presentation with a short reminder about the existing knowledge on the relationship between institutions and development, in particular those institu- tions that are related to the governance of a country. I will then explicitly introduce elites into that description of the role of institutions in development, which, in effect, is equivalent to introducing political economy factors under the assumption that institutions are fixed. In that framework, I will finally show that a major factor in explaining economic development or stagnation is the capacity of a national politico- economic system to generate institutional changes. I will then conclude by briefly analyzing a few historical examples of institutional changes or lack thereof, and drawing some implications for the role of external actors, in particular the inter- national development community. Institutions, Governance, and Growth: The Basics Figure 1 gives a very schematic representation of the evolution of macroeconomic thinking in the field of development over the past 50 years or so, under the simplify- ing assumption that development is equivalent to economic growth. It is convenient to divide that evolution into three stages, with the most recent and elaborated one being based on the role of institutions. In a first stage, growth was seen as resulting mechanically from the accumula- tion of productive factors (physical and human capital), as well as exogenous factors like technological progress. This approach is described by the arrows with FIGURE 1. Successive Approaches to Economic Growth (b,c) (c) Accumulation Policy Institutions productivity (B) (a,b,c) (C) Income growth (a,b,c) (B,C) Exogenous factors technology Source: Author. DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 13 the label (a) in figure 1. In a second stage, the focus shifted to policies that may create an environment more or less favorable to the accumulation process and to gains in total factor productivity. Some of these policies may be quite close to accumulation itself, such as those concerned with education and human capital. However, the relationship may be much less direct for the various policies that affect investment behavior or the efficient allocation of resources. The quality of macroeconomic policies (as summarized by inflation or the budget deficit), the size of the government sector, or the outward orientation of the economy are examples of indirect determinants of investment and productivity gains. Solid arrows with the label (b) in figure 1 describe that broader view of development. The dotted arrows with the label (B) correspond to a "reduced form" specification of the underlying structural model. What ultimately matters for economic growth is the nature of the policies being implemented, as well as domestic and external exogenous factors that could modify their effect. In the third stage, policies themselves are taken to be endogenous and essentially determined by institutions, or in other words by the rules of the game behind public decision making and possibly by exogenous characteristics of the countries under study. At this stage of growth modeling, the structural model is now much more intricate. As shown by the solid arrows labeled (c), the causality chain leading to growth now includes the way in which existing institutions and exogenous charac- teristics of societies lead to policy decisions of some type, which in turn determine the process of accumulation and productivity gains ultimately responsible for growth. At this level, too, it is also logically possible to use a reduced-form view--as shown by the dotted arrow labeled (C)--that bypasses all these intermediary steps and where growth is essentially determined by the nature of institutions in a country and exo- genous characteristics of that country. It is interesting that our understanding of growth has evolved by successively deeming as endogenous those factors or economic processes that at a previous stage were considered the ultimate determinants of growth. In a way, this presentation is about the continuation of this process. Yet it is not so much about trying to make institutions endogenous as it is about understanding how institutions change by themselves or can be modified through the intervention of third parties. Empirically, considerable efforts have been devoted over the last 15 years or so to testing the various reduced-form models that correspond to the dotted arrows in figure 1. Following the well-known pioneering paper by Barro (1991), the analysis has essentially been of the cross-country type, often with several observations of a single country over different time periods. An impressive number of policies and exogenous parameters have been analyzed, yielding results of varying robustness.1 Most of this literature is about the role in generating growth of a large number of policies such as trade openness, security, or the nature of the tax system, and exoge- nous factors describing a country, such as geographical location, ethno-linguistic composition, or the inequality of the distribution of resources. By comparison, less has been done with variables that describe the nature and quality of institutions. In effect, several studies combine both institutional and policy variables as determinants of economic growth, without really controlling for the 14 | FRANÇOIS BOURGUIGNON endogeneity of the latter or trying to explore the relationship between the two sets of variables. It is often more the quality of the government, to borrow from the title of the paper by La Porta and others (1998), that is being considered, than institutions per se. In Knack and Keefer (1995), La Porta and others (1998), and Rodrik (1999), for instance,2 variables used include the democratic nature of decision making, the rule of law, political rights, or protection of property rights. These may indeed be considered as true institutional characteristics in the sense that they describe the rules of the game, as Douglass North (1990) defined institutions.3 But alongside these vari- ables, these studies also include tax/GDP ratios, social spending, or the quality of bureaucracy, which seem to be under the direct control of the government and must thus be considered as policy variables. The correct reduced-form model to be used to explore the role of institutions in economic development should include only variables that truly describe existing institutions in a country, and exogenous characteristics that may combine with insti- tutions to determine the nature and pace of development. This is what is done, for instance, in Acemoglu, Robinson, and Johnson (2001) and Acemoglu, Johnson, and Robinson (2005), which present rather solid evidence of the role of institutions on development in the long run. Figure 2 gives some idea of that relationship--even though no allowance is made there for the possible endogenous nature of the insti- tutional variable and the influence of other exogenous country characteristics. The figure relates GDP per capita in 2004 to the indicator of the rule of the law proposed by Kaufman, Kraay, and Mastruzzi (2005). Because GDP per capita in figure 2 is considered in levels rather than in growth rates, the implicit view of growth is defi- nitely very long run. Figure 3, taken from Knack (forthcoming), shows the same rela- tionship within a medium-run context, where it is the 20-year growth rate of GDP per capita that is correlated with an institutional indicator: namely, an indicator elab- orated by the International Country Risk Guide. It can be seen that the fit is much less satisfactory, even though the positive relationship is still statistically significant. The comparison of these two charts suggests that the influence of institutions on growth is indeed very long run. FIGURE 2. Simple Cross-Country Correlation between Income Level and Rule of Law Indicator 11 LUX 2004 USA IRL NOR NLDFINSWE ISL AUT in FRAJPN BEL HKG DNK 10 SGP ISR KOR GRC NZL CZE SLV MLT PRT HUN SVK POL SAU EST PPP, ARG CRI LTU LVA MAR 9 RUS URYBWA MYS ROM IRN THA TKMBLR KAZ BRABGR COL DOM TUR TUN VEN UKR NAM PERCHN PRY ALB LBN PHL capita, AZE ECUARM SYRGUY EGY 8 IDN NIC GEO IND HNDPNG VNM LSO per SDN LAO UZBGINKGZ PAK GHA BGD MNG HTI MDASEN UGANPL 7 TJK MOZ NGA KEN GDP MLI SLE YEM NER MDGZMB GNB TZA MWI Log6 ­2 ­1 0 1 2 Av. `Rule of Law,' 1996­2004 Source: Author's calculations using the rule of the law indicators proposed by Kaufman, Kraay, and Mastruzzi (2005). DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 15 FIGURE 3. Partial Cross-Counry Correlation between Twenty-Year GDP Growth Income Level and the ICRG Institutional Indicator, 1982­2002 0.08 CHN 0.06 KOR BWA SGP 0.04 THA IND wth,org IDN TUN CHL HKG MYS PAK UGACYP MLT BGDEGY SDN 0.02 KWT DOM LKATUR BHR IRN ECU SLV GHA MLI CRI ISL (residual) BRA GTMPOLDZA MEXCOL SEN MWIGMB PNG 0.00 SYR BRNJAM HND ISR URY income GUY KEN BOL JOR PAN TGOTTO HUN CMRBGR ZAF PHL ARG PRY PER ZWE COG NER ­0.02 HTI ROM VEN ZMB capita NIC 1982­2002 erP SLE ­0.04 ZAR ­0.06 ­0.08 ­20 ­10 0 10 20 30 ICRG index (residual) Source: Knack (forthcoming). Note: The ICRG institutional indicator controls for initial income and school attainment. Empirical work on institutions and development is made difficult because of the nature of the institutional indicators available and because most available indicators cover only the recent past. The best indicators should be objective--or in other words, based on observed behavior. Practically, however, statistical observation in this field is difficult. Whether the objective is to describe the degree of protection of property rights, or even the extent of democracy, no simple statistics are available across countries. This is the reason why most indicators rely on the perception of some aspects of institutions by a set of agents. These agents may be the general pub- lic, as in the indicators that can be derived from barometers available on a regular basis in several regions. They may be firm managers, as in the indicators that can be derived from the Investment Climate Surveys, conducted by the World Bank and the European Bank for Reconstruction and Development.4 However, most indicators are based on experts' opinion on particular institutional aspects. These indicators are compiled by several business agencies and NGOs such as Business Environment Risk Intelligence, the International Country Risk Guide, Business Monitor International, or Transparency International. In the World Bank, Daniel Kaufmann, Aart Kraay, and Massimo Mastruzzi regularly compile a kind of consensus statistics based on these original indicators.5 A governance database is also compiled every year by country teams. The World Bank systematically uses the resulting Country Perfor- mance and Institution Assessments to allocate aid among low-income countries.6 The problem with these various perception-based indicators is that their definition is not always clear, even when it is based on several sub-indicators in a given field. They are probably fine when rough comparisons are made across countries. They may be less so when the comparison is made over time within a given country, which is 16 | FRANÇOIS BOURGUIGNON ultimately what should matter for the understanding of the relationship between the quality of institutions and development. This problem is especially severe given that most of these indicators are available only over the past few years. Elites and Institutions in the Process of Development It is now time to explicitly introduce economic and political agents into our model to complete our description of the way countries make policy decisions that affect development. This is what is done in figure 4. The population is divided into two groups: the "elite" and the "people." There may be various subgroups within each of these classes, but it simplifies the exposi- tion to consider only two groups. The dotted arrows represent the way that policy decisions are made (P) and the distribution of income between the two groups (d). They correspond to the political economy system of the hypothetical country being considered. Policy decisions depend on the interest of the various parties (the "dis- tribution" (d) arrows) and on the institutions in place (the (c) arrows). In other words, institutions determine the rules of the political economy game leading to policy decisions, which in turn impact the nature and pace of economic growth and the way it is distributed between population groups. If institutions included a fairly democratic public decision-making process, then the influence of both the elite and the people on policies would be more or less proportional to the demographic size of these two groups. Presumably, the arrow originating in the box labeled "people" (Pp) would convey more power than the arrow originating in the box labeled "elite" (Pe). The opposite case is depicted in figure 4, with the elite having much more power to control policies than the people. That people still have some influ- ence on policies reflects some kind of imperfect democracy, rather than a pure autocracy. Yet institutions include many other dimensions than democracy. For instance, some policies that would partly expropriate the people's property or extract massive rents from them could not be imposed by the elite in the presence FIGURE 4. Institutions, Policies, and Development: The Role of Agents (b) Accumulation Policy Institutions productivity (c) (b) (Pp) (Pe) (i) Income Elites growth (d) People (b) Exogenous factors technology Source: Author. DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 17 of a judicial system with some minimum effectiveness in protecting property rights or preventing corruption. As is apparent in figure 4, institutions are not exogenous. They may be modified by society, and both elites and the people can play a role in that process; see the arrows labeled (i). In the most basic model of elite control, where information is per- fect, elites choose institutions that maximize their payoff. That payoff may be their direct economic interest, or if elites are somewhat benevolent, it may be broader social welfare. Note in particular that elites are not necessarily monolithic. They may comprise groups with interests closer to those of the people, which might have some power on the overall elite in imperfect democratic systems. As democratic practices gain strength, the power to reform institutions progressively shifts to the people. Figure 5 shows a simple example of how different institutional choices may result from different structures of power, and may lead to different economic outcomes. The institution being considered here is the control of corruption practices or rent- seeking by economic agents: in other words, some component of the rule of the law. It is assumed that only the elite have the power to extract rent from other people in the population, and that rent-seeking reduces the efficiency of the economy at an increasing rate. As the intensity of rent-seeking increases, GDP per capita goes down, so that the elite face a kind of Laffer curve. The amount the elite can actually extract from the people first goes up with rent-seeking, but goes down after a threshold has been reached where efficiency losses are more important than the additional rent. If the elite has political power and can oppose any institutional reform that would in some way prohibit rent-seeking, the equilibrium will be in A, corresponding to a low level of GDP per capita. If, on the contrary, the people are able to impose an institu- tional reform that will ban rent-seeking, the economy will end up in B, at a much higher level of GDP per capita. This example shows how bad institutions, supported by an unequal structure of political power, may generate inefficient economic results. In effect, the situation described in figure 5 is observed, in one form or another, in many countries. It is another example of the general argument developed in the 2005 World Development Report (World Bank 2005b), according to which equity and development are complementary. The lack of equity lies here in the inequality of FIGURE 5. Economic Equilibria With and Without Elite's Rent-Seeking Income per capita No rent-seeking B Rent (per capita of the elite) A GDP per capita Intensity of elite's rent-seeking Optimal rent- seeking intensity Source: Author. 18 | FRANÇOIS BOURGUIGNON political rights, which does not permit establishing institutions that control corruption or rent-seeking by the elite, and generates inefficient economic outcomes. What we would like to understand is how to move from one type of institution that leads to an imperfect economic equilibrium to another institution that leads to a better economic equilibrium. In terms of figure 5, the question is to know how to go from a situation like A, the maximum rent situation, to B, where GDP per capita is the greatest--or at least to some intermediate situation. This question of the per- sistence of bad institutions imposed by powerful elites has been explored in depth by Acemoglu and various coauthors.7 Most of their analysis relies on theoretical models that are shown to fit some stylized historical facts, and on cross-country regression work. In effect, the paper referred to earlier--Acemoglu, Robinson, and Johnson (2001)--goes much beyond estimating the effect of institutions on long-run growth. It also considers the historical origin of the institutions that are observed today. In the case of developing countries, in particular, the authors find an extreme persistence of the effects on development of natural conditions that go back several centuries, such as the mortality of settlers in the fifteenth century. Their conclusion is that the influence of these initial conditions on development has been transmitted mostly through their impact on the structure of power between elites and the peo- ple and the institutions it may generate. Even though this particular result has been the subject of a rather heated debate, this is certainly an interesting hypothesis.8 It also is a rather disturbing proposition, which would tend to suggest a very strong historical determinism in the formation of institutions, and therefore in the development performances of countries. As a matter of fact, both theory and direct observation also suggest that institutions can change under diverse circumstances, even when they are initially supported by apparently powerful elites. Following some exogenous shock that may disturb some initial political economy equilibrium, some changes may appear Pareto-superior and be implemented spontaneously. Others may result either from conflicts within the elite, collective action by the people, or possi- bly the simple threat of such a collective action when repression is found to be too costly by the elite.9 Those circumstances that make institutional change possible are what economists and social scientists working on development should strive to identify. We conclude this presentation by mentioning some interesting examples of spontaneous nonviolent examples of institutional change and what they suggest in terms of possible interven- tions by outside actors willing to ease those changes and the acceleration of develop- ment in slow-growing developing countries. Some Examples of Institutional Changes Dramatic institutional changes are being observed in many parts of the world. In Russia and in Eastern and Central Europe, democratization, marketization, and privatization have taken place at an unexpected speed. There is a drive toward democratization, or at least universal suffrage, in Africa. Democratization in Chile at DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 19 the end of the military regime has come with important and successful economic reforms. Land titling in some Sub-Saharan African countries or affirmative action against the caste institution in India are no less important institutional changes toward establishing and protecting private property rights or rebalancing the structure of political and economic power. All these institutional changes foreshadow dramatic changes in economic policies and development outcomes where they have occurred, and suggest that comparable changes are possible where they are needed for devel- opment. In the context of this presentation, the question is which factors triggered those reforms, and which factors are responsible for the persistence of inefficient institutions where such reforms did not occur. We briefly mention some examples in what follows. As an example of the persistence of institutions and the role of initial conditions, the seminal work by Engerman and Sokoloff (1997) that compares the evolution of institutions in North and South America is very instructive. The presence of economies of scale in the most important commercial products in the southern part of North America and in South America promoted the emergence of a rich class of large planters employing poor natives or imported slaves. In the northern part of North America, the absence of commercial crops with large economies of scale led on the contrary to a large number of smaller farms run by a relatively homogeneous immigrant population. The composition of elites in the two regions was thus funda- mentally different, and institutional changes proceeded at a different pace. A rela- tively equitable distribution of resources and a relatively homogeneous population in the North fostered a quick extension of suffrage to all social strata10 and the devel- opment of mass education. On the contrary, with a very unequal initial distribution of wealth and human capital, the Southern elite resisted all institutional changes that would have threatened its control of power, including mass education. Eventually, democratic mechanisms were introduced, but at a slower pace and with frequent interruptions. This in turn resulted in a much slower economic development. The sequence of reforms that finally led to the dismantlement of the Soviet Union and the dramatic political and economic changes in that part of the world give another example of elite-led institutional changes: in this case, of reform that was initially resisted by another part of the elite. According to major contributors to the analysis of the last years of the Soviet Union (see, for instance, Åslund 1995), the initial reason for reform was the realization by Soviet leaders that the economy was running inefficiently, and that this resulted in an increasing technological and mili- tary gap with the United States. A first set of reforms aimed at a partial liberalization of the domestic economy through the introduction of some market mechanisms. Yet this initiative by the Gorbachev government was met with hostility by part of the Nomenklatura, for whom this reform meant the disappearance of some of their rents. To overcome this obstacle, a second set of reforms was then launched with the aim of bypassing the apparatchiks by democratizing the choice of business managers. It is this introduction of democratic mechanisms that eventually led to the independ- ence of some Soviet republics and then to the dismantlement of the entire Soviet Union, as well as to the full liberalization and marketization of the economy. 20 | FRANÇOIS BOURGUIGNON The institutional reforms launched in China at the end of the 1970s were also elite-led. As in the case of the Soviet Union, they aimed at increasing economic effi- ciency. Interestingly enough, however, they were not followed by political reforms of the same magnitude as in the Soviet Union. In effect, the general characteristics of the political regime stayed more or less unchanged at the same time that the economy settled on an extremely fast growth path. One important difference in the case of China is that there was no strong division in the elite about the reforms that progres- sively liberalized economic mechanisms. Since they kept some control over part of the economy, bureaucrats and apparatchiks were able to maintain their rents, and even to increase them both because of the general economic growth and also because growth was creating new niches for enrichment or rent-seeking. A hypothesis about institutional economic reforms in China and their success in an otherwise unchanged political environment would thus be that, overall, those reforms benefited the whole population. In other words, they were roughly Pareto-improving, something which may not have taken place in the Soviet Union because the Nomenklatura that opposed the Perestroika estimated, rightly or wrongly, that the net gains it could draw from it were negative. In the latter case, the liberalization of the economy even- tually occurred because of the second sequence of reforms in the political sphere. But this was obtained only after some political turmoil and a severe economic recession, and it is not yet clear whether these reforms have put the economy on a fast and sustained growth path. Governance and institutions have also been subject to important reforms in a number of African countries. As in the preceding cases, some reforms were a response by elites to the economic inefficiency inherited from the past--very often in a post-conflict context. For instance, the recent evolution in Ethiopia and Madagascar very much consisted in reestablishing market mechanisms, and thus shares similarities with the reforms in transition economies in Eastern Europe some 10 or 15 years ago. In countries like Mozambique, Rwanda, and Uganda, conflicts that generated a change in the governing elites made institutional reforms easier, in some sense, because of a common will to break with the preconflict func- tioning of the economy and the society. Similar stories may be told for some other Sub-Saharan African countries, although reforms did not always lead to a sus- tained acceleration of growth--far from it. In many other countries, ransacking elites continue to stifle growth and development, and conflicts among elites often escalate to involve non-elites. Reform in those countries is very slow, and in many is simply absent. A handful of countries like Botswana, Mauritius, and Senegal are known for the remarkable stability of their democratic institutions. Yet, unlike the remark- able economic success of the first two countries, GDP per capita regressed in Senegal for practically 30 years after independence. This suggests that other reforms necessary for development were not undertaken in the country, or have been undertaken only recently. In the same perspective, the acceleration in the move toward universal suffrage that followed the signing of the Cotonou Convention between the African, Caribbean, and Pacific Countries (ACP) and the European Union shows that some institutional changes can be imposed from outside, but DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 21 this is generally not sufficient to guarantee progress on other dimensions of democratization and in other institutions that condition development. Many other cases could be mentioned: the achievement of unilateral trade liberal- ization in Chile, for instance, or the deregulation and opening of the Indian economy in the 1990s. Who was perceived to win and who was perceived to lose because of these institutional changes, and which political economy factors facilitated them? What has been the actual effect of these policy changes? These are some of the fun- damental questions that the economic development literature needs to deal with more frequently. Conclusion We understand why institutions are important, and we have some rough evidence to illustrate this. Yet, too little is known at this stage about how institutions emerge and evolve. Conflicts do not always arise when institutional reforms are overdue. The literature suggests that when elites know that their costs in an impending conflict are greater than the costs of the reforms, they may adopt desirable changes. But they will not do so if the cost of a potential conflict or of the repression is low enough. The problem is that information about these costs and benefits is extremely imperfect. Practically, existing institutions correspond to some kind of political economy equi- librium and tend to persist, even when there is another more efficient equilibrium. Although weak institutions in some countries stymie development, external inter- vention to change institutions is difficult because it violates the principle of national sovereignty. At most, there is room only for general recommendations like in the Cotonou Convention. So what can be done, especially in international development institutions and agencies? Information and analysis may be their first contribution. Data collection should lead to improvements in institutional indicators so as to permit reforms and institu- tional changes to be monitored more closely. Analysis may help determine the win- ners and losers of a specific reform. It may happen that such an analysis will point to some Pareto-improving reforms that should be adopted after the information is being made available. In cases where there is no easy win-win, all members of a society need clear information on the direct and indirect effects of institutional change, how bene- fits may compensate for losses, and what commitment device may ensure that com- pensation. It may often be that reforms are not undertaken because such information is not available, or is possibly jammed by the advocates of the status quo. A second line of action may lie in the loose conditionality that may be exerted in low-income countries through the allocation of Official Development Assistance by the donors' community. In the Poverty Reduction Strategy (PRS) Process, donors and countries exchange views about policies for poverty reduction, and aid is made dependent over time on some assessment of the performances of recipient countries. Institutional changes are part of the PRS, and part of the performance assessment. The weakness of that line of external intervention, however, is that it requires the coordination of donors around the PRS and the performance assessment, and time 22 | FRANÇOIS BOURGUIGNON consistency of donors. Practically, those two conditions may prove difficult to meet. In Sub-Saharan Africa, the New Partnerships for Africa's Development initiative, which includes peer review of a country's governance institutions by other countries, may play a similar role by relying on "reputational" incentives rather than economic ones. Even within such a contractual framework, the experience of the World Bank in advising our partner-countries demonstrates that critical information is missing. If it were possible to identify with some precision those who gain and those who lose in a reform, as well as the size of the gains and losses--and possibly mechanisms that could ensure some compensation--my experience convinces me that very substantial progress would be made. Thank you very much. Notes 1. For a recent useful survey of this voluminous literature, see Durlauf, Johnson, and Temple (2005). 2. Other contributors include Mauro (1995) and Hall and Jones (1999). 3. North (1990) defines institutions as "the rules of the game in a society, or, more formally, the humanly devised constraints that shape human interaction." 4. Data from 76 countries and 51,000 firms are available at http://www.enterprisesurveys. org/. 5. Kaufmann, Kraay, and Mastruzzi (2005). See also the related World Bank Web site: http://www.worldbank.org/wbi/governance/pubs/govmatters4.html. 6. World Bank (2005a). "Country Policy and Institutional Assessments: 2005 Assessment Questionnaire." Available at http://siteresources.worldbank.org/IDA/Resources/CPIA2005 Questionnaire.pdf. Doing Business, an annual World Bank report, also includes publiciz- ing this kind of subjective indicator on governance institutions in many countries. 7. See Robinson and Acemoglu (2006); Acemoglu, Ticchi, and Vindigni (2007). 8. For other tests of this nature, see Acemoglu, Johnson, and Robinson (2005). 9. On the dynamics of institutions, see Greif (2006). 10. By 1828, practically all states had universal suffrage for adult white males. References Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2005. "The Rise of Europe: Atlantic Trade, Institutional Change, and Economic Growth." American Economic Review 95 (June): 546­79. Acemoglu, Daron, James A. Robinson, and Simon Johnson. 2001. "The Colonial Origins of Comparative Development: An Empirical Investigation." American Economic Review 91 (December): 1369­1401. Acemoglu, Daron, David Ticchi, and Andrea Vindigni. 2007. "Emergence and Persistence of Inefficient States." Working Paper 12748. National Bureau of Economic Research, Cambridge, Mass. Åslund, Anders. 1995. How Russia Became a Market Economy. Washington, DC: Brookings Institution Press. DYNAMICS OF INSTITUTIONS, DEVELOPMENT, AND ELITES | 23 Barro, Robert J. 1991. "Economic Growth in a Cross Section of Countries." The Quarterly Journal of Economics 106 (2): 407­43. Durlauf, Steven N., Paul A. Johnson, and Jonathan R. W. Temple. 2005. "Growth Econo- metrics." In Handbook of Economic Growth, vol. 1, ed. Philippe Aghion and Steven Durlauf, 555­677. Elsevier. Engerman, Stanley L., and Kenneth L. Sokoloff. 1997. "Factor Endowments, Institutions, and Differential Paths of Growth among New World Economies: A View from Economic His- torians of the United States." In How Latin America Fell Behind: Essays on the Economic Histories of Brazil and Mexico, 1800­1914, ed. Stephen H. Haber. Stanford, Calif. Stan- ford University Press. Greif, Avner. 2006. Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. New York: Cambridge University Press. Hall, Robert E., and Charles I. Jones. 1999. "Why Do Some Countries Produce So Much More Output Per Worker Than Others?" The Quarterly Journal of Economics 114 (1): 83­116. Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2005. "Governance Matters IV: Governance Indicators for 1996­2004." Policy Research Working Paper 3630, World Bank, Washington, D.C. Knack, Stephen. Forthcoming. "Governance and Growth." In Proceedings of the Symposium on the 2006 KfW-University of Giessen Developing Countries Prize on Good Governance. Knack, Stephen, and Philip Keefer. 1995. "Institutions and Economic Performance: Cross- country Tests Using Alternative Institutional Measures." Economics and Politics 7 (3): 207­28. La Porta, Rafael, Florencio Lopez-de-Silano, Andrei Shleifer, and Robert Vishny. 1998. "The Quality of Government." Working Paper 6727, National Bureau of Economic Research, Cambridge, Mass. Mauro, Paulo. "Corruption and Growth." 1995. Quarterly Journal of Economics 110: 681­712. North, Douglass C. 1990. Institutions, Institutional Change, and Economic Performance. New York: Cambridge University Press. Robinson, James A., and Daron Acemoglu. 2006. "Persistence of Power, Elites, and Institu- tions." Discussion Paper 5603, Center for Economic Policy Research. http://ssrn.com/ abstract=913068. Rodrik, Dani, 1999. "Where Did All the Growth Go? External Shocks, Social Conflict, and Growth Collapses." Journal of Economic Growth 4 (4): 385­412. World Bank. 2005a. "Country Policy and Institutional Assessments: 2005 Assessment Ques- tionnaire." http://siteresources.worldbank.org/IDA/Resources/CPIA2005Questionnaire.pdf. ------. 2005b. World Development Report 2005­A Better Investment Climate for Everyone. http://go.worldbank.org/WVDAOSZJ20. Keynote Address The Eurasian Growth Paradox ANDERS ÅSLUND AND NAZGUL JENISH In the first decade of postcommunist transition, multiple growth regressions showed that the more radical and comprehensive the market economic reform, the earlier a country returned to economic growth and the more vigorous its growth--and Central Europe took the lead. Since 2000, however, annual growth in the Commonwealth of Independent States (CIS) countries has been more than 4 percentage points higher than in the Central European countries. A regression analysis for 20 postcommunist countries shows, with strong significance, that a reduction of public expenditures has most effect- ively stimulated economic growth. As expected, oil exports are also positive and signifi- cant. The distance from the European Union (EU) is also positive and significant: that is, the further a country is from the EU, the higher its economic growth. The effect of corruption is negative for growth but only marginally significant. Neither the laggard effect nor investment reveals any significant effect. The conclusion is that at least among postcommunist countries, more emphasis should be given to the need to reduce public expenditures to boost economic growth. Since the collapse of communism in 1989, economic output in different regions of the former socialist camp has developed in starkly contrasting fashions. Initially, output fell sharply all over. From 1992, however, Poland recorded growth, and then one country after the other followed, though Moldova and Ukraine remained in the doldrums as late as 1999. The lesson from 1989 to 1998 was that economic reform worked. The more radical and the earlier the economic reform efforts were, the sooner a country would return to economic growth and the greater the upturn would be. Central Europe and the Baltics shone, while the countries of the Commonwealth of the Independent States (CIS) underperformed badly. Yet even the growth rates of the leaders were mediocre. We shall discuss these lessons in detail later. Anders Åslund is senior fellow at the Peterson Institute for International Economics, and Nazgul Jenish is a doctoral candidate at the University of Maryland, College Park. The authors express their gratitude for valuable comments made primarily by participants at the ABCDE in St. Petersburg, notably François Bourguignon, Cheryl Gray, Shigeo Katsu, Hans Rausing, and Lucio Vinas de Souza. The authors benefited greatly from the excellent research assistance of Mathew Gibson. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / The World Bank 25 26 | ANDERS ÅSLUND AND NAZGUL JENISH FIGURE 1. GDP Growth Rates in the CIS-11, CE-4, and Baltic-3, 1998­2004 10 CE-4 9 Baltic-3 CIS-11 8 7 (%) tear 6 5 wthorg 4 GDP 3 2 1 0 1998 1999 2000 2001 2002 2003 2004 Year Source: World Bank World Development Indicators 2005; UNECE online statistics, http://www.unece.org/stats/data.htm. Strangely, everything was turned upside down from 1999 on. From 1999 to 2004, eleven CIS countries (CIS-11) had an average annual growth of 7.8 percent,1 while the four Central European Visegrad countries (CE-4, the Czech Republic, Hungary, Poland, and the Slovak Republic) recorded an average annual growth of only 3.6 percent. The three Baltic countries (Baltic-3, Estonia, Latvia, and Lithuania) came closer to the first group, with 7.1 percent growth, and Bulgaria and Romania came closer to the Central Europeans, with 5.4 percent (see figure 1). We limit our investigation to these twenty countries. How can this growth paradox be explained? Why did the pioneers of market reforms so quickly become the laggards in growth? This paper seeks to answer these questions. In the second and third sections, we investigate the facts and suggest variables that warrant further exploration. In the fourth section, we undertake a regression with the most interesting variables. Lessons from 1989 to 1998: Transition to a Market Economy Works When the transition to a market economy started, recorded output plummeted in all countries, though the Soviet economy had already been in free fall. In 1990, only Hungary and Poland launched their transitions. The sudden declines in their regis- tered production caused a shock, and their relative economic performance set the stage of the early debate. When other countries in Central and South-East Europe entered the transition in 1991, their output plummeted even more, but these falls were nothing in comparison with the CIS countries, several of which experienced real collapse. Not only were the declines in output huge, but they lasted for years. Poland took an early lead by returning to growth in 1992. By 1994, the whole of Central Europe and THE EURASIAN GROWTH PARADOX | 27 South-East Europe registered growth, and three of the most vigorous reformers in the former Soviet Union had also arrived at growth: Armenia, Latvia, and Lithuania. In 1995, they were followed by other reformers: namely, Estonia, Georgia, and Kyrgyzstan. However, several former Soviet republics experienced prolonged decline followed by stagnation, particularly Kazakhstan, Moldova, Russia, and Ukraine. Only at the end of the decade did they return to economic growth. The total fall in output was staggering. According to official statistics, the aggregate decline in GDP was 19 percent in Central Europe and 29 percent in South-East Europe. In the former Soviet Union, the collapse was truly stunning, reaching 44 percent in the Baltics and 53 percent in the CIS (UNECE 2000). No doubt, these figures are exaggerated. Perhaps half of the decline can be discarded as the result of statistical misrepresentation (Åslund 2002, chapter 4). The old system exaggerated output for the sake of fulfilling plan targets, while the new system stimu- lated underreporting for purposes of tax evasion. The inherited socialist statistical systems could not capture new decentralized enterprise development, and the under- ground economy mushroomed, especially in partially reformed countries (Berg 1994; Johnson, Kaufmann, and Shleifer 1997). Terms of trade or implicit trade subsidies changed sharply, and substantial Soviet subsidies to Central Asia were abolished (Orlowski 1993, 1995; Tarr 1994; Rosati 1995). Shortages soon disappeared, qual- ity improved greatly, and the structural changes were huge. Therefore, the statistical problems are substantial. In fact, many of the initial output declines have been reduced in later statistical revisions, which have succeeded in capturing more of real output. For our purposes, however, we have little choice but to use the official statistics, making the assumption that the later growth rates have been less dis- torted--though like most, we dismiss Turkmenistan's statistics as sheer fiction. Soon, a huge literature on the causes of output changes evolved.2 By and large, it concluded: the more radical and comprehensive was the market economic reform, the earlier a country returned to economic growth and the more vigorous its growth. The three foci of the transition were macroeconomic stabilization, deregulation, and privatization. As convinced reformers usually pursued all three aims in parallel, it is statistically difficult to disentangle these effects because of covariation. Almost all transition countries started out with high inflation, and output continued to fall until inflation had been brought under control. Stanley Fischer, Ratna Sahay, and Carlos A. Végh (1996b, p. 89) concluded: "The simple--but essential--message that emerges . . . is that real GDP rebounds following inflation stabilization, which in turn appears highly correlated with the improvement in the public finances." In a broader international regression, Michael Bruno and William Easterly (1998) found the critical threshold was relatively high: 40 percent inflation a year. In addition, Peter Christoffersen and Peter Doyle (2000, p. 439) stated: "There is no evidence that disinflation necessarily incurs significant output costs, even at moderate inflation rates." Moderate inflation did not impede growth significantly. Deregulation was the basis for the formation of a market economy, and over time regression analysis shows the rising importance of deregulation for growth (Berg and others 1999). Privatization was always more controversial, but the regressions that 28 | ANDERS ÅSLUND AND NAZGUL JENISH included the share of GDP arising from the private sector showed that privatization had a clear positive impact on growth (Berg and others 1999; De Melo, Denizer, and Gelb 1997; EBRD 1999). The standard causes of long-term economic growth (Barro and Sala-i-Martin 2004) were of little or no importance. Surprisingly, Andrei Illarionov showed that the investment ratio in GDP was negatively correlated with economic growth: that is, the less a country invested, the higher its growth (see figure 2).3 The explanation is probably that high investment reflected the maintenance of a soft budget con- straint, a large public sector, wasteful public investment, and outright theft. Human capital was ample and underemployed, so there is little reason even to investigate it. Overall technology, research, and development appeared similarly irrelevant. Sens- ibly, nobody paid much attention to these factors. The issue was rather how to utilize the existing physical capital and import foreign technology to ease bottlenecks (Åslund 2002). Apart from the transition indicators, growth was correlated with the expansion of exports. Imports took off slightly later. The countries that were about to join the European Union (EU) benefited from privileged access to the large EU market. As a result, the share of their exports to the EU of 15 members rose from half in 1989 to two-thirds in 2000. The CIS countries, by contrast, suffered from severe discrimina- tion by the EU, and the share of their exports to the EU stayed constant at around one-third (Åslund and Warner 2004). Covariation made it difficult to ascertain whether this was really a positive effect of market access or whether it was a result of the EU accession countries adopting many of the systemic features of the EU countries. A corollary was that the closer a country was to Brussels, the higher its economic growth. With regard to politics, the 1990s evidenced a strong positive correlation between democracy, comprehensive market reforms, and economic growth (Berg and others FIGURE 2. Investment Rate as a Proportion of GDP vs. GDP Growth Rate in Russia, 1993­2004 15 10 (%) ater 5 wthorg 0 GDP ­5 ual Ann ­10 ­15 0 5 10 15 20 25 30 Investment rate/GDP (%) Source: UNECE online statistics, http://www.unece.org/stats/data.htm. THE EURASIAN GROWTH PARADOX | 29 1999; EBRD 1999; Åslund 2002) because in the early transition, the threat against successful market reforms did not come from the many losers, but from the few winners who engaged in rampant rent-seeking (Hellman 1998). A corollary of the prior observations was that corruption was negatively correlated with economic growth (EBRD 1999). In conclusion, radical market reform, macroeconomic stabilization, privatization, EU accession, export expansion, democracy, and reasonable governance all went together. Analytically, one problem was that the covariance was overwhelming. Another problem was that the growth rates remained anemic, and only Poland had convincingly exceeded its economic level of 1989. A third problem was that the com- parative standards--the CIS countries--were performing truly miserably. Thus, although one decade had passed, we could not really say all that much about the causes of economic growth, apart from the obvious point that a critical mass of market eco- nomic elements was vital. In particular, Kazakhstan, Moldova, Russia, and Ukraine appeared stuck in an under-reform trap (Åslund, Boone, and Johnson 2001). A World of Opposites, 1999­2004: The Winners are the Prior Losers Strangely, whatever had been true until 1998 was false from that year onward. The starkest contrast evolved between the four Central European countries and the eleven CIS countries.4 The latter group grew more than twice as fast as the former year after year (see figure 1). This could not be explained by sheer chance. The dividing event was the Russian financial crash of August 1998, which had many repercussions for the whole CIS region. Several other countries underwent similar crises at approximately the same time, and the patterns were very similar from country to country. Most postcommunist countries maintained higher public expenditures than they could finance domestically for years. They had high tax rates, but they failed to collect much of the taxes. Instead, they ran up excessive foreign debts. Sooner or later they lost international creditworthiness as the inter- national financial institutions refused to provide more credits. On the verge of external default, or in default, CIS governments had no choice but to minimize their budget deficits. They could no longer borrow money abroad or from their population, and tax revenues could not be boosted in haste. Therefore, they were left with no choice but to cut expenditures severely. Several countries slashed their public expenditures by about one-tenth of their GDP in a year or two, often when their GDP was falling sharply. These cuts amounted to one-quarter or more of total public expenditures. Bulgaria cut its public expenditures as a share of GDP by 11 percent in 1997, Moldova by 10 percent from 1998 to 2000, Kyrgyzstan by 9 percent from 1995 to 1997, and Russia by 8 percent in 1999 (Åslund 2002; EBRD 2005). Such drastic cuts are very different from ordinary budget trimming. Governments fight desperately to avoid disaster, which means that budget politics change com- pletely. Whatever was politically impossible is suddenly accepted as economically vital. The big budget post that most transition countries cut drastically in crisis was 30 | ANDERS ÅSLUND AND NAZGUL JENISH enterprise subsidies, which often involved rent-seeking schemes, such as barter. As a consequence, enterprises' budget constraints were sharply hardened and their playing field became more level. Another effect was that many former state managers who had seized control over their old enterprises but did not know how to run them under capitalism were persuaded to sell them to new entrepreneurs in order not to lose everything. After public expenditures had been cut down to size in many countries, the multitude of taxes and their high rates made little sense, since a broad understanding developed that these taxes could not possibly be collected. Then, tax reforms introducing ever fewer taxes as well as lower and flatter taxes spread throughout the CIS and to the verges of the EU. As tax rates fell, tax administration could be simplified and tax collection improved. In hindsight, the Russian financial crash can be considered the crucial event that rendered the CIS countries full-fledged market economies. Their fiscal sys- tems were put in reasonable order, and ever since, inflation has been moderate. Most CIS countries derive at least 60 percent of GDP from their private sectors. Markets, albeit encumbered, drive their economies. A critical mass of market economy and private enterprise has been achieved, although the CIS countries continue to lag behind the EU accession countries, according to the EBRD transition indicators, which have changed little since 1998, recording only a light convergence (see figure 3). Their minimal movement amidst major structural changes suggests that these transition indicators might not be very relevant as a measurement of actual structural developments. The Russian financial crash of 1998 was followed by several other dramatic devel- opments. Russia devalued the ruble by three-quarters in 1998, and most other CIS countries subsequently devalued their currencies by about 50 percent, which benefited exporters. Soon afterward a commodity boom started, driven by Chinese imports of FIGURE 3. Composite Transition Index for CE-4 and CIS-9, 1990­2003 1.00 CE-4 0.90 CIS-9 0.80 x inde 0.70 0.60 ansitiontr 0.50 0.40 0.30 Composite 0.20 0.10 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Sources: De Melo, Denizer, and Gelb 1997; Havrylyshyn and Wolf 1999, p. 34; Åslund 2002; EBRD 2001, pp. 12, 14; EBRD 2002, p. 20; EBRD 2003, p. 16. THE EURASIAN GROWTH PARADOX | 31 commodities, allowing the CIS countries to boost their exports in spite of stagnant EU markets and EU protectionism. The expansion was driven by supply rather than demand, as evident from the failure of all forecasts based on demand to predict the CIS resurgence. Although only four of twelve CIS economies were significant energy exporters (Azerbaijan, Kazakhstan, Russia, and Turkmenistan), growth rates across the CIS were similarly strong. Commodity-poor Armenia has registered the highest growth rate. Nonetheless, to check the effect of major energy exports, we introduce a dummy for the three major energy exporters in our sample (Azerbaijan, Kazakhstan, and Russia). The export boom has been followed by increased investment, as would be expected, which has further reinforced the economic growth. Much of the growth in Azerbaijan and Kazakhstan has been spurred by extremely high foreign direct investment (FDI), which has been motivated by potential oil production. It is generally acknowledged that countries with a lower level of economic devel- opment, everything else being equal, grow faster than wealthier countries. Thus, one would expect the CIS countries to grow faster than the Central Europeans after they have caught up somewhat with regard to transition reforms. However, this "laggard effect" would hardly explain a difference of more than 1 to 2 percent annual growth between these two groups of countries (Åslund and Warner 2004). The laggard effect measured against GDP per capita in purchasing power parities (PPP) must be assessed. Clearly, additional factors are needed to explain a steady difference in economic growth of over 4 percentage points each year for half a decade. One possibility is that this is simply recovery growth, and that the main explanation is the huge unused capacity in many post-Soviet economies after an official decline in output of about half of GDP, as especially Yegor Gaidar (2005) has argued. However, striking systemic differences have developed in recent years. Most con- spicuously, the CIS countries have drastically cut their public expenditures to about one-fifth less as a share of GDP than in Central Europe, where it turned out to be possible to collect quite high taxes and international financing has remained accessible. This implies that economic freedom has increased in the CIS countries in a fashion that is not captured by the EBRD transition indicators. A simple plot of growth against government expenditure as a share of GDP points to a negative correlation between these variables (see figure 4). Although care should be exercised when interpreting this type of graph, more accurate regression analysis below confirms the strong negative association between growth and government spending. Thus, public expenditures appear a plausible explanation of the observed differences in economic growth between Central European and CIS economies. Similarly, the CIS has adopted a low-tax regime, while Central Europe has taken only limited steps in that direction as yet. Low and flat taxes are proliferating in the East, while most of Central Europe still has comparatively high and progressive income taxes. Russia has had a flat income tax of 13 percent since 2000, and Ukraine since 2004. Admittedly, the Slovak Republic chose a flat income tax of 19 percent in 2004 and Romania one of 16 percent in 2005, as tax competition stings, but Poland still has progressive taxes peaking at 40 percent. Corporate profit taxes are declining in 32 | ANDERS ÅSLUND AND NAZGUL JENISH FIGURE 4. Public Expenditure and Growth in CIS-11, CE-4, and Baltic-3, 1999­2004 16 14 12 10 8 (%) th worg 6 4 2 GDP 0 ­2 ­4 ­6 0 10 20 30 40 50 60 70 Public expenditure (% of GDP) Source: EBRD (various years). the whole region, but payroll taxes are being reduced much more in the CIS than in Central Europe. In addition, Central European countries have developed a habit of running budget deficits of about 6 percent of GDP, while the CIS countries have nearly balanced budgets, with an average budget deficit of barely 1 percent of GDP for the last half decade (EBRD 2004). The CIS countries also have de facto freer labor markets than the Central European countries. Unfortunately, no easy measure is at hand, and this is another aspect of economic freedom in the CIS countries that the EBRD transition indicators ignore. The same could be said about agricultural policies. While growth and democracy were nicely correlated in the 1990s, we see an opposite picture after 1998. A simple plotting of growth against a Freedom House democracy index suggests a negative correlation between these two indicators. The CIS countries, which are by and large authoritarian, have grown faster than the democratic countries in Central Europe. But what lies behind this? Have the Central Europeans just relaxed, while the CIS governments were shaken up by the Russian financial crash of 1998? The EU is probably part of the explanation. The first parts of the common legislation, the acquis communitaire, were undoubtedly useful, helping to build market institutions, while the last parts included new regulations such as the Common Agricultural Policy. It is also possible that the old idea of authoritarian advantage has some relevance when the main risk to economic development becomes popular pressures for regula- tion of labor markets in favor of insiders and excessive taxes on the rich to the benefit of social transfers for the majority. The dominant risk during the first decade was rent-seeking by elites, which was best checked by democracy. As before, neither human capital nor technology is likely to have had much impact on growth, as free resources have remained ample. THE EURASIAN GROWTH PARADOX | 33 Regression Analysis, 1999­2004 Specification To investigate more accurately the relative contribution of the major factors discussed above to the differences in growth between the CEE and CIS countries, we estimate the following panel data model: uti yti = + 1 GDP ti + 2GDPt G + 3 GDPti +1Oili +2Corrupi +3CISi + t +uti; I ­1,i = i + vti, t =1, K, T; i =1, K, N; where the dependent variable, yti, is annual GDP growth rate for country i in year t. There are six explanatory variables: 1. (G/GDP)ti, government expenditure as a share of GDP 2. GDPt­ , lagged per capita GDP (in logs), to control for the "catch-up" effect 1,i 3. (I/GDP)ti, fixed investment as a share of GDP, a measure of the physical capital 4. Oil, oil-producing country dummy, to account for the effect of surging energy exports 5. Corrupi, a corruption index, a proxy for the quality of the political institutions 6. CIS, regional dummy. The CIS dummy serves as a proxy for the distance from the European Union and stands in for other time-invariant, structural factors that differ between the CEE and CIS regions, such as labor market regulation. We control for common shocks reflecting global and regional economic conditions by including fixed-year effects, t. The error term is composed of two parts: µi, an unobserved individual effect; and vti, an idio- syncratic component. Unlike the standard growth literature, our regression does not include any meas- ures of human capital, since all postcommunist countries enjoy relatively high levels of education, which do not vary considerably across countries and over the sample period, and hence cannot explain the observed variation in growth. Nor does it con- tain specific labor market indicators, since the data, even when they are available, are plagued with severe measurement errors. Moreover, as it is now recognized in the econometric literature, simply increasing the number of right-hand-side variables in growth regressions is unlikely to take away the omitted variables bias problem. Therefore, we do not here strive for the maximum generality and completeness of explanatory variables but rather focus on a few principal variables and robust ways of evaluating their impact. The error-component specification is well-suited for this purpose. It allows us to exploit variation both across countries and over time, as well as to reduce the omit- ted variables bias. For instance, any differences in human capital across countries are captured by country-specific effects. 34 | ANDERS ÅSLUND AND NAZGUL JENISH Data Our sample consists of observations for twenty transition economies--eleven CIS countries, three Baltic states, and six Central European and South-East European countries--over the period from 1999 to 2004. The GDP per capita (PPP) data comes from the World Bank's 2005 World Development Indicators (WDI) report (World Bank 2005). The annual GDP growth rate and government expenditure data are drawn from the 2004 EBRD Transition Report. Investment figures are from the UNECE databases. Finally, we use the corruption perception index scores constructed by Transparency International (2006). Higher scores correspond to lower levels of perceived corruption. The indicator displays little variation over time, and therefore the period averages, rather than individual year estimates, enter the regression. All the data are expressed in terms of ratios obviating the need to control for population and country sizes. Estimation Procedure There are two major possible sources of estimation bias. The first is that lagged GDP per capita, investment, and government spending are likely to be correlated with the unobserved individual effects. The other is the potential endogeneity of investment and government spending--that is, investment and government spending--may be correlated with the contemporaneous idiosyncratic error term. We first estimate the model with the fixed-effects procedure, commonly used in panel data regressions. This technique is robust to the presence of correlation between regressors and unobserved individual effects, as it removes the country- specific effects by subtracting time averages before applying the OLS procedure. However, it does not take care of the second problem. Another shortcoming of the method is that it cannot consistently estimate coefficients on time-invariant regressors such as Oil, CIS, and Corruption. Nevertheless, it provides a useful benchmark for the time-varying regressors. An alternative strategy that addresses both estimation concerns is to difference the regression and then estimate jointly the transformed equation and the equation in levels with the two-step efficient general method of moments (GMM) procedure pro- posed by Arellano and Bover (1995). In this procedure, investment and government spending are instrumented with their second lags. The GMM estimator is consistent, asymptotically normal, and invariant to the choice of transformation. Results Overall, the regression results support our main predictions. Government spending and energy exports are the key to explaining the differences in growth in the transi- tion countries. Economic regulation and corruption seem to have moderate impact, while the laggard effect and investment seem to play a negligible role. Table 1 summarizes the findings. Column 1 reports the fixed-effects method esti- mates. Column 2 presents the GMM estimates. We carried out a series of sensitivity checks using different right-hand-side variables. One of them incorporates potential THE EURASIAN GROWTH PARADOX | 35 TABLE 1. Regression for GDP Growth Rate General method of moments (GMM) Fixed effects No spillovers Spillovers Explanatory variable (1) (2) (3) Constant n.a. 0.089 0.084 (0.065) (0.133) G/GDP ­0.229* ­0.136* ­0.138* (0.087) (0.044) (0.058) Lagged GDP per capita ­0.165 ­0.009 ­0.007 (0.112) (0.017) (0.017) I/GDP 0.075 0.047 0.042 (0.068) (0.057) (0.057) Oil n.a. 0.024* 0.025* (0.012) (0.012) Corruption n.a. 0.008 0.008 (0.005) (0.005) CIS n.a. 0.015 0.015 (0.015) (0.015) Year 2000 ­0.010 ­0.004 ­0.004 (0.006) (0.004) (0.004) Year 2001 0.000 0.002 0.002 (0.005) (0.004) (0.003) Year 2002 0.003 0.001 0.001 (0.005) (0.005) (0.004) Year 2003 0.019* 0.014* 0.014* (0.006) (0.004) (0.004) Year 2004 0.029* 0.019* 0.019* (0.008) (0.005) (0.005) Spillovers 0.001 (0.034) R --2 0.69 0.51 0.51 Source: Authors' calculations. Note: Robust standard errors are shown in parentheses. Sample size is 120. n.a. = not applicable. * Significant at 5 percent. spillovers from neighboring countries. We re-ran the GMM regression by adding the weighted average of the log per capita GDP for a country's neighbors and big trading partners. The results are shown in column 3. Throughout all the regressions, the coefficient on the government expenditures is negative and strongly significant. It is significant at 1 percent level in the fixed effects and GMM regressions, shown in columns 1 and 2, and is significant at 5 percent in the regression with spillovers. Not only does it have the predicted sign but it has also the largest effect, in terms of magnitude, among all the variables entering the regression. The estimate implies that a reduction of 1 percent of GDP in government spending, everything else being equal, gives rise to about a 0.14 percent increase in the GDP growth rate. 36 | ANDERS ÅSLUND AND NAZGUL JENISH As anticipated, energy exports boost growth. The coefficient on Oil is positive and significant at 5 percent level. The distance from the EU seems also to have a positive effect on growth, as suggested by a positive coefficient on CIS. The coefficient on corruption, which proxies for the quality of institutions, is pos- itive and marginally significant, indicating that low levels of corruption are associated with higher growth. This is consistent with the preceding finding about a negative correlation between government spending and growth, since high government spend- ing, as a rule, goes hand in hand with high corruption. The laggard effect does not appear to be a major factor accounting for the big gap in the growth rates. The coefficient on lagged GDP per capita has a negative sign, but it is not statistically significant. Nor do these regressions reveal a significant relationship between investment and growth, though the estimated coefficient is positive. One possible explanation is that most postcommunist countries started transition with high initial levels of physical capital. Therefore, the marginal effect of additional investment is small. Another explanation is that it takes time for improvements in investment to translate into growth, and the time series is simply too short to detect any stable relationship between the two variables. However, some previous studies using longer series for a larger country sample also obtained insignificant estimates (see Barro and Sala-i- Martin 2004). The effect of spillovers from neighboring economies turns out to be statistically insignificant. The weights are constructed based on geographic distances between countries. Though not perfect, this weighting system accounts reasonably well for eco- nomic linkages between countries, such as trade and the costs of transporting goods. Most of the CIS countries, especially the oil-exporting states, enjoy considerably lower internal energy prices than the CEE economies. The sizeable energy price dif- ferentials across the two regions are therefore deemed by some analysts to be the leading explanation of the observed growth differences. To test this hypothesis, we reran the same regressions with an additional explanatory variable: internal gaso- line prices in each of the sample countries. Gasoline prices serve as a proxy for domestic energy prices. Although we do not report here the estimates for those regressions because of some data problems, the preliminary estimates, nevertheless, suggest that our main results are robust to the inclusion of energy prices. It affects neither the sign nor the significance of the coefficients. Moreover, the magnitudes of the coefficients on the major explanatory variables such as government expenditure, oil, corruption, and lagged GDP do not change or change only negligibly. Thus the lower internal energy prices do not seem to be a major source of vigorous growth in the CIS region. A substantial literature on economic growth and the size of the state exists, but it contains no agreement. La Porta and others (1999) showed with empirical material from 200 countries that bigger government is usually better, but such a regression does not say anything about causality. The Scandinavian countries had very small and efficient states in the1930s, which were therefore allowed to grow, for instance. More- over, corruption takes a long time both to develop and to dwindle (Treisman 2000). THE EURASIAN GROWTH PARADOX | 37 The postcomunist region offers a particular starting position of states that are both large and highly corrupt. Our regression suggests that with those initial conditions, a sharp reduction in public expenditures is the best way of boosting economic growth. Naturally, it would be desirable to reduce corruption swiftly, but knowledge and capability of how to do so are very limited. Yet we do know that corruption usually falls with rising income. It should be emphasized that the postcommunist state was no average state but extreme in most regards. First of all, by any measurement it was much larger than the state in other countries at that level of development, whether measured in terms of taxation, public distribution, degree of regulation, or share property owned by the state. As a natural consequence, it was less subject to checks and balances than in most other states and it was severely overstretched. Second, the postcommunist state was pretty parasitical. It did the wrong things, hindering economic development rather than promoting it, while anti-socially redistributing from the poor to the rent- seeking elites (Milanovic 1998; Hellman 1998). Third, the postcommunist state was ineffective and inefficient because of a high degree of corruption in comparison with other states (Transparency International 2006). Thus, regardless of what one may think of the role of the state in general, in postcommunist countries it would be rather surprising if economic growth would not be boosted by a reduction of the size of the state by any measurement. Some of the transition indicators measure the role of the state in regulation and ownership. We suggest adding the redistributive func- tion of the state. Conclusions The main conclusion arising from our analysis of economic growth in the postcom- munist region since 1999 is that the sharp rise in the growth rate in the CIS countries can mainly be explained by a drastic reduction in public spending and budget deficits in these countries. A second explanation is unsurprisingly that growth has been boosted by the commodity boom on world markets. Contrary to common views, the impact of the laggard effect is not conclusive from our regression. The coefficient on the lagged per capita GDP is negative, but statisti- cally insignificant. Growth in the CIS countries is tempered by higher corruption than in Central Europe, which is also born out by the regressions, though it is only marginally significant. Greater distance from Brussels also seems to have a positive effect on growth. In effect, the CIS countries have adopted the highly successful East Asian growth model--lock, stock, and barrel--while the less dynamic Central European countries have adopted the EU model, which has not been conducive to high economic growth, even if some countries--mainly Ireland, the three Baltic countries, and the Slovak Republic--have managed to go against the current. That one model is generally supe- rior does not mean that all its parts are superior. Depressingly, the CIS countries that have generated impressive growth are largely authoritarian. 38 | ANDERS ÅSLUND AND NAZGUL JENISH International institutions designed to promote growth in the postcommunist world, notably the World Bank and the European Bank for Reconstruction and Development (EBRD), need to incorporate these insights in their advice. For years, the EBRD has shown how Central Europe has scaled its transition indicators, but it fails to explain why Central Europe has only achieved a growth rate of 3­4 percent in recent years. By contrast, Janos Kornai (1992) noticed that the Central European states had devel- oped a premature Western European social welfare system. This has turned out to be a social welfare trap with Western European tax rates, social transfers, and labor mar- ket regulations. These countries' membership in the EU has reinforced these negative features and reduced their inclination to reform. Meanwhile, they are ignoring the Maastricht restriction that is supposed to limit budget deficits to 3 percent of GDP, maintaining steady budget deficits on the order of 6 percent of GDP. The obvious conclusion is that high public expenditures and taxes are bad for eco- nomic growth--at least in the postcommunist countries, saddled with excessively large governments of poor quality. Unsurprisingly, liberal economic policy or greater economic freedom does promote economic growth. Consequently, international financial institutions should advocate cuts in public expenditures in postcommunist countries with poor growth. Notes 1. Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan, Ukraine, and Uzbekistan. Turkmenistan is excluded because of its extremely poor and unreliable statistics. All averages used here are unweighted because we are inter- ested in the comparative performance of the different countries. If weighted averages were used, we would be preoccupied with the relative performance of Poland versus Russia, since these economies dominate in their respective regions. 2. See Åslund, Boone, and Johnson (1996); Berg (1994); Berg and others (1999); Christof- fersen and Doyle (2000); De Melo and Gelb (1996); De Melo, Denizer, and Gelb (1997a), De Melo, Denizer, Gelb, and Tenev (1997b); EBRD (1999); Fischer and Sahay (2000); Fischer, Sahay, and Végh (1996a, 1996b, 1997); Havrylyshyn and Wolf (1999); Popov (2000); Sachs (1996); Selowsky and Martin (1996). 3. Illarionov presented his findings in a lecture at the Higher School of Economics on April 3, 2002 and many times elsewhere. 4. We ignore Turkmenistan because of its substandard statistics. References Arellano, Manuel, and Olympia Bover. 1995. "Another Look at the Instrumental Variables Estimation of Error-Components Models." Journal of Econometrics 68: 30­51. Åslund, Anders. 2002. Building Capitalism: The Transformation of the Former Soviet Bloc. New York: Cambridge University Press. Åslund, Anders, and Andrew Warner. 2004. "The EU Enlargement: Consequences for the CIS Countries." In Beyond Transition: Development Perspectives and Dilemmas, ed. Marek Dabrowski, Ben Slay, and Jaroslaw Neneman, 231­52. Aldershot: Ashgate. THE EURASIAN GROWTH PARADOX | 39 Åslund, Anders, Peter Boone, and Simon Johnson. 1996. "How to Stabilize: Lessons from Post-Communist Countries." Brookings Papers on Economic Activity 26 (1): 217­313. ------. 2001. "Escaping the Under-Reform Trap." International Monetary Fund Staff Papers, Special Issue. Barro, Robert J., and Xavier Sala-i-Martin. 2004. Economic Growth, 2nd ed. Cambridge, Mass.: MIT Press. Berg, Andrew. 1994. "Supply and Demand Factors in the Output Decline in East and Central Europe." Empirica 21 (1): 3­36. Berg, Andrew, Eduardo Borensztein, Ratna Sahay, and Jeronim Zettelmeyer. 1999. "The Evolution of Output in Transition Economies: Explaining the Differences." Working Paper WP/99/73, International Monetary Fund, Washington, DC. Bruno, Michael, and William Easterly. 1998. "Inflation Crises and Long-Run Growth." Journal of Monetary Economics 41: 3­26. Christoffersen, Peter, and Peter Doyle. 2000. "From Inflation to Growth: Eight Years of Transition." Economics of Transition 8 (2): 421­51. De Melo, Martha, and Alan Gelb. 1996. "A Comparative Analysis of Twenty-eight Transition Economies in Europe and Asia." Post-Soviet Geography and Economics 37 (5): 265­85. De Melo, Martha, Cevdet Denizer, and Alan Gelb. 1997. "From Plan to Market: Patterns of Transition." In Macroeconomic Stabilization in Transition Economies, ed. Mario I. Blejer and Marko Skreb, 17­72. New York: Cambridge University Press. De Melo, Martha, Cevdet Denizer, Alan Gelb, and Stoyan Tenev. 1997. "Circumstance and Choice: The Role of Initial Conditions and Policies in Transition Economies." Policy Research Working Paper 1866, World Bank, Washington, DC. EBRD (European Bank for Reconstruction and Development). 1999. Transition Report 1999. London: EBRD. ------. 2001. Transition Report 2001. London: EBRD. ------. 2002. Transition Report 2002. London: EBRD. ------. 2003. Transition Report 2003. London: EBRD. ------. 2004. Transition Report 2004. London: EBRD. ------. 2005. Transition Report 2005. London: EBRD. Fischer, Stanley, and Ratna Sahay. 2000. "The Transition Economies after Ten Years." Working Paper 30, International Monetary Fund, Washington, DC. Fischer, Stanley, Ratna Sahay, and Carlos A. Végh. 1996a. "Stabilization and Growth in Transition Economies: The Early Experience." Journal of Economic Perspectives 10 (2): 45­66. ------. 1996b. "Economies in Transition: The Beginnings of Growth." American Economic Review 86 (2): 229­33. ------. 1997. "From Transition to Market: Evidence and Growth Prospects." In Lessons from the Economic Transition: Central and Eastern Europe in the 1990s, ed. Salvatore Zecchini, 79­101. Norwell, Mass.: Kluwer. Gaidar, Yegor T. 2005. Dolgoe vremya (The Long Term). Moscow: Delo. Havrylyshyn, Oleh, and Thomas Wolf. 1999, "Growth in Transition Countries. 1991­1998. The Main Lessons." Paper presented at the conference, "A Decade of Transition," Inter- national Monetary Fund, Washington, DC, February 1­3. Hellman, Joel S. 1998. "Winners Take All: The Politics of Partial Reform in Postcommunist Transitions." World Politics 50: 203­34. 40 | ANDERS ÅSLUND AND NAZGUL JENISH Johnson, Simon, Daniel Kaufmann, and Andrei Shleifer. 1997. "The Unofficial Economy in Transition." Brookings Papers on Economic Activity 27 (2): 159­239. Kornai, Janos. 1992. "The Postsocialist Transition and the State: Reflections in Light of Hungarian Fiscal Problems." American Economic Review 82 (2): 1­21. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny. 1999. "The Quality of Government." Journal of Law, Economics and Organization 15 (1): 222­79. Milanovic, Branko. 1998. Income, Inequality, and Poverty during the Transition from Planned to Market Economy. Washington, D.C.: World Bank. Orlowski, Lucjan. 1993. "Indirect Transfers in Trade among Former Soviet Union Republics: Sources, Patterns and Policy Responses in the Post-Soviet Period." Europe-Asia Studies 45 (6): 1001­24. ------. 1995. "Direct Transfers between the Former Soviet Union Central Budget and the Republics: Past Evidence and Current Implications." Economic Planning 28 (1): 59­73. Popov, Vladimir. 2000. "Shock Therapy versus Gradualism: The End of the Debate." Com- parative Economic Studies 42 (1): 1­57. Rosati, Dariusz K. 1995. "The Impact of the Soviet Trade Shock on Central and East European Economies." In Output Decline in Eastern Europe: Unavoidable, External Influence or Homemade? ed. Robert Holzmann, G. Winckler, and János Gács, 131­59. Boston: Kluwer. Sachs, Jeffrey D. 1996. "The Transition at Mid Decade." American Economic Association, Papers and Proceedings 86 (2): 128­33. Selowsky, Marcelo, and Ricardo Martin. 1996. "Policy Performance and Output Growth in the Transition Economies." American Economic Review 87 (2): 349­53. Tarr, David G. 1994. "The Terms-of-Trade Effects of Moving to World Prices on Countries of the Former Soviet Union." Journal of Comparative Economics 18 (1): 1 ­24. Transparency International. 2006. "Corruption Perceptions Index." http://transparency.org/ policy_research/surveys_indices/cpi/2006. Treisman, Daniel S. 2000. "The Causes of Corruption: A Cross-National Study." Journal of Public Economics 76: 399­457. UNECE (United Nations Economic Commission for Europe). 2000. Economic Survey of Europe, No. 1. New York: United Nations. World Bank. 2005. World Development Indicators. CD-ROM. Washington, DC: World Bank. Keynote Address The Transition of Economics: The Cases of Israel and Russia GUR OFER The announced topic for this address is "Transition in Economic Thinking." The "advanced search" engine of Google provides the client with the following options: With all the words With the exact phrase With at least one of the words Without the words Originally I thought that there was a mistake and that Boris Pleskovic wanted me to talk about "thinking about economics in transition." I then considered talking about "economic transition in thinking" or "economic thinking in transition." My final choice fell on "The Transition of Economics: The Cases of Israel and of Russia." This is the option of "some of the words but a few additional ones." In a paper on the transfer of modern economics to Russia via the New Economic School (NES) that I presented at a recent conference in Budapest (Ofer 2007), I eluded to what I considered to be a somewhat similar case--for me, certainly a model to follow--when I found myself deeply engaged in the NES project. It so happened that in December 2005 the economics community in Israel gathered to remember Don Patinkin and to mark the tenth anniversary of his death. Among others, people discussed his role, how he single-handedly established modern eco- nomics in Israel. It so happened that Patinkin came to teach at NES in its first year and then came back for another two; he died just a few months later in 1995. NES (or in Russian, RESH, the Russian Economic School), a graduate school of economics, was established in Moscow in 1991 by a group of economic professors from Russia and the West. Its goal was to introduce modern economics to Russia, to train academic and professional economists along the teachings of Western economics, in order to meet the requirement of the transition to a market economy. NES opened its gates in the fall of 1992 and since then has graduated more than 400 economists in a two-year masters--indeed, graduate--program, along a curriculum Gur Ofer is emeritus professor of economics, The Hebrew University of Jerusalem. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / The World Bank 41 42 | GUR OFER similar to that in the leading U.S. universities (but without PhD work). Teaching during the first years was performed almost entirely by visiting professors from the West, including from the Hebrew University of Jerusalem, but an indigenous faculty was gradually developed, made from Russians who came back with PhD degrees from the West, among them graduates of NES, and a small number of domestic professors who trained themselves in modern economics. Today NES has a domestic faculty of 15 who run the school and conduct most of the teaching. NES is run as a Western university, with a tenure track appointment system based on publications in leading Western journals. Policy-oriented work on Russia and on transition is concentrated mainly in CEFIR, the Center of Economics and Financial Research, a think tank at NES.1 A number of papers at the end of the last century described and analyzed the spread of modern economics in a great number of countries.2 Ofer (2007) mentions some similarities and differences between these experiences and the NES project. The Israeli experience, which in many ways seems to be more similar to that of NES, was not included in the above-mentioned volumes;3 therefore I thought it appropri- ate to present it today. The similarities and differences between the two experiences-- one already completed, and with great success, and the other, that of Russia and NES, on its second stretch, following a successful while challenging first stage of mountain climbing--justify in my view spending a few minutes on looking into the Israeli experience in greater detail while keeping an eye on the Russian experience so far. The memory of Don Patinkin and the more relaxed atmosphere around lunch tables add justification to this choice. Let me start then with the story of the transfer of modern economics into Israel. The Introduction and Development of Modern Economics in Israel4 The Hebrew University was established in 1925, mostly by a group of professors who came to Palestine from East and Central Europe (Germany), but also a few from the United States. They adopted the continental model of higher education. Albert Einstein was the keynote speaker at the opening ceremony on Mount Scopus. The university did not have a faculty of social sciences until after World War II, let alone a department of economics: this, despite efforts to establish one since the 1930s. A few courses in social sciences, including statistics, the economics of the Middle East, and history of economic thought, mostly in a Marxian vein, were taught under the umbrella of the faculty of humanities. A number of candidates to fill the gap in "theoretical economics" were considered over the years both from Western Europe and from the United States, among them Abba Lerner. The highly respectable Senate of the Hebrew University, with figures like Martin Buber, was faced with a difficult choice. Established scholars refused to commit (you can imagine the poor state of Palestine and the Hebrew University during that period, as well as the very low level of the salaries offered, not to speak about tenure) and young scholars presented a great risk. Following much hesitation, a position was offered to a fresh PhD from Chicago, in his twenties, with strong Jewish and Zionist attachments. His name was Don Patinkin, THE TRANSITION OF ECONOMICS: THE CASES OF ISRAEL AND RUSSIA | 43 unknown in the field, but "very promising" according to a letter by Jacob Marschak, who included in his list of potential candidates Abba Lerner and even Paul Samuelson. In addition to hesitations on the basis of his young age, the Senate was also worried about the fact that the work of Patinikin was highly mathematical, and thus unfit for the needs of an undeveloped country. The Senate finally assured itself that "since he is young he can still be flexible enough to accommodate the needs of the university (defined as "political economy of Palestine and the Middle East")." Patinkin came in 1949, at the age of 27 and initiated the development of a new curriculum and department. He acted in three main directions. First, he assembled a group of current students to be sent for PhD studies in the West (the United States and England) with the hope that they would come back and form the new depart- ment. Second, he developed and wrote down lecture notes for new courses, the most important of which was the now legendary Ec. 1 (introduction to econom- ics), following the emerging "Chicago tradition." Until the new faculty came, he taught almost single-handedly both undergraduate and graduate courses. Finally, he took steps to establish an infrastructure for the study of and research on the Israeli economy. Patinkin's economics, brought from Chicago, was the analytical positivistic version of "neoclassical economics" and it came to replace both the old continental version of "institutional economics" as well as the "political economy" concentrated mostly around the teaching of Marx. It is significant that the Hebrew University made simultaneous efforts during the early 1950s to establish a chair for "labor and the cooperation movement" following socialist teaching. Indeed, the topic was taught for a few years. It has to be remembered that the political leadership during the first three decades of Israel was in the hands of labor parties, among them Marxist parties with strong connections to the Soviet Union, and that the economic system was highly interventionist, even "Etatist," and it supported a high level of equality and of redistribution (not much to redistribute then). The notion of "economic effi- ciency" that was divorced in principle from considerations of the redistribution of income and wealth and that could be shown to work in conditions of "perfect competition" and "free markets" was clearly novel and challenging (see more on these below). No less challenging and novel were the concepts of "shortage," "alter- native costs," "marginal costs," "utility," and many others. The Chicago school and indeed Patinkin himself devoted much effort to demonstrating that markets can eliminate unemployment without government intervention, as claimed and articu- lated by Keynes. The method of teaching was by solving problem sets by students ahead of the discussion in class. At least under Patinkin, most of the time in class was devoted to going through the solutions. This was in contradiction to the continental method of reading lectures by the professors and reciting them for the exams. To this day, there are assignments of weekly problem sets in most courses. To be counted, homework had to be handed in by students before classes. A story tells of a student who was late and went to Patinkin's home to deliver his problem set before class. When a youngster opened the door, our student asked him if he would be so kind as to hand the assignment to his father. Needless to say, the youth was Don Patinkin himself. 44 | GUR OFER A side but important issue was that of the language of teaching. None of the above terms, and hundreds of others, existed in biblical Hebrew and they had to be invented. Teaching in English was out of the question for the old-young nation that had just established its national independence in the land of Israel. A few years later a major publishing house rushed to translate Samuelson's text and it had to be scrapped, as nobody could understand the meaning of the new invented terms. It had to be retranslated using the language developed slowly in Patinkin's classrooms. (He himself struggled hard to express himself in Hebrew.) While most of the readings, Samuelson excepted, were in English, English literacy of the students has remained a problem to this day. The level of English teaching in high schools had been mixed at best, and many students graduated, even some with a master's degree, with no ability to continue to read economics in English. As the English language is the language of economics no less than mathematics, this state is a formula for a fast depletion of the human capital of professional economists.5 The American pattern of combining teaching and research, imported from nine- teenth century Europe and perfected later, was also transferred to Israel. Research was pursued in two directions: general research in economics, directed at the global community of economists via Western refereed journals, sabbatical leaves, confer- ences, and the like; and research, much of it policy-oriented, on the main issues of the Israeli economy. Patinkin provided a personal example for both. On top of the heavy burden of teaching a number of courses and running the department, he continued to produce first-rate theoretical research, which culminated in the mid 1950s with the publication of his Money, Interest and Prices (1965), a work that came close to getting him the Nobel Prize. Since most of the young faculty members who joined the department during the 1950s completed their PhD work in the West, they also were trained and motivated to publish abroad. Work on the Israeli economy was considered by Patinkin to be no less, and possibly even more important: especially the application of modern economics to the grave problems of the Israeli economy at that stage. While some infrastructure in terms of data was inherited from the pre-state period and the British Mandate, a systematic system of national accounting and related data had to be built from nearly scratch--let alone dealing with the economic consequences of the mass migra- tion, reconstruction, and economic development. Patinkin initiated the establishment of a research center devoted to the Israeli economy, invited experts from abroad to set it out, including Simon Kuznets, and encouraged members of the department and graduate students to join in and devote time to it. He himself, a theoretical macro- economist (remember, with mathematical inclinations), left for a year of study at Johns Hopkins University (with Kuznets) to train himself in economic development, national accounting, and the like. Upon returning, he became the academic head of the Falk Institute and devoted much time to study, write, and direct research on the Israeli economy. Following a number of basic studies on the Israeli economy by himself and other young faculty members, Patinkin published the now classic study The Israeli Economy: The First Decade (1959). A Hebrew language journal, The Economic Quarterly, was established in 1953 as the main arena for works on the THE TRANSITION OF ECONOMICS: THE CASES OF ISRAEL AND RUSSIA | 45 Israeli economy. While faculty members contributed papers, despite what was said above, they were seldom considered for promotion purposes. The above raises the issue of the appropriate balance between outside and inside orientation of the academic work. In principle, openness and the participation in the global academic research and interaction is the necessary condition for productive research and policy work on the Israeli economy. The question is about the balance. Early on there was a discussion on whether or not to establish a full-blown Israeli academic journal in English for refereed academic papers. It was decided not to follow this path, in order to encourage faculty members to publish abroad. A related preference, relaxed only partially in later years, was to send the best students for PhD work abroad. All these decisions resulted in less academic work on the Israeli economy over the years by faculty members, which drew significant criticism. This, however, was assumed to be compensated, at least partially, by the high level of general academic work and the ability of faculty members to provide and disseminate the theoretical foundations and developments in economics in general to the expand- ing community of professional economists working on the Israeli economy across the country. Later on, however, the partial dichotomy between theoretical economics and policy-oriented research on the Israeli economy was considerably bridged over when the experience in Israel in the areas of macroeconomics, indexing and living with and control of inflation, as well as with economic growth and migration, became the top- ics of high-level academic work published in first-rate journals. In addition, leading academic economists served all along as chairs and members of key public commis- sions that initiated important reforms in the Israeli economy, most notably the stabi- lization plan of 1985 that ended a period of dozen years of high inflation and no growth, and steered the Israeli economy back to a path of economic growth. The development of the economics faculty at the Hebrew University, the only one in Israel in the 1950s, was rather slow. By the early 1960s, the faculty had grown to about ten members, mostly upon coming back from studies abroad--and indeed, most did come back. Growth was much faster during the 1960s, when the Hebrew University was also involved in the establishment of additional departments of eco- nomics in other universities. Per contra, the number of graduates increased faster and they assumed mid-level positions in the government, the ministry of finance, the cen- tral bank, the Central Statistical Office, other government offices, and public and private institutions. They were known as "Patinkin boys" and they improved the level of economic analysis and work at mid levels, though not immediately the major aspects of economic policy (see more on this below). Quantitatively, there are today seven university-level departments of economics, including many research and policy centers devoted to economic theory, as well as to the Israeli economy, a similar number of business schools, with thousands of students, tens of thousands of graduates occupying almost all the positions for economists, including at the top of government, public, and private institutions. Furthermore, many of Patinkin's descendants, boys, grandchildren, etc. . . conduct research and perform high-level policy work on all aspects of the Israeli economy in a great number of academic and nonacademic institutions, notably the research 46 | GUR OFER department of the Bank of Israel, at all major banks, government ministries, and private organizations. As mentioned, the initial tension or dichotomy between out- ward- and inward-looking economic work has been largely eased and bridged over. Qualitatively, economics in Israel excels. It ranks high in terms of publication in top international journals and in the level of training of students. Israeli academic economists are invited for sabbatical years to the best universities in the West and top Israeli students are likewise admitted with full fellowships for PhD studies. Israeli academic economists served as the chief economists at the IMF and the World Bank and served as presidents of the International Economic Association. Many more academic economists than before have chosen to locate abroad, mostly in the United States, and most are highly esteemed. At the same time a significant number of academics and students from other countries (true, mostly Jewish) came to teach and study in Israeli universities and to work in Israel, including the new recruit to head the Israeli central bank. The so-called "brain drain" is therefore partially offset by brain gain in a variety of forms. Until recently there was little worry that brain drain could become a real threat, but cuts in the budgets for higher education and reluctance, mostly by the government, to engage in needed reforms may tilt the balance in that direction. True, the two (actually three) Israeli Nobel Prize winners in economics came from other spheres at the Hebrew University, but they clearly benefited from externalities of the economics department. Economics in Israel is clearly more advanced (and was advanced earlier) than in many continental countries, some much larger and all with a longer academic history in economics. A word must be said about the nature and content of the economic policy agenda and advice by the economic community in Israel over the years. Indeed, this topic deserves a paper of its own. As mentioned, the Israeli economy, including during the pre-state period, was dominated by the public sector. It was the major entrepreneur of economic development as well as the owner of a significant share of the economy. In addition, there was a very high level of intervention in the economic activities of the private sector: price controls, heavy foreign trade taxation and exchange rate controls, heavy regulation, subsidization, and direct economic support; a very strong trade union; and a labor-dominated government. While economists could understand, even justify, this high level of government involvement at the start on the basis of what one would term a series of "market failures" and the need for a critical concerted effort for economic takeoff, they nevertheless fought for a more rational analysis and policy, on both the micro and macro levels. David Ben-Gurion, the founder of Israel and its first prime minister, claimed all along that economic laws do not apply to Israel. Later on, as the economy emerged from the deep crisis of the early 1950s, economists increased their pressure on the government to allow more openness, more competition, freer markets and less intervention, more respon- sible monetary and fiscal policies, and budgets with lower deficits. Progress in all these directions did take place over the years, with ups and downs, albeit at a much slower pace that seemed appropriate to the economists. More recently, as Israel has become much more open and "liberal," questions about the appropriate teaching of economics and the policy implications have been THE TRANSITION OF ECONOMICS: THE CASES OF ISRAEL AND RUSSIA | 47 raised mostly by outsiders to the field, but for the first time also within the economic community. Among them: Is the "separation" between considerations of efficiency and redistribution as per the neoclassical teaching valid? Did the basic economic courses, especially Ec. 1, sufficiently emphasize the conditions under which markets are efficient and the prevalence and importance of market failures--and hence the need for government intervention and regulation? Is privatization always beneficial? How about globalization? Even more fundamental, are the assumptions about "Homos economicus" and personal utility maximization justified? Are they correct? Don't they increase the materialistic tendencies of society and reduce its level of solidarity? (That assumes at least some incorporation of the utility of others in one's own utility function.) In short, is there a degree of overkill in the way we teach economics? The least that economists should do is pay more attention to social issues. These claims originated to some extent from parallel developments abroad, but also from domestic processes, especially the trend of rising income inequality and the incidence of poverty, and the recent tendency of the government to somewhat limit its large redistributive role. A full discussion of these issues must be deferred. My feeling is that while economics as we study it on paper contains appropriate responses to most if not all these claims, there may be a need to better emphasize the balance of forces and the social neutrality of economics, at least in order to be more persuasive with other professions and groups in society. Economics in Israel is a success story in terms of achievements and quality, and a near optimal balance between theory and policy, of being outward and inward looking, of becoming a respected member of the global community of economics, but also having significant influence in shaping the economic policy of Israel and sharing in its successes and problems. Some Parallels with Russia and a Concluding Note Is the Israeli experience a model for Russia, and for other transition countries? No two cases can be identical, but I am sure that you can see many potential parallels. As I mentioned above, this model served as a guide and inspiration to me when I got involved with others in the project of NES, of offering modern economics to Russia. The model and the original blueprints were of course modified by the collective knowledge and experience of the "founding fathers" of NES on the nature of the Soviet and the Russian economy and polity. The problems faced by modern economics as it tries to make its way into Russia are discussed in great detail in Ofer (2007) and related works cited there. Many similarities are revealed; I trust that you can see them. Let me mention only a few. First, it could have been assumed that the resistance in Russia to the importation of modern economics would be stronger, despite the transition. Unlike in Israel, in Russia there was a well-established and entrenched field of "political economy." While the carpet was pulled out from under its feet, it did control the entire profes- sion in the universities and institutes. Therefore, going through a flagship university 48 | GUR OFER like Moscow State University, as was done in Israel with The Hebrew University, was out of the question. Likewise, possibly it should have been assumed beforehand that there would be difficulties for NES to take part in policy work. It must be admitted that part of the hesitation and doubt at the start were not on the demand, but on the supply side: would the young economists, coming back to Russia or trained at NES, be able or willing to engage in Russian-oriented policy research and work? This fear was proved to be mostly misdirected. Rather, there was more resistance on the demand side. It also took quite a long time in Israel, but it seems to be further delayed in Russia, despite excellent policy work that is being done at CEFIR and NES. Gradually, this glass ceiling has been lifted and there is more and more listening as time goes on.6 The same is true about making more efforts on a national level to bring back pro- fessionals trained abroad and to reform the system of higher education. Israel has a special fund for this; in Russia it still has not happened. In Russia there also seems to be less enthusiasm, national and professional, with the new and fascinating expe- rience that Russia has been going through following the fall of the communist regime, compared to Israel in its pioneering days. Back in 1991, when NES, as well as the High School of Economics, the Institute of Economic in Transition, the Leon- tief Center in St. Petersburg, and other schools and think tanks teaching and using modern economics were established, there seems to have been a different perception on what would happen, as compared with what really took place. Back in 1991 there was the assumption (or hope) and the enthusiasm that a new leadership would take over with ideas and plans about the economy in which projects like NES would fit in better. This did not fully materialize then, and nowadays there seems to be even a feeling of retreat. This current environment, political as well as economic, does not encourage people to come back to Russia. Hence one may expect, among other consequences, a larger loss of brain drain than was the case in Israel. It may sound a bit paradoxical, but these eventualities, whether anticipated or not, emphasize two things: the importance of the continued development of entities like NES, and added justification of the strategic decision taken at NES to follow a policy of maximum openness, of study abroad, of intensive travel in both directions, and of joining the international economics profession by working in similar areas and publish- ing in leading journals; and, at the same time, of applying this modern economics to issues of Russia, aiming to reach a balance between the two, similar to the balance that was described earlier for Israel. Indeed, as in Israel, there are in Russia and in transi- tion countries in general a number of topics of paramount significance to economics, like the new institutional economics and political economics that are being presented during the transition. The high proficiency in mathematics is also very important here. Also, as in Israel, there are similar discussions and disagreements on the position of economics on issues of social justice and solidarity and on the right level of government intervention, openness, competition, and so on. While some demands for restraint and gradualism in the movement in these directions in Russia may be justified, there is the danger, as during the early period in Israel, that such demands might have been misinterpreted by the economic and political leadership and will be used adversely to retreat to unwarranted interventions. THE TRANSITION OF ECONOMICS: THE CASES OF ISRAEL AND RUSSIA | 49 Like Israel then, Russia needs modern economics and can gain from it more than established market economies. A basic change in the system was called for. Like Israel in its early years, Russia is in a potential position to compete successfully and move ahead of some of the countries on the continent that are changing very slowly to advance modern economics in their midst. With its high level of human capital and the advanced structure of the economy, Russia like Israel or even more, can become an important contributor to the advance of economics in general, and around the issues of transition in particular. In these two respects, the introduction of modern economics to Russia is different than that to most developing countries. A few Latin American countries and India may be exceptions. We at NES try to contribute our small share to these developments. Notes 1. More information about NES and its development is available in Ofer (2007) and the ref- erences there, and the Web site of NES, www.nes.ru. 2. See, among others, Portes (1987); Kolm (1988); Frey and Frey (1995); Coats (1996, 2000). 3. However, see Barkai 1993. 4. This section draws on the following: Liviatan ( 2007); Gross (2004, 2005); Barkai (1993); Michaeli (2005); Patinkin (1994). 5. A similar problem arose with the translation to Russian of a number of texts, including the second edition of Patinkin's Money Interest and Prices (1965). Faculty and students at NES had to struggle with an earlier draft with many problems. Also a number of returning PhDs hesitated to teach advanced courses in Russian for fear of lack of lan- guage. NES is encouraging teachers to offer courses in English. 6. Arkady Dvorkovich, a graduate of the first class at NES (class of 1994), was a deputy minister of economics and is now serving as the head of the presidential Experts' Direc- torate. Ksenia Eudaeva, also a graduate of the first class and a PhD from MIT, was just appointed the academic director of the Center for Strategic Research (the research center under the Ministry of the Economy that works on economic reform). References Barkai, Haim. 1993. "Don Patinkin's Contribution to Economics in Israel." In Monetary Theory and Thought: Essays in Honor of Don Patinkin, ed. Haim Barkai, Stanly Fischer, and Nissan Liviatan, 3­14. London: The Macmillan Press Ltd. Coats, William, ed. 1996. The Post-1945 Internationalization of Economics. Durham, North Carolina and London: Duke University Press. ------. 2000. The Development of Economics in Western Europe Since 1945. Oxford: Routledge Publishing. Frey, Bruno S., and Rene L. Frey. 1995. "Is There a European Economics?" Kyklos 48 (2) (special issue): 185­6. Gross, Nachum. 2004. "The Economics Department at the Hebrew University during the 1950s" (in Hebrew). Discussion Paper 04.06, The Falk Institute, Jerusalem. ------. 2005. "Social Sciences at the Hebrew University until 1948/9: Plans and Begin- nings." In The History of the Hebrew University in Jerusalem: Establishment and Growth, Volume A (in Hebrew), ed. Chagit Lavsky, 503­41. Jerusalem: Magnes Press of the Hebrew University. 50 | GUR OFER Kolm, Serge-Christophe. 1988. "Economics in Europe and in the US." European Economics Review 32: 207­12. Liviatan, Nissan. 2007. "Don Patinkin." In the New Palgrave Dictionary of Economics, 2nd edition, forthcoming. Michaeli, Michael. 2005. "The Department of Economics at the Hebrew University: The Early Days from a Personal Angle" (in Hebrew). Mimeo. Ofer, Gur. 2007. "Teaching and Research in Modern Economics in the Russian Federation: The Experience of the New Economic School." In Capacity Building in Economic Educa- tion and Research, ed. François Bourguignon, Boris Pleskovic, and Yehuda Elkana, 51­74. Washington, DC: World Bank. Patinkin, Don. 1965. Money, Interest and Prices. 2nd ed. Evanston, Illinois: Row Peterson. ------. 1959. The Israeli Economy: The First Decade. Jerusalem: The Falk Institute. ------. 1994. "From Chicago to Jerusalem." Economic Quarterly 43 (2). Portes, Richard. 1987. "Economics in Europe." European Economics Review 31: 1329­40. Keynote Address Interaction of Political and Economic Transition YEGOR GAIDAR The interaction of political and economic development in the course of postcommunist transition and reform is a very complex and controversial topic that has a serious impact on the development of events in Russia today. When the team of people that is now in power came to the Kremlin, I had my own hypothesis about how they viewed the situation and the development of events. And during the past five years, this hypothesis has been borne out by facts. The essence of the hypothesis is as follows. These people understood that Russia needed a market economy. Moreover, they understood that Russia needed liberal market economic reforms: the reforms that had not been completed in the 1990s because most governments of the 1990s were not supported by the parliamentary majority. They also believed that democracy in Russia­­a real, functioning democracy-- is still not needed for Russia, that Russia had not yet grown enough to witness a true democracy. Democracy would eventually exist in Russia, they believed, but not at that particular moment. And so, proceeding from that belief, they developed the economic and political measures that were brought about at that time. Rather serious, positive economic reforms were carried out, starting with the issues concerning private property in land and taxation policy, which allowed the government to decrease the tax burden while making it possible to collect much more for the country's budget. But at the same time, the imperfect­­the young, but still functioning­­system of democratic institutes in Russia was being dismantled. This system, which existed by the beginning of the 2000s, had an influential parliament. There was a certain independence and freedom of the press. The regional authorities were more or less autonomous in the solving of their local problems. There were influential entrepre- neurial structures, which were able to take part in the decision making. All this existed by the year 2000, but by the year 2004, all this had disappeared. The basis for such a movement is easy for me to understand because I have had a chance to read hundreds of books and thousands of articles that describe how right Deng Xiaoping was when he split the economic and the political reforms in Yegor Gaidar is professor, director of the Institute for the Economy in Transition. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / World Bank 51 52 | YEGOR GAIDAR China and how wrong Gorbachev was when he brought those together in Russia. And these books and articles all said that first the economic ground should be established for democratic development, and only then the movement toward democracy should be started. If you believe that those who work today in the Kremlin have never read any of these articles or books, or at least extracts from them, you are mistaken. And this point of view definitely has had an effect on the practical development of events. But in my opinion, what has been written is a simplification of the actual development process. In my opinion, we should not employ this simplistic approach. Socialism is a system within which politics and economy, the state structure, and the functioning of the everyday economic life are all intertwined. For those people who have never lived under socialism, it's a bit difficult to understand this, and it's difficult to explain this to them. Within the framework of the stable socialist system, which was formed in the 1930s, a nearby shop sells bread, bread which you are able to buy. And this fact depends not on the interests of those who bake the bread and not on the interests of those who sell the bread, but on the very simple fact that those who make it and those who sell it know very well that they are going to be punched if this bread is not available in the shops. This situation is very much contrary to what Adam Smith espoused. The producer knows very well that this bread is going to be taken away from him, and that the price paid will not be the price which would make the amount of supply equal. And if he doesn't give it away, he might end up in a gulag. Those who work in the economic sphere are absolutely sure that the author- ities will employ violence, and that they will force people to do what people under any other conditions­­under the conditions of a market economy­­would never do; people would never work without any remuneration. People would never sell bread at prices that do not comply with a production cost, and authorities would not be able to send people to their deaths without any punishment at all. But as soon as people begin to doubt the ability of the authorities to apply as much violence as nec- essary, then the fear-based system stops working, and bread disappears from the shops because the fear disappears. Unfortunately, my experience shows that what I'm saying is true. However, to explain it to the people who haven't lived through it would be impossible. Fortunately for them. The problem with the fear-based system is as follows. It, by its own devel- opment, undermines the basis of its own stability. Killing several hundred thousand people and sending millions to camps in an agrarian country where the overwhelm- ing majority of people live in villages, have no education, and are illiterate­­provided there is political will, severity and missing ideology are tasks that can be resolved. Implementing such a system in a literate and developed urbanized country with well- educated citizens, as world experience shows, would be much more difficult because those in authority would not be convinced of their right to employ violence. In 1989 in Beijing, during the suppression of the strikes at Tiananmen Square, in a country that at the time was quite agrarian and not well educated, troops from Beijing were not thought to be reliable enough to use tanks . . . to fight the people, so troops that were considered reliable had to be brought in from Tiananmen College. And that was in an agrarian and not well-developed state. INTERACTION OF POLITICAL AND ECONOMIC TRANSITION | 53 In the Soviet Union, in 1991, troops who were ready to use tanks against their com- patriots were difficult to find, and that fact was recognized by the Soviet authorities. It was recognized, not in 1989, but much earlier: in 1962, after a growth in prices that had been quite moderate. However, the increases violated an implicit contract between the people and the authorities that was formed in the 1950s. Its essence was as follows. You would guarantee stable prices and stable social programs to us, and for that we would endure you. You understand that you are not elected, and so we would endure you, and you shouldn't interfere with our lives. When the Soviet authorities, facing real economic problems and difficulties, increased prices by 30 percent, that increase proved to be a cause for very serious disorders in one of the Russian towns in Novocherkassk. And the most unpleasant thing for, and what most dismayed, the authorities was that, at the first stage of these disorders, the troops didn't want to use weapons against the citizens. The Soviet authorities remembered how they came to power in 1917, a result of a process that started with shots at people who participated in demonstrations against food shortages in Petrograd. If it could happen in Novocherkassk before there were severe indications of the need to move loyal inner troops there, it could well happen in Moscow the next time. Recognition of this fact would influence the evolution of events in the Soviet Union in the last decades of its existence: recognition of the fact that the entire structure of authority was based on the ability of those in authority to exert unlimited violence against the people, and that ability was being eroded. It was being undermined by the evolution and development of society. The society was becoming ever more literate and educated. Alongside this was a question that permeated all the sessions of the political bureau: how much grain could be mobilized and how much was really needed? As soon as it was understood that the ability to exert violence was limited, it appeared that less corn could be mobilized than was needed. This assessment is based on all the archival material that is currently available. So what next? The country is no longer selling corn­­and you know that Russia was the world's largest corn exporter before 1917; exports were about 2.5 times higher than those of the United States. But in the 1980s Russia became the world's largest corn importer. That transition of the country from being the world's largest exporter to being the largest importer happened against the background of the diminishing ability of the authorities to exert violence. To mobilize the amount of corn that was needed, millions of peasants would need to die, as happened in the 1930s. That was a political and economic change that would pre- determine the history of the Soviet Union and the evolution of events. As it became apparent that it would not be possible to obtain as much corn as was needed because of the deep crisis in agriculture, and a world record volume of corn would have to be purchased elsewhere, another characteristic of the socialist economy came to the fore. It was apparent that the products of the processing industry­­an industry that Soviet authorities set up at the peasants' expense, exporting corn when millions of people in the country were starving to death, as happened in 1932­33, and purchasing complete imported equipment­­could not be sold for convertible currency. When the crisis with currency in the Soviet Union became obvious, the possibility of increasing exports of the machine-building industry was not even discussed, because it was understood that this would not be realistic. 54 | YEGOR GAIDAR So you have the urbanized society's growing demand for food products, including corn. You have a chronic crisis in agriculture stemming from how you once proceeded with industrialization and the many peasants you killed. And you have a noncompeti- tive machine-building sector. What should you do? And then a magic wand appeared: the opening of the largest oil fields dis- covered in the country, in western Siberia, with unique yields of boreholes and new production conditions that are not deep seated, and with enormously high prices for oil. You can compensate for the chronic inefficiency of the agriculture produc- tion by relying on abnormally high growth rates in petroleum exports. In nominal values, the volume of real exports would increase tenfold and the country would become the world's largest importer of food and corn. But then you have to under- stand that you will be endlessly dependent on a factor that is not predictable: the status of the world's petroleum market. No one can predict this market. As of 1985, the Soviet empire­­the political system, the economy, everything­­was hanging on three small nails: weather, on which the harvest would depend; the state of the art at the largest petroleum fields; and oil prices. After the Soviet invasion of Afghanistan, Saudi Arabia understood that it might need the help of the United States, and the United States needed to reduce the prices for oil. The history of how Saudi Arabia increased oil production on a monthly basis a few times within one year, how oil prices dropped four times, how the Soviet economy started to disintegrate because it relied on oil prices that were abnormally high by historical measures (as in the early 1980s): that is one of the most interesting political detective stories of the 20th century. However, the essence is not related to the detective evolution of the events. The fact is that the socialist model of industrialization made the country objectively dependent on a long-term basis on parameters that fluctuated uncontrollably. And when you have such foreign economic shocks, of such a type and to such an extent, and when a regime has few stable resources, the consequences are significant. The Soviet Union was not the only state in the world that experienced these external shocks. The 1980s were not an easy period for all oil-producing countries. Adaptation was quite difficult. Mexico was confronted with a series of economic crises. How- ever, a crisis to the extent of that in the Soviet Union was not recorded elsewhere. And that is not surprising. It is difficult for people to live in an economy that relies on one particular commodity, because when trade conditions radically deteriorate by 10 percent, and more than once, it affects the financing of their education, health care, and culture. What should be done when the resources for borrowing are exhausted? Nevertheless, many countries did adapt to it. However, for the socialist regime, legitimization is based on the idea that the authorities know better than the people what should be done, implying that they are the cleverest, they are equipped with the most up-to-date ideology. The people shouldn't interfere with development matters, and the authorities will bring us to a happy future. That is the basis of the legitimization of this regime. And for such a regime to tell the people that it appears that we have brought you in the wrong direc- tion, and now we have external economic shock for which you will have to pay. That would be impossible, that would be beyond the limits of any political reality. INTERACTION OF POLITICAL AND ECONOMIC TRANSITION | 55 The Soviet authorities­­and I know this from archival material­­never even discussed it when there was a large-scale, unprecedented foreign economic shock. They understood that the people would tell them the following. "You think you are so clever. You explained to us for a long time that you didn't need our advice, that you better understood where we were proceeding. You want us to tighten our belts more. You should tighten your belts." This is the only idea that explains the fact that the Soviet regime, having experienced a large-scale foreign economic crisis, proceeded toward its own bankruptcy and disintegration as a victim, without trying to do anything to stop the catastrophe, just observing how the currency reserves were becoming exhausted, how we were not paying our external debts, how the catastrophic crisis in finance was increasing, how the crisis in the consumer market was developing, and how the corn reserves were disappearing. The regime just observed these changes and did nothing about them. An attempt to combine the economic and political liberalization of the Soviet Union did not occur and was not the basis for the country's disintegration. Instead, a political and economic structure was formed that was not stable internally and that was based on violence. As the level of development increased, it undermined the ability of the authorities to use unlimited violence against their own people. Mean- while, there was a long, deep, clinical crisis in agriculture and competitive interna- tional markets in the processing industry. The economy of the country depended on world prices that were prone to fluctuation. This was the basis of the catastrophe that happened in the Soviet Union. Growth After Transition: Is Rising Inequality Inevitable? Increasing Inequality in Transition Economies: Is There More to Come? PRADEEP MITRA AND RUSLAN YEMTSOV Inequality has generally been increasing in the transition economies of Eastern Europe and the former Soviet Union but, as predicted by a number of theoretical models, the increase differed substantially across countries. This paper decomposes changes in inequality both by income source and socioeconomic group, with a view to under- standing the determinants of inequality and assessing how it might evolve in the future. The empirical analysis relies on a set of comparable inequality statistics put together for the recent World Bank study, Growth, Poverty and Inequality in Eastern Europe and the Former Soviet Union: 1998­2003 (World Bank 2005b). The paper argues that further evolution of inequality in Eastern Europe and the former Soviet Union will depend on various factors. Some of these factors relate to transition, such as the evolution of the education premium; a bias in the investment climate against new private sector firms, leading to an excessive dispersion of labor market outcomes; and regional impediments to mobility of goods and labor. But other factors are increasingly important, such as technological change and globalization. The paper also contrasts key features of inequality in Russia with trends in inequality observed in China, where rapid economic growth has been accompanied by a steep increase in inequality. It argues that China's experience is largely a developmental phenomenon rather than a transition-related one, deriving from the rural-urban divide; thus it is of limited relevance for predicting changes in inequality in Russia. Introduction Consider the evolution of GDP per capita and inequality in per capita consumption in Poland and Russia, the two largest transition economies of Eastern Europe and the former Soviet Union, respectively. Poland, shown in the upper panel of figure 1, experienced a relatively shallow transitional recession and a decline in inequality, and then a more gradual increase in inequality with some temporary reversals. This Pradeep Mitra is chief economist, Europe and Central Asia Region, and Ruslan Yemtsov is senior economist, Poverty Reduction and Economic Management Unit, Europe and Central Asia Region, at the World Bank. The authors thank discussant Jan Svejnar and two anonymous referees for useful comments. Annual World Bank Conference on Development Economics 2007 ©2007 The International Bank for Reconstruction and Development / The World Bank 59 60 | PRADEEP MITRA AND RUSLAN YEMTSOV FIGURE 1. Poland and Russia: Real Per Capita GDP and Gini Index, 1990­2003 a. Poland 150 140 130 120 1985=100 x, Inde110 100 90 1985 1987 1989 1991 1993 1995 1997 1999 2001 Year Real GDP per capita Gini index b. Russia 175 150 125 1990=100 x, 100 Inde 75 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year GDP per capita Gini index Sources: For Poland, Keane and Prasad (2002a) for 1985­97 (Gini for consumption per capita without durables); authors' estimates for Gini 1998­2002. For Russia, simulations based on published expenditure distributions by Goskomstat for 1990­96 and authors' estimates for 1998­2003. pattern exemplifies developments in Eastern Europe more generally. This was fol- lowed by a sharper increase during the late 1990s and early 2000s, however, to the point that the Gini coefficient of inequality was more than 25 percent higher in 2003 compared to 1989. In contrast, Russia, shown in the lower panel of figure 1, broadly exemplifies developments in the countries of the Commonwealth of Independent States (CIS). Russia experienced a wrenching transitional recession accompanied by an explosive increase in inequality, which peaked in the mid-1990s. However, this was moderated to some extent during the very rapid growth that occurred after the 1998 financial crisis, so that the Gini coefficient was 10 to 15 percent higher in 2003 compared to 1991. Since 1999, the transition economies of the former Soviet Union have grown at rates approximating China's extraordinary performance. Together with the transition INCREASING INEQUALITY IN TRANSITION ECONOMIES | 61 economies of Eastern Europe, they surpassed the pre-transition levels of GDP per capita for the region in 2004. While these developments are encouraging, they have occurred in the shadow of the realization that rapid growth in China, shown in figure 2, has been accompanied by a steep increase in income inequality, as measured by the increase in the Gini coef- ficient of income inequality by 2 percentage points a year between 1990 and 2001. The Gini coefficient was nearly 50 percent higher in 2003 compared to 1981. For countries in Eastern Europe and the former Soviet Union, which share a socialist legacy with China, this could be seen as a harbinger of things to come. Will improved economic performance in Russia and other transition countries in Eastern Europe and the former Soviet Union come at the expense of a further widen- ing of income disparities? Has the transition to a market economy moved these coun- tries irreversibly to a higher inequality path, on which other factors not related to transition, such as globalization, will be superimposed, possibly generating even more unequal distributions? And is economic policy capable of influencing these processes? These are the key questions addressed in this paper. In attempting to pro- vide answers, the paper: · Reviews the extensive literature on the determinants of inequality in transition, focusing on the stylized facts, · Uses a consistent and comparable consumption aggregate for the transition economies of Eastern Europe and the former Soviet Union, which aims to overcome deficiencies in existing data and provide a firmer foundation for those stylized facts, and · Decomposes inequality by sources of income and household groups, with a view to understanding the role of key determinants of inequality in different countries. FIGURE 2. China: Real Per Capita Income and Gini Index, 1981­2001 350 300 250 1981=100 200 x, Inde 150 100 50 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 Year Real mean per capita income Gini index Source: Ravallion and Chen 2004. 62 | PRADEEP MITRA AND RUSLAN YEMTSOV The paper is organized in seven sections. The second section, following this introduction, raises the question of what is really known about inequality in the transition countries by examining the quality of available data. The third section summarizes the construction of and presents a data set more amenable to within and across country comparisons. The fourth section reviews the guidance available from theoretical models of transition on the key determinants of inequality. The fifth section presents the decomposition of inequality by income source and by household groups, which constitutes the key contribution of the paper, and assesses the outlook for inequality in the future. The sixth section compares the experience of the countries of Eastern Europe and the former Soviet Union with regard to growth and inequality with what is known from published sources about China in order to assess whether rising inequality in the latter portends the future of the former set of countries. The seventh section concludes with implications for policy and areas for further research. Increasing Inequality in Transition: What Do We Actually Know? Table 1, based on most widely used published data, suggests that all the coun- tries in Eastern Europe and the former Soviet Union experienced an increase in inequality. However, despite an apparently common legacy (see Alexeev and Gaddy 1993), countries experienced very different degrees of increased inequality. On the one hand, as already seen in the example of Russia, a rapid increase in inequality occurred in the middle-income and low-income CIS countries, followed by some moderation. On the other hand, as the example of Poland illustrates, at least until the mid-1990s, the new member states of the European Union (the EU-8), appear to have experienced a more gradual but steady increase in inequality. Table 1 makes clear that, by the early 2000s, the region exhibited the full spectrum of inequality outcomes, ranging from fairly unequal to fairly equal distributions of income. To what extent can the data presented in table 1 be taken at face value? A flavor of the controversies surrounding the "stylized facts" depicted in the table is provided in figure 3, which depicts a wide range of alternative inequality estimates for one country, Russia, drawn from different well-documented sources. The figure shows that, for the most recent period, depending on which source of data is chosen, Russia could be classified as anything from a moderately high to a high inequality country or as anything from a country exhibiting rising to falling inequality. The example clearly illustrates the point that published data on income distribu- tion should be treated with great care--and this for at least five different reasons, following Atkinson and Micklewright (1992). First, published data from different countries rely on different imputation and adjustment procedures. In Ukraine, for example, significant and rather unusual imputations are undertaken with reported in-kind components. In some countries, total incomes include imputed rents, while in others they do not; which option is chosen can have large effects. In Russia, inclusion of proper imputed rents in the full TABLE 1. Selected Countries, Eastern Europe and the Former Soviet Union: Gini Indices for Per Capita Incomes from "Official" Sources, 1987­2003 Country 1987­90 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Armenia 0.269 0.570 0.537 0.428 Azerbaijan 0.345 0.440 0.373 Belarus 0.233 0.280 0.253 0.244 0.249 0.253 0.235 0.247 0.245 0.246 0.249 Bulgaria 0.245 0.344 0.340 0.384 0.357 0.366 0.345 0.326 0.332 0.333 0.370 0.351 Croatia 0.251 0.333 0.29a Czech Rep. 0.197 0.228 0.270 0.258 0.230 0.239 0.212 0.232 0.231 0.237 0.234 Estonia 0.240 0.395 0.350 0.370 0.361 0.354 0.361 0.389 0.385 0.393 0.402 Georgia 0.313 0.430 0.469 Hungary 0.214 0.231 0.242 0.246 0.254 0.250 0.253 0.259 0.272 0.267 0.268 Kazakhstan 0.297 0.330 0.35 Kyrgyz Rep. 0.308 0.353 0.470 0.411 0.399 0.414 0.377 0.382 0.342 Latvia 0.240 0.310 0.326 0.321 0.327 0.358 0.379 Lithuania 0.248 0.350 0.347 0.309 0.332 0.343 0.355 0.354 0.357 0.318 Macedonia, FYR 0.349 0.369 0.367 0.34a 0.34a Moldova 0.267 0.365 0.360 0.420 0.437 0.435 0.436 0.411 Poland 0.255 0.265 0.274 0.285 0.320 0.328 0.334 0.326 0.334 0.345 0.341 0.353 0.356 Romania 0.232 0.290 0.312 0.302 0.305 0.298 0.299 0.310 0.353 0.349 0.352 Russia 0.259 0.260 0.289 0.398 0.409 0.381 0.375 0.381 0.398 0.399 0.394 0.396 0.398 0.404 Slovak Rep. 0.186 0.237 0.249 0.262 0.249 0.264 0.263 0.267 0.299 (Continues on next page) 63 64 TABLE 1. continued Country 1987­90 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Slovenia 0.220 0.227 0.282 0.250 0.302 0.305 0.298 0.299 0.310 0.353 0.22a 0.22a Tajikistan 0.334 0.470 Turkmenistan 0.308 0.360 Ukraine 0.240 0.282 0.288 0.290 0.277 0.271 Uzbekistan 0.351 0.330 Sources: Data from UNICEF TRANSMONEE 2005 edition, www.unicef-icdc.org/research, except for selected countries and years from Milanovic (1998) for Estonia 1992, Georgia 1989, Kyrgyz Republic 1993, Latvia 1994, Lithuania 1994, Slovak Rep. 1989; World Bank (1999) for Azerbaijan 1995, Estonia 1994, Georgia 1997 and 2000, Uzbekistan 1994; World Bank (2002a) for Armenia 1999; World Bank (2003a) for Kyrgyz Republic 1997; Vecerník (1995) for Czech Republic 1992 and 1994; and Eurostat (2005) for Croatia 2003, Macedonia FYR 2002­3, Slovenia 2002­3. Note: The reference countries, Poland and Russia, appear in bold. For Russia, data for 1992 and earlier years refer to total incomes; for later years, they refer only to money incomes. Empty cells are for years with no available data. a. Data are from Eurostat (2005) or EC (2005) and rely on an OECD equivalence scale. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 65 FIGURE 3. Russia: Evolution of Gini Index, Various Sources, 1992­2004 0.50 0.45 x 0.40 inde Gini 0.35 0.30 0.25 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Year Officially published, nominal per capita incomes Per capita expenditures in RLMS, direct data est. Per capita consumption corrected for regional price differencesa Sources: Goskomstat; World Bank 2005d. Note: RLMS = Russian Longitudinal Monitoring Survey; HBS = Household Budget Survey; NOBUS = National Survey of Social Programs and Participation. a. For 1998­2002, based on the HBS data; for 2003, using the NOBUS data. consumption in 1993 reduced the Gini index from 0.42 to 0.35 (Buckley and Gurenko 1997). Thus different rows in table 1 cannot be compared with one another, and a higher country Gini does not necessarily translate into higher inequality for a compa- rable concept of welfare. Second, in all the EU-8 countries, wages account for over 60 percent of household income. In contrast, among the low-income countries of the CIS, wages represent less than 15 percent in some cases. At the same time, while public transfers are a much more important component of income in the EU-8, where they comprise 25 to 30 percent of total income, their importance has shrunk dramatically in the low-income CIS countries. Public transfers in Georgia and Moldova, for example, represent less than 10 percent of GDP. Wages and transfers can be measured quite well by household surveys, whereas other sources of income, such as from informal self-employment, are notoriously hard to measure with any precision. Such compositional effects have serious implications for the accuracy with which inequality is measured. For this reason, table 1 is a poor guide to describing inequality in the case of low-income CIS countries such as Armenia, Georgia, the Kyrgyz Republic, and Moldova. Third, there are serious issues of underreporting and nonresponse. Richer house- holds, for example, tend to be increasingly missed by sample surveys. In practice, countries undertake different degrees of adjustment to correct for nonresponse but, in doing so, make a number of assumptions that can undermine comparability. In Russia, unlike in any other country, the increasing gap between reported incomes and estimates from macroeconomic sources is arbitrarily assigned to the top decile of 66 | PRADEEP MITRA AND RUSLAN YEMTSOV households as "undeclared" incomes (World Bank 2005d), limiting comparability with other data on income distribution. Fourth, correction for regional price differences is not normal practice in many statistical offices.1 Fifth, the use of equivalence scales has not been converging toward a single standard.2 All of this implies that, while official data can suggest that inequality has increased in all countries in transition, the magnitude of such increases is less certain. Despite these limitations, data such as those reported in table 1 are used to generate "stylized facts" and draw far-reaching conclusions about the evolution of inequality in transi- tion (Ivaschenko 2003). Toward Comparable Data on Inequality in Transition The lack of consistency of "official" data on inequality prompted the creation of comparable and consistent inequality statistics based on primary records from household surveys across the transition countries of Eastern Europe and the former Soviet Union.3 Most of these surveys are conducted by statistical offices and are, in that sense, "official." But the way in which primary data were used led to indexes that are different from the numbers reported in table 1. First, the preferred measure of welfare is consumption rather than income. The choice of consumption was dictated by practical considerations. While data on incomes remain particularly difficult to collect in transition countries, practice has shown that data on consumption can be gathered with considerable accuracy. Survey consumption modules have become more detailed over time and are better able to capture the various dimensions of consumption, including informal payments. Second, unlike the practice of simple aggregation undertaken by many statistical offices of the region, a distinction was made between different components of consumption. Since consumer durables and housing are consumed over a long period of time, it is customary to include the imputed value of the consumption flow associated with the possession of consumer durables (including housing) but to exclude the expenditure on the purchase of such goods. The lack of data, however, limits the application of this approach to all countries. It was therefore decided not to include estimates of the flow of services of durables or of durable purchases or rents. Third, given the significance of spatial differences in the transition countries, an adjustment for spatial price differences was made, using Paasche price indices based on survey data in all countries. In cases where data were collected over a long period of time, it was also necessary to adjust for changes in prices over time. Quarterly Consumer Price Index (CPI) indices taken from International Monetary Fund (IMF) data were used to compute real values. Fourth, households in the transition countries have coped with poverty by relying on an array of nonmarket strategies, including producing their own food and engag- ing in reciprocal exchange with other households and institutions. A consistent approach was used to assign a monetary value to these components of consumption. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 67 Fifth, the same procedure, which conforms to methods used in other international household survey data depositories such as the Luxemburg Income Study, was used to clean the data of outliers across all data sets. Since a consistent approach was used across all data sets, one can be reasonably confident that differences across countries in the final consumption measure arise from differences in the primary data and are not due to the method of aggregation. Results for all countries with available primary records are presented in table 2. The table clearly shows that there are discontinuities and that the evidence is of variable quality. However, the difference in country experiences regarding the evolu- tion of inequality is striking, even with data that are as comparable as possible. It dispels the notion that countries would converge to some common level of inequality that prevails in the long run in market economies and provides motivation for the analysis undertaken in this paper.4 The new data confirm the overall picture that had emerged from the data on income inequality. Specifically, table 2 underscores four points. First, all the transi- tion countries have become more unequal. Second, there were rapid increases in inequality in many CIS countries, followed by some stabilization, or even subse- quent moderation. Third, there was a much more gradual increase in Central Europe, with continued change up to the most recent year for which data are available. Fourth, there was a wide diversity of experience, even among countries within the same subgroup of countries. For example, the Baltic states experienced inequality paths similar to that of Russia, whereas in Belarus, which retains many features of a command economy, the evolution of inequality more closely resembled that in Central Europe. That said, the magnitude of increase and ranking of each country with respect to inequality usually differs, at times dramatically, from that provided by the income- based data in table 1. Income-based and consumption-based measures of inequality appear to be fairly consistent with each other only in some cases, typically in the EU-8 countries. This is clearly not the case in the low-income CIS countries and in some middle-income CIS and South-East European countries. For the reasons explained above, the new consumption-based data are believed to be more accurate. Indeed it is consumption inequality, based on the new data, which is reported in figure 1 on Poland and Russia used to introduce this paper.5 The data in table 2 are also helpful in illustrating the evolution of inequality over time and decomposing its sources by household groups in countries that are deemed broadly representative of four clusters in the region: that is, Hungary, Latvia, and Poland for the EU-8 countries; Romania for South-East Europe; Kazakhstan and Russia for the middle-income CIS countries; and Georgia, Moldova, and Tajikistan for the low-income CIS countries. Table 3 presents key data on the Gini index of inequality for those countries. Having consistent data is the first step toward understanding the drivers for the increase in inequality and attempting to predict future evolution. The next section puts together six main drivers for the inequality increase often used to contrast the experiences of various countries. 68 TABLE 2. Gini Index for Per Capita Consumption Country 1988­92 1993­95 1996 1997 1998 1999 2000 2001 2002 2003 Albania 0.291 0.319 Armenia 0.444 0.321 0.325 0.310 0.285 Bosnia 0.263 0.295 Belarus 0.228 0.287 0.291 0.299 0.293 0.301 0.292 Bulgaria 0.234 0.283 0.350 0.337 0.277 Estonia 0.230 0.395 0.376 0.339 0.332 0.335 0.330 Georgia 0.28 0.370 0.404 0.386 0.393 0.397 0.383 0.390 0.391 Hungary 0.210 0.232 0.250 0.259 0.254 0.251 0.250 Kazakhstan 0.257 0.327 0.353 0.346 0.330 0.318 Kyrgyz Republic 0.260 0.537 0.523 0.405 0.360 0.346 0.299 0.290 0.292 0.276 Latvia 0.225 0.310 0.316 0.317 0.336 0.340 0.350 Lithuania 0.224 0.373 0.323 0.303 0.304 0.306 0.305 0.305 0.325 Macedonia FYR 0.340 0.368 0.373 Moldova 0.241 0.343 0.371 0.365 0.350 0.357 0.345 0.328 Poland 0.235 0.264 0.268 0.277 0.296 0.302 0.305 0.307 0.320 Romania 0.255 0.282 0.274 0.283 0.282 0.286 0.294 0.289 Russiaa 0.238 0.395 0.353 0.369 0.357 0.349 0.339 0.338 0.332 Serbia 0.292 Tajikistan 0.289 0.327 Ukraine 0.233 0.325 0.285 0.293 0.303 0.274 0.268 Uzbekistan 0.250 0.333 0.453 0.355 0.326 0.354 Sources: Data in bold are based on comparable consumption indicator from ECAPOV II (World Bank 2005b). Data in italics are from direct survey data estimates from other sources-- ECAPOV I (World Bank 2000), World Development Indicators, and Milanovic (1998)--and are based on grouped data. Data for Poland in italics are from Keane and Prasad (2002a), (consumption per capita without durables) and refer to 1990 for the period 1989­92. Only figures from ECAPOV II are consistent across time. Note: Empty cells are for years with no available data. a. Based on HBS, except for 2003, where NOBUS data are used. TABLE 3. Representative Countries: Inequality Indices (Gini) for Comparable Per Capita Consumption Gini index, and years Georgia Hungary Kazakhstan Latvia Moldova Poland Romania Russia Tajikistan Initial year 1999 1993 2001 1998 1998 1998 1998 1997 1999 End year 2002 2002 2003 2002 2002 2002 2002 2002 2003 Gini index, initial year 0.397 0.232 0.346 0.336 0.371 0.296 0.274 0.353 0.289 Gini index, end year 0.391 0.250 0.318 0.350 0.328 0.320 0.289 0.338 0.327 Source: Authors' estimates. 69 70 | PRADEEP MITRA AND RUSLAN YEMTSOV Main Drivers of Inequality in Transition To what extent can one appeal to the literature on inequality in transition for guid- ance on explanations of these disparate trajectories of growth and inequality? That literature suggests that the principal determinants of inequality in transition were: · Wage decompression and growth of the private sector · Restructuring and unemployment, reverting to subsistence economy · Changes in government expenditure and taxation · Price liberalization, inflation, and arrears · Asset transfer and growth of property income · Technological change, increased mobility, and globalization. Driver 1. Wage decompression and growth of the private sector Transition involved the emergence of a private sector, which was to grow over time. By 2004, over 60 per- cent of GDP was produced in the private sector (EBRD 2005). This shift changed the process of wage setting by introducing a tighter link between productivity and wages. It is usual to associate inequality outcomes more closely with labor market conditions, primarily to inequality in wages, which in turn depends on the level of returns to human capital and changes in endowments. Indeed, wage inequality is a major driver of overall inequality. At first glance, data on wage inequality appear to mirror those for inequality of consumption. Contrasting different data sources, based both on enterprise records and household surveys, figure 4 reports available Gini indexes for wages in Russia and Poland. While dispersion between different sources is indeed very large, the levels and patterns closely resemble the trends depicted in figure 1. Why did wage inequality in Russia increase so rapidly? Returns to education alone seem to be insufficient to explain it. Cross-country studies find that the returns to education increased from the "pretransition" period to the "early transition" period. The meta-study by Fleisher, Sabirianova, and Wang (2004) suggests that the sharpest increases occurred during the early years of transition. Flabbi, Paternostro, and Tiongson (2005) examine the evolution of the skills premium in transition economies through the late 1990s or the period thereafter through 2002 or 2003 using ISSP data, which is an internationally comparable survey. However, neither that study nor the other sources reported in table 4 produce any evidence that Russia stands out as having particularly large or distorted patterns of returns to education compared to Poland (or to other economies in transition such as Hungary; see Campos and Jolliffe 2003). Indeed it started with a much lower level of returns but by the mid-1990s had already converged to Polish levels. This factor therefore cannot be used to explain the excess inequality of Russian wages; other explanations are required. Arrears, as reported by Lehmann and Wadsworth (2001), were responsible for up to a third of the "excess" inequality in wages in Russia. At the peak of wage arrears in November 1998, 64 percent of workers were owned back wages and the Gini INCREASING INEQUALITY IN TRANSITION ECONOMIES | 71 FIGURE 4. Russia and Poland: Gini Index for Monthly Wages, Various Sources a. Russia 0.55 0.50 0.45 x 0.40 inde 0.35 Gini 0.30 0.25 0.20 1989 1991 1993 1995 1997 1999 2001 2003 Year Official wage, GKS RLMS ISSP b. Poland 0.55 0.50 0.45 x 0.40 inde 0.35 Gini 0.30 0.25 0.20 1989 1991 1993 1995 1997 1999 2001 Year Official, GUS HBS LFS ISSP Sources: Russia and Poland, ISSP, from Flabbi, Paternostro, and Tiongson (2005). For Russia, GKS from Goskomstat; RLMS from Lukianova (2005) for contractual wages (cleaning out the effect of arrears). For Poland, HBS from Keene and Prasad (2002b); LFS from Newell and Socha (2005); GUS from World Bank (2004). Note: GKS = Russian Statistical Office (Goskomstat) Survey of Wages; GUS = Polish State Statistical Office; ISSP = International Social Survey Program; LFS = Labor Force Survey; RLMS = Russian Longitudinal Monitoring Survey. index for wages actually paid was as high as 0.58 (Lehmann, Wadsworth, and Acquisti 1999). By 2004 the share of workers who were owed wages fell to 15 per- cent6 and the Gini index for paid wages fell to around 0.44: that is, by just less than a third (Lukianova 2005). But even at this level, wage inequality was considerably higher than in Poland or other countries in Central and Eastern Europe. Therefore this factor, while providing a partial explanation for the inverse U-shape of the evo- lution of wage inequality in Russia, does not fully account for excess inequality in the distribution of earnings. 72 | PRADEEP MITRA AND RUSLAN YEMTSOV TABLE 4. Poland and Russia: Returns to Education Poland 1991 1992 1993 1994 1995 1996 1998 2001 2002 Years of educationa 0.060 0.071 0.081 0.080 0.079 0.070 0.081 0.092 0.106 Dataset ISSP ISSP ISSP ISSP ISSP ISSP ISSP ISSP ISSP Russia 1985 1990 1991 1996 1997 1998 2000 2001 2002 Years of educationa 0.028 0.039 0.028 0.079 0.072 0.091 0.094 0.084 0.097 Dataset RLMS RLMS ISSP RLMS ISSP RLMS RLMS ISSP RLMS Sources: ISSP data from Flabbi, Paternostro, and Tiongson (2005); RLMS data from Gorodnichenko and Sabrianova Peters (2004). Note: ISSP = International Social Survey Program; RLMS = Russian Longitudinal Monitoring Survey. a. Controls include gender, location, age, and family status. Another explanation is provided by distinct differences in minimum wages, which were set at around 40 percent of the average wage in Central and Eastern Europe, as opposed to at 10 percent of the average wage in Russia (see Rutkowski 2001 and World Bank 2005c). This allowed Russian firms to maintain low-paid jobs that otherwise would have been economically unviable, so that low minimum wages were a very important policy-induced factor contributing to higher wage dispersion. As opposed to relatively stable sectoral and interindustry wage differentials, regional variation in real wages, relative to the national average, almost tripled in Russia between 1995 and 2003 (World Bank 2005d). Segmentation of labor markets is a common feature of many transition economies, but in Russia this dispersion takes particularly extreme forms due to institutional, infrastructure, and geographical realities (Earle and Sabirianova 2002). Increasing wage inequality in the transition countries of Eastern Europe and the former Soviet Union reflects a rising education premium, minimum wage policies, and increased divergence of wages across sectors, regions, and occupations. But wages, although important, were not the only determinant of inequality outcomes. The following factors played a role as well. Driver 2. Restructuring and unemployment The closure and restructuring of firms, together with the entry of new firms, is central to transition, as resources are reallocated to more productive uses. Associated labor market developments have manifested themselves in a combination of open unemployment, lower labor force participation, and low-productivity employment, such as subsistence agriculture or informal sector activities. Ex ante there was little insight into what the incidence of job losses and its distribution across households would look like. Ex post there are indeed important variations across countries and regions in the implied effects on inequality. The role of employment status as a contributor to inequality is examined below, in the decomposition of inequality among households partitioned by labor market status. Driver 3. Changes in government expenditure and taxation The system of social transfers was a sizeable factor initially thought to act to countervail increasing INCREASING INEQUALITY IN TRANSITION ECONOMIES | 73 inequality. But in practice it had its intended effect only in a few EU-8 countries, particularly Hungary, where social assistance programs expanded in real terms.7 In contrast, low-income CIS countries, faced with fiscal stringency, drastically reduced coverage of their safety nets to focus on the most needy. Other CIS countries aimed at retaining key benefits but compressed levels to a simple per capita distribution among the claimants. The role of transfers as a contributor to inequality is examined below, in the decomposition of inequality by source of income. On the revenue side, the transition induced a dramatic shift in the composition and incidence of taxes, such as the introduction of value added tax, while tax compliance declined. Limited empirical evidence suggests that most changes worked in favor of greater equality, but with significant variation across countries and time periods.8 Driver 4. Price liberalization, inflation, and arrears All the socialist economies embarked on the process of transition with a substantial monetary overhang (Flem- ming and Micklewright 1999). Hence when prices were liberalized, they jumped and inflation rates tended to persist. Experience from other high inflation episodes, such as in Latin America, points to strong redistributive effects. Aggregate data indeed indicate that the inflation tax in Russia appears to have had a powerful effect. In 1992, for example, it has been estimated that households were hardest hit by inflation, los- ing about 12 percent of GDP through this tax on financial assets (Commander and Lee 1998). This amounted to roughly a quarter of household income and is likely to have been regressive. Similar if not more redistribution took place in Belarus, Bulgaria, Georgia, and Ukraine, but did not substantially affect the EU-8 countries. Arrears on pensions and social benefits payments appeared in the inflationary environment of several countries in the CIS and South-Eastern Europe. Arrears were concentrated in the bottom part of the distribution and, in a highly inflation- ary environment, resulted in a cut in real wages in a highly unequalizing way (Lehmann and Wadsworth 2001). Similar effects have been found by Klugman (1998) for Uzbekistan. These factors, however, were largely transitory in nature and affected the shape of the distribution only in certain time periods. Driver 5. Asset transfer and growth of property income Perhaps the most visible sign of transition everywhere has been the large-scale transfer of previously publicly owned assets into the hands of private agents, a development that has produced a long-term shift in the distribution of wealth. The increase in the share of entrepreneurial income, and the share of families receiving financial income, was an immediate result common to all transition economies. In Russia, for example, the share of property, interests, and profits in the cash receipts of households increased from around 4 percent in 1989 to 20 percent in 2003 (Goskomstat). These sources of income are known to be unequalizing (Milanovic 1998). The role of entrepreneurial income as a contributor to inequality is examined below, in the decomposition by source of income. Many privatization programs are therefore believed to have worsened the distri- bution of assets and income, at least in the short run.9 As against this, it should be 74 | PRADEEP MITRA AND RUSLAN YEMTSOV noted that a large part of national wealth was transferred in a rather equitable way through privatization of housing to tenants at below-market prices. In Russia, by early 1996 nearly 50 percent of the housing stock was in private hands, a proportion that had grown to 70 percent by 2005. Imputing an economic value to subsidized goods and assigning it to households in different parts of the distribution shows that this had mitigating effects on inequality (see Flemming and Mickewright 1999). Driver 6. Technological change and globalization Technological change and mod- ernization of the economy in a broad sense have been important in the evolution of inequality in many countries. Atkinson (2003) shows that transition economies were not alone in experiencing growing inequality: there has been an increase in inequal- ity in many OECD countries because of the change in technology associated with globalization: that is, a rise in the premium for skilled workers and a decline in the relative wage of unskilled workers. Figure 5 shows the extent of inequality increases in Austria, China, Georgia, Hungary, Mexico, Poland, Russia, Sweden, the United Kingdom, and the United States, using what is believed to be the most reliable indi- cator of dispersion in living standards for each country. The figure demonstrates that the increase in inequality in transition economies indeed occurred against the back- drop of a global increase in inequality--with, however, important variations across FIGURE 5. Selected Countries: Gini Index Changes between 1980s and 2000s 50 40 x inde Gini 30 20 Austria China Georgia Hungary Mexico Poland Russia Sweden United United Kingdom States Mid-1980s Early 2000s Sources: ECAPOV II (World Bank 2005b) for Georgia, Hungary, Poland, Russia; Ravallion and Chen (2004) for China; LIS (www.lispoject.org) for other countries. Note: Levels of Gini index are not comparable across countries, as different concepts and definitions of welfare are used. For Europe and Central Asia for the early 2000s, current consumption per capita without housing rental values, correcting for regional price differences and without outliers. Data for the early 1980s come from published sources and refer to total expenditures, not correcting for price differences. For China, total incomes per capita, correcting for price differences. For OECD countries, per equivalent adult total money incomes, correcting for regional price differences without outliers. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 75 countries. It is therefore inherently very difficult to separate transition-related determinants from the global factor of technological progress. How different drivers interact with one another is very much a question of particu- lar country circumstances, initial conditions, and most importantly, policy choices. The taxonomy of drivers offers some basic insights into the reasons behind the variation across countries with respect to increases in inequality. Models of transition provide further guidance regarding the role of policies. Models of Restructuring Aghion and Blanchard (1994) proposed a theoretical model of transition dynamics describing the reallocation of productive resources in transition. The transition is for- malized as a reallocation of labor and capital across state and private sectors, with unemployment as a transient step between the two,10 and highlights ways in which endowments and policies in transition affect the distribution of income (Commander and Tolstopiatenko 1996). In the CIS, with an ungenerous benefits regime and low initial values for closure and restructuring, the reallocation of labor to the private sector is protracted, and inequality rises gradually and steadily to high levels. More generous benefit regimes with higher probabilities of restructuring, as in Central Europe, lead to unemploy- ment peaking at higher levels, but given a rapid movement of workers into the pri- vate sector and a generous floor in the form of unemployment benefits, the rise in inequality is less pronounced, reaching a hump at a lower level than that observed in the first scenario (figure 6). An attractive feature of the model is the conceptualization of restructuring: not as a one-time shift in the behavior of agents, but as a whole array of outcomes with different degree of rent appropriation by insiders in partly restructured enter- prises. The empirical study of the first ten years of transition revealed the coexis- tence of new, partly restructured, and unrestructured firms as a defining feature of the move from a command to a market economy (World Bank 2002e). This introduces an additional source of variability and hence inequality, which is captured by the model. However, a comparison of predictions from the model with empirical evidence shows a surprising reversal of patterns between Central and Eastern Europe (CEE) and the CIS. The inverse U-shaped trajectory of inequality seems to emerge not in the CEE, but in some CIS countries. The failure of the model to predict the actual evo- lution of inequality may be a result of its limitations,11 or alternatively may reflect the effect of mitigation offsetting policy measures. Despite these limitations, the ability of the model to portray a large variation in the levels and shape of the development of inequality is instructive. It suggests that there is probably no single "transition" story as far as the evolution of inequality is con- cerned. Furthermore, the model results are broadly consistent with the story emerging from the earlier description of different drivers as being one that can yield a wide vari- ety of outcomes across countries and over time. A number of these factors are directly or indirectly influenced by policies, which also differed across countries. 76 | PRADEEP MITRA AND RUSLAN YEMTSOV FIGURE 6. Simulation Results from the Restructuring Model with Different Configuration of Parameters: Two Typical Trajectories for Inequality Indices a. CIS 0.50 0.45 0.40 Gini 0.35 0.30 Theil_L 0.25 Inequality 0.20 0.15 0.10 0.05 0 0 5 10 15 20 25 30 35 40 45 Time b. CEE 0.50 0.45 0.40 0.35 Gini 0.30 0.25 Theil_L Inequality 0.20 0.15 0.10 0.05 0 0 5 10 15 20 25 30 35 40 45 Time Source: Commander and Tolstopiatenko 1997. Note: Time is in model years. Decomposing Inequality Change in Transition The most direct approach to capturing the relative importance of the drivers empir- ically is to decompose inequality into its components and associate each component with a particular channel of redistribution. The structure of inequality by income INCREASING INEQUALITY IN TRANSITION ECONOMIES | 77 source can be looked at in two ways: as inequality coming from between and within economic group differences. Following Shorrocks (1982), the contribution of each component of income to total inequality can be obtained from the product of the concentration coefficient for each component and the respective weights of those components in total income. Concentration coefficients in turn depend on how unequally an income source is distributed ("own Gini") and how closely it is correlated with total income.12 The product of the share of a particular type of income and its concentra- tion coefficient equals the contribution of that income component to the Gini index. The sum of these contributions equals the Gini index. Following Milanovic (1999), the main income sources taken to represent key drivers of the level and changes in inequality in transition are: wage income, pensions, social transfers and non-wage income (a combination of all other income sources, ranging from in- kind subsistence income, farm incomes, and remittances to property income and incomes from self-employment and entrepreneurial activities). This stylized frame- work, by focusing on the relative importance of structural shifts versus own (or within) inequality effects, is helpful in understanding how inequality levels changed during transition. Given its potential for an analytical description of changes in the sources of inequal- ity, it is somewhat surprising that there are only a few studies that use this frame- work.13 This section of the paper performs such empirical decomposition exercises going back 15 years, as opposed to the hypothetical exercises presented by Milanovic (1999), to understand the implications for inequality of employment reallocation between shrinking state and growing private sectors. The subsection that follows looks at the evolution of the structure of income over time and the contribution of each component to inequality in Russia and Poland over the period 1987 through 2002. The choice of these countries for analysis was dictated by data limitations. The subsection after that presents group-based decompositions using data on consumption inequality for a larger group of countries. These are the key contributions of this paper. Decomposition of Inequality by Income Sources Tables 5 and 6 report levels and changes over time in the structure of incomes, the concentration coefficient of each component of income, and its contribution to the Gini index of income inequality in Poland and Russia, respectively. The following points may be noted. First, by the end of the period, the Gini index was seven points larger in Russia (0.41) than in Poland (0.34). Second, the directions of change in the income structure in both countries were generally similar, reflecting transition-related drivers: that is, falling share of wages and a rising share of both entrepreneurial incomes (profits and income from self- employment) and pensions. The changes are consistent across periods of economic decline and growth, with wages and transfers moving in opposite directions. But the distributional outcomes were very different. Only to a limited extent were the differences due to changes in the composition of the source of income, such as the much steeper rise in social transfers (including 78 | PRADEEP MITRA AND RUSLAN YEMTSOV TABLE 5. Poland: Contribution of Income Sources to Total Inequality, 1987­2002 1987 1994 1998 2002 1994­87 2002­1994 Income structure: Percent Work income 60 54 56 55 ­6 +1 Of which wages 55 46 48 47 ­9 +1 "Entrepreneurial" 5 8 8 8 +3 +0 Income from farm 13 10 7 4 ­3 ­6 Old-age pension 17 24 26 24 +7 +0 Social transfers 5 5 6 9 +0 +4 Other income 5 6 4 8 +1 +2 Total income 100 100 100 100 Inequality: Concentration coefficients Work income 0.260 0.330 0.388 0.431 +27% +31% Of which wages 0.251 0.302 0.350 0.394 +20% +30% "Entrepreneurial" 0.360 0.488 0.613 0.650 +35% +33% Income from farm 0.415 0.390 0.471 0.575 ­6% +47% Old-age pension 0.171 0.175 0.204 0.263 +3% +50% Social transfers ­0.100 0.080 ­0.017 ­0.011 +180% ­86% Other income 0.340 0.283 0.450 0.263 ­17% ­7% Decomposition: Gini index, contributions Gini, per capita 0.250 0.280 0.320 0.343 +0.030 +0.063 income Work income 0.156 0.178 0.217 0.237 +0.022 +0.059 Of which wages 0.138 0.139 0.168 0.185 +0.001 +0.046 "Entrepreneurial" 0.018 0.039 0.049 0.052 +0.021 +0.013 Income from farm 0.054 0.039 0.033 0.023 ­0.015 ­0.016 Old-age pension 0.029 0.042 0.053 0.063 +0.013 +0.021 Social transfers ­0.005 0.004 ­0.001 ­0.001 +0.009 ­0.005 Other income 0.017 0.017 0.018 0.021 +0.000 +0.004 Sources: Milanovic (1999) for 1987; ECAPOV I (World Bank 2000) for 1994; World Bank 2004; authors' calculations based on HBS data for 2002. pensions) in Poland compared to Russia. To examine to what extent these differences matter, it is not sufficient to simply compare actual changes across countries because the observed change is a complex result of interactions between drivers pulling inequality in different directions. What is needed is a counterfactual. However, producing a fully satisfactory counterfactual distribution is difficult and requires building a model of household income (Bourguignon, Ferreira, and Leite 2004), which goes beyond the scope of this paper. But it is feasible to conduct simulations using either base period concentration coefficients or income shares. It is recognized that such a counterfactual is purely hypothetical because share and concentration often change for the same reason. With this caveat in place, simple simulations show that the differences in the pace of structural change in income sources do not fully explain the inequality differential INCREASING INEQUALITY IN TRANSITION ECONOMIES | 79 TABLE 6. Russia: Contribution of Income Sources to Total Inequality, 1989­2004 1989a 1992 1996 1998 2004b 1998­89 2004­98 Income structure: Percent Work income 79 67 48 55 62 _24 +7 Wages 74 61 34 49 54 ­25 +5 "Entrepreneurial"c 5 6 14 6 8 +1 +2 Income from farm 4 8 15 11 8 +7 ­3 Old-age pension 8 10 18 20 17 +12 ­3 Social transfers 7 6 2 2 2 ­5 +0 Other income 2 9 17 13 11 +10 ­1 Total income 100 100 100 100 100 Inequality: Concentration coefficients Work income 0.285 0.540 0.679 0.540 0.515 +90% ­5% Wages 0.280 0.531 0.644 0.514 0.454 +84% ­12% "Entrepreneurial"c 0.360 0.633 0.764 0.750 0.925 +108% +23% Income from farm 0.300 0.350 0.440 0.573 0.375 +186% ­35% Old-age pension ­0.200 ­0.140 0.111 0.025 0.094 +113% +276% Social transfers 0.086 0.317 0.500 0.450 0.150 +425% ­67% Other income 0.200 0.833 0.512 0.492 0.373 +146% ­24% Decomposition: Gini index, contributions Gini, per capita 0.22 0.47 0.51 0.45 0.41 +0.206 -0.024 income Of which: Work income 0.225 0.362 0.326 0.297 0.319 +0.072 +0.022 Wages 0.207 0.324 0.219 0.252 0.245 +0.045 ­0.007 "Entrepreneurial"c 0.018 0.038 0.107 0.045 0.074 +0.027 +0.029 Income from farm 0.008 0.028 0.066 0.063 0.030 +0.055 ­0.033 Old-age pension ­0.016 ­0.014 0.020 0.005 0.016 +0.021 +0.011 Social transfers 0.006 0.019 0.010 0.009 0.003 +0.003 ­0.006 Other income 0.004 0.075 0.087 0.059 0.041 +0.055 ­0.018 Source: 1992­98 from Commander, Tolstopiatenko, and Yemtsov (1999); RLMS (2004). a. Figures for 1989 from Milanovic (1999) based in HBS. b. Authors' estimates based on RLMS data. c. Includes in-kind and cash incomes from nonagricultural self-employment, informal work, and property income. between Russia and Poland. In fact, an application of Poland's income structure to Russia's concentration coefficients yields a Gini index for Russia that is just 1 point lower than it actually was. And application of Russia's income structure to Poland would have led to a Gini index about 3 points higher than actual in the latter country. The results are very different for the simulations focusing on the impact of chang- ing concentration coefficients. Application of end-period concentration coefficients to Russia's original income structure would have resulted in inequality exceeding its actually observed level by at least 5 percentage points. For Poland the result is striking: application of Polish end-period concentration coefficients to the original income 80 | PRADEEP MITRA AND RUSLAN YEMTSOV structure would have resulted in a Gini coefficient of about 0.45, a level observed in Russia during this period and 10 points higher than the actual outcome in Poland. This exercise thus shows that changes in structure and in concentration coeffi- cients offset each other--more so in Poland--but that the factors that increase inequality within income sources clearly dominate. Among these sources of change, three need to be mentioned. First, labor income is the main source of livelihood, and distribution of earnings is the main determinant of overall inequality. But the shape of the distribution is also determined by concentration. Increasing concentration coefficients of wages drove up the overall Gini coefficient in both countries, contributing around 25 points to inequality in Russia, and 18.5 points in Poland. The difference between these contri- butions, which is 6.5 Gini points, is almost the entire difference between the Gini indices for Poland (0.34) and Russia (0.41). At the same time, the concentration coefficient for wages in Poland (0.39 in 2002) is surprisingly large and not much lower than that for Russia (0.45 in 2004), despite the "own" Gini indices for wages being significantly lower, as shown in figure 4. This is due to different degrees of polarization of labor incomes in the two countries: in Poland a share of households as large as 47 percent (in 2002) did not receive any wage income, compared to 35 percent in Russia (in 2004), reflecting a more sizeable adjustment in employment in Poland compared to Russia, among other things (see also Rutkowski 1996). The second determinant of changes in inequality are transfers, pensions, and other social benefits. The effect of transfers on inequality was not uniform, and changes were mostly driven by changes in the size and the distribution of pensions. In both countries, changes in the distribution of pensions played a significant role as contrib- utors to the increase in inequality. But since their concentration coefficients were below those of market income sources, this expansion reduced inequality compared to potential levels. Had there be no increase in transfers in Poland, inequality would have been fully 3 Gini points (or 10 percent) higher. The effects would indeed have been more progressive had there been no unequalizing change in the concentration coefficients of pensions.14 Other social transfers, on the other hand, played a dramat- ically different role in Poland and Russia: thus, for example, the failure to target social benefits in Russia, as shown by a rapid increase in their concentration coeffi- cient in early transition, is in sharp contrast to the situation in Poland. The third broad driver of inequality is private sector growth, combined with increasing informality. The latter is difficult to measure with precision since the data cover reported incomes, which are known to underestimate informal incomes signif- icantly (Yemtsov 2001). In particular it is important to distinguish between survival- type activities, new entrepreneurial incomes, and incomes from property. Informal income in various guises features in different parts of the income spectrum: in farm income, in the form of in-kind consumption from own land plots; in entrepreneurial income, as many businesses are not registered, or in the form of "side" wages reported as a result of freelancing; or in "other income," especially in the CIS, where this term is often used as a euphemism for not fully legal or untaxed income. In terms of sheer size, its effects were large. It is also quite remarkable that in the post-1998 INCREASING INEQUALITY IN TRANSITION ECONOMIES | 81 crisis period in Russia, some income sources with a strong informal component, such as farming and other incomes, show a fall in their concentration coefficients. To summarize these results briefly, the comparison of Poland and Russia from the late 1980s to early 2000s finds that there is no single determinant of inequality. Dif- ferent drivers, at times working in opposite directions, combined to create a complex patchwork, which is rich enough to allow a wide variety of outcomes. The analysis now turns to a more in-depth examination of spatial and other group-based factors of inequality. Decomposition of Inequality by Groups A notable drawback of inequality decompositions based on components such as those presented in the previous subsection is their reliance on income data, with the atten- dant problems of accurate reporting discussed in the second section of this paper. However, with the population divided into groups affected by transition, total inequality can also be represented as the sum of inequality from within each of the groups, and part of the inequality comes from differences in means between these groups. Decompositions of inequality by groups allow one to move to indicators of inequality in consumption, which is superior to income in terms of data quality. This subsection decomposes consumption inequality in seven representative tran- sition countries into the contribution of inequality "between" groups and inequality "within" groups using the Theil entropy measure of inequality (Bourguignon 1979; Shorrocks 1980). The sum of the within- and between-group contributions equals 1. Table 7, panels a­c, shows to what extent inequality can be explained by inequal- ity between groups, such as rural residents versus city dwellers, high school graduates versus those with less education, and working families versus jobless households.15 The choice of these partitions is designed to capture some key dimension of transition such as the emergence of new social classes and changing distribution within those classes. While none of them corresponds as neatly to a set of drivers of inequality as the distribution of income by source, they complement the story emerging from the decomposition of income in an important way. First, they identify winning and losing groups more clearly than is possible with decomposition by income source. Second, differences by educational attainment help assess the magnitude and dynamics of inequality effects related to technological change. Third, they add location effects, which account for a significant share of inequality in virtually every country, since transition resulted in changes in the concentration of economic activity and migration. Urban-Rural (Location) (Table 7a) Changes in structure. There were significant changes in the distribution of population across locations. In Hungary, the share of rural areas dropped from 38 percent to 35 percent of population during the period under review. In Latvia, the share of the capital city of Riga increased from 33 to 38 percent, and in Tajikistan the share of rural areas dropped from 78 to 73 percent in just five years. People have been migrating to higher-income areas. As a result, this driver reduced inequality. Had the initial distribution of population by location stayed the same, inequality in Latvia, for 82 TABLE 7a. By Location (for Inequality Measured by Consumption Per Capita) Latvia Hungary Poland Romania Russia Moldova Tajikistan 1998 2002 1993 2002 1998 2002 1998 2002 1997 2002 1998 2002 1999 2002 Theil entropy measure 0.198 0.254 0.149 0.126 0.206 0.217 0.167 0.178 0.205 0.186 0.240 0.209 0.142 0.190 Decomposition of Theil inequality measure (percent) 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Of which between locations 6 16 5 3 5 7 3 11 3 4 9 9 3 4 Of which within capital city 37 50 30 19 5 14 nd nd 8 7 22 27 9 15 Of which within other urban areas 34 19 58 51 56 58 57 70 18 14 18 20 Of which within rural areas 23 16 28 29 31 28 42 31 32 19 50 51 69 62 Theil entropy index for per capita consumption Capital city 0.189 0.241 0.194 0.127 0.187 0.266 nd nd 0.182 0.178 0.212 0.214 0.151 0.217 Other urban areas 0.189 0.177 0.124 0.124 0.198 0.195 0.157 0.160 0.177 0.181 0.230 0.176 0.158 0.213 Rural areas 0.184 0.189 0.125 0.116 0.189 0.190 0.170 0.154 0.258 0.168 0.218 0.184 0.131 0.169 Population shares (percent) Capital city 33 38 19 17 4 8 nd nd 6 6 17 18 6 9 Other urban areas 37 32 43 48 55 52 55 54 68 68 19 19 16 18 Rural areas 31 30 38 35 41 40 45 46 27 26 63 63 78 73 Real means, relative to national mean per capita consumption = 1.00 Capital city 1.193 1.357 1.251 1.149 1.420 1.386 nd nd 1.504 1.148 1.467 1.427 1.385 1.370 Other urban areas 0.976 0.852 0.993 1.024 1.093 1.088 1.085 1.181 0.980 1.065 0.998 0.882 1.032 0.999 Rural areas 0.824 0.698 0.885 0.897 0.834 0.805 0.898 0.783 0.944 0.801 0.873 0.911 0.962 0.952 TABLE 7b. By Education of the Household Head (for Inequality Measured by Consumption Per Capita) Latvia Hungary Poland Romania Russia Moldova Tajikistan 1998 2002 1993 2002 1998 2002 1998 2002 1997 2002 1998 2002 1999 2002 Theil entropy measure 0.198 0.254 0.149 0.126 0.206 0.217 0.167 0.178 0.201 0.181 0.240 0.209 0.142 0.189 Decomposition (percent) Decomposition: 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Of which between education group 7 22 11 12 13 26 11 20 2 5 8 8 4 4 Of which within primary education 15 9 23 19 18 14 32 25 5 2 24 21 28 22 Of which within second education 54 34 5 5 26 19 25 25 31 29 32 38 24 35 Of which within vocational education 4 6 38 40 24 21 19 18 39 37 23 18 27 18 Of which within tertiary education 20 28 22 23 19 20 13 13 22 27 14 15 17 21 Theil entropy index for per capita consumption Within primary education 0.158 0.171 0.136 0.105 0.165 0.181 0.155 0.145 0.256 0.159 0.219 0.166 0.126 0.202 Within secondary education 0.189 0.189 0.110 0.095 0.184 0.130 0.139 0.138 0.193 0.176 0.237 0.214 0.141 0.175 Within vocational education 0.267 0.195 0.113 0.109 0.163 0.158 0.139 0.139 0.207 0.177 0.220 0.189 0.136 0.176 Within tertiary education 0.183 0.224 0.188 0.125 0.208 0.195 0.169 0.153 0.176 0.164 0.203 0.189 0.148 0.178 Population shares (percent) Primary education 23 21 30 28 24 22 40 39 5 3 31 30 34 22 Secondary education 58 48 7 7 27 28 28 33 40 39 35 40 26 42 Vocational education 4 11 52 49 38 37 25 20 34 34 23 19 27 18 Tertiary education 15 19 12 17 11 13 8 8 22 24 11 11 13 18 Real means, relative to national mean per capita consumption = 1.00 Primary education 0.814 0.644 0.844 0.824 0.827 0.789 0.884 0.789 0.902 0.794 0.842 0.866 0.922 0.919 Secondary education 0.980 0.944 1.138 1.079 1.150 1.263 1.084 0.981 0.904 0.880 0.922 0.930 0.942 0.902 83 Vocational education 0.902 0.761 0.962 0.962 0.817 0.692 0.915 1.123 0.997 0.986 1.092 1.061 1.034 1.108 Tertiary education 1.377 1.669 1.483 1.365 1.619 1.647 1.621 1.804 1.174 1.217 1.502 1.514 1.239 1.221 84 TABLE 7c. By Household Labor Market Status (for Inequality Measured by Consumption Per Capita) Hungary Poland Romania Russia Moldova Tajikistan Georgia 1993 2002 1998 2002 1998 2002 1997 1999 2002 1998 2002 1999 2002 1999 2002 Theil entropy measure 0.087 0.107 0.152 0.181 0.128 0.149 0.218 0.214 0.193 0.24 0.209 0.149 0.187 0.279 0.271 Decomposition (percent) 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Of which between groups 2 2 1 0 4 1 0 1 5 9 8 2 0 3 1 Of which within the group of wage earners 52 53 38 40 55 69 62 63 67 43 38 13 40 17 19 Of which within the group of self-employed 16 15 43 38 18 7 5 9 14 10 18 19 39 41 39 Of which within the group of subsistence farmers 2 2 3 3 11 8 26 19 8 29 19 66 14 29 28 Of which within the group of nonworking 27 27 15 18 13 14 6 8 6 8 16 1 6 11 13 Theil entropy measure Within "formal" wage earners 0.084 0.104 0.145 0.171 0.118 0.142 0.194 0.207 0.184 0.239 0.208 0.171 0.21 0.228 0.205 Within "informal" of self- employed 0.077 0.096 0.157 0.189 0.156 0.243 0.172 0.2 0.203 0.277 0.213 0.159 0.17 0.239 0.247 Within "informal" subsistence farmers 0.086 0.089 0.146 0.202 0.121 0.133 0.339 0.233 0.156 0.187 0.134 0.138 0.16 0.339 0.354 Within LM group of nonworking 0.092 0.114 0.147 0.180 0.126 0.149 0.196 0.221 0.18 0.196 0.243 0.433 0.24 0.368 0.325 (Continues on next page) TABLE 7c. continued Hungary Poland Romania Russia Moldova Tajikistan Georgia 1993 2002 1998 2002 1998 2002 1997 1999 2002 1998 2002 1999 2002 1999 2002 Population shares (percent) "Formal" wage earners 66 56 41 43 61 56 68 62 67 36 32 11 35 20 23 "Informal" self-employed 8 15 43 37 14 18 7 9 12 7 16 15 42 50 45 "Informal" subsistence farmers 1 3 2 2 13 13 17 20 14 48 38 74 17 20 20 LM group of nonworking 26 27 15 17 12 13 7 8 8 9 13 0 5 10 12 Real means, relative to national mean per capita consumption =1.00 "Formal" wage earners 1.022 0.97 0.988 0.992 1.059 1.077 1.02 1.042 1.055 1.222 1.193 1.071 1.011 1.046 1.086 "Informal" self-employed 1.089 1.172 0.967 0.988 0.824 0.816 0.929 1.037 1.116 1.152 1.095 1.17 1.018 0.942 0.941 "Informal" subsistence farmers 0.845 1.041 1.249 1.263 0.911 0.899 0.99 0.865 0.693 0.784 0.774 0.956 0.955 1.212 1.081 LM group of nonworking 0.921 0.968 1.092 1.01 1.006 1.023 0.903 0.963 0.891 1.161 1.071 1.05 0.931 0.785 0.916 Source: Authors' calculations based on World Bank ECA regional data archive. Note: nd = no data. 85 86 | PRADEEP MITRA AND RUSLAN YEMTSOV example, would have been 15 percent higher and that in Tajikistan 12 percent higher by 2003. Changes in "between" inequality. In general, consumption in rural areas is lower than in urban areas. Capital cities have much higher living standards. In most cases, consumption is about 40 percent higher in capital cities than the national average, but inequality is also higher. Over time, the relative position of rural areas has deteriorated in most countries, sometimes quite sharply. This is common across all countries shown in table 7a. As a result, the "between" component of consumption inequality went up everywhere except in Moldova, where it remained unchanged, and in Hungary, where it fell: the only case of clear convergence between locations. Changes in "within" inequality. "Within" capital city inequality increased every- where, except for Russia, where it stayed virtually the same, and Hungary. In rural areas, "within" location inequality fell for the most part, but remained broadly the same in Hungary and Moldova. There was no clear pattern for changes in inequality within other urban areas. Some part of the decline in inequality within rural areas may be linked to land owner- ship or use rights reforms. A broadly similar redistribution of land occurred in low- income CIS countries such as Armenia and Moldova, which also have labor-intensive agriculture and where it is reasonable to expect the effects to have been equitable. Table 7a also shows a large fall in the Theil entropy index in rural Moldova, as opposed to rural Tajikistan, where land reform has been much less comprehensive. Regional factors. While the urban/rural dichotomy is small, the role of regional differences may be much greater. Thus Yemtsov (2003), using official per capita income data series, shows that "between-regional" factors among Russia's 80-plus regions accounted for about a third of the overall inequality, with the increase in the between-regions component being a key driver of the change in inequality between 1995 and 2000. However, direct survey measurements instead of official data reveal much smaller roles for regional variations: only about 15 percent of overall inequality can be ascribed to the between-regional differences in means, with stability between 1997 and 2002 (World Bank 2005d). Thus, while the persistence of regional factors is evident, their role as drivers of inequality change is not. Lack of convergence across Russian regions in mean real incomes is also presented as a major factor influencing the outlook for inequality going forward by Dolinskaya (2002) and by Fedorov (2002). Summing up, while locational factors play a role as a driver of inequality, it is unlikely that they will strongly influence the dynamics of inequality going forward. It is therefore necessary to focus on within-urban and, for low-income CIS countries at any rate, within-rural drivers as key factors that will determine the evolution of inequality in the future. Education (Table 7b) Changes in structure. The shift toward higher skills is clear and universal. As the structure is changing in favor of groups with higher incomes, while groups with INCREASING INEQUALITY IN TRANSITION ECONOMIES | 87 depressed incomes are becoming smaller, the effect is a reduction in inequality. Rus- sia has a different system of classification with regard to levels of education that is not easily reconciled with those prevailing in the other countries and hence is not fully comparable. Changes in "between" inequality. The share of the "between" component presents a picture strikingly different from that seen in the case of the urban/rural divide. Not only is it much larger, but it also clearly and consistently increases throughout the region. In Latvia and Poland, it accounts for up to a quarter of all inequality. There were also large increases in the "between" component in Romania and in Russia. In Russia, however, the contribution of the "between" component remains small, as it does in Tajikistan. This is because the returns to education, as measured by the relative mean consumption of those with higher education (the bottom panel of table 7b), are low when compared with those in countries of Central and South-East Europe that are further advanced in the transition. Groups with specific skills, such as vocational education, lost in relation to other groups, measured again by returns to education, especially in rapidly restructuring economies such as Latvia and Poland.16 Changes in "within" inequality. Changes in "own" inequality by education group are informative. The contribution of the "within" component in primary education fell everywhere, mostly reflecting its fall in the share of this group within the popula- tion. However, inequality within the group remains large, reflecting the presence of very large losses for some of these individuals. At the same time, the role of the "within" component in tertiary education went up virtually everywhere, particularly in Latvia, Russia, and Tajikistan, contributing between a fifth and a quarter to total inequality in all countries except Moldova and Romania. It is likely that this reflects rapid technological change but also possibly revealed differences in the adaptability of skills in the face of exposure to global competition. Most importantly, inequality among those with the highest skill levels, as measured by the Theil entropy index, exceeds inequality among other education groups in the quickly globalizing economies of Central Europe, where demand for skills is likely to have been shifting rapidly. In Russia, Moldova, and Tajikistan, by contrast, much of the inequality arises in the middle or bottom of the skills distribution--most likely a transitional phenomenon--with these countries lagging behind those in Central Europe with respect to both size and intensity of change. The evolution of consumption inequality by level of education is clearly a complex product of many factors, including policy. Specifically, the extensive use of transfer payments in Central Europe targeted to the unemployed, who are more likely to have the lowest level of skills, might have resulted in their consumption inequality being "artificially" low. Labor Market (Table 7c) Table 7c focuses on the market for labor, dividing up households into groups char- acterized by wage employment, entrepreneurial activities, subsistence activities, and nonemployment (retirement, unemployment, and so on).17 The choice of this parti- tion reflects what is important in transition economies and has been developed by 88 | PRADEEP MITRA AND RUSLAN YEMTSOV one of the authors of this paper for the first time in a recent World Bank study about Growth and Poverty in ECA (2005 b).18 A few broad generalizations emerge. First, the effects of restructuring on income distribution operated not so much through the relative size of between-sector differ- ences, but through the variation in the role of "employment/nonemployment" types and inequality within the group of unemployed or marginally employed (e.g., those who consumed themselves a large part of their output). Table 7c shows that, even by 2002­3, as much as 20 to 40 percent of the population in Georgia, Kazakhstan, and Moldova were in families reliant on subsistence farming and that, in Georgia and Moldova, a further 10 percent had no employed family members. Over the period of analysis, more people moved into employment and fewer people remained in subsistence employment, but, with the exception of Moldova, Russia, and Tajikistan, the shift was not large enough. The allocation of population between employment as a whole and unemployment in more advanced economies is comparatively steadier, suggesting that the transition-induced reallocation is much farther advanced there. Second, the growth of entrepreneurship has been a major contributor to an increase in inequality in many countries. This is because as a group it is associated with higher inequality in outcomes than wage employment or subsistence activities, and its share in total population has generally been rising. There are exceptions to this finding, however: notably Georgia, where a decline in the share of households characterized by entrepreneurial activity has resulted in a falling contribution of this group to inequality. Third, the rise in the contribution to inequality of the nonemployed (transfer recipients) is an important factor behind rising inequality, particularly in the advanced transition economies of Central Europe, but in Romania as well. The increase is due to growing inequality within this group, accompanied, in many cases, by its rising share in total population. Growing inequality among the nonemployed may be a reflection of the increasingly poor opportunities for those who are unem- ployed or out of the labor force to sustain their standard of living (relative to national mean per capita consumption) and can be related to the failure to increase the share of the employed in total population. Beyond these generalizations, how different factors come together is very much a country-specific matter. In Russia, in particular, where overall inequality has somewhat receded during the period under review, the main factor is the shift from self-employment (whether entrepreneurial or subsistence) to wage employ- ment between 1999 and 2002, accompanied by a decline in inequality among wage earners. One factor explaining this decline is the reduction in arrears which, as discussed in the fourth section, has been a feature of the economic recovery after the financial crisis in 1998. Overall inequality declined in Moldova as well. However, this is not due to chang- ing shares of different groups, but a decline in within-group inequality for all major groups: that is, wage employees, entrepreneurs, and subsistence farmers. The reduction in wage inequality may be due to a reduction in arrears. However, the reduction in inequality among agricultural self-employed and rural residents engaged in subsistence INCREASING INEQUALITY IN TRANSITION ECONOMIES | 89 farming is a likely outcome of somewhat delayed, but equitable land reform. In contrast, in Poland and Romania, upward pressure from nonworkers has been rein- forced by rising inequality among wage earners. This is no doubt related to the further decompression in wages in those countries (World Bank 2003a, 2004, 2005a). Sectoral effects. Many survey datasets analyzed in table 7 do not contain detailed sector identifiers that would allow households to be allocated to particular activities. Despite these limitations, it is important to present at least a partial account of the role of sectoral reallocation in the evolution of inequality. This is closely related to changes in the sectoral composition of employment in transition. Intersectoral differ- ences during the 1998 to 2002 period increased their contribution to overall inequality in Russia from 2 to 6 percent, but remained stable in Poland at around 6 percent. The share of services in overall inequality expanded in both Russia and Poland, but whereas the services sector is the most unequal in Poland, it is the second most unequal after manufacturing in Russia.19 Agriculture does not seem to play an active role in those countries. The taxonomy presented in this section can be used to assess what course possi- ble changes in inequality might take for a particular country compared to other coun- tries in the region. Should one expect inequality in Russia, for example, to increase further? This could happen to some extent, reflecting increases in education premia and possibly, but not necessarily, worsening of interregional inequality. So far Russia has been lagging behind countries such as Hungary, Poland, and Romania in the size of the wage premium for education (Rutkowski 2001). There is therefore the potential for some widening of wage differentials between skilled and unskilled labor. While between-regional inequality--which explains up to a third of inequality in Russia--might persist, it need not aggravate an increase in inequality. On the con- trary, to the extent that such inequality has roots going back to central planning, it can be mitigated through freer movement of goods and labor across Russia's regions. In addition, depending on societal attitudes to inequality, intergovernmental fiscal transfers can play an equalizing role as well. Unfortunately, comparable consumption aggregates are not available beyond the 1998­2002 period for all countries in table 7. For this reason it is necessary to rely on published studies and different sources to examine the extent to which the decompo- sition exercises presented in this section can be used to look forward. Extending the time horizon of available data, figure 7 presents a set of results from available studies using group decompositions for Hungary, Poland, and Russia, where the graphs show the contribution of each component to overall inequality. Well in line with priors regarding the increase in education premia, there is a large and increasing contribu- tion from differences between education groups. But alongside the increase in differ- ences across different levels of education attainment, the importance and persistence of locational effects in Russia--and even in Hungary and Poland--is striking. Summing Up The decomposition of changes in inequality over time by income source and socioeconomic group helps identify the forces behind the direction and magnitude of 90 | PRADEEP MITRA AND RUSLAN YEMTSOV FIGURE 7. Russia, Hungary, and Poland: Relative Importance of Between-Groups Inequality over Time percent of total inequality a. Russia 1992 2001 Region 10.2 Region 15.2 Employment 8.7 Employment 6.2 Age 6.6 Education 6.2 Location (U/R) 6.3 Age 6.1 Gender 2.6 Location (U/R) 5.8 Education 1.9 Dependency 1.3 Family size 1.3 Family size 1.1 Dependency 0.9 Gender 0.9 b. Hungary 1992 2003 Education 17.9 Education 27.0 Employment 15.1 Employment 14.0 Location (U/R) 6.0 Location (U/R) 12.0 Age 3.1 Dependency 5.0 Dependency 2.9 Age 3.0 Gender 1.9 Gender 1.0 c. Poland 1994 2002 Education 13.0 Education 19.1 Region 7.9 Region 12.2 Employment 4.0 Employment 6.4 Age 1.7 Age 1.7 Sources: For Hungary, Töth (2004a, 2004b). For Poland, World Bank (2004), Szulc (2003). For Russia, Popova (2005). Note: Decompositions of inequality for Hungary rely on mean log deviation index of inequality based on money incomes. For Poland, Theil entropy index is used as a measure of inequality, and consumption per equivalent adult was used instead of income. For Russia, Theil index of inequality based on which decompositions were made was computed for disposable resources per equivalent adult. For location, U/R = urban/rural. changes in inequality across the transition countries. Although the theoretical frame- work developed to explain changes in inequality in transition does not allow rigorous testing of hypotheses and exact identification of various effects, it allows broad qualitative conclusions to be drawn. These conclusions are summarized in table 8. They show that each of the drivers of inequality, especially the ones specific to tran- sition (1­5), operates through a specific channel and can be mapped by looking at the components of inequality in a particular way. However, the role--and, in some instances, the direction of influence--of each effect differs across countries, depending INCREASING INEQUALITY IN TRANSITION ECONOMIES | 91 Table 8. Identifying the Role of Inequality Drivers with Decomposition Results = inequality increasing, = inequality decreasing Drivers Decomposition by sources Decomposition by groups 1. Wage decompression and · "Own" wage inequality increase · Increase in private sector the growth of private sector · Fall in the share of wages and in unemployment in incomes 2. Restructuring, unemployment, · Coefficent of concentration for · Increase in the number of or reverting to subsistence wages subsistence farmers economy · Increase in share of informal · Increase in the number of incomes unemployed 3. Fiscal adjustment affecting · Changes in the real value of Inequality among transfer government expenditure transfers recipients and taxation · Changes in targeting 4. Price liberalization, · "Excess" inequality in wages · Excess inequality among inflation, and arrears fixed-income recipients (transfer and state sector workers) 5. Asset transfer and property · Property incomes increase · Increase in the number of incomes · Entrepreneurial incomes increase self-employed · Imputed rents 6. Technological change and · Returns to education · Migration to urban areas expansion of knowledge · Increase in variation of returns · Education expansion economy, migration · Premium for highly skilled · Inequality among the skilled Source: Authors. on how advanced they are in the process of transition to a market economy, a consideration not captured in the table. This complexity of interactions between the determinants of inequality results in a clear conclusion: there are no common, all-encompassing explanations for the increase and, in some cases, subsequent decline in inequality in the transition countries across periods of economic decline and growth. The analysis of the paper also suggests that initial conditions and policy choices have been important in shaping the outcomes. Changing Inequality in China This paper started with a comparison of rising inequality between China and Russia and the suggestion that these two phenomena may be more closely linked than usually thought. Table 9 demonstrates that increasing inequality in China is as firmly established a fact as rising inequality in the transition countries of Eastern Europe and the former Soviet Union.20 In what follows, it should be noted, however, that unlike in the case of the Eastern Europe and the former Soviet Union region, microeconomic data from China are not available for the analysis of this paper. The key determinants of inequality in China are very different from what has been observed in Eastern Europe and the former Soviet Union. As figure 8 clearly shows, 92 | PRADEEP MITRA AND RUSLAN YEMTSOV Table 9. China: Increases in Gini Coefficients for Per Capita Incomes, Various Studies Rural Urban Data Study source 1988 1995 2001­02 1988 1995 2001­02 Ravallion and Chen (2004), SSB 0.297 0.334 0.365 0.211 0.283 0.323 Chen and Wang (2002) Wu and Perloff (2004) SSB 0.300 0.338 0.343 0.201 0.221 0.269 Li (2000) SSB 0.301 0.323 0.230 0.280 Khan and Riskin (1998, 2005) CASS 0.338 0.416 0.375 0.233 0.332 0.318 Gustaffson and Li (2001) CASS 0.228 0.276 Wagstaff (2005) CHNS 0.395a 0.419a Meng (2000) CASS 0.234 0.282 Sources: As noted. Note: CASS = Economics Institute of the Chinese Academy of Social Sciences Survey; CHNS = China Health and Nutrition Survey; SSB = State Statistical Bureau based on Household Budget Survey. Empty cells are for years with no available data. a. All China, 1989 to 1997. FIGURE 8. Selected countries in Eastern Europe and the Former Soviet Union versus China: Gap in Per Capita Consumption between Urban and Rural Areas, 1993­2002 90 (%) 80 70 areas 60 urban in 50 40 premium 30 20 10 consumption 0 capita erP ­10 ­20 Moldova Armenia Poland Hungary Romania Russia China 1981 1993 1996 1998 2000 2002 Sources: Authors' estimates. For China, data on real incomes are from Ravallion and Chen (2004). Note: All means include a cost-of-living adjustment. China stands out as a country with an extremely large rural-urban gap. Indeed, Shorrocks and Wan (2005) report that, at an estimated 37 percent in 2000, China has the highest "between" urban-rural component of inequality in the world. A sig- nificant determinant of China's inequality derives from the rural-urban divide: that is, migration from rural to urban areas and rapid changes in the sectoral composition of output, a classic development phenomenon. In contrast, the turbulent early years INCREASING INEQUALITY IN TRANSITION ECONOMIES | 93 of transition in some countries of Eastern Europe and the former Soviet Union witnessed a reversal of this gap, as some sources of livelihood became available in rural areas at a time when unviable enterprises were being restructured or closed in industrial cities. Given the nature of economic development and the comparatively rudimentary nature of safety nets in China, changes in the distribution of wages are an important determinant of the evolution of inequality. As was the case in other transition economies, China had an extremely compressed wage structure in the prereform period, a feature that changed following the onset of reforms. However, the level of inequality remained low until the early 1990s, more than a full decade after economic reforms began (Li 2003), increasing rapidly since then.21 Gustaffson and Li (2001) report that, between 1988 and 1995, the Gini index for urban earnings increased from 24.0 to 30.4. Urban wages were highly and, almost certainly, artificially equalizing in 1988, with a concentration ratio of only 0.178 (Khan and Riskin 2005). In line with the slow pace of reforms, it rose gradually to 0.198 in 1995 (compared to 0.302 in Poland in 1994 and 0.644 in Russia in 1995), and to 0.245 in 2002 (compared to 0.394 in Poland in 2002 and 0.454 in Russia in 2004). While the rapid growth of private, foreign, and mixed-ownership enterprises contributed to this increase, the comparatively slow restructuring of state-owned enterprises is likely to have arrested the pace of change (Knight and Song 2003). In this connection, it is useful to be reminded that, in 2003, over 80 million out of 250 million urban employed in China were working in state-owned enterprises (China Statistical Yearbook 2003). Most importantly, a comparison of wage inequality in China and Russia suggests that, while returns to education were negligible in China--but not in Russia--in 1989, subsequent developments led to an increasing education premium becoming a stronger driver of wage increases in China, albeit from a lower base. In Russia, in con- trast, it played a less prominent role in explaining the evolution of wage inequality. The analysis of regional differences, which played a dominant role in explaining the development of inequality in China, suggests that there are significant impedi- ments to the operation of market forces. Thus Shi, Sicular, and Zhao (2002) explore the question of rural-urban inequality in greater detail for nine different provinces using the China Health and Nutrition Survey (CHNS). Once differences in living costs are taken into account, the authors conclude that the apparent labor market distortion in the form of registration system and other impediments for migration amounts to a rate of apparent taxation on rural wages of 81 percent. Shi (2002) finds that 28 percent of the rural-urban wage difference can be explained directly via the coefficient on registration. Inasmuch as impediments to migration reflect distortions inherited from the command economy, the large role of regional factors as drivers of wage inequality in China and Russia is a phenomenon related to transition. Further reforms in product and labor markets in both countries can be expected to lead to greater equalization of wages across regions. Would faster growth in the transition countries of Eastern Europe and the former Soviet Union be accompanied by increasing inequality on a scale similar to that in China? Inasmuch as the latter derives from the rural-urban divide--namely, migration 94 | PRADEEP MITRA AND RUSLAN YEMTSOV from rural to urban areas and rapid changes in the sectoral composition of output, a classic development phenomenon for which there is no obvious analogue for the transition countries discussed earlier in the paper--the answer is negative. However, looking forward, it is also likely that transition-related factors will become less important in the evolution of inequality in Eastern Europe and the former Soviet Union compared to factors such as technological progress, global changes in skills premia, the effects of demographic changes, and migration. To the extent that China's income distribution is influenced by its increasing integration in world markets, its experience is relevant for Eastern Europe and the former Soviet Union, which have also been integrating into the global economy, in pointing to the role of such long-term factors. That analysis remains to be done. Conclusions and Policy Implications By the early 2000s, the transition countries of Eastern Europe and the former Soviet Union exhibited the full spectrum of inequality outcomes, from fairly unequal to fairly equal. Indeed, developments in economic growth and income inequality over different time periods have been sufficiently rich and varied in Poland and Russia--illustrative cases, respectively, for Eastern Europe and the Commonwealth of Independent States--as well as in China, to cast doubt on any easy generalization about the relationship between growth and inequality. The paper has demonstrated that inequality is the result of complex interactions between initial conditions, country circumstances, and--importantly--policy choices, which need careful analysis. Before turning to the implications of this analysis for policy, it is important to distinguish between equality of opportunity and equality of outcomes. The World Development Report 2006 (World Bank 2005a) makes a persuasive case for policies that promote equality of opportunity, defined as opportunities to pursue a life of an individual's choosing and to be spared from extreme deprivation in outcomes. However, it cites the examples of decollectivization of agriculture in China in the late 1970s and wage decompression in Central and Eastern Europe following the onset of transition in those countries as cases where a history of repressed inequality precludes using the resulting inequality of outcomes to infer inequality of opportu- nities. Indeed, since income differences provide incentives to invest in education, to work, and to take risks, any policy that is cognizant of trade-offs between efficiency and equity will result in inequality of outcomes. A dominant driver of inequality common to Central Europe, China, and Russia has been wage decompression. While the share of wages has declined in the transition economies of Eastern Europe and the former Soviet Union and, more modestly, in urban China, their concentration coefficient--which depends both on how unequally wage incomes are distributed and how closely they are correlated with total income--has increased significantly in all cases. And although wages became less unequally distributed in Russia in the late 1990s and early 2000s, reversing the trend of increasing inequality in earlier years, that reversal is due in part to a reduction of wage arrears, which is a one-time phenomenon. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 95 Could inequality in wages increase further in the transition countries of Eastern Europe and the former Soviet Union? A recent examination of the evidence (Yemtsov, Cnobloch, and Mete 2006) shows that, while rates of return to schooling are low in the transition countries, they are starting to increase and, furthermore, that there is a positive association between progress with market reforms as meas- ured by EBRD transition indicators and returns to schooling. Hence, to the extent that the evolution of wage inequality is a reflection of the education premium, it is certainly possible to envisage greater inequality of wage outcomes as market reforms fully take hold in lagging reformers. An important issue in the CIS countries is the reduction of the informal economy. While self-employment, including subsistence agriculture, played the role of a safety net following the deindustrialization and retrenchment that occurred in the early years of transition, and hence were welfare-improving relative to the potential unem- ployment that would otherwise have occurred, an important policy issue now is how to create more productive jobs. That would also mitigate the unequalizing effect of wage decompression, among other things. The creation of new jobs could be accom- plished through the removal of those elements in the investment climate that confer a disadvantage on new private sector firms, which are important in employment creation (World Bank 2005c). Surveys of the business environment (EBRD 2005) indicate that beyond simplification of firm registration and licensing and reform of tax administration, the creation of a level playing field between state and privatized firms on the one hand and new private sector firms on the other would require "second generation" reforms in the areas of competition policy, the regulatory regime, and institutions that protect property rights, such as the court system. Leveling the play- ing field would lead to restructuring and the exit of unviable firms, accompanied by job destruction, which would need to be managed through more active use of the social safety net. This paper has shown that location is an important determinant of inequality in Russia and that it exerts an influence in Hungary and Poland as well. While this might persist, it need not lead to a further increase in inequality. On the contrary, such inequality, to the extent that it has roots going back to central planning, can be mitigated through freer movement of goods and labor brought about through product and labor market reform. In addition, depending on societal attitudes to inequality, intergovernmental fiscal transfers can play a role as well. The size and targeting of public transfers has had large and persistent effects on income distribution: broadly equalizing in Central Europe, and unequalizing in the CIS. While the absence of pensions--the most significant component of public trans- fers--would have aggravated inequality, pensions were not markedly egalitarian in their incidence, even in Poland, because most pensioners do not fall into the lower end of the income distribution. However, as the contrasting experience of Poland and Russia show, improved targeting of "other social transfers," mainly social assistance, can play a significant role in reducing income inequality and remain a policy instru- ment that can be used in line with a country's preference for inequality, provided they are fiscally sustainable. Finally, while an assessment of the available evidence suggests that further increases in inequality in the transition countries of Eastern Europe and the former 96 | PRADEEP MITRA AND RUSLAN YEMTSOV Soviet Union are not inevitable, the paper identifies several gaps in the understanding of inequality on which future research might profitably focus. Such an agenda would include an in-depth exploration of the nonincome dimensions of inequality and inequality of opportunities; the role of technological change and globalization; housing policies, subsidies, and imputed rents; and the effect of tax policies on the distribution of income. Notes 1. When such corrections are made, they tend to reduce inequality as measured by the Gini index by between 1 and 3 percentage points (Yemtsov 2003). 2. The use of the Eurostat equivalence scale rather than per capita typically reduces the value of Gini index by about 2 percentage points (see for example Forster, Jesuit, and Smeeding 2003). 3. Copies of much of the survey data conducted in the region are stored in the World Bank Europe and Central Asia (ECA) regional data archive. At the time of writing the archive contained primary unit record data from recent household surveys for 24 countries spanning the period 1998­2004. 4. Ravallion (2001), quoting Benabou (1996), argues that countries are expected to converge to the same distribution and proposes a test for such convergence, but due to data limi- tations transition economies have not been fully incorporated in his analysis. 5. The data used in figure 2 for China come from Ravallion and Chen (2004). 6. The data are taken from RLMS. 7. See Aghion and Commander (1999); Förster and Tóth (1997); Garner and Terrell (1998); Keane and Prasad (2002a). 8. See Kattuman and Redmont (2001) for Hungary; Garner and Terrell (1998) for the Czech Republic and Slovak Republic; and Commander and Lee (1998), followed by Decoster (2005) for some evidence for Russia. 9. For a review, see Birdsall and Nellis (2003); Davies and Shorrocks (2005). 10. This state can also be reinterpreted as subsistence or informal sector employment. 11. This is essentially a model of labor reallocation, which omits capital (or mixed) income, thereby bypassing one of the most important features of transition. Furthermore, param- eters of the distribution within each sector are taken as exogenous and constant. 12. Note that whenever the concentration coefficient of income source k is greater than the overall Gini coefficient, an increase in the income source k (holding everything else constant), will increase inequality. See Cowell and Jenkins (1995). 13. See Commander, Tolstopiantenko, and Yemtsov (1999); Szulc (2003); Kattuman and Redmont (2001); and Kyslitsyna (2003). 14. This increase might seem somewhat counterintuitive, as transfers are often regarded as fac- tors mitigating against inequality increase (Keane and Prasad 2002a). Paradoxically, it was largely a result of the increased pensions and greater reliance on pension payments by recip- ients. Before transition, inadequate pension payments were often supplemented by individ- ual work after the pension age, and their recipients were as likely to be in the bottom of the distribution as in the top. After the changes in pension policy and indexation, most pension- ers moved to the middle of the distribution, while having to forego additional earnings with tighter labor markets. This created a stronger positive correlation between income level and pensions and hence larger concentration coefficients. Gustaffson and Nivorozhkina (2005) used a unique study of households in Taganrog in 1989 and a follow-up study in 2000 to arrive at the same conclusions: the main beneficiaries from expanding public transfers have INCREASING INEQUALITY IN TRANSITION ECONOMIES | 97 been households in the middle of the income distribution, and that outcome also positively contributed to the increase of income inequality in Russia. 15. Table 7 relies on per capita equivalence scale, as do tables 3 and 4, where data availability dictated the choice. However, results on decomposition by groups presented in table 7 are also available on per equivalent adult basis (with constant degree of returns to scale, 0.5 or 0.75). Results are broadly in line with reported here in table 7 (with the exception of Moldova), and are available from authors on request. 16. Returns to vocational education in Hungary were unchanged between 1993 and 1999, but the country had reformed its vocational education very early on in transition, as reported in Kertesi and Köllo (2001). 17. The definitions used are as follows. A household was classified as belonging to a group of wage earners when at least one household member was a salaried employee and there were no working members who were self-employed with only minimal supplementary income obtained from self-production (<5 percent). A household was classified as belonging to a group of entrepreneurs when it owned a business or at least one adult was self- employed, with only minimal self-production (<5 percent), making such self-employment truly market-oriented. A household was classified as belonging to a group engaged in subsistence activities when at least one adult was in self-employment and significant income accrued from self-production (>5 percent). Finally, a household fell into the group of nonemployment when no adult was in employment or self-employment. 18. The values reported for the Theil entropy index for per capita consumption in table 7c are different from those in tables 7a and 7b. This is because labor market information is available for a number of countries but often only for a subset of households and/or for a subset of time periods, such as only one quarter of a year; inequality can be decom- posed only for that particular quarter instead of the full year as in tables 7a and 7b. Furthermore, in surveys, questions on employment are subject to a much larger nonre- sponse than questions about location or education levels. Most figures and directions of change are similar, with the exception of Hungary, where labor market information is poorly reported in household budget surveys. 19. The results of sectoral decompositions are not reported in table 7 but are available from the authors on request. 20. The coefficients in table 9 are widely believed to be underestimates for urban areas (see for example Khan and Riskin 2005). This is because the urban sample of the national survey includes only permanent residents and migrants with permits (hukous) registered in urban areas. The estimates of unregistered migrants differ and are as large as 150 million. Unreg- istered migrants are believed to earn significantly lower salaries and their omission from the sample definitely underestimates urban inequality in China. 21. Gustaffson and others (2001) find surprisingly similar earnings profiles for Chinese and Russian urban workers in 1989. References Aghion, Philippe, and Olivier Blanchard. 1994. "On the Speed of Restructuring." Massachu- setts Institute of Technology and European Bank for Reconstruction and Development, London. Aghion, Philippe, and Simon Commander. 1999. "On the Dynamics of Inequality in the Tran- sition." The Economics of Transition 7 (2): 275­98. Alexeev, Michael, and Clifford Gaddy. 1993. "Income Distribution in the U.S.S.R. in the 1980s." Review of Income and Wealth, Series 39: 23­36. 98 | PRADEEP MITRA AND RUSLAN YEMTSOV Atkinson, Antony. 2003. "Income Inequality in OECD Countries: Data and Explanations." Revised version of paper prepared for the CESifo conference, "Globalization, Inequality, and Well-Being," Munich, November 8­9, 2002. Atkinson, Anthony B., and John Micklewright. 1992. Economic Transformation in Eastern Europe and the Distribution of Income. Cambridge: Cambridge University Press. Benabou, Roland. 1996. "Inequality and Growth." Working Paper 5658, National Bureau of Economic Research, Cambridge, Mass. Birdsall, N., and J. Nellis. 2003. "Winners and Losers: Assessing the Distributional Impact of Privatization." World Development 31 (10): 1617­33. Bourguignon, Francois. 1979. "Decomposable Income Inequality Measures." Econometrica 47 (4): 901­20. ------. 2003. "The Growth Elasticity of Poverty Reduction: Explaining Heterogeneity across Countries and Time Periods." In Inequality and Growth: Theory and Policy Implications, eds. T. Eicher and S. Turnovski, 3­26. Cambridge, Mass.: MIT Press. Bourguignon, François, Francisco H. G. Ferreira, and Phillippe G. Leite. 2004. "Beyond Oaxaca-Blinder: Accounting for Differences in Household Income Distributions Across Countries." Background paper for "Inequality and Economic Development in Brazil," World Bank, Washington, DC. Buckley, Robert, and Eugene Gurenko. 1997. "Housing and Income Distribution in Russia: Zhivago's Legacy." World Bank Research Observer 12 (1): 19­32. Campos, Nauro F., and Dean Jolliffe. 2003. "After, Before, and During: Returns to Education in Hungary (1986­1998)." Economic Systems 27: 377­90. Chen, Shaohua, and Yan Wang. 2002. "China's Growth and Poverty during the 1990s: Trends Between 1990 and 1999." Policy Research Working Paper 2651, World Bank, Washington, DC. Commander, Simon, and Une Lee. 1998. "How Does Public Policy Affect the Income Distri- bution? Evidence from Russia, 1992­1996." European Bank for Reconstruction and Development and World Bank. Commander, Simon, and Andrei Tolstopiatenko. 1996. "Restructuring and Taxation in Transition Economies." Policy Research Working Paper 1625, World Bank, Washington, DC. ------. 1997. "Unemployment, Restructuring, and Inequality in the Transition." European Bank for Reconstruction and Development, London. Commander, Simon, Andrei Tolstopiatenko, and Ruslan Yemtsov. 1999. "Channels of Redis- tribution: Inequality and Poverty in the Russian Transition." Economics of Transition 7 (2): 411­47. Cowell, F., and S. Jenkins. 1995. "How Much Inequality Can We Explain? A Methodology and an Application to the United States." Economic Journal 105: 421­30. Davies, James, and Antony Shorrocks. 2005. "Wealth Holdings in Developing and Transition Countries." Paper prepared for the Luxembourg Wealth Study conference on "Construc- tion and Usage of Comparable Microdata on Wealth," Perugia, January 27­29. Dolinskaya, Irina. 2002. "Transition and Regional Inequality in Russia: Reorganization or Pro- crastination?" IMF Working Paper 169. International Monetary Fund, Washington, DC. Decoster, Andre. 2005. "How Progressive and Indirect Taxes in Russia?" Economics of Transition 13 (4): 705­29. Earle, John S., and Klara Z. Sabirianova. 2002. "How Late to Pay? Understanding Wage Arrears in Russia." Journal of Labor Economics 20 (3): 661­707. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 99 EBRD (European Bank for Reconstruction and Development). 2005. Transition Report. London: European Bank for Reconstruction and Development. EC (European Commission). 2005. Report on Social Inclusion 2005: An Analysis of the National Action Plans on Social Inclusion (2004­2006) Submitted by the 10 New Member States. European Commission Directorate-General for Employment, Social Affairs, and Equal Opportunities. Luxembourg: European Communities. http://europa.eu.int. Eurostat. 2005. "Statistics in Focus: Population and Social Conditions." In Income Poverty and Social Exclusion in the EU-25, ed. Anne-Catherine Guio. Series­ 13/2005, Luxembourg. Fedorov, L. 2002. "Regional Inequality and Regional Polarization in Russia, 1990-1999." World Development 30 (3): 443­56. Flabbi, Luca, Stefano Paternostro, and Erwin R. Tiongson. 2005. "Returns to Education in the Economic Transition: 1986­2002." World Bank, Washington, DC. Fleisher, Belton M., Klara Sabirianova, and Xiaojun Wang. 2004. "Returns to Skills and the Speed of Reforms: Evidence from Central and Eastern Europe, China, and Russia." Working Paper 703, William Davidson Institute, University of Michigan Business School, Ann Arbor. Flemming, J., and J. Micklewright. 1999. "Income Distribution, Economic Systems, and Tran- sition." In Handbook of Income Distribution, eds. A. B. Atkinson and F. Bourguignon, ch.14. Amsterdam: Elsevier Science BV. Förster, Michael F., and István György Tóth. 1997. "Poverty, Inequalities, and Social Policies in the Visegrad Countries." In Economics of Transition 5 (2): 505­10. Förster, M., D. K. Jesuit, and T. M. Smeeding. 2003. "Regional Poverty and Income Inequality in Central and Eastern Europe." Discussion Paper 2003/65, United Nations University/World Institute for Development Economics Research (UNU/WIDER), Helsinki. Garner, Thesia, and Katherine Terrell. 1998. "A Gini Decompositon Analysis of Inequality in the Czech and Slovak Republics during the Transition." The Economics of Transition 6 (1): 23­46. Gorodnichenko, Yuri, and Klara Sabirianova Peters. 2004. "Returns to Schooling in Russia and Ukraine." Discussion Paper 1325, Institute for the Study of Labor (IZA), Bonn. Goskomstat (various years). Russian Statistical Yearbook. Moscow: Goskomstat. Gustafsson, Bjorn, and Shi Li. 2001. "The Anatomy of Rising Earnings Inequality in Urban China." Journal of Comparative Economics 29: 118­35. Gustafsson, Bjorn, and Ludmila Nivorozhkina. 2005. "How and Why Transition Made Income Inequality Increase in Urban Russia: A Local Study." Journal of Comparative Eco- nomics 33: 772­87. Gustafsson, Bjorn, Shi Li, Ludmila Nivorozhkina, and Katarina Katz. 2001. "Rubles and Yuan: Wage Functions for Urban Russia and China at the End of the 1980s." Economic Development and Cultural Change 50: 1­17. Ivaschenko, Oleksiy. 2003. "Growth and Inequality: Evidence from Transitional Economies." In Inequality and Growth: Theory and Policy Implications, eds. T. Eicher and S. Turnovski, 155­98. Cambridge, Mass: MIT Press. Kattuman, Paul, and Gerry Redmont. 2001. "Income Inequality in Early Transition: The Case of Hungary 1987­1996." Journal of Comparative Economics 29 (1): 40­65. Keane, M., and E. Prasad. 2002a. "Inequality, Transfers, and Growth: New Evidence from the Economic Transition in Poland." Discussion Paper 448, Institute for the Study of Labor (IZA), Bonn. ------. 2002b. "Changes in the Structure of Earnings during the Polish Transition." Working Paper 135, International Monetary Fund, Washington, DC. 100 | PRADEEP MITRA AND RUSLAN YEMTSOV Kertesi, Gábor, and János Köllo. 2001. "Economic Transformation and the Revaluation of Human Capital­Hungary, 1986­1999." Budapest Working Papers on the Labour Market 2001/4, Institute of Economics, Hungarian Academy of Sciences, Budapest. Khan, Azizur Rahman, and Carl Riskin. 1998. "Income Inequality in China: Composition, Distribution, and Growth of Household Income, 1988 to 1995." The China Quarterly 154: 221­53. ------. 2005. "China's Household Income and its Distribution, 1995 and 2002." The China Quarterly 182 (June): 356­84. Klugman, Jeni. 1998. "Wage in Transition: The Case of Uzbekistan." PhD dissertation, Aus- tralian National University, Department of Economics, Canberra. Knight, John, and Lina Song. 2003. "Increasing Urban Wage Inequality in China: Extent, Elements, and Evaluation." Economics of Transition 11 (4): 597­619. Kyslitsyna, Olga. 2003. "Income Inequality in Russia during Transition: How Can It Be Explained?" Working Paper 03/08, Economics Education and Research Consortium (EERC), Moscow. Lehmann, Hartmut, and Jonathan Wadsworth. 2001. "Wage Arrears and the Distribution of Earnings in Russia." Working Paper 421, William Davidson Institute, University of Michigan Business School, Ann Arbor. Lehman, Hartmut, Jonathan Wadsworth, and Alessandro Acquisti. 1999. "Grime and Punish- ment: Job Insecurity and Wage Arrears in the Russian Federation." Journal of Comparative Economics 27: 595­617. Li, Shi. 2000. "Changes in Income Inequality in China Transition." Chinese Academy of Social Sciences, Beijing. ------. 2003. "A Retrospect and Prospect of Individual Income Distribution in China." China Economic Quarterly 2 (2): 379­404. Lukianova, Anna. 2005. "Wage Differentiation and Phenomenon of Working Poor in Russia." World Bank, Washington, DC. Meng, Xin. 2000 "Economic Restructuring and Income Inequality in Urban China." Research paper, Australian National University, Canberra. Milanovic, B. 1998. "Income, Inequality, and Poverty during the Transition from Planned to Market Economy." Regional and Sectoral Studies, World Bank, Washington, DC. ------. 1999. "Explaining the Increase in Inequality during the Transition." Policy Research Working Paper 1935, World Bank, Washington, DC. National Bureau of Statistics. China Statistical Yearbook 2003. Beijing. Newell, A. 2001. "The Distribution of Wages in Transition Countries." Discussion Paper 67, Institute for the Study of Labor (IZA), Bonn. Newell, A., and M. Socha. 2005. "The Distribution of Wages in Poland, 1992­2002." Discus- sion Paper 1485, Institute for the Study of Labor (IZA), Bonn. Popova, D. 2005. "Dynamics of Income Inequality in the Context of Globalization." In Russia's Accession to WTO: Imaginary and Real Consequences, ed. L. Ovcharova for the Indepen- dent Institute for Social Policy (IISP). Moscow: Pomatur (in Russian). Ravallion, Martin. 2001. "Inequality Convergence." Policy Research Working Paper 2645, World Bank, Washington, DC. Ravallion, Martin, and Shaohua Chen. 2004. "China's (Uneven) Progress against Poverty." Policy Research Working Paper 3408, World Bank, Washington, DC. Rutkowski, J. 1996. "High Skills Pay Off: The Changing Wage Structure during the Economic Transition in Poland." Economics of Transition 4: 89­112. INCREASING INEQUALITY IN TRANSITION ECONOMIES | 101 ------. 2001. "Earnings Inequality in Transition Economies of Central Europe: Trends and Patterns during the 1990s." Social Protection Discussion Paper 0117, World Bank, Washington, DC. Shi, Xinzheng, 2002. "Empirical Research on Urban-Rural Income Differentials: The Case of China." China Center for Economic Research, Beijing University. Shi, Xinzheng, Terry Sicular, and Yaohui Zhao. 2002. "Analyzing Urban-Rural Income Inequality in China." Paper presented at the International Symposium on Equity and Social Justice in Transitional China, July 11­12, Beijing. Shorrocks, Anthony. 1980. "The Class of Additively Decomposable Inequality Measures." Econometrica 48 (3): 613­25. ------. 1982. "Inequality Decomposition by Factor Components." Econometrica 50 (1): 193­211. Shorrocks, Anthony, and Guanghua Wan. 2005. "Spatial Decomposition of Inequality." Journal of Economic Geography 5 (1): 59­81. Svejnar, Jan. 2002. "Transition Economies: Performance and Challenges." Journal of Economic Perspectives 16 (1): 2­28. Szulc, Adam. 2003. "Inequality during the Economic Transition in Poland: 1993­1999, Evidence." Institute of Statistics and Demography, Warsaw School of Economics. Tóth, István György. 2004a. "Assessing the Effects of Economic Transition on Income Distribution: The Case of Hungary, 1987­2001." Paper presented at Fifth Annual Global Development Conference, "Understanding Reform," New Delhi, January 28. ------. 2004b. "Income Composition and Inequalities in Hungary, 1987­2003." In TÁRKI Social Report 2004, 72­92. Budapest: TÁRKI. Vecerník, Jiri. 1995. "Changing Earnings Distribution in the Czech Republic: Survey Evidence from 1988­1994." Economics of Transition 3: 355­71. Wagstaff, Adam. 2005. "Decomposing Changes in Income Inequality into Vertical and Hori- zontal Redistribution and Reranking, with Applications to China and Vietnam." Policy Research Working Paper 3559, World Bank, Washington, DC. World Bank. 1999. Georgia: Poverty and Income Distribution. Volumes I and II. Poverty Reduction and Economic Management Unit, ECA Region, World Bank, Washington, DC. ------. 2000. ECAPOV 1. Making Transition Work for Everyone: Poverty and Inequality in Europe and Central Asia. Washington, DC: World Bank. ------. 2002a. "Armenia Poverty Update." World Bank, Washington, DC. ------. 2002b. "Georgia Poverty Update." World Bank, Washington, DC. ------. 2002c. "Hungary: Long-Term Poverty, Social Protection, and the Labor Market." World Bank, Washington, DC. ------. 2002e. Transition: The First Ten Years. Analysis and Lessons for Eastern Europe and the Former Soviet Union. Washington, DC: World Bank. ------. 2003a. "Kyrgyz Republic: Enhancing Pro-poor Growth." World Bank, Washington, DC. ------. 2003b. "Romania Poverty Assessment." World Bank, Washington, DC. ------. 2004. "Poland: Growth, Employment, and Living Standards in Pre-Accession Poland." World Bank, Washington, DC. ------. 2005a. World Development Report 2006, Equity and Development. Washington, DC: World Bank. ------. 2005b. ECAPOV II. Growth, Poverty, and Inequality in Eastern Europe and the For- mer Soviet Union, 1998­2003. Washington, DC: World Bank. 102 | PRADEEP MITRA AND RUSLAN YEMTSOV ------. 2005c. "Labor Markets in Europe and Central Asia: Toward Creating More and Better Jobs." World Bank, Washington, DC. ------. 2005d. "Russian Federation: Reducing Poverty through Growth and Social Policy Reform." World Bank, Washington, DC. Wu, Ximing, and Jeffrey M. Perloff. 2004. "China's Income Distribution over Time: Reasons for Rising Inequality." Working Paper 977, Department of Agricultural and Resource Economics (CUDARE), University of California, Berkeley. Yemtsov, Ruslan. 2001."Georgia: Income Distribution." Discussion Paper 252, Institute for the Study of Labor (IZA), Bonn. ------. 2003. "Quo Vadis Inequality and Poverty Dynamics across Russian Regions." Dis- cussion Paper 2003/67, United Nations University/World Institute for Development Economics Research (UNU/WIDER), Helsinki. Yemtsov, R., S. Cnobloch, and C. Mete. 2006. "Evolution of the Predictions of Earning during Transition." World Bank, Washington, DC. Trade Liberalization, Inequality, and Poverty Reduction in Latin America GUILLERMO PERRY AND MARCELO OLARREAGA Latin America's trade liberalization in the late 1980s and early 1990s was accompanied in some countries by increases in skill premiums, wage and income inequality, and even in poverty: results unexpected by many. This paper argues that this was mainly the result of four factors. First, most Latin American countries are rich in natural resources (which in general are complementary with capital and skills) and more abundant in capital than other developing countries with large pools of unskilled labor, such as China and India, that were integrating into the world economy while Latin America was liberalizing. Second, dynamic effects of trade led to new goods being produced in the region through outsourcing, an acceleration of skill-biased technical change, and Schumpeterian creative destruction, resulting in an increase in demand for skills in most industries. Third, initial conditions and contemporary events make predictions based on a simple factor abundance model difficult to generalize. As an example of the latter, the pre-reform structure of protection was biased toward unskilled-intensive sectors in some Latin American and Caribbean countries and tariff reductions naturally led to a relative increase in demand for skills. Differences in consumption bundles across income groups, exchange rate movements, and the growing importance of a nontradable service sector also make predictions less straightforward. Fourth, imperfectly functioning labor markets--such as potential transitions in and out of unemployment and informality--as well as income volatility are likely to affect and sometimes change the direction of the impact of trade reforms on income inequality and poverty. Finally, the paper argues that the effect of trade on poverty (and income inequality) depends on complementary policies being implemented. The impact of trade on poverty reduction can be significantly enhanced (and the effects on inequality mitigated) by policies that increase the provision and access to skills and other productive assets to the poor. Trade reforms in Latin America have often been associated, in the popular debate, with increases in income inequality, poverty, and skill premiums. On the other hand, Guillermo Perry is chief economist for Latin America and the Caribbean Region at the World Bank and Marcelo Olarreaga is senior economist in the Office of the Chief Economist for Latin America and the Caribbean Region at the World Bank. The authors are greatly indebted to Daniel Lederman, Humberto Lopez, William Maloney, Guido Porto, and Maria Fernanda Rosales for their inputs and contributions to many of the sections in this paper. They are also grateful to Jaime de Melo, Andres Solimano, Jan Svejnar, participants at the LACEA Network of Inequality and Poverty (NIP) meeting in ITAM, Mexico, and two anonymous referees for their comments. Annual World Bank Conference on Development Economics 2007 © The International Bank for Reconstruction and Development / The World Bank 103 104 | GUILLERMO PERRY AND MARCELO OLARREAGA many in the economics profession, based on the Heckscher-Ohlin framework and the Stolper-Samuelson theorem, expected trade liberalization in Latin America, as in other developing countries, to reduce income inequality through an increase in the relative demand for unskilled labor (the presumed abundant endowment of developing countries). This paper reviews the facts, attempts to explain the apparent puzzles, and explores what governments can do to ensure that international trade becomes an instrument for poverty reduction. The distributional impacts of trade liberalization are difficult to assess ex ante as trade reforms, through their changes on factor and good prices, will naturally lead to losses to some agents and gains to others. In the presence of effective distributive policies (taxes, subsidies, and transfers), it is possible in principle to redistribute the overall gains to achieve Pareto efficiency.1 In the absence of such effective redistribu- tive policies--as is the case in Latin America, where the tax and transfer systems accomplish little redistribution (Perry and others 2005)--little can be said in general terms regarding the welfare implications of trade at the aggregate level, or for groups of individuals. Whether trade openness will benefit poor households becomes an empirical question and data analysis is needed to provide clues regarding the impacts of trade across different households. This paper examines these issues in the context of the trade liberalization episodes that took place in most countries of Latin America from the mid-1980s (Chile and Mexico) to the early 1990s (figure 1).2 Not only tariffs, but also quanti- tative restrictions, were substantially reduced during this period across Latin America. As we will see, their impact on income inequality and poverty varied across countries within the region. The paper is organized as follows. The second section begins by reviewing the evidence on the effects of these trade reform episodes on income inequality, wage inequality, skill premiums, and demand for skills. It then turns to explaining the apparent puzzles through a characterization of Latin America factor endowments relative to other regions, a description of dynamic effects that may have resulted in a relative increase in the demand for skilled factors, and a study of the role played by initial conditions and contemporary events such as the initial tariff structure and exchange rate movements. The third section focuses on the impact of trade reform on the incomes of the poorest individuals. It also discusses the effects of transitions in and out of unemployment and informality, and the impact of trade on income volatility. The fourth section analyzes the importance of policy complementarities in ensuring that the poor benefit from trade reforms. The fifth and final section offers concluding remarks. Trade Opening and Inequality: Latin America's "Puzzle"? Many economists, following the predictions of neoclassical Heckscher-Ohlin and Stolper Samuelson trade models (HO/SS henceforth), expected trade liberalization in developing countries to reduce income inequality through an increase in the relative demand for unskilled labor. Under the assumption that capital and skills TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 105 FIGURE 1. Trade Openness Indicators, Latin America, 1987 and 1998 a. Average tariffs 60 50 40 eighted 30 unw 20 % 10 0 Argentina Brazil Chilea Colombia Mexicoa Peru 1987 1998 b. Quantitative restrictions 60 50 40 eighted 30 unw 20 % 10 0 Argentina Brazil Chilea Colombia Mexicoa Peru 1987 1998 Source: Perry, Lederman, and Suescun 2002. Note: The y-axis shows the unweighted average across all goods. a. For Chile and Mexico, the preliberalization data is for the year 1984, as they started their trade reforms earlier than the rest. were complements, trade liberalization was expected to reduce skill premiums and hence wage inequality. This seemed to be confirmed by the experience of the Asian Tigers' trade liberalization in the 1960s, where wage inequality declined after trade liberalization (figure 2).3 Many analysts also expected reductions in inequality and skill premiums when most Latin American countries sharply reduced their tariffs and nontariff trade barriers in the mid-1980s and beginning of the 1990s, although several previous empirical studies had not found a consistent relation between trade liberalization in developing countries and domestic income distribution.4 What was observed in Latin America, however, was an increase in wage inequality (with some exceptions) and skill premiums (between workers with tertiary and secondary education), and in some cases an increase in overall income inequality (figure 3). The increase in skill premiums for workers with tertiary education is explained by the fact that there were significant increases in relative demand for these skilled workers in most countries just after trade liberalization, which were not met by increases in supply. There was also increased relative demand for workers with secondary education vis-à-vis those with just primary education, although relative supply shifts led to reduced skill premiums in this case in several countries.5 106 | GUILLERMO PERRY AND MARCELO OLARREAGA FIGURE 2. Wage Inequality Before and After Trade Liberalization, East Asian Tigers 52 50 48 Singapore 46 Malaysia coefficient 44 Gini Korea, Rep. of 42 40 t t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t is the year of trade liberalization Sources: University of Texas inequality project. Time at which trade reforms occur is from Wacziarg and Welch (2003). Note: Wage inequality is computed using a regression between the Deininger & Squire (D&S) inequality measures and the UTIP-UNIDO pay inequality measures (which is measured as the wage dispersion between 28 three-digit manufacturing sectors) controlling for the data source characteristics in the D&S dataset and for the share of manufacturing in total employment. Nevertheless, the observed increases in skill premiums wage and income inequality may not have been caused by trade opening, as many other things were going on at the same time. Indeed, many countries engaged simultaneously in capital account opening, financial sector liberalization, privatization of public enterprises, and domestic markets deregulation. It is extremely difficult to identify and separate the effects of all these policy reforms, and the few studies that have attempted to do so come to different conclusions. Behrman, Birdsall, and Szekely (2000) examined the joint and separate effects of reforms on wage differentials among workers with primary, secondary, and tertiary education over time, using comparable wage data and the index of reforms in Lora (1997),6 as extended and modified by Morley (2000). They found that the package of reforms had a strong but temporary effect on wage differentials (a negative effect on wages of workers with primary education, a slightly positive effect for workers with secondary education, and a strongly positive effect for workers with tertiary education), but that this was due to the effects of financial market reform, capital account opening, and tax reform, while trade reform results were not statistically significant and the effect of privatization went in the opposite direction. The stronger effects were due to capital account opening and financial liberalization, although they faded away rapidly over time. They hypothesized this was probably due to complementarities between capital and skills, and thus concluded that "technological progress, rather than trade flows, appears to be the channel through which reforms are affecting inequality" (2000, p. 3). Morley (2000) examined the joint and separate effects of reforms on income distri- bution using household survey data and the same reform indexes, and he found strikingly different results: the overall effect was statistically not significant, while trade and tax reform appeared to increase income inequality, and capital account opening to reduce it. The effects of financial liberalization and privatization were not TRADE LIBERALIZATION, FIGURE 3. Skill Premiums, Wage Inequality, and Income Inequality Before and After Trade Liberalization in Latin America INEQUALITY, a. Wage inequality (Gini index) 54 55 53 52 51 50 49 AND 47 48 45 coefficient 46 coefficient POVERTY 43 Gini44 Gini 41 39 42 37 REDUCTION 40 35 t t t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t­10 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t is the year of trade liberalization t is the year of trade liberalization Argentina Brazil Chile Colombia Mexico Peru Bolivia Costa Rica Ecuador El Salvador Uruguay IN Sources: University of Texas inequality project. Time at which trade liberalization occurs is from Wacziarg and Welch (2003). LATIN Note: Inequality in wages computed from a regression relationship between the D&S measures and the UTIP-UNIDO pay inequality measures (which is measured as the wage dispersion in 28 three-digit manufacturing industries) controlling for the source characteristics in the D&S data and for the share of manufacturing in total employment. AMERICA | (Continued on the next page) 107 108 | GUILLERMO FIGURE 3. continued b. Income inequality (Gini index) PERRY 65 54 52 60 AND 50 55 MARCELO 48 coefficient coefficient 46 50 Gini Gini44 45 OLARREAGA 42 40 40 t t t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4t+5t+6 t+7 t+8 t+9t+10 t is the year of trade liberalization t is the year of trade liberalization Argentina Brazil Chile Colombia Mexico Peru Costa Rica Ecuador El Salvador Uruguay Sources: Dollar and Kraay 2002; De Ferranti, Perry, Ferreira, and Walton 2003. Time at which trade liberalization occurs is from Wacziarg and Welch (2003). (Continued on the next page) TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 109 FIGURE 3. continued c. Skill premium 0.9 0.8 0.7 0.6 (%) 0.5 0.4 0.3 premium 0.2 skill 0.1 0 ­0.1 ­0.2 t t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t is the year of trade liberalization Argentina Brazil Chile Colombia Mexico Sources: Manacorda, Sánchez-Páramo, and Schady 2005. Time at which trade liberalization occurs is from Wacziarg and Welch (2003). Note: The panel shows changes in skill premiums for workers with tertiary versus secondary education. statistically significant. The same happens with studies using global samples with respect to the effect of trade openness on inequality. While Dollar and Kraay (2004) found no significant effects of trade openness on inequality, Barro (2000) found a pos- itive association between the two variables, surprisingly stronger for poorer countries.7 Differences in results in these studies may be due to differences in samples, indexes of inequality and reform or trade openness, control variables, and estimation proce- dures. However, it is notable that none of them found a statistically negative effect of trade opening or openness on wage or household income inequality (that is, of reducing inequality), as conventional wisdom would have suggested. This is also true for most available country studies using microeconometrics. For example, Nicita (2004) found that trade liberalization increased income inequality in Mexico, Feliciano (2001) found a similar effect on wage inequality also in Mexico, and Galiani and Sanguinetti (2003) and Galiani and Porto (2005) found similar results for Argentina. In contrast, Porto (2006) estimated a decrease in overall inequality in Argentina due to Mercosur, a trade liberalization experiment within developing countries. Further, the notoriety of trade opening as the backbone of the pro-market reform program in Latin America has led to the generalized belief that it was indeed the main culprit behind the observed increases in wage and income inequality. We examine potential explanations for such results in the discussion that follows. First, we take a better look at relative factor endowments in Latin America and discuss what one should expect from a more informed analysis. Then we look at the role played by the introduction of new goods produced in the region through outsourcing, skilled-bias technological change, and entry and exit of firms subject to more competition from abroad. We then look at the importance of initial conditions (such as the initial tariff structure) and other contemporary events. Finally, we explore other potential channels through which trade liberalization can affect income inequality. 110 | GUILLERMO PERRY AND MARCELO OLARREAGA Is There a Puzzle? A Better Look at Factor Endowments The answer to the apparent puzzle may actually be related to Latin America's relative factor abundance and their complementarities.8 Latin America is rich in natural resources. By the time of trade liberalization in the region (late 1980s and early 1990s), other developing countries with large pools of unskilled labor endowments, lower capital per unskilled worker ratios, and hence lower wages, such as China and India, were already emerging in the world trade scene (see figure 4). The East Asian Tigers of the 1960s and 1970s were relatively poor in natural resources and China, India, and Vietnam were by then not integrated into the world economy (nor was most of Latin America to a large extent), so the Tigers were at the time among the least capital-abundant economies integrating in the world economy. Note that there is significant heterogeneity within Latin America, which can help explain differences in outcomes across countries. For example, in the year 2000 Haiti had a capital to unskilled worker ratio of $150, whereas Uruguay had a ratio close to $80 thousand. Similarly, the unskilled to skilled ratio ranges from 0.12 in Haiti to 1.6 in Barbados. It is 1.03 in Uruguay, which is the third country in the ranking after Barbados and Trinidad and Tobago. Thus, given that Uruguay and Haiti are at opposite ends of the scale in terms of skilled labor and capital abundance in the region, it may not be surprising that after trade reform one observes a very different outcome in terms of wage and income inequality, with Uruguay experiencing an increase in wage inequality (and a more modest increase in income inequality), as shown in figure 3. Similarly, Jamaica has net exports of natural resources per worker of $650,9 whereas Venezuela has net exports of natural resources per worker close to $2,600. The skilled to unskilled ratio is 0.6 in Venezuela and 1 in Jamaica. The consequences of (somehow similar) trade reforms are likely to be very different across countries, given the differences in endowments within Latin America. Implications of LAC´s Intermediate Unskilled Labor Abundance It has been shown that, even in a world with two factors of production, if one assumes that factor endowment differences in the world are too large to allow for full global factor price equalization through free trade (as is evidently the case), what matters is not factor abundance compared to the global economy, but relative to the cone (a sub- set of countries with similar relative endowments) within which a given country competes on the same products (Davis 1996). In this two-factor and n country model, countries that have high unskilled labor abundance when compared to the global economy, but high capital/skill endowments within its cone will suffer increased inequality from trade liberalization.10 If Latin America was indeed more skilled-abun- dant than other competing developing areas when it liberalized to trade, most notably China and India, as figure 4 indicates, such a simple model would indeed predict a rise in demand and returns for capital and for more skilled workers, assuming capital/skill complementarities and that countries are in the same diversification cone. Robertson (2004) provides some indirect evidence. He examines the relationship between relative goods prices and relative wages on Mexico after its accession to the General Agreement on Tariffs and Trade (GATT) in 1986, and after its accession to TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 111 FIGURE 4. Factor Endowments by Region, 1990 a. Capital to unskilled labor index 20 15 10 5 0 ­5 ies ica ica India Asia Asia Tigers ibbean Asia Afr China al countr th an Afr Car East South High-incomeAsian Nor Centr the OECDEast and and and ly Sub-Sahar ica East Ear Amer n Europe Middle Latin Easter Source: Barrow and Lee 2000 dataset. Note: A positive number indicates that the region relative to the world is well-endowed in skilled labor. It is calculated as the ratio of the region to the world relative skilled labor abundance minus 1. b. Index of natural resource comparative 100,000 advantage by region 50,000 0 ­50,000 ­100,000 ­150,000 ­200,000 ica ica ies Asia Asia Afr ibbean Asia Afr China India al Tigers th an Car East countr Nor Centr South Asian the and and OECD and East Sub-Sahar ly East ica n Europe Ear Middle Amer High-income Latin Easter Source: World Development Indicators 2000. Note: The natural resource index is calculated as the trade balance (exports minus imports) in ores, mineral, fuel, agricultural raw materials, and food divided by the labor force. Units are US$ per worker. 112 | GUILLERMO PERRY AND MARCELO OLARREAGA the North American Free Trade Agreement (NAFTA) in 1994. After GATT accession, the relative price of skilled-intensive goods and the relative wage of skilled workers rose. After NAFTA, results are reversed, but this is consistent with Mexico's integration with the United States and Canada--skill-abundant countries--rather than with the whole world. This is perfectly consistent with the prima facie evidence shown for Mexico in figure 2 (panel a), where there is an increase in wage inequality starting at the time of entrance to the GATT, which coincides with the start of trade reforms, according to Wacziarg and Welch (2003). Similarly, wage inequality starts declining eight years after the reform--around 1994--which coincides with Mexico's entrance into NAFTA. Incorporation of new products/processes (or rapid growth of some specific products/ processes) in the margin also played a significant role in some Latin American coun- tries. Feenstra and Hanson (2003) provide evidence for Mexico, where outsourcing may have led to an increase in relative demand for skilled workers both in the United States and Mexico. Indeed these activities were below the average skill intensity in developed countries but above the average skill intensity in receiving developing countries. There are no studies of this sort available, unfortunately, for other Latin American countries, but similar effects are also likely to be important for Central American and Caribbean countries (such as Costa Rica) that enjoy similar location advantages and special trade arrangements with the United States. Implications of Latin America's Abundance of Natural Resources Trade opening in economies that are abundant in natural resources should lead to increased rents to landowners (and land is heavily concentrated in most of Latin America), oil and mine owners (normally the state, so distributional effects would depend on what they do with the increased rents), and holders of exploitation rights (which are mostly large capital-intensive companies). Effects on income distribution should also depend on the degree of complementarities with capital and skills, which is expected to be high in the case of mining, forestry, fisheries, and agricultural raw materials.11 Table 1 shows that net exports of mining and agricultural raw materials tend to be positively correlated with capital, while net exports of food are correlated with unskilled labor in Latin America.12 Further, capital and skilled labor also seem to be correlated within Latin America. Models considering three factors of production grow enormously in complexity and predictions become more difficult (and sometimes impossible).13 Nevertheless, and consistent with the simpler description above, Leamer, Rodriguez, and Schott (1999) explain Latin American inequality as a natural consequence of a development path determined by its rich natural resources endowments, with the use of a simple three-country, three-goods, three-factor framework (in which skills and capital are complementary and thus aggregated into a single factor). The model is illustrated in figure 5. The corners of the triangle represent endow- ments of three production factors: raw labor, natural resources, and capital (human and physical). The three arrows represent three different development paths as coun- tries with different endowments accumulate capital, holding fixed their relative sup- plies of land and labor. For example, for a country abundant in raw labor, as capital TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 113 Table 1. Partial Correlations between Capital, Skilled Labor, and Natural Resources Dependent variable: capital/unskilled labor Coefficient t-statistic Skilled/unskilled 1.4916 2.38* Food ­0.0003 ­2.84* Mining 0.0003 1.85*** Fuel ­0.0000 ­0.28 Agricultural raw materials 0.0021 4.04* Number of observations 199 R-Squared 0.57 Source: Authors' calculations. Note: The regression includes country and year fixed effects. * Significant at the 1 percent level. ** Significant at the 5 percent level. *** Significant at the 10 percent level. FIGURE 5. Natural Resource Development Paths for Latin America Natural resources primitive extraction forestry capital-intensive extraction, mining, permanent crops E F farms, food products, G wood products peasant farming, tourism pulp and paper woodworking agribusiness A B C D handicrafts Raw labor apparel machinery Physical and human capital Source: Leamer, Rodriguez, and Schott 1999. accumulates, it moves from diversification cone A, where it produces handicrafts, farming, and apparel, to cone C, where it produces apparel, food products, and machinery. As the country gets closer to the capital vertex factor, returns to raw labor and natural resources increase, while returns to capital decline. Natural resource­rich countries without a large pool of unskilled labor (such as Venezuela) would follow a development path from primitive extraction of natural resources to capital-intensive extraction and to capital- and skill-intensive manufac- ture (path E-F-G in the diagram). Such countries may never compete in the same products (such as apparel) with countries that are abundant in unskilled labor and poor in natural resources (China and India would follow the A-B-C-D path in the 114 | GUILLERMO PERRY AND MARCELO OLARREAGA diagram) and would maintain a higher capital intensity and skill premium, leading to higher inequality all along their development path. To understand this, note that as countries develop along the A-B-C-D path, the returns to unskilled labor increase, reducing income inequality, whereas as countries develop along the E-F-G path, it is the returns to natural resources that increase, and the latter is complementary to capital and skills, as discussed earlier.14 Developing countries that are abundant in both natural resources and unskilled labor (Brazil and Mexico) would follow an intermediate development path (exempli- fied in the diagram by the points from "peasant farming/woodworking" to "food products" and finally to "agribusiness"). The impact of capital accumulation along these paths will lead to increases in skills premiums and income inequality that are somewhere in between those observed for the other two paths. The diagram can also be used to illustrate a key point for trade opening in Latin America. As trade protection in some Latin American countries was especially high in unskilled-intensive sectors, countries kept a product mix that balanced some natural resource­intensive exports (that were nonetheless taxed) with unskilled- intensive protected production for the domestic sector (and some exports, especially for developed markets in which they enjoyed trade preferences). Trade opening would then mean shifting further from the raw labor vertex toward the natural resource vertex, reducing the demand for unskilled labor. When natural resources are complementary with capital and skills, skill premiums and capital returns would increase, leading to higher inequality. Thus a more thorough analysis of the factor endowment model suggests that the impact of trade openness on income inequality will vary across countries with different endowments. But the type of factor endowment may also matter, as food production and agricultural raw materials are likely to be intensive in unskilled labor, whereas mining and fuel are more likely to be intensive in capital and skills. In a search for some systematic evidence along these lines, we ran Lopez's (2003) model, including explanatory variables that interact trade openness with indexes of natural resource abundance.15 Results are shown in table 2. In all cases, trade openness continues to increase inequality.16 More interestingly, the magnitude of the effect rises with net exports of minerals and fuels (the interac- tion of trade openness with abundance in mining and fuel increase inequality beyond the average effect of trade openness on inequality).17 On the other hand, in countries that are abundant in factors associated with food production and agricul- tural raw materials, trade openness leads to a below average impact on inequality and can even reverse signs, which implies that openness may reduce inequality in these countries. Is There a Puzzle? Schumpeter's Destructive Creation and Skill Biased Technological Change Most studies do not observe large labor reallocations across industries after Latin America's trade liberalization,18 in sharp contrast to findings for the United States that indicate higher sensitivity for employment than for wages in the face of trade TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 115 Table 2. Differential Effects of Trade Openness on Inequality According to Factor Endowments Variable (1) (2) (3) (4) (5) Openness to trade 2.37 1.69 1.67 1.79 2.02 t-stat 3.04* 1.93** 2.00* 1.69 2.57* Openness to trade x food index -0.31 t-stat -2.54* Openness to trade x mining index 0.50 t-stat 1.97* Openness to trade x fuel index 0.03 t-stat 1.00 Openness to trade x raw material index -0.57 t-stat -2.99* Sargan p-val 0.77 0.75 0.75 0.75 0.73 SOC p-val 0.42 0.38 0.39 0.4 0.42 Source: Authors' estimations based on Lopez (2003). Note: The table reports the results of regressing the five-year changes in the Gini coefficient on the following five- year average variables: secondary education, financial development, government consumption, infrastructure, gov- ernance, inflation, banking crisis, terms of trade, output volatility, and exchange rates. The estimation method is GMM, as in Lopez (2004). * Significant at the 5 percent level. ** Significant at the 10 percent level. shocks.19 At the same time, several studies have found that in the last two decades the share of skilled workers has increased substantially in most industries. Moreover, a larger share of the observed increase in aggregate demand for skilled workers in Latin America after trade liberalization is explained by intraindustry increases in rel- ative skilled labor employment, rather than interindustry reallocation of workers from unskilled-intensive industries to skilled-intensive industries.20 The substantial intra-industry allocation that led to skill deepening in most industries can be partly explained by changes in product mix within sectors, shifting production toward those that are relatively more skilled-intensive, as found by Feenstra and Hanson (2003) for Mexico and discussed above. Such a fact, however, may also be explained by two dynamic effects induced by trade opening. First, an acceleration of Schumpeter's destructive creation, as a consequence of increased competition, would lead to a significant reallocation of labor toward more productive (and more skilled-intensive) firms within industries subject to trade liber- alization.21 There is indeed evidence that trade (or exchange rate) shocks have caused significant productivity increases and reallocation of labor across firms within the same sector, for example after the exchange rate devaluation of 1995 in Mexico (see Verhoogen 2004). Second, the acceleration of Skill Biased Technical Change (SBTC)--which has been identified by several studies as a major force behind increases in inequality in OECD countries--might also help explain a general increase in skill intensities in Latin America. Suggestive evidence in support for this hypothesis in Latin America and the Caribbean is presented in Closing the Education and Technology Gap 116 | GUILLERMO PERRY AND MARCELO OLARREAGA FIGURE 6. Latin American Capital Equipment Imports/GDP Before and After Trade Reforms 8 7 6 (%) 5 /GDP 4 ts 3 Impor 2 1 0 t t­10t­9 t­8 t­7 t­6 t­5 t­4 t­3 t­2 t­1 t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 t+9t+10 t is the year of trade liberalization Argentina Brazil Chile Colombia México Source: World Development Indicators. (De Ferranti, Perry, Gill, Guasch, Maloney, Sánchez-Páramo, and Schady 2003), based on Sánchez-Páramo and Schady (2003). They find that those sectors in which more skill upgrading to tertiary education took place after trade liberalization were pretty much the same across different Latin American countries. This suggests that the increase in skill intensities in the 1990s had a common external origin (SBTC could have accelerated through trade opening). They also find that there is more skill upgrading in countries and industries with higher import penetration--especially of research and development­intensive imports--and foreign direct investment (FDI) stocks. The sharp rise of capital equipment imports after trade opening (figure 6) would suggest one potential channel for a faster transmission of SBTC.22 In support of this view, Harrison and Hanson (1999) found that those firms in Mexico that, within each industry, import more machinery and materials are more likely to employ a higher share of white-collar workers. Similar results relate skill upgrading with the degree of exposure of firms to foreign technology in Chile (see Pavcnik 2002).23 Is There a Puzzle? Initial Conditions and Contemporary Events Initial conditions matter. Some Latin American countries used to protect more unskilled labor-intensive sectors; trade opening reduced protection more sharply in these sectors, further reducing the demand for unskilled labor. Goldberg and Pavcnik (2004) cite studies for Brazil, Colombia, and Mexico that found such a pattern of protection before reform.24 Figure 7 provides some evidence that this was indeed the case in Mexico and Brazil, two large Latin American countries where trade liberaliza- tion was accompanied by increases in wage inequality, as shown in figure 3, panel a. As Davis and Mishra (2005) point out, this would make no sense in the standard HO/SS model, as such goods would be exports, not imports; however, it does in a more complex model in which countries that are intermediate in unskilled labor abundance import different kinds of goods from countries with both higher and TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 117 FIGURE 7. Changes in (Production Weighted) Average Tariffs in Skilled- and Unskilled-Intensive Sectors Brazil Mexico 0 ­5 ­10 points ­15 rcentageeP­20 ­25 ­30 Unskilled Skilled Sources: For Brazil, UNCTAD TRAINS data. For Mexico, Hanson and Harrison (1999). Note: In the case of Brazil, we are comparing tariff changes between 1989 and 2004. In the case of Mexico, we are comparing tariff changes between 1984 and 1990. The following ISIC rev. 2 two-digit sectors are considered as unskilled-intensive: ISIC 31 (food, beverages, and tobacco); ISIC 32 (textile, wearing apparel, and leather industries); ISIC 33 (wood and wood products); and ISIC 39 (other manufacturing industries). lower capital (skill) intensity. In addition, import substitution industrialization in Latin American countries led to tariff structures that gave positive effective protec- tion to most manufacturing sectors at the expense of natural resource­based activi- ties (which had negative effective protection). Thus trade liberalization should have been expected to significantly increase the rents of holders of property rights on natural resources. This contrasts with the earlier East Asian experience, which involved increased incentives to the export-competing sector (and unskilled-intensive sector) rather than sharp reductions in protection levels. This may also help explain the different patterns observed in the two regions (Wood 1997). As industries with higher tariff cuts were more unskilled-intensive, lack of labor mobility--which is alleged to be behind the observed lack of labor reallocation between industries--could lead to changes in industry wage premiums, which would translate into increases in relative skill premiums. Some studies for Colombia (Attanasio, Goldberg, and Pavnick 2004) and Mexico (Feliciano 2001) find evidence of such a relation, though with small quantitative effects, while others do not (Pavnickand others 2004, in a study of Brazil). In addition, trade opening in Latin America was generally accompanied by a period of appreciation and overvaluation of the currency, which cannot be attributed to trade liberalization. On the contrary, we would have expected from theory to observe a compensatory devaluation of the exchange rate. Rather, appreciation and overvaluation of the exchange rate were caused by the sharp rise in capital flows-- which took place mostly as a consequence of capital account opening, at a time in which there was a global increase of capital flows to all emerging markets--as well as to monetary and exchange rate policies in some countries (notably in Argentina and Brazil). This unexpected phenomenon led to a period of increase in growth and relative prices of the nontradable sector, as well as of cheap capital imports. The latter, in 118 | GUILLERMO PERRY AND MARCELO OLARREAGA addition to the effect of reductions in tariffs of capital goods imports, caused biased capital deepening and additional increased demand for skills. The increased share of the nontradable sector also led to an increase in informality, as discussed in the next section. On the other hand, currency overvaluation delayed some of the expected increased rents to natural resource owners and mitigated the increase in demand for skills, as nontradable sectors are usually less skilled-intensive. Thus currency overvalu- ation had both temporary positive and negative effects on income distribution, though probably the latter were larger in most countries. These effects were transitory; they ended abruptly with the currency crises and devaluations prompted by the generalized capital flows reversals after the Russian crisis of 1998. Finally, the growing importance of the (nontradable) service sector in some devel- oping countries, induced by changes in consumer preferences as countries developed, may also help explain the increase in wage premium.25 For this to be true, the service sector would need to be more skilled-intensive than the rest of the economy. Sanguinetti, Arim, and Pantano (2001) provide some evidence for Argentina and Uruguay, where the skilled to unskilled labor ratio in services is around 50 percent higher than in the rest of the economy. Further Effects of Trade Liberalization on Income Distribution Pervasive labor market rigidities in Latin America mitigated or delayed the realloca- tion of labor in response to the price changes induced by trade liberalization. At the same time, they might have contributed to temporary increases in unemployment in some countries, and to observed increases in informality, adding to inequality effects. There might also have been effects of trade liberalization on increased labor partici- pation--especially of female workers who typically had lower skill levels--which might have accentuated wage inequality trade effects, but reduced overall household income inequality.26 There has also been a debate around the effects of trade opening on macroeconomic and household incomes volatility. Evidence on this is mixed. Macroeconomic volatility was actually reduced in the 1990s with respect to the two previous decades (De Ferranti and others 2000), although this might have been mostly the effect of better macroeconomic policies. To the extent that the poor are less pro- tected against macro and idiosyncratic labor market risks, any increase in such risks would lead to a more unequal distribution of welfare and of income distri- bution in the short run. (It might also have some long-term effects on income distribution through lower human capital accumulation of the poor.)27 Moreover, as trade liberalization imposes competitive pressure and requires quick adaptation to changing market conditions, it may lead to a move toward informality where labor and business regulations are rigid, leading to unprotected and perhaps lower-paid workers. Trade liberalization affects households' income and welfare through changes in factor returns and employment, but also through changes in prices of the goods they consume. Goñi, López, and Sérven (2005) found that in Latin America, price changes TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 119 in the 1990s strongly benefited the poor and hurt the more well-to-do. Figure 8 shows the case of Brazil, but Goñi, López, and Sérven found similar results for other countries. This is the other side of the coin of the reduction of former protection that mostly affected unskilled labor-intensive activities, which mostly produced basic con- sumption goods. Commonly used income inequality measures do not correct for the heterogeneity of the consumption bundle by deciles. Thus actual increases in inequality in Latin America in the 1990s may not be as large as indicated by commonly used figures (which used the aggregate CPI as a deflator). Similarly, the effects of trade reforms on increased inequality may have been exaggerated. A more comprehensive discussion of the impact of trade liberalization on unem- ployment, income volatility, and informality is left for the next section, which focuses on poor households. Finally, trade liberalization may also affect households' incomes and welfare through changes in government taxes and expenditures in response to the effects of trade liberalization on government revenues. Actually, most countries in Latin America compensated for tariff revenue reductions by VAT increases, with no obvious effect on inequality. On the other hand, in those countries in which taxes and royalties from oil and mineral extraction are important, the medium-term effect on inequality of increased public expenditures was probably quite heteroge- neous, ranging from being progressive in Chile (where public expenditures have been progressively focused in favor of the poor) to regressive in oil-producing countries (in which high subsidies to energy, tertiary education, and pensions tend to make overall public expenditures regressive).28 Given that the larger effects of trade liberalization on increased wage inequality operated through increases in skill premiums, government responses in increasing education coverage--through both investments in education supply and conditional cash transfers to increase effective demand by the poor--are key for determining overall long-term effects (see the fourth section on policy complementarities). FIGURE 8. Individual and Average Annual Inflation, Brazil, 1988­96 734 733 732 731 average 730 inflation of 729 elv 728 Le727percentile specific 726 725 724 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 Income percentiles Source: Goñi, López, and Servén 2005. Note: The x-axis shows the income percentiles and the y-axis shows the level of inflation over the period 1988­96 experienced by the average individual in each percentile of the Brazilian income distribution. 120 | GUILLERMO PERRY AND MARCELO OLARREAGA Trade Reforms and Poverty in Latin America The discussion in the second section made clear that regardless of whether trade reforms generate aggregate gains, not everyone in the economy will benefit, and that Latin America's trade liberalization in the late 1980s and early 1990s was actually accompanied by increases in income inequality in many countries. However, indexes of income inequality are aggregate measures, and increases in income inequality may be consistent with declines in poverty. This section focuses exclusively on the impact that trade reforms in Latin America had on the region's poor. Recent cross-country evidence (Dollar and Kraay 2004) shows no statistically significant impact of trade on the relative income of the poor in a sample of 72 devel- oping countries and 24 developed countries. This is not surprising, given that as argued in the previous section, the impact of trade on income inequality is inherently a function of relative endowments and initial conditions, such as tariff structure and other regulations and policies. To find a statistically significant relationship would have meant that a homogeneous relationship exists between trade reform and house- hold income regardless of factor endowments.29 A casual look at the evolution of the income of the poor in Latin America before and after trade reforms gives an ambiguous picture. Of the ten countries for which we have data on the income of the poor before and after trade liberalization, six countries show an increase in real income, but four countries experience a decline in the income of the poor after trade reforms (figure 9). The absence of a clear pattern is confirmed when we combine our different exploratory exercises regarding the evolution of poverty, income inequality, and wage inequality before and after trade reforms. Results are summarized in table 3. Costa Rica is the only country in the region for which we have data available that experienced a significant declined in wage inequality. As expected, this was accom- panied by an increase in the income of the poor. In Brazil and Mexico, however, FIGURE 9. Income of the Poor Before and After Trade Reform in Latin America , 1600 1400 quintile PPP 1200 at 1000 bottom in USD 800 1985 600 incomes 400 age constant 200 erv A 0 y azil ru xico Br Chile Pe Rica ador Me ugua Colombia Salv Ecuador Ur Costa El Before After Sources: Dollar and Kraay (2002) for the income of the poor. Wacziarg and Welch (2003) for the timing of trade reform. TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 121 Table 3. Wage and Income Inequality and Poverty Before and After Reform Wage inequality Income inequality Increased by more Changed by less Decreased by more than 5 percent than 5 percent than 5 percent Increased by more Bolivia, El Salvador, Argentina, than 5 percent Guatemala, Mexico Colombia Changed by less Brazil, Ecuador, Chile, Peru Costa Rica than 5 percent Uruguay Decreased by more than 5 percent Source: Authors' analysis based on the material presented above. Note: Countries in bold are those in which the income of the poor declined by more than 5 percent after trade reforms. Countries underlined are those that did not experience a change in the income of the poor (within a 5 percent margin) after trade reforms. Countries in italics are those in which the income of the poor increased by more than 5 percent after trade reforms. If the country name is not in bold, underlined, or in italics, it means that we had no information regarding the income of the poor before and after liberalization. where wage inequality increased, the income of the poor declined after trade reforms. But even in some countries (Bolivia, Ecuador, and El Salvador) where wage inequality increased significantly, there were some increases in the income of the poor. This illustrates the difficulty of inducing the evolution of poverty by looking at the evolu- tion of wage or income inequality. Note, however, that this prima facie evidence does not imply causality, but simply that these changes happened simultaneously. In the case of Mexico, microeco- nometric studies confirm the increase in wage inequality after the reform (Robertson 2000; Nicita 2004). Nicita (2004) shows that in spite of a decline in poor house- holds' wages and agricultural income associated with Mexico's trade reforms in the 1990s, their welfare increased due to a sharp decline in the cost of their consumption bundle.30 Similarly, Argentina seemed to have experienced significant increases in income inequality at the time of trade reforms, as suggested by Porto (2006). However, when Porto measured the impact of trade policy changes on individual wages and established causality, results suggest that the poor experienced a 6 percent increase in real labor income that can be attributed to the trade reforms of the 1990s. But changes in wages, agricultural income, and the cost of the consumption bundle are only part of the story when measuring the impact of trade reforms on the poor. As trade reforms get implemented, firms adjust, jobs are lost in some activities, and employment opportunities are made available in other firms within or across sectors. Changes in the employment status or transitions into or out of formality induced by trade reform, as well as changes in individual income volatility, will affect the income of the poor in the medium term. But perhaps more importantly they may affect their investment decisions in physical and human capital, which in turn will affect their long-term income levels. In the next sections, we first look at the impact that trade reforms had on the unemployment of poor individuals in Latin America. We then turn to the impact of trade liberalization on formal-informal employment, and finally on income volatility. The section closes by looking at the medium-term implications of trade reforms on poverty through some of these channels. 122 | GUILLERMO PERRY AND MARCELO OLARREAGA Trade-induced Changes in Unemployment of the Poor The evolution of unemployment before and after trade reforms varies across countries in Latin America. While unemployment increased after a couple of years in Argentina, Brazil, and, transitorily in Peru, it actually fell in Chile and Colombia, and did not change much in Mexico. These observed trends do not imply causality, however. Some studies have attempted to establish causality with the help of microeconomic data. Goldberg and Pavnick (2005) could not find a statistically significant impact of trade reforms on urban unemployment in Colombia. But they argue that data con- straints (a short time dimension) and the partial equilibrium nature of their exercise may partly explain the absence of impact. Moreover, they do not allow for differ- ences on the impact of trade on unemployment across individuals, and therefore provide little evidence of whether the poor (or the unskilled) had stronger or weaker movements in and out of unemployment. Casacuberta and Gandelman (2006) look for differences across types of workers (unskilled versus unskilled) in the adjustment process of the labor demand of Uruguayan firms before and after the trade reforms that started in 1989. They find that trade reforms accelerated the adjustment process when firms are creating jobs for both skilled and unskilled workers. However, the adjustment process is also much faster when firms are cutting jobs of unskilled workers after the reforms. They did not observe any impact for skilled workers. This seems to suggest that unemploy- ment of the skilled was reduced by Uruguay's trade reforms, whereas the impact on the unskilled remains ambiguous. However, unskilled workers are more likely to be able to move out of unemployment and benefit from the more rapid creation of jobs when labor mobility is facilitated by labor market reforms. Porto (2005) takes a more direct look at the potential impact of trade expansion on labor income. He estimates the impacts of eventual world agricultural trade liberali- zation on wages, employment, and unemployment in Argentina. In the estimation of these wage and unemployment responses, the empirical model allows for individual labor supply responses and for adjustment costs in labor demand. It finds that a 10 percent increase in the price of agricultural exports would cause an increase in the Argentine employment probability of 1.36 percentage points, matched by a decline in the unemployment probability of 0.75 percentage points and an increase in labor market participation of 0.61 percentage points. Further, the unemployment rate would decline by 1.23 percentage points (by almost 10 percent). Expected wages would increase by 10.3 percent, an effect that is mostly driven by higher employment probabilities. Interestingly Porto finds no statistically significant differ- ences between the poor and the nonpoor, but he observes that the employment effect contributes a larger proportion to the increase in the expected wage of the poor, whereas the wage effect dominates in the case of nonpoor. Krivonos and Olarreaga (2005) find similar effects in the case of increases in export prices in Brazil. The study estimates the impact of an eventual 10 percent increase in world price of sugar (an important export commodity in Brazil), as a con- sequence of global trade liberalization, on labor income, wages, and employment. The results suggest that labor income would increase by an average 1 percent. The impact is 10 percent larger at the bottom of the income distribution. As in Porto TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 123 (2005), the contribution to changes in income of moves out of unemployment is much larger at the bottom of the income distribution: 45 percent versus 28 percent of the total change in expected income at the top of the income distribution. The latter two exercises suggest that even though trade may lead to increases in wage inequality, the poor may be better off, not only because their real wages may increase (due to reductions in cost of their consumption bundle), but also because in some cases of trade expansion their employment opportunities could increase.31 Such increases in labor demand for the unskilled may not necessarily translate into higher wages because there is a large pool of unskilled unemployed in most countries that puts downward pressure on unskilled wages. To provide some further evidence on the role of trade openness on the unemploy- ment of the poor in the region, we ran the regressions presented in table 4. They explain unskilled and skilled labor unemployment (proxied by those unemployed with primary and secondary education for unskilled workers and those with tertiary edu- cation for skilled workers) with country and year dummies, and GDP per capita as controls. The indicator of trade openness is borrowed from Wacziarg and Welch (2003).32 We also interact trade openness with an indicator of abundance in natural resources to explore the potential heterogeneity in a region abundant in natural resources. The sample is composed of 27 countries in the Latin America and the Caribbean region for which data are available in the World Development Indicators. The results suggest that GDP per capita and trade openness have a statistically sig- nificant and negative impact on unemployment of both skilled and unskilled workers. However, countries that have a stronger comparative advantage in natural resources tend to have a smaller impact of trade openness on unemployment, and the sign may even be reversed: countries very rich in natural resources may experience increases in unemployment when they open up to trade. Table 4. The Impact of Trade Openness on Unskilled and Skilled Labor Unemployment Unskilled labor Skilled labor unemployment unemployment Log of GDP per capita -0.26* -0.45* (0.06) (0.11) Trade openness -0.32* -0.98* (0.06) (0.23) Index of natural resources 0.07** -0.76* (0.03) (0.07) Trade openness * Index of natural resources 0.23* 0.54* (0.03) (0.20) R-squared 0.63 0.64 Number of observations 126 126 Source: Authors' estimations. Note: The estimator is OLS with country and year. Standard errors corrected for correlation before and after trade reforms are provided in parenthesis. * Significant at the 1 percent level. ** Significant at the 5 percent level. 124 | GUILLERMO PERRY AND MARCELO OLARREAGA Impact on Formal-Informal Employment and the Consequences on Poverty Almost half the Latin American labor force works in the informal sector. Moreover, informal employment rose in some countries in the region, such as Colombia and Peru, in the 1990s, while trade reforms were implemented. However, it is unlikely that trade liberalization contributed to the increase in informality, as has been claimed by many analysts.33 Sectors with a larger exposure to trade (measured by the ratio of export and imports to GDP) tend to have higher rates of formality (De Ferranti and others 2002). Moreover, natural resource­based sectors, in which many Latin Amer- ican countries have a comparative advantage, do not tend to show higher rates of informality, and there is little evidence of a shift toward outsourcing with informal firms after trade opening (De Ferranti and others 2002). Given that informality is con- centrated in the nontraded sector, the increase in informality is more likely to be due to currency overvaluation that pushed the economy toward nontraded activities in the 1990s. Figure 10 presents evidence for the cases of Brazil and Mexico. The only available micro-evidence on the link between trade reform and transi- tions between formal and informal sectors is the paper by Goldberg and Pavnick (2003), which focuses on the trade liberalization experiences of Brazil and Colombia. They find no significant impact of trade liberalization on informal employment in Brazil. For Colombia they obtain some weak evidence that the trade liberalization of the late 1980s resulted in increased informality, but after the labor market reforms of the 1990s, they find that the continuous opening of the Colombian economy did not lead to more informality. This result highlights the importance of considering the different institutional setups under which reforms take place. Regardless of whether trade reform leads to more informal employment or not, one may wonder whether increases in informality are associated with more poverty. Marcouiller, Ruiz de Castilla, and Woodruff (1997) find evidence of a formal sector wage premium in El Salvador and Peru, but not in Mexico once they control for individual worker characteristics (and selection). In the case of Mexico, they find a wage premium in the informal sector. This inconclusive evidence is complemented by a study of employment transitions in Mexico by Maloney (1999). He finds little evidence of segmentation between the informal and formal sectors. Only 30 percent of workers who moved into the informal sector did so involuntarily. Those who voluntarily moved saw an increase in income of close to 30 percent. The nonincome aspect of the decision to move between formal and informal employment is also important and sometimes neglected when purely focusing on wage income (Marcouiller, Ruiz de Castilla, and Woodruff 1997). On the one hand, the presence of benefits in the formal sector may increase the value of formal employment. On the other hand, independence and flexibility in the informal sector may increase the value of informal employment for those transitioning, even though wages may be lower (Bosch and Maloney 2005). To conclude, regardless of whether trade reform contributed to shifting workers into informality or not, it is unlikely that this would have had any significant direct impact on poverty. It may, however, have an indirect impact on poverty due TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 125 FIGURE 10. Evolution of Relative Formal/Informal Sector Sizes and Wages, Brazil and Mexico a. Brazil 2.6 4.5 2.4 4.0 2.2 ied 3.5 2.0 salar ages w e 1.8 3.0 mal or inf/ 1.6 Relativ 2.5 1.4 malroF 2.0 1.2 1.0 1.5 19821983198419851986198719881989199019911992199319941995199619971998 real exchange rate wage/formal/informal formal/informal b. Mexico 2.8 4.3 2.6 4.1 2.4 3.9 ied 2.2 3.7 salar ages w 2.0 3.5 e mal or 1.8 3.3 nfi/ Relativ1.6 3.1 1.4 2.9 malroF 1.2 2.7 1.0 2.5 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 real exchange rate wage/formal/informal formal/informal Source: De Ferranti and others 2002. Note: The left y-axis measures the formal wage to informal wage ratio (no units); the right y-axis measures the ratio of formal to informal employees (no units). The y-axis also measures changes in the real exchange rate. to the higher income volatility and lower protection against adverse shocks in the informal sector. Impact on Income Volatility and Poverty There was a decline in Latin America's aggregate income volatility that coincided with the period of trade reforms in most countries.34 But this was probably due more 126 | GUILLERMO PERRY AND MARCELO OLARREAGA to better macroeconomic policies than to trade liberalization, as argued earlier. The theoretical literature has suggested various channels through which trade reform might affect individual income risk. Most of them, but not all, suggest that trade liberalization may increase income risk. First, the increase in foreign competition leads to a reallocation of factors of production across firms and sectors that can increase income risk. Second, some have argued that increased foreign competition increases demand elasticities for goods produced domestically and therefore the derived labor demand elasticities; this implies that any given shock may lead to larger variations in wages and employment. On the other hand, a given shock may be better absorbed in a bigger market. Therefore, a more open economy may experience lower income volatility than a closed economy--unless the closed economy is subject to fewer shocks than the global economy. Thus, the answer to whether trade liberaliza- tion leads to more or less income volatility is necessarily empirical. Until very recently there was only anecdotal evidence that linked trade reforms to individual income volatility (see Winters, McCulloch, and McKay 2004). Krebs, Krishna, and Maloney (2005) fill this gap. They estimate time-varying parameters of individual income risk for Mexico during the 1987­8 period, built industry level and time-varying estimates of income risk, and correlated them with tariff levels and changes in tariffs. They find no statistically significant impact of tariff levels on income risk. In other words, workers in sectors with lower tariffs do not necessarily experience higher or lower income risk. However, they find that changes in tariffs do generate higher income risk: a tariff reduction of 5 percent raises the standard deviation of the persistent shocks to income by about 25 percent. Such an increase in income risk can have important consequences on already vulnerable households, increasing the need for safety nets at the time of trade reforms. This again underlines the importance of accompanying trade reforms with adequate policies and institutions. The medium- and long-run consequences of increased income volatility are discussed in the next section. Medium- and Long-Run Effects of Trade Policy on the Income of the Poor In the longer run, the economy will continue to adjust to the new incentives. It is likely that these adjustments will become increasingly important to determine the long-run impact of trade reforms on poverty. These adjustments refer, for example, to human capital investment decisions. Faced with changed wages for skilled and unskilled labor, individuals may review their decisions. For instance, if the skill premium increases with trade, there will be incentives to increase the average educational attain- ment. This implies that individuals will become more educated, with positive private and aggregate effects. But education policies need to follow to ensure that everyone can indeed invest in education as trade reform increases the incentives to invest on it. In terms of production decisions, if prices of certain crops remain higher and pro- duction is sustainable and profitable, farmers may decide to acquire the knowledge that is needed for effective cultivation or may decide to upgrade the capital stock required to start production (such as oxen charts, ploughs, or high-yield seeds). Similarly, firms may exploit the new trading opportunities by investment in physical capital TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 127 and investing in research and development (R&D) to increase productivity. Or firms may adjust the quality of their goods. These responses may lead to further growth in labor demand and further changes in wages and household welfare. The potential increases in income risk may also have long-run consequences in the absence of credit and insurance markets. Households facing higher income volatility are likely to invest less in education and may be more likely to require children to drop out of school when faced with important negative shocks. This can lead to persistence in low education levels among poor families (Perry and others 2005). Well-developed credit and insurance markets and conditional cash transfers--such as Oportunidades in Mexico--can help solve these problems (De Janvry and others 2006). Some of these long-run impacts are often ignored in the discussion. This is proba- bly because there is not enough variation in the data to have a sense of how long-run decisions respond to changed trade opportunities. But even if scarce documentation of these impacts exists, one needs to emphasize their importance. Long-run changes imply further household adjustments, and household adjustments call for policies that cushion the exposure to negative shocks and boost the impact of positive ones. This is especially true if the aggregate effects of trade liberalization on growth are positive. Thus, even if trade liberalization may increase inequality and even poverty in the short run, the positive long-run effects on growth are likely to lead to poverty reduction in the long run. Some evidence in a large panel to support this point is pro- vided in Perry and others (2005). The Importance of Complementary Policies The literature on trade and growth has long argued that trade reform is not a silver bullet; other policies are needed for countries to develop. More recently, the litera- ture has argued that the impact of trade reform on growth itself may depend on other policies, regulations, and institutions. For example, Banerjee and Newman (2004) present a model in which lack of financial development and sluggish factor mobility result in losses in poor countries when opening up to trade, as unproductive sectors are wiped out by foreign competition but the capital and labor attached to them fail to divert to more efficient uses. Acemoglu and Zilibotti (2001) show that access to imported intermediate inputs and capital goods does not lead to productivity improvements in developing countries that fail to improve their human capital (to adopt the new technologies) and to enforce intellectual property rights (to encourage the development of technologies best suited to their skill mix). The empirical evidence quickly followed. Bolaky and Freund (2004) show that trade does not lead to higher growth in economies with excessive business and labor regulations. Increased openness is, if anything, associated with a lower stan- dard of living in heavily regulated economies. Excessive regulations restrict trade- induced growth because resources are prevented from moving into the most productive sectors and to the most efficient firms within sectors. In addition, in highly regulated economies, increased trade is more likely to occur in the "wrong" goods: that is, goods where comparative advantage does not lie. Their results imply 128 | GUILLERMO PERRY AND MARCELO OLARREAGA that regulatory reform is not only beneficial per se, but it also enhances the benefits of trade liberalization. Chang, Kaltani, and Loayza (2005) present some interesting panel evidence (Bolaky and Freund present a cross section) on how the growth effect of openness depends on a variety of structural characteristics. For this purpose, they use a growth regression specification that interacts a proxy of trade openness with proxies of educational investment, financial depth, inflation stabilization, public infrastructure, governance, labor-market flexibility, ease of firm entry, and ease of firm exit. They find that the growth effects of openness are positive and econom- ically significant in the presence of a relatively flexible labor market and firm entry and exit regulations. The estimates for other interactions were not robust across specifications. Perhaps surprisingly, the literature on how other policies, institutions, or regula- tions may affect poverty and income inequality outcomes before and after trade reforms is almost nonexistent. It is clear, however, that as in the case of GDP growth, the impact of trade reform on the income of the poor could be boosted or even change sign in the presence of different institutional setups and complementary reforms. De Ferranti and others (2005) present some evidence.35 In the discussion that follows, we illustrate the key interrelationship between trade reforms and other policies by undertaking two exercises. First we provide micro-level evidence that suggests that the presence of other reforms may affect the way in which trade liberalization affects wage inequality. This is done by focusing on the differen- tial impact of tariff changes on wages and on wage inequality in two different episodes of trade liberalization of the Argentine economy: the liberalization of the 1970s and early 1980s on the one hand, and the trade liberalization of the 1990s on the other. Second, we provide some cross-country evidence by merging the datasets of Chang, Kaltani, and Loayza (2005) with those of Dollar and Kraay (2002). This allows us to see the extent to which other policies affect the impact of trade reforms on the relative income of the poor within Latin America and the Caribbean.36 Complementary Reforms and Wage Inequality37 Argentina undertook two trade reforms in the last three decades. The first (failed) episode occurred in the late 1970s, and the second in the early 1990s. These two trade reforms have particular features that make them different, such as initial levels of tariffs. But the two major differences, perhaps, were the nature of the other reforms that took place at the same time,38 and the entry of China and India in world markets as an important competitor for unskilled labor-intensive products. In what follows, we exploit these differences in concurrent reforms and entry of a potential competitor in world markets to better understand how a given reduction in tariffs can affect the skill premium in different ways, depending on other circumstances that characterize the economy. The analysis builds on a recent paper by Galiani and Porto (2005). They built an unusual historical dataset of tariffs and wages spanning the 1974­2001 period in Argentina. This comprises almost 30 years of data on sector-level tariffs (at the TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 129 three-digit level of the ISIC classification) and individual wages. They regress the log of the wage of individual i, in industry j at time t (ln wijt) on a number of individual characteristics (xijt) and the log of the tariff in industry j at time t, ln jt. That is, ln Wijt = xijtt + t EDUCijt + ln jt + ln jt EDUCijt + Ij + Yt + ijt, (1) where EDUC are educational dummies (unskilled, semiskilled, skilled), Ij is an indus- try fixed effect, and Yt is a survey effect. In this formulation, both the returns to schooling and the returns to age are allowed to vary from year to year. Three defini- tions of skills are proposed. Unskilled labor comprises workers with at least primary education. Semiskilled labor includes workers with secondary education. Skilled labor is represented by college graduates. By including survey effects and industry dummies, they control for changes in exchange rates (devaluations and apprecia- tions) and industry-specific shocks so that the impacts of tariffs are not confounded by specific shocks or by aggregate shocks (related to policy or business cycle). These fixed effects help control for unobservable effects that may produce a spurious cor- relation between tariffs and wages. By comparing their different estimates of and in the 1970s and 1990s, we are able to assess how the impacts of trade on skill premiums depend on other concurrent policies and/or differences in the external environment. The main findings are reported in table 5. The first column lists the results for the liberalization of the 1970s and the second, for the liberalization of the 1990s. The impacts of the trade reforms were different in the two periods. During the first liberalization episode, the average wage reacted positively to tariffs, so that the tariff cuts led to a decline in wages (for potential explanations, see the second section above). In addition, we do not find any impact of the tariff changes on the skill premium, either for the semiskilled or the skilled workers (the interactions of the tariff with the educational dummies are not statistically significant). During the second episode of trade liberalization, during the 1990s, the positive association of tariffs and average wages is still observed. However, Table 5. Tariff Reforms, Complementary Reforms, and the Skill Premium, Argentina Liberalization of the 1970s Liberalization of the 1990s Log tariff 0.824* 0.128* (0.364) (0.043) Log tariff * semiskilled ­0.032 ­0.128* (0.098) (0.048) Log tariff * skilled 0.21 ­0.442* (0.241) (0.111) Time-varying returns to schooling Yes Yes Time-varying return to age Yes Yes Time effects Yes Yes Industry effects Yes Yes R-squared 0.86 0.35 No. of observations 7,922 11,131 Source: Galiani and Porto (2005). * Significant at the 1 percent level. 130 | GUILLERMO PERRY AND MARCELO OLARREAGA in this second period, there are negative and significant interaction terms for semi- skilled and skilled workers. The estimates suggest the following dynamics of wages after the two tariff reforms. During the 1970s, trade liberalization led to an average decline in wages, with no overall distributional consequences on wage inequality. However, the reforms of the 1990s led to a decline in the wages of the unskilled workers, no significant change in the wages of the semiskilled workers, and an increase in the returns to highly skilled workers. While part of the estimated differences between the two periods may be due to differences in the tariff changes between the two liberalization episodes or changes in the external environment (the entry of China and India into world markets), another part of the story can be attributed to the differences in concur- rent policies that were taking place during the 1970s and 1990s. For example, the openness of the 1990s, together with the privatization of most of the Argentine service sector, could have led Argentine firms to upgrade the quality of their exportable products, thus leading to an increase in the relative demand of skills and in the skill premium. Similarly, the financial reforms could have comple- mented the new trading opportunities to induce firms to finance and engage in skilled biased technical change. Also, the competition imposed by Chinese products may have pushed Argentina into specializing in more skilled-intensive products and kept the pressure on the wage of unskilled workers. These stories, while not formally proven by our analysis, clearly suggest that the interaction of tariff reforms with other concurrent reforms and events may play a role in the impacts on wage inequality. The Role of Complementary Policies in Explaining the Impact of Trade Reform on the Income of the Poor: Cross-country Evidence for Latin America and the Caribbean We combined the datasets of Dollar and Kraay (2002) and Chang, Kaltani, and Loayza (2005) to try to explore the role played by complementary policies in explaining the heterogeneity of the impact of trade reform in Latin America and the Caribbean on the income of the poor. We considered as complementary policies labor market flexibility, entry of firm flexibility, closing of firm flexibility, a gover- nance indicator, and secondary education enrollment (as a proxy for education poli- cies). All these variables were borrowed from Chang, Kaltani, and Loayza (2005). The variable to be explained is the log of average income in the bottom quintile, borrowed from Dollar and Kraay (2002). Control variables include the log of GDP per capita, and country and year dummies. The trade openness indicator is as before the one provided by Wacziarg and Welch (2003). The sample is given by matching of the two datasets, and it contains 18 Latin American and Caribbean countries (Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico, Nicaragua, Panama, Peru, Uruguay, and Venezuela). There is a maximum of six observations per country. Thus the sample is small and inferences should be made with caution. TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 131 Table 6 summarizes the results. The first column suggests that trade openness had a positive but not very significant impact on the income of the poor (condi- tional on GDP per capita). But as we explore the impact of trade reforms on the income of the poor under different policy environments in the other columns, the coefficient on openness sometimes changes sign and becomes statistically signifi- cant. The second column suggests that openness will have a positive impact on the income of the poor in countries with flexible labor laws, but not necessarily in countries with more rigid labor laws.39 Similarly, the third column suggests that openness will have a stronger and more positive impact on the income of the poor in countries with low barriers to firm entry, whereas flexibility to close firms (fourth column) has no statistically significant impact. Similarly, openness is more likely to have a positive impact on the income of the poor in countries with higher levels of education (fifth column). Also, trade reforms are more likely to positively affect the income of the poor when accompanied by better governance (sixth column). The last column introduces all variables simultaneously, and suffers from multi- collinearity, but confirms that openness is more likely to positively impact the Table 6. The Role of Complementary Policies in Explaining the Impact of Trade Openness on the Income of the Poor Log Log Log Log Log Log Log income income income income income income income of the of the of the of the of the of the of the poor poor poor poor poor poor poor Log GDP 0.93** 0.93** 0.92** 0.92** 0.92** 0.90** 0.90** per capita (0.06) (0.06) (0.06) (0.07) (0.07) (0.08) (0.07) Openness 0.03 -0.29 -0.48** 0.04 -0.43** -0.30* -0.39** (0.08) (0.17) (0.23) (0.08) (0.19) (0.15) (0.17) Openness*labor 0.83** 0.57* market flexibility (0.30) (0.29) Openness*firm 0.85** 0.42 entry flexibility (0.34) (0.80) Openness*firm 0.02 -0.07 closing flexibility (0.21) (0.25) Education 0.02 0.01 (0.06) (0.06) Openness* 0.13** -0.09 education (0.04) (0.14) Openness* 0.84** 0.77 governance (0.36) (0.48) R-squared 0.62 0.63 0.64 0.60 0.63 0.64 0.98 No. of observations 74 68 72 68 72 72 68 Source: Authors' estimation based on the merger of datasets from Dollar and Kraay (2002) and Chang, Kaltani, and Loayza (2005). Note: All regressions are with-in country and year dummies (using deviations to the country and year mean and then adding up the overall mean) and standard errors provided in parenthesis are corrected (nonparametrically) for corre- lation in the error term before and after trade reform in each country. ** Significant at the 5 percent level. * Significant at the 10 percent level. 132 | GUILLERMO PERRY AND MARCELO OLARREAGA income of the poor in countries with labor market flexibility, low barriers to firm entry, and good governance.40 Summing Up: Short- and Long-term Effects of Trade Liberalization on Inequality and Poverty Although the impact of trade reform on poverty, wage inequality, and income inequality was quite varied within Latin American countries, a broad view emerged that trade reforms in the region in the late 1980s and early 1990s were often accom- panied by increases in wage inequality and more moderate increases in income inequality. Regardless of whether or not wage and income inequality were rising, trade reform tended to be accompanied by reductions in poverty, mainly through reductions in the cost of the consumption bundle of the poor, as well as moves out of unemployment. This is important because as the income of the poor increases with trade reform, they may be more able to undertake the necessary investments to adjust in the presence of market failures, such as the absence of credit or insurance. Policy complementarities matter (with education, access to credit, insurance, and flexible entry and exit of firms and labor markets) and can also help the poor maximize the new economic opportunities offered by trade reforms. The increase in wage inequality in Latin America at the moment of trade reform contrasts with what apparently happened when the East Asian Tigers opened up at the end of the 1960s, and with what could have been expected from a simple HO/SS model. We argue in this paper that there is actually no puzzle. Most of the static effects of trade liberalization examined should have been expected to go in the direc- tion of increased demand for skills and thus of wage inequality, given both Latin America's initial structure of protection (that benefited mostly unskilled-intensive sectors) and relative endowments. The region is rich in natural resources, which are often complementary to capital and skills, and not necessarily abundant in unskilled labor. Latin America also has intermediate levels of relative capital endowments (as measured by capital to unskilled labor ratios), especially when compared with coun- tries such as China and India that have a large pool of unskilled workers and were integrating into the world economy at the time of Latin America trade reforms. Quantitatively more important, labor reallocation took place within sectors as product mix changed--more skilled-intensive products/processes appeared and grew at the margin, through outsourcing--and with the emergence of more productive and innovative firms, which normally demand higher skills, in an environment of heightened competition. Faster transfer and adaptation of skill biased technical change further strengthened relative demand for skills. It is thus not surprising to observe a generalized increase in relative demand for skills (especially at the tertiary level) in the years following trade liberalization, which led to higher skill premiums for workers with tertiary education (although not for workers with secondary edu- cation, given the sharp increases in secondary enrollments in many countries). Beyond skill premiums and wage inequality, the direction of other effects discussed is mixed, and they might have had some equalizing effects on household TRADE LIBERALIZATION, INEQUALITY, AND POVERTY REDUCTION IN LATIN AMERICA | 133 income distribution. Increases in rents for holders of property rights on natural resources probably contributed to income inequality in most countries, although they also benefited small farmers. Changes in consumer prices appear to have benefited the poor to a greater extent, as would be expected if many of the unskilled-intensive and formerly protected activities were producers of basic goods. This explains why the real income of the poor increased substantially after trade reforms in many coun- tries. In some countries there were temporary spells of unemployment that tended to increase inequality, while in others reductions in unemployment and/or some effects on increased labor participation might have had the opposite effect. Importantly, observed increases in informality cannot be attributed to trade opening. They were mostly due to other events and policies (exchange rate appreciation caused by increased capital flows, capital account opening, and monetary policies) that led to an unexpected initial increase in prices and share of nontradables. Policy complementarities can ensure that the poor benefit from trade reforms that would otherwise increase poverty. The evolution of adverse effects on skill premiums and wage inequality over time will depend on supply responses in education and training by workers, firms, and governments. A fast response in skill upgrading would reduce inequality effects over time and probably augment positive growth effects. Government actions to increase supply and quality of public education, to help overcome liquidity and informational constraints by poor families and workers, and to evolve toward more competitive and efficient training and remedial education services would play a key role in such a response. Increased access to credit and insurance would also be likely to allow poor households to adjust better to reforms and take full advantage of the new opportunities offered by trade liberalization. More flexible labor markets and entry and exit regulations for firms would also allow productive firms to enter and increase their share in potentially more profitable activities more easily. Finally, increased access to infrastructure, financial services, and technical assistance would facilitate small farmers' adjustment to new challenges and opportunities. Notes 1. Assuming that net gains exist, which may not be the case in the presence of market failures (such as access to credit or imperfect competition). 2. Chile first liberalized in 1975­6, although this was followed by backtracking after the debt crisis during 1983­5 and liberalization since then. For the purposes of figures showing developments before and after liberalization, we use 1985 here as the date of definitive liberalization. Similarly, we use 1989 as the date for Argentina's definitive liberalization, and not the failed liberalization of the late 1970s. 3. See Wood (1997) for a similar point. Data on skill premiums start in 1983, more than a decade after the Asian Tigers liberalize. It suggests, however, a decline in skill premium for the Rep. of Korea and unchanged skill premiums for Singapore and Hong Kong from 1983 onward (see Willem te Velde and Morrissey 2004). 4. See Krueger (1978); Bhagwati (1978); Choksi, Michaely, and Papageorgiu (1991). 5. See estimates for Argentina, Bolivia, Brazil, and Colombia in Manacorda, Sánchez- Páramo, and Schady (2005). 134 | GUILLERMO PERRY AND MARCELO OLARREAGA 6. Lora measures trade reform as the average of the average level of tariffs and the average tariff dispersion, although he obtains the same result with a measure of trade openness. 7. Interacting trade openness with GDP per capita. 8. Wood (1995, 1997) was the first to observe this as a potential explanation. It is surpris- ing, though, that apart from Davis, most of the already vast literature (see the survey by Goldberg and Pavcnik (2004)) attempting to explain this apparent puzzle ignores this basic fact and concentrates almost exclusively on other issues that certainly helped explain the puzzle, but were probably less important. 9. Our index of natural resources is based on trade in goods and ignores tourism exports. 10. Results are even more heterogeneous within developing countries in a three-factor world, with some experiencing improved income distribution and some worsened income distribution. Leamer and Levinson (1995) point out that there is currently no empirical identification of international production cones, and hence there is no way to anticipate distributional consequences of trade liberalization in a Davis framework. 11. As discussed above, there is considerable heterogeneity across different natural endow- ments by countries. 12. Complementarities of oil intensity with capital and skill were not significant, though this may be due to underestimation of capital investments in the sector (especially in exploration). 13. Ethier (1984) shows that in a world with more goods than factors, it is impossible to predict production levels (once zero profits and factor price equalization are assumed). The reason is that the solution to this problem involves f equations and g unknowns, where f is the number of factors and g is the number of goods, and f50 percent dividends, and liquidity increase significantly after ownership), 18 were partial sales, 12 were sold to employee privatization, while leverage, employment, and financial risk shareholding associations (ESAs), and 6 were sold to anchor investors. (measured as the inverse of times interest earned) decline significantly. Performance changes are pervasive across subgroups, but there is some evidence that full privatization works better than partial privatization, and that sales to ESAs work better than others. Omran (2002) Performs similar study to Omran (2001), but also compares performance Find that the performance of SOEs also improves significantly of privatized companies to a matched set of 54 firms that remained during the postprivatization period, and that privatized state owned. firms did not perform any better than SOEs. Sun, Jia, and Compares pre- versus postprivatization financial and operating Finds that privatized companies triple their absolute level of Tong (2002) performance of a sample of 24 Malaysian firms divested via public share profits, more than double real sales, and significantly offering by the end of 1997. Employs MNR tests first; then panel data increase dividends and reduce leverage. Results are robust regression to further examine sources of performance changes. across various subsamples. Stocks of privatized firms earn normal returns (insignificantly different from market index). Regression analysis shows that institutional investors and directors have a positive impact on the performance of privatized firms , and that option schemes, rather than direct remuneration, give better incentives to managers. Boubakri and Examines pre- versus postprivatization performance of 16 African firms Documents significantly increased capital spending by Cosset (2003) privatized through public share offering during the period 1989­96. privatized firms, but finds only insignificant changes in Also summarizes findings of three other studies pertaining to profitability, efficiency, output, and leverage. privatization in developing countries. Sun, Fang, and Tests whether privatization improves financial and operating performance Finds no significant change after privatization in any variable Tong (2004) of 31 Singaporean companies divested through public share offering except output (significant increase) using MNR methods. between 1975 and 1998. Employs MNR tests first; then panel data Then uses regression analysis to show that output and regression to further examine sources of performance changes. leverage improve, but that efficiency deteriorates, after privatization. Concludes that there is little performance improvement after the change because Singaporean SOEs were unusually well managed before divestment. Boubakri, Cosset, Investigates the role of ownership structure and investor protection in Documents that private ownership tends to concentrate over and Guedhami corporate governance using a sample of 170 firms from 26 developing time after divestment, and that privatization results in a (2005) countries that were privatized over the 1980­97 period. Specifically relinquishment of control by the privatizing government over examines what ownership structure results from privatization, and how three years after the initial sale. Much of the decrease in it evolves subsequently; how the level of ownership protection impacts state ownership is absorbed by foreign and local institutional postprivatization ownership structure; and how ownership structure and investors, while the average stake held by individuals is less investor protection relate to firm performance. important. Also finds that interaction between legal protection and ownership concentration has a significant negative effect on firm performance, suggesting that ownership concentration matters more in countries with weak legal protection. Source: Megginson 2005, chapters 3 and 4; Chong and López-de-Silanes 2005. 261 262 TABLE 2. Summary of Empirical Studies of Privatization: Developed Countries Study Sample description, study period, and methodology Summary of empirical findings and conclusion Galal, Jones, Tandon, Compares actual postprivatization performance of Documents net welfare gains in 11 of the 12 cases that equal, on and Vogelsang (1994) 12 large firms (mostly airlines and regulated utilities) in average, 26 percent of the firms' predivestiture sales. Finds no Britain, Chile, Malaysia, and Mexico to predicted performance case where workers were made worse off, and three where of these firms had they remained SOEs. workers were made significantly better off. Green and Vogelsang Provides a historical overview of British Airways evolution Shows that British Airways suffered severely during the airline (1994) as a state-owned enterprise through its first years as a depression of the early 1980s, but that the operational changes fully privatized company. Also analyzes how operating and restructuring that the management team executed during the and financial performance evolved before and after mid-1980s paved the way for the successful sale of the the company's sale. government's 100 percent ownership in 1987. Martin and Parker (1995) Using two measures (ROR on capital employed and annual Mixed results. Outright performance improvements after growth in value-added per employee-hour), examines privatization found in less than half of firm-measures studied. Several whether 11 British firms privatized during 1981­88 improved improve before divestiture, indicating an initial "shake-out" performance after divestment. Also attempts to control effect upon the announcement of privatization. for business cycle effects. Price and Measures the technical efficiency of the U.K. natural gas Shows that the industry's rate of productivity growth increased Weyman-Jones (1996) industry before and after its 1986 privatization and significantly after privatization--though not as much as it could associated regulatory changes using Malmquist indices have if the industry had been restructured and subjected to direct and nonparametric frontier analysis. competition and more appropriate regulation. Newberry and Performs a cost-benefit analysis of the 1990 restructuring and The restructuring/privatization of CEGB was "worth it," in that there Pollitt (1997) privatization of Britain's Central Electricity Generating Board has been a permanent cost reduction of 5 percent per year. (CEGB). Compares the actual performance of the privatized Producers and shareholders capture all this benefit and more. firms to a counter-factual assuming that CEGB had Consumers and the government lose. Also shows that alternative remained state owned. fuel purchases involve unnecessarily high costs and wealth flows out of the country. Laurin and Bozec (2000) Compares productivity and profitability of two large Canadian Total factor productivity of Canadian National was much lower than rail carriers, before and after the 1995 privatization of that of the privately owned Canadian Pacific (CP) during the Canadian National (CN). Compares accounting ratios for 1981­91 period, but became just as efficient during the entire 17-year period 1981­97 and for three subperiods: the preprivatization (1992­95) period, and exceeded it after 1995. CN fully state-owned era (1981­91), the preprivatization period stock price out-performed CP, the transportation industry, and the (1992­95), and the postprivatization era. Also examines stock Canadian market after 1995. Both firms shed workers after 1992, returns from 1995 to 1998. Creates a six-firm comparison but CN's employment declined by more (34 percent vs.18 percent) group of Canadian privatizations, and computes accounting as average productivity almost doubled (97 percent increase). ratios and stock returns for these firms as well. CN's capital spending increased significantly, though CP's increased more. A six-firm Canadian privatization comparison group also experienced significant increases in investment spending and productivity, and a significant decline in employment. Villalonga (2000) Examines the effect of privatization on efficiency for 24 Spanish Finds insignificant changes in level and growth rate of efficiency firms fully divested between 1985 and 1993. Tests for after privatization. The significant positive effect found for the separate effects of ownership change, once other political business cycle suggests that government sold firms during and organizational factors and time period (state of the recessions. Capital intensity, foreign ownership, and size also are business cycle) effects are accounted for. positively related to efficiency improvements. Privatization seems to decrease efficiency for five and six years after divestiture, but increase efficiency seven and eight years after, and four and three years before, suggesting the importance of time effects. Florio (2001) Presents an analysis of the welfare impact of the U.K. Concludes that privatization has modest effects on efficiency of privatization program from 1979 to 1997. Considers the production and consumption, but has important effects on impact on five types of agents: firms, employees, distribution of income and wealth. Acknowledges fiscal benefits, shareholders, consumers, and taxpayers. lower prices in most areas, and productivity growth, but asserts these would have been achieved under continued state ownership (due to extrapolation of existing trends). Calculates that, at best, the net present value of the welfare change for each British consumer is less than £1,000, and would be lower if distributional issues were accounted for. (Continues on next page) 263 264 TABLE 2. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusion Dumontier and Investigate the value that is created or lost during the state Finds that the government created value in nationalized firms, Laurin (2002) ownership period for each firm nationalized during 1982 and but state and taxpayers did not benefit because of the premium then re-privatized between 1986 and 1995. Then tests paid to shareholders upon nationalization (20 percent) and whether privatization improved performance over that underpricing of the initial public offering at privatization. Financial achieved during post-1982 nationalized period. Some 44 and operating performance of companies improved during companies (39 banks and 5 industrial firms) were nationalized nationalization phase, then improved even more after privatization. and then re-privatized. Profits and sales increased after privatization, while efficiency improved over all three periods. Employment fell during the nationalized period, but increased after privatization, due to higher sales. Capital spending was highest during nationalized period, due to government subsidies. Leverage declined during nationalized period, but increased after privatization. Dividends declined during the nationalized period, but increased after privatization. Saal and Parker (2003) Examines the productivity and price performance of the Finds no significant evidence that productivity growth, measured by privatized water and sewerage companies of England and growth in total factor productivity, is improved by privatization-- Wales after the industry was privatized and a new regulatory despite reductions in labor usage. Also finds that increases in regime imposed in 1989. Examines the joint impact of output prices have outstripped increased input prices, leading to privatization and new economic regulatory environment on significantly higher economic profits after privatization. performance. Source: Megginson 2005, chapters 3 and 4; Chong and López-de-Silanes 2005. TABLE 3. Summary of Empirical Studies of Privatization in Latin America Study Sample description, study period, and methodology Summary of empirical findings and conclusions Hachette and Analyzes the difference in 10 performance indicators of 144 private, Finds no significant differences in behavior among public, Luders (1993) public, and privatized firms in Chile during the period from private, and privatized firms that operate under similar sets 1974 to1987. of rules and regulations. Sanchez and Uses a descriptive case study approach to analyze the privatization Finds great differences in the effects of privatization in the Corona (1993) experiences of Argentina, Chile, Colombia, and Mexico. Focuses on countries covered by the study. Concludes that firms, the preparatory measures taken prior to privatization; on valuation, institutions, and regulations need sufficient time to prepare sale mechanisms, regulation and supervision, and on the fiscal and for the privatization process to be successful. macroeconomic impact of privatization. Galal, Jones, Compares postprivatization performance of 12 large firms from Finds net welfare gains in 11 of 12 cases covered. Gains are Tandon, and Chile and Mexico. The companies covered are mostly airlines and on average equal to 26 percent of the firms' predivestiture Vogelsang (1994) regulated utilities. sales. Finds no case where workers were made worse off, and three cases in which workers' conditions improved. Petrazzini and Using International Telecommunications Union data through 1994, Deregulation and privatization are both associated with Clark (1996) tests whether deregulation and privatization impact the level and significant improvements in the level and growth of growth of teledensity, prices, service quality, and employment. teledensity, but have no consistent impact on the quality The sample covers 26 developing countries, including some of service. Deregulation is associated with lower prices Latin American nations. and increased employment; privatization has the opposite effect. Pinheiro (1996) Analyzes the performance of 50 former Brazilian SOEs before and after Concludes that privatization has improved the performance privatization. Uses data up until 1994. The variables used are net sales, of the firms. Rejects the null hypothesis of no change in net profits, net assets, investment, employment, and indebtedness. behavior for the production, efficiency, profitability, and investment variables. It finds a significant negative impact on employment. Ramamurti (1996) Surveys studies of four telecom, two airline, and one tollroad privatization Concludes that privatization is very positive for telecoms, programs in Latin America during the period 1987­91. Also discusses partly due to the scope for technology, capital investment, political economic issues and methods used to overcome bureaucratic and attractiveness of offer terms. Finds much less scope for and ideological opposition to divestiture. productivity improvements for airlines and roads; observes little improvement. 265 (Continues on next page) 266 TABLE 3. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Ramamurti (1997) Examines restructuring and privatization of Ferrocarilla Argentinos, the Documents a 370 percent improvement in labor productivity national railroad, in 1990. Tests whether productivity, employment, and and a 78.7 percent decline in employment (from 92,000 to need for operating subsidies (equal to 1 percent of GDP in 1990) 19,682). Services were expanded and improved, and change significantly after divestiture. delivered at lower cost to consumers. Need for operating subsidies largely eliminated. Chisari, Estache, Assesses macroeconomic and distributional effects of privatization in Concludes that effective regulation translates into annual and Romero (1999) Argentina's gas, electricity, telecommunications, and water sectors. gains of about 1.25 billion of GDP. Privatization cannot It uses a computable general equilibrium model. be blamed for increased unemployment as it may be due to ineffective regulation. La Porta and Tests whether performance of 218 SOEs privatized through June 1992 Output of privatized firms increased 54.3 percent, while López-de-Silanes improves after divestment. Compares performance with industry- employment declined by half (though wages for remaining (1999) matched firms, and splits improvements documented between industry- workers increased). Firms achieved a 24 percentage point and firm-specific influences. increase in operating profitability, eliminating need for subsidies equal to 12.7 percent of GDP. Higher product prices explain 5 percent of the improvement; transfers from laid-off workers, 31 percent, and incentive-related productivity gains account for the remaining 64 percent. Ros (1999) Uses ITU data and panel data regressions to examine the effects of Countries with at least 50 percent private ownership in the privatization and competition on network expansion and efficiency. main telecom firm have significantly higher teledensity levels The study covers 110 countries during the 1986­95 period. and growth rates. Both privatization and competition increase efficiency. However, only privatization is positively associated with network expansion. Birch and Haar (2000) A descriptive study of the privatization experience in the last two Finds sizeable effects of privatization on the macroeconomic decades in Argentina, Brazil, Chile, Colombia, Mexico, Peru, Venezuela, conditions (both in the short and long run). Also shows a and some Caribbean countries. positive effect of privatization on productivity and a negative effect on employment. Clarke and Cull (2001) Uses evidence from the privatization program of provincial banks in Finds that provinces with high fiscal deficits were willing Argentina during the 1990s. Tests econometrically how political to, first, accept layoffs; and second, to guarantee a larger constraints affect transactions during bank privatization. part of the privatized bank's portfolio in return for a higher sale price. Pombo and Performs ex post measuring and econometric analysis of 30 large Panel data analysis finds very positive results for privatized Ramirez (2001) Colombian manufacturing firms and 33 power generation plants manufacturing firms. Total factor productivity indices increase privatized during the 1993­98 period. Employs both panel data from 0.27 to 0.50 points, while profit rates increase by 1.2 regression analysis and Megginson, Nash, and van Randenborgh percentage points. Productive efficiency in power production (MNR) matched pre- versus postprivatization tests. is not systematically related to ownership changes, once other factors accounted for. Galiani, Gertler, Examines the impact of privatizing water services on the mortality of All three measures show that child mortality fell 5 to 8 percent and Schargrodsky young children in Argentina. Between 1991 and 2000, 30 percent in areas that privatized their water services. The increase in (2001) of Argentina's public water companies covering 60 percent of the access to and quality of water caused the reduction in population were privatized. Estimates the impact of privatization on mortality. Investment increased, service provision became child mortality using three different measures. more efficient, and quality improved. The number of people connected to the network increased dramatically, but prices did not. Wallsten (2001) Analyses the effect of telecommunication reforms. It explores the Indicates that competition is significantly associated with impact of privatization, competition, and regulation on telecom firms' increases in per capita access to telecommunication services performance. This study covers 30 African and Latin American countries and with decreases in its costs. Privatization is helpful only during the 1984­97 period. if coupled with effective, independent regulation. Concludes that competition combined with privatization is best. Privatizing a monopoly without regulatory reforms should be avoided. (Continues on next page) 267 268 TABLE 3. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Estache (2002) Asks whether Argentina's 1990s utilities privatization program was a cure Finds that privatization, per se, was quite successful: it raised or a disease. Certainly, the privatizations of Argentina's electricity, gas, significant revenues for the state and the new private water and sanitation, and telecommunications utilities are today the operators, increased efficiency and service levels object of intense anger within the country, but the study attempts to significantly--without significantly raising the rates they determine whether this anger is appropriate. It notes that privatization charged. The rates charged to consumers, however, increased occurred just before the country was gripped by a massive political significantly, since the government exploited the new and economic collapse and tries to separate the impact of privatization ownership structure to impose indirect taxes that it could from the overwhelming impact of the collapse. not impose through direct levies. Once the economic crisis began, government actions discriminated against the privatized companies and foreign operators were vilified as exploiters when they tried to raise fees in line with inflation and devaluation. Trujillo, Martín, Uses pooled and panel data with fixed and random effects to examine Finds that private sector involvement in utilities and transport Estache, and the macroeconomic effects of private sector participation in have minimal positive effects on GDP. There is crowding out Campos (2002) infrastructure. Uses a sample of 21 Latin American countries from of private investment, private participation reduces 1985 to 1998. recurrent expenditures--except in transport, where it has the opposite effect. The net effect on the public sector account is uncertain. Chong and A detailed analysis of the contractual arrangements of privatizations Concludes that clear, homogeneous, transparent, and Sanchez (2003) and concessions in infrastructure. It covers four countries: Brazil, Chile, credible institutional processes during privatization yield Colombia, and Peru. positive outcomes. Source: Megginson 2005, chapters 3 and 4; Chong and López-de-Silanes 2005. TABLE 4. Summary of Empirical Studies of Privatization in the Transition Economies of Central and Eastern Europe Study Sample description, study period, and methodology Summary of empirical findings and conclusions Claessens, Djankov, Examines the determinants of performance improvements for a sample Documents that privatized firms do prosper, primarily and Pohl (1997) of 706 Czech firms privatized during 1992­95. Using Tobin's Q, tests because of the concentrated ownership structure that whether concentrated ownership structure or the presence of an results. Finds the more concentrated the postprivatization outside monitor (bank or investment fund) improves Q more than ownership structure, the higher the firm's profitability and dispersed ownership. market valuation. Large stakes owned by bank-sponsored funds and strategic investors are particularly value-enhancing. Dyck (1997) Develops and tests an adverse selection model to explain the Documents that privatized East German firms are much Treuhand's role in restructuring and privatizing eastern Germany's more likely to transfer Western (usually German) state-owned firms. In less than five years, the Treuhand privatized managers into key positions than are companies that remain more than 13,800 firms and parts of firms and, uniquely, had the state owned. Also finds that Treuhand emphasizes sales resources to pay for restructuring itself--but almost never chose open to all buyers rather than favoring Eastern Germans. to do so. Instead, it emphasized speed and sales to existing Western The main message is that privatization programs must firms over giveaways and sales to capital funds. The paper carefully consider when and how to affect managerial rationalizes the Treuhand's approach. replacement in privatized companies. Plans open to Western buyers, and which allow management change, are most likely to improve firm performance. Pohl, Anderson, Compares the extent of restructuring achieved by over 6,300 private Privatization dramatically increases the likelihood and Claessens, and and state-owned firms in seven Eastern European countries during success of restructuring. Firm privatized for four years Djankov (1997) 1992­95. Uses six measures to examine which restructuring strategies will increase productivity three to five times more improve performance the most. than a similar SOE. Finds little difference in performance. based on the method of privatization, but finds that ownership and financing effects impact restructuring. Smith, Cin, and Using a sample with 22,735 firm-years of data drawn from period of Finds that a percentage point increase in foreign Vodopivec (1997) "spontaneous privatization" in Slovenia (1989­92), examines the ownership is associated with a 3.9 percent increase in value- impact of foreign and employee ownership on firm performance. added, and for employee ownership, with a 1.4 percent increase. Also finds that firms with higher revenues, profits, and exports are more likely to exhibit foreign and employee ownership. (Continues on next page) 269 270 TABLE 4. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Weiss and Analyzes the effects of ownership by investment funds on the Finds that ownership concentration and composition jointly Nikitin (1998) performance of 125 privatized Czech firms during the period affect the performance of privatized firms. Concentration of 1993­95. Assesses these effects by measuring the relationship ownership in the hands of a large shareholder, other than between changes in performance and changes in the composition an investment fund or company, is associated with of ownership at the start of privatization. Uses robust estimation significant performance improvements (for all techniques, in addition to OLS, since data strongly reject measures of performance). Concentrated ownership normality. by funds does not improve firm performance. Preliminary post-1996 data suggest that changes in investment fund legislation may improve firm performance. Berg, Using macroeconomic data from 26 transition countries for 1990­96, Results point to the preeminence of structural reforms over Borensztein, examines the relative roles of macroeconomic variables, structural both initial conditions and macroeconomic variables in Sahay, and policies, and initial conditions in explaining the large observed explaining cross-country differences in performance and Zettelmeyer differences in output performance after transition began. the timing of recovery from the sharp recession that hit (1999) every transition economy in the early 1990s. Claessens and Studies the effect of management turnover on changes in Finds that the appointment of new managers is associated Djankov (1999a) financial and operating performance of 706 privatized Czech with significant improvements in profit margins and labor firms over the period 1993­97. Examines changes in profitability productivity, particularly if the managers are selected by and labor productivity. private owners. New managers appointed by the National Property Fund also improve performance, though not by as much. Claessens and Examines the relationship between ownership concentration and Finds that concentrated ownership is associated with Djankov (1999b) corporate performance for 706 privatized Czech firms during the higher profitability and labor productivity. Also find that period 1992­97. Uses profitability and labor productivity as foreign strategic owners and non-bank­sponsored indicators of corporate performance. investment funds improve performance more than bank- sponsored funds. Frydman, Gray, Compares the performance of privatized and state-owned firms in Privatization "works," but only when the firm is controlled Hessel, and the transition economies of Central Europe, and asks the question by outside owners (other than managers or employees). Rapaczynski (1999) "when does privatization work?" Examines the influence of Privatization adds over 18 percentage points to the annual ownership structure on performance using a sample of growth rate of a firm sold to a domestic financial company, 90 state-owned and 128 privatized companies in the Czech Republic, and 12 percentage points when sold to a foreign buyer. Hungary, and Poland. Employs panel data regression methods to Privatization to an outside owner also adds about isolate ownership effects. 9 percentage points to productivity growth. Further, the gain does not come at the expense of higher unemployment; insider-controlled firms are much less likely to restructure, but outsider-controlled firms grow faster. Shows the importance of entrepreneurship in reviving sales growth. Frydman, Gray, Examines whether the imposition of hard budget constraints is alone Finds privatization alone adds nearly 10 percentage points Hessel, and sufficient to improve corporate performance in the Czech Republic, to the revenue growth of a firm sold to outside owners. Rapaczynski Hungary, and Poland. Employs a sample of 216 firms, split between Most important, finds that the threat of hard budget (2000) state-owned (31 percent), privatized (43 percent), and private constraints for poorly performing SOEs falters, since (26 percent) firms. governments are unwilling to allow these firms to fail. The brunt of the lower creditworthiness of SOEs falls on state creditors. Frydman, Examines whether privatized Central European firms controlled by Documents that all state and privatized firms engage in Hessel, and outside investors are more entrepreneurial--in terms of their similar types of restructuring, but that product restructuring Rapaczynski ability to increase revenues--than firms controlled by insiders by firms owned by outside investors is significantly more (2000) or the state. Study employs survey data from a sample of 506 effective, in terms of revenue generation, than by firms with manufacturing firms in the Czech Republic, Hungary, and Poland. other types of ownership. Concludes that the more entrepreneurial behavior of outsider-owned firms is due to incentive effects, rather than human capital effects, of privatization--specifically greater readiness to take risks. Lizal, Singer, Examines the performance effects of the wave of break-ups of Finds an immediate positive effect on the efficiency and and Svejnar Czech and Slovak SOEs on the subsequent performance of the profitability of small and medium size firms (both master (2000) master firm and the spin offs. The regressions use data for and spin-offs) and a negative effect for the larger 373 firms in 1991 and 262 firms in 1992. firms in 1991. The results for 1992 are similar but not statistically significant. (Continues on next page) 271 272 TABLE 4. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Angelucci, Analyzes the effect of ownership and competition on firm Finds that competitive pressure (measured by market Estrin, Konings, performance, measured by total factor productivity (TFP), in three structure) is associated with higher productivity in all three and Zolkiewski transition economies for the years 1994 to 1998. Uses reported countries; increased import penetration is positively (2001) company accounts data for 1994 and 1998 for 17,570 Polish associated with performance in Poland, but negatively in companies, and for 1997­98 for 1,500 Bulgarian and Bulgaria and Romania; competitive pressure has stronger 2,047 Romanian companies. Tests whether private foreign-owned effects in private firms and privatization is associated with firms outperform private domestic companies, and whether these higher performance in more competitive sectors; both outperform SOEs. privatization is associated with better firm performance and privatized firms outperform SOEs in all three countries. Overall, finds that there are complementarities between competitive pressure and ownership. Carlin, Fries, Uses data from a 1999 survey of 3,305 firms in 25 transition Finds that competition has an important and nonmonotonic Schaffer, and countries to examine the factors that promote restructuring by effect on the growth of sales and labor productivity, with Seabright (2001) firms and enhance subsequent performance--as measured by performance improving more for firms facing one to three growth in sales and in sales per employee over a three-year competitors than for firms facing many competitors or for period. Survey includes about 125 companies from each of 25 monopolists (one-fourth of SOEs face no competition for countries, with larger samples from Poland and Ukraine their main products in their domestic markets). (200+ firms) and Russia (500+ firms). Just over half were newly Controlling for other factors, finds no significant established firms, 8 percent were privatized to insiders, 22 percent relationship between privatization and performance. were privatized to outsiders, and 16 percent remained state owned. Newly created firms generally outperform all other categories. Old firms (privatized and SOEs) are much more likely to cut employment than new entrants, but the paper finds some evidence that private firms (new entrants and privatized) are more likely to engage in new product development. Overall, finds competition to be a more powerful influence on performance than ownership, per se. Coricelli and Identifies the presence of soft budget constraints and analyzes their Finds that hard budget constraints do promote passive Djankov (2001) impact on enterprise restructuring in Romania during the initial restructuring, in the form of labor shedding, but not new transition period. Employs a simple analytical model and a sample investment. Active restructuring requires access to external of 4,429 enterprises with data from 1992 to 1995 to test whether financing. Tightened bank credit can induce hard budget hardening budget constraints promote beneficial restructuring and constraints and raise enterprise efficiency in the short run, new investment or whether access to external financing is required but at the cost of curtailing investment. to promote new investment normality. Earle and Examines the impact of privatization--and the method of Shows consistently positive, highly significant effects of Telegdy (2001) privatization--on firm performance in Romania over the period private ownership on labor productivity growth, with the 1992­99. Employs a dataset of 2,354 firms owned by the point estimate implying an incremental 1.0 to 1.7 State Ownership Fund (SOF) in 1992, and traces the evolution percentage point growth in productivity for a 10 percent of ownership over the next six years. Most of these (77%) still had rise in private shareholding. Insider transfers some state ownership (50.9% median) in 1998. and mass privatizations have smaller, but still significantly positive effects. Fidrmuc and Uses a sample of 178 Czech firms privatized during first wave Finds that efficiency and profitability declined after Fidrmuc (2001) of voucher privatization (1992­94) to test whether ownership privatization, and that changes in firms' operations do not change promoted increased efficiency and profitability. Uses vary significantly by size or ownership--but do vary by MNR pre- versus postprivatization comparison techniques to industry type, with nonmanufacturing firms experiencing test for performance changes. more positive (or less negative) changes. Harper (2001) Examines the effects of privatization on the financial and operating Finds that the first wave of privatization yielded performance of 174 firms privatized in the first--and 380 firms disappointing results. Real sales, profitability, efficiency, divested in the second--wave of the Czech Republic's voucher and employment all declined dramatically (and significantly). privatizations of 1992 and 1994. Compares results for privatized However, second-wave firms experienced firms to those that remained state owned. Employs Megginson, significant increases in efficiency and profitability and the Nash, and van Randenborgh (MNR) (1994) methodology and decline in employment--though still significant-- variables to measure changes. was much less drastic than after the first wave (­17 percent vs. ­41percent). Lizal and Examines strategic restructuring and new investment performance Finds that foreign-owned companies invest the most and Svejnar (2001) of 4,000 Czech companies during 1992­98. Dataset includes over (domestically owned) cooperatives the least; private firms 83,000 quarterly observations. Develops and tests a dynamic do not invest more than state-owned firms; cooperatives model of restructuring and investment, allowing the authors to and small firms are credit rationed; and SOEs operate examine the separable impact of private versus public and under a soft budget constraint. domestic versus foreign ownership on restructuring, as well as the importance of access to credit and a soft budget constraint 273 on firm investment. (Continues on next page) 274 TABLE 4. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Zinnes, Eilat, Employs a unique panel dataset of macroeconomic, ownership Regardless of the performance measure employed, finds that and Sachs structure, and indicator variables measuring the depth and economic performance gains come only from deep (2001) breadth of reform and privatization for 24 transition countries privatization--meaning that change of title reforms only to determine whether "change of ownership" (privatization) yield economic gains after key institutional and agency- alone is enough to promote improved economic performance related reforms have exceeded certain threshold levels. over the 1990­98 period, or whether "deep privatization" By themselves, change of title reforms never have a involving improved corporate governance, enhanced prudential significant impact on performance, but the higher the regulation, and hardening of budget constraints is also required. OBCA level a country has, the more positive the impact Develops an OBCA indicator variable for each country of an increase in change of title on economic measuring the breadth and depth of reforms, and includes performance. While ownership matters, institutions this variable in regressions. Uses four measures of economy-wide matter just as much. macroeconomic performance as dependent variables. Cull, Matesova, Examines the incentive of managers of voucher-privatized Czech Controlling for size, industry, capital intensity, and initial and Shirley companies to "tunnel" (strip assets out of companies at the leverage, finds that voucher-privatized JSCs perform (2002) expense of outside shareholders) and "loot" their companies. significantly worse than firms with concentrated ownership Looting occurs when firms face a soft budget constraint and that are purchased for cash. Investment fund­controlled managers are able to borrow heavily, extract funds from the firm, JSCs underperform all other firms, including other JSCs. and then default on the debt without penalty. Employs a dataset Fund-controlled JSCs also take on liabilities at a much with 1,017 observations from 392 companies spread nearly evenly faster rate than other firms, indicating that they are between 1994 and 1996. Half the firms are voucher-privatized operating under a soft budget constraint. Though not joint stock companies (JSCs), while half are limited liability able to measure directly, evidence indirectly shows that companies (LLCs). looting is a widespread occurrence for many JSCs. Claessens and Examines changes in the performance of 6,354 privatized and Finds that privatization is associated with significantly Djankov (2002) state-owned firms in seven transition economies over the 1991­95 increased sales and productivity growth and, to a lesser period, and tests whether privatization improves performance extent, with fewer job losses. In six of seven countries, (as measured by increased sales and labor productivity). Sample privatized firms show higher sales growth or smaller includes all manufacturing firms that are registered as state owned declines in sales than SOEs, and privatized firms reduce in 1991 and have more than 25 employees and have full balance their sales forces by an average of 6.11 percent, compared sheet and income statements for 1992­5. Constructs panel data to 7.42 percent for SOEs (a significant difference). The showing evolution of ownership over the period. positive effect of privatization is stronger in economic magnitude and statistical significance as the time elapsed since privatization increases. Grosfeld and Examines whether competition and corporate governance are Finds that product market competition has a positive and Tressel (2002) substitutes or complements with respect to promoting performance significant impact on performance. The effect of ownership improvements in Poland's transition. Uses the available data for all concentration, which is quite high in Poland, turns out to be 200 nonfinancial firms listed on the Warsaw Stock Exchange from U-shaped. Firms with dispersed ownership and those 1991 to 1998. First studies the separate effects of competition and where one shareholder owns more than 50 percent of ownership concentration on productivity growth at the firm level, voting shares have higher productivity growth than those and then examines their interaction. with intermediate levels of ownership concentration. Competitive pressure does not affect newly created firms, but does significantly improve performance of privatized companies. Presence of a large foreign owner increases productivity growth significantly. Concludes that good corporate governance and competitive pressures are complements. Lizal and Uses a panel of over 83,500 quarterly observations from 4,000 Finds that foreign owners unambiguously improve long-term Svejnar (2002) medium and large Czech companies over the 1992­98 period performance (measured several ways, including profits to assess the effects of mass privatization on firm performance. and investment) of privatized companies, but domestic owners do not. (Continues on next page) 275 276 TABLE 4. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Kocenda and Analyzes the effect of ownership on postprivatization performance Find that concentrated foreign ownership improves Svejnar (2002) using a dataset of 2,529 to 2,949 observations on an unbalanced economic performance, but domestic private ownership panel of 1,371 to 1,540 medium and large Czech firms. Defines does not, relative to state firms. Foreign-owned firms six categories of owners and examines the impact of each. engage in strategic restructuring by increasing sales and profits, while domestic firms reduce sales and labor costs without increasing profits. Ownership concentration is generally associated with improved performance. Overall, concludes that state ownership plays a much more economically and socially beneficial role in this transition economy than theory would predict. Glennerster Using a panel dataset on 470 formerly state-owned firms in the Finds weak but significant evidence that privatization can (2003) former Yugoslav Republic of Macedonia (FSRM) for 1996­99, yield benefits even with predominantly insider sales and in examines whether privatization increases profitability of divested an environment of weak corporate governance. On companies. Uses a fixed effects panel data regression to address average, privatization leads to a 30 percent increase in selection bias in both the timing and method of privatization. revenues and costs, a 16 percent increase in the number of workers employed, and a $1,200 increase in profits per worker. Firms sold to outsiders and those with more concentrated ownership expand more than other, similar firms after privatization. Employee buyouts perform relatively poorly. Also finds that lack of access to capital is an important reason why insider privatizations perform poorly, since those firms where new owners bring in new capital see particularly high growth rates after privatization. Source: Megginson 2005, chapters 3 and 4; Chong and López-de-Silanes 2005. TABLE 5. Summary of Recent Empirical Studies Examining Whether the Imposition of Hard Budget Constraints and Improved Incentives for Managers Improve Corporate Performance Study Sample description, study period, and methodology Summary of empirical findings and conclusions Groves, Hong, Using a sample of data for 769 Chinese state-owned enterprises Finds that new positive and negative incentives were McMillan, and over the years 1980­89, examines the impact of developing a effective in promoting improved performance, and that Naughton competitive managerial labor market on firm performance and management contracts were widely adopted as part of (1994) management productivity. reform process. Poorly performing managers were more likely to be replaced, and managerial pay was linked to firm sales and profits. Output per worker rose 67 percent in real terms between 1980 and 1989 for sample firms. Competition improved performance without ownership changes. Bertero and Employing a sample of 150 Italian manufacturing SOEs, with Find that the SOE firms' response to increased debt Rondi 1,278 firm-year observations, examines whether imposition of during the hard budget constraint period, 1988­93, was (2000) a hard budget constraint can improve SOE performance. consistent with financial pressure, but was not during the Exploits the fact that the fiscal environment became much soft budget constraint period of 1977­87. Only during tighter for Italian state enterprises in the late 1980s. the later period did firms respond to financial pressure by increasing total factor productivity and reducing employment. Imposition of a hard budget constraint improves performance without ownership change. Frydman, Examines whether the imposition of hard budget constraints Finds that privatization alone added nearly 10 percentage Gray, Hessel, is alone sufficient to improve corporate performance in the Czech points to the revenue growth of a firm sold to outside and Rapaczynski Republic, Hungary, and Poland. Employs a sample of 216 firms, owners. Most important, finds that the threat of hard (2000) split between state-owned (31 percent), privatized (43 percent), budget constraints for poorly performing SOEs falters, and private (26 percent) firms. since governments are unwilling to allow these firms to fail. The brunt of the lower creditworthiness of SOEs falls on state creditors. Privatization is required to improve performance; the threat of a hard budget constraint is not credible. Source: Authors' compilations. 277 278 | SERGEI GURIEV AND WILLIAM MEGGINSON effect on the net price, while "investment" and "efficiency" programs actually reduce the price. Chong and López-de-Silanes (2002) also show that even labor force retrench- ment programs are counterproductive, as they all too often lead to adverse selection in the employees being let go. Methods of Sales Megginson and others (2004) study the determinants of the choice between asset sales and share issue privatization. The study shows that the choice depends on both market institutions and firm-specific factors. Larger and more profitable firms are more likely to be sold via public capital markets. Better protection of property rights leads to a higher chance of privatization via asset sale. The existing research on share issue privatization (summarized in Megginson 2005, chapter 6) shows that these issues are substantially underpriced. Investors who buy the privatization share issues earn statistically and economically significant excess returns (about 30 percent!) both in the short and long term. This is especially striking given that the corporate finance literature documents negative long-term excess returns for private sector share offerings. The underpricing probably reflects the fact that privatizing governments pursue multiple goals rather than just revenue maximization (see Boycko, Shleifer, and Vishny 1995; Biais and Perotti 2002). Welfare Effects Studies of the effects of privatization on social welfare and inequality have tra- ditionally focused on divestments of utilities. These studies measure the effect of privatization on the access to services and generally find substantial benefits, especially for lower-income groups.3 Galiani, Gertler, and Schargrodsky (2001) show that water privatization in Argentina has resulted not only in substantial productivity growth but also in reduction in child mortality (saving the lives of about 500 infants and young children each year). Similar results are obtained in the studies of telecom privatizations (see tables 4.12 and 4.13 in Megginson 2005 for a summary of the results). Stock Market Development Another important impact of privatization is the development of financial markets. Privatizations have contributed not only to the rise of the global capital markets but, more importantly, have increased capitalization and liquidity of almost all national stock markets outside the United States. Boutchkova and Megginson (2000) calculate the turnover ratios (total value of trading over the market capitalization) for individ- ual financial markets and regress these on the number of privatization deals in a given country in a particular year. Controlling for country fixed effects and first-order auto- correlation, they find that each privatization raises the stock market liquidity (proxied by the turnover ratio) by 2.3 percent in the next year and by further 1.7 percent the year after that. The relationship between privatization and stock market development seems to be well understood by the governments: Megginson and others (2004) show that governments are more likely to privatize through share issues in countries with less developed capital markets, apparently in order to foster stock market development. PRIVATIZATION: WHAT HAVE WE LEARNED? | 279 When Does Privatization NOT Work? The most thorough panel study of mass privatization (Brown, Earle, and Telegdy 2006) shows that privatization substantially improves productivity in Romania and Hungary (by about 20 to 30 percent) but has no positive effect in Ukraine and even a negative effect in Russia. If the general lesson from privatization research is that privatization usually "works," how should one explain the failure of privatization in Russia and other CIS countries? The evidence suggests that privatization succeeds, but only if the relevant institutional environment is in place--including private prop- erty rights protection, rule of law, hard budget constraints, competition, and regulation (see table 6, as well as tables 1­5). In this respect, Russia and other CIS countries did not have the benefit of prospective EU accession to force the pace of necessary reforms. Also, EU accession made privatization irreversible in Central and Eastern Europe, while in the CIS policy reversal was indeed an important risk (which did in fact materialize in Belarus and Russia, and almost materialized in Ukraine); hence a fast mass privatization was needed. The other major problem in the CIS arose from the decision to rule out foreign participation in privatization for ideological reasons. Given all the constraints, Russian and other CIS privatizers had to adopt noncash privatizations. Research shows that noncash privatization is inferior to trade sales and share issue privatization. The intuition is straightforward. First, noncash privatization results in insider ownership, which implies that demand for institutional reforms develops very slowly. Since market institutions are not in place, secondary market trading results in ownership concentration in the hands of a few politically connected owners.4 The larger the insider's ownership stake, the more the insider is protected from expropri- ation and regulation, and the more market power the company has. Therefore it is not surprising that in Russia, Ukraine, and other CIS countries, the postprivatization redistribution results in economic domination by a few large business groups (Guriev and Rachinsky 2005). The role of these so called "oligarchs" is not clear. On one hand, they improve performance of their own firms and provide the only counter- weight to a predatory government. They also represent the only significant con- stituency for whatever pro-market institutional change might take place in Russia (Boone and Rodionov 2002; Guriev and Rachinsky 2005). On the other hand, their dominance subverts institutions in their own favor at the expense of competition policy and entry of new firms.5 The other implication of noncash mass privatization is the resulting fragility and ambiguity of private property rights. As the owners have paid relatively little for the assets, voters believe that privatization is not fair, and the politicians can always find support for expropriation. This risk undermines incentives to invest. Moreover, the negative attitude to privatization may eventually result in a nationalization backlash (as observed in Russia since early 2004). Given all the problems with privatization in Russia, was there a better alterna- tive? Boycko, Shleifer, and Vishny (1995) and Nellis (1999) argue that the voucher privatization was the lesser evil. Unlike Central and Eastern Europe, Russia lacked an outside anchor of EU accession, had a divided government, and therefore could 280 TABLE 6. Summary of Empirical Studies of Privatization in Russia and Former Soviet Republics Study Sample description, study period, and methodology Summary of empirical findings and conclusions Earle (1998) Investigates the impact of ownership structure on the (labor) productivity OLS regressions show a positive impact of private (relative to of Russian industrial firms. Using 1994 survey data, examines differential state) share ownership on labor productivity, with this result impact of insider, outsider, and state ownership on the performance primarily due to managerial ownership. After adjusting for of 430 firms--of which 86 remain 100 percent state owned, 299 are selection bias, however, finds that only outsider ownership is partially privatized, and 45 are newly created. Adjusts empirical significantly associated with productivity improvements. methods to account for tendency of insiders to claim dominant Stresses that leaving insiders in control of firms--while ownership in the best firms being divested. politically expedient--has very negative long-term implications for the restructuring of Russian industry. Earle and Using a sample very similar to that used by Earle (1998) above, examines Finds that a 10 percentage point increase in private share Estrin (1998) whether privatization, competition, and the hardening of budget ownership raises real sales per employee by 3­5 percent. constraints play efficiency-enhancing roles in Russia. Subsidies (soft budget constraints) reduce the pace of restructuring in state-owned firms, but the effect is small and often insignificant. Djankov (1999a) Investigates the relation between ownership structure and enterprise Shows that foreign ownership is positively associated restructuring for 960 firms privatized in six newly independent states with enterprise restructuring at high ownership levels between 1995 and 1997. Employs survey data collected by the World (>30 percent), while managerial ownership is positively Bank in late 1997 from Georgia, Kazakhstan, the Kyrgyz Republic, related to restructuring at low (<10 percent) or high levels, Moldova, Russia, and Ukraine. but negative at intermediate levels. Employee ownership is beneficial to labor productivity at low ownership levels, but is otherwise insignificant. Djankov (1999b) Using the same survey data as in Djankov (1999a) above, studies effects Privatization through management buy-outs is positively of different privatization modalities on restructuring process in Georgia associated with enterprise restructuring, while voucher (92 firms) and Moldova (149 firms). Georgia employs voucher privatized firms do not restructure more rapidly than firms privatization, while the majority of Moldovan firms are acquired by that remain state owned. Implies that managers who gain investment funds--and numerous others are sold to managers for cash. ownership for free may have less incentive to restructure, as their income is not based solely on the success of the enterprise. Estrin and Uses a random sample of 150 Ukrainian firms with data from 1996 to Finds that privatization, per se, is not significantly associated Rosevear (1999) test the relationship between enterprise performance and ownership. with improved performance, and finds no benefit to outside Explores whether privatization yields improved company performance (versus insider) ownership. Does find clear positive effects and whether specific ownership forms lead to differentiated associated with insider ownership. Outside owners are never performance at the enterprise level. able to deliver performance superior to SOEs, and insider ownership does not yield a better profit performance than in nonprivatized companies. Anderson, Lee, Examines effects of competition and ownership on the efficiency of Finds that competition has qualitatively large effects; and Murrell (2000) newly privatized firms using a sample 211 Mongolian companies with perfectly competitive firms have nearly double the efficiency (survey-derived) ownership data in 1995. Mongolia's privatization of monopolies. Enterprises with residual state ownership program is being implemented in a country lacking the basic institutions appear to be more efficient than other enterprises, reflecting of capitalism. an environment where the government is pressured to focus on efficiency and institutions gave little voice to outside owners. Djankov and Uses data for over 6,600 Kazakh enterprises during 1996­99 to examine Finds that newly created (de novo) private enterprises, Nenova (2000) "why did privatization fail in Kazakhstan?" Tries to explain rapid established after 1992, perform markedly better than declines in output for all sectors except oil and gas. privatized firms or those that remain SOEs. Privatized firms perform as badly as, or worse than, SOEs. Privatization fails to improve performance because divested firms are used as short-term vehicles for extracting private benefits. Grigorian (2000) Examines the relationship between ownership and operating performance Concludes that privatization has brought significant using a dataset of 5,300 small, medium, and large Lithuanian companies performance improvements overall. Also finds a negative with data over the 1995­97 period. Performance defined as increased bias in selecting firms for privatization; once this is accounted revenues and improved export performance. Also uses regression for, performance improvement is even more dramatic analysis to study a subsample of 618 companies which were fully (there is a nine-fold increase in the coefficient on private state owned in 1995; roughly half of these were partially privatized ownership). Expected subsidies contribute negatively to over the next two years. performance, but the study finds no significant impact regarding market competition. (Continues on next page) 281 282 TABLE 6. continued Study Sample description, study period, and methodology Summary of empirical findings and conclusions Andreyeva (2001) Examines empirically the responsiveness of firm performance to Finds that firm efficiency improves significantly with ownership and market structures, sector and regional specificity, and privatization. Also documents a significant influence of varying degrees of soft budget constraints. Uses a panel of 524 medium industry affiliation and regional location in shaping firm and large firms with performance data for 1996­98. performance; more concentrated markets perform better. Concludes that a policy of attracting strategic investors capable of pushing restructuring and bringing new investment to privatized firms should become a priority for policy makers. Pivovarsky (2001) Using data on 376 medium and large Ukrainian firms, investigates the Finds that ownership concentration is positively associated relationship between ownership concentration and enterprise with enterprise performance, and that concentrated performance. ownership by foreign companies and banks is associated with better performance than concentrated domestic ownership. Concludes that the privatization method has lasting impact on ownership structure; privatization methods that grant significant ownership stakes to single parties have greater efficiency gains than methods that create dispersed ownership. Jones and Uses fixed effects production function models estimated on a random Finds that, relative to state ownership, private ownership is Mygind (2002) sample of 660 Estonian firms with data from 1993 to 1997. Privatization 13­22 percent more efficient; all types of private ownership in Estonia created a widely varied ownership structures, and the study are more productive, though concentrated managerial attempts to estimate the relationship between ownership and ownership has the biggest effect (21­32 percent productive efficiency. improvement) and ownership by domestic outsiders the smallest (0­15 percent improvement), with ownership by foreigners (21­32 percent) and employees (24­25 percent) yielding intermediate levels of improvement. Source: Authors' compilations. PRIVATIZATION: WHAT HAVE WE LEARNED? | 283 not credibly commit to reforms. The privatizers therefore rushed to use the narrow window of opportunity. In order to make the reform irreversible, they had to sell tens of thousands of enterprises within a very few years; a case-by-case approach could have taken decades. It was not an issue of maximizing revenue from the auc- tions, since it would have been hard to run all the auctions in a fair and transparent manner.6 Additionally, while transition in Russia is taking longer than that of Central and Eastern Europe, it is still happening, while in Belarus and a few other CIS countries the reforms have been delayed indefinitely. It is also important to put the Russian experience in perspective. Grosfeld and Hashi (2003) show that even though the Czech Republic and Poland have pursued very different privatization policies, the ultimate ownership structures are actually quite similar and are driven by the same factors. This is also consistent with the studies in table 6: privatization works in Russia as well whenever it results in concentrated ownership (in particular, foreign ownership). By design, the Russian voucher privatization program did not generate such ownership structure right away, but the postprivatization reallocation should eventually produce an efficient ownership structure. Because of the underde- veloped financial markets and legal system it has taken much longer than expected, but the recent comprehensive study of ownership structure in Russian industry (Guriev and Rachinsky 2005) shows that this pattern is finally emerging. Yet even though there is concentrated ownership within firms that strengthens incentives, the illegitimacy of property rights undermines incentives to invest. This negative implication of the haste of privatization is likely to last for years. The "Great Outlier": China Another case study often used by the opponents of privatization is the transition expe- rience of China. Allegedly, China is growing very fast without mass privatization--or even because of the decision not to privatize. However, the existing evidence suggests that privatization works in China as well (see table 7). Foreign ownership and foreign listing (in particular, listing in Hong Kong) also positively affect performance. It is also not true that China has not privatized, although initially the government decided to try to improve SOE performance without privatization. As these hopes faded, China began privatizing smaller SOEs or leasing them to managers in an exchange of a fixed share of the resulting profit (Kikeri and Kolo 2005). This arrange- ment can be likened to partial privatization. Much privatization has also occurred via foreign direct investment into China. As a result, employment at Chinese state-owned industrial enterprises fell by half during the 1990s. China has also undertaken case-by-case privatizations of minority blocks for a few hundred large SOEs. According to the World Bank Privatization Database, there were about 200 large privatization deals between 1991 and 2003 that yielded revenues of more than $US18 billion--about as much as the entire Russian privatization. How- ever, partial privatization did not result in substantial efficiency improvements; one reason was the prevalence of soft budget constraints resulting from state banks that made nonperforming loans to SOEs. While further privatization is certainly needed, the Chinese government is delaying full privatization of all SOEs out of fear of high unemployment and the attendant negative political implications. 284 TABLE 7. Summary of Empirical Studies of Privatization in China Study Sample description, study period, and methodology Summary of empirical findings and conclusions Jia, Sun, Examines whether privatization through listing of Chinese companies Finds that real net profits are unchanged after privatization, and Tong (2002) in Hong Kong causes performance to improve. Uses a sample of 41 and that return on sales declines significantly. Output Chinese H-share SIPs from 1993­98. Uses MNR and pooled regression increases and leverage decline significantly. Regressions panel data methodology. show that state ownership is negatively related to performance. H-share ownership has significant, positive effect on performance. Sun and Tong (2003) Evaluates the performance of 634 Chinese SOEs listed on stock Using MNR methods, finds significant improvements in exchanges during the period 1994­98. Uses both Megginson, Nash, return on sales and the level of real earnings, real sales, and and van Randenborgh (MNR) preprivatization vs. post privatization employee productivity after partial privatization. Also finds comparisons and panel data regression methods to examine whether that more recently privatized companies are of higher partial divestment improves firm's earnings, output, and efficiency quality--and perform better after divestment--than do (real output per employee). Also examines differential effect of state those divested earlier. Panel data regressions verify basic and "legal person" shareholdings. findings that privatization improves performance, and find that different ownership structures have opposite effects on a firm's performance. State shareholdings hinder performance, while "legal person" shareholdings promote improvements. Tian (2003) Examines the ownership and control structure of 826 partially privatized Finds that government shareholding remain very large in companies listed on Chinese stock exchanges from 1994­98 and tests partially privatized companies, and that the relationship the relationship between ownership structure and firm value--as between state holdings and firm value is U-shaped. Going measured by Tobin's Q. from state ownership levels of 0 to 30 percent, increasing ownership, causes firm value to decline, but after that Tobin's-q increases with increasing state ownership. Wei, Varella, Uses MNR methods to test whether performance improved for 208 Documents significant improvements in real output, assets, D'Souza, and Chinese companies partially privatized through public share offering sales, sales efficiency, the level of real profits, and Hassan (2003) between 1990 and 1997. leverage. Firms in which more than 50 percent voting control is conveyed to private investors improve performance more than do those that remain state controlled. Bai, Li, and Using a comprehensive panel of 15,496 enterprises for 1998­2003, Finds that the layoffs are not large, and that those who Tao (2006) examines how privatization affects corporate performance and the remain employed benefit. Consumers benefit from lower stakeholders' payoffs. All enterprises in the sample were state owned prices, and tax payments do not change. Performance in 1998; 18 percent of them were privatized by 2003. improves via lower costs of finance and administration. Privatizing more than 50 percent of the firm has a stronger effect than privatizing a minority stake. Source: Authors' compilations. 285 286 | SERGEI GURIEV AND WILLIAM MEGGINSON These fears have not materialized yet. The recent study by Bai, Li, and Tao (2006) examines a comprehensive dataset of 15,000 Chinese firms from 1998 to 2003. All these firms were state owned in 1998, and 18 percent of them were privatized dur- ing the period. The panel nature of the dataset allows including firm-level fixed effects, which takes care of many methodological problems. It turns out that lay-offs are smaller than in other countries, and those workers who remain employed receive higher wages and nonwage benefits. Other stakeholders benefit from privatization as well. Consumers enjoy lower prices, yet tax payments do not change. The effects on firm performance are positive and mostly come from the lower costs of finance and administration. Also, consistent with the discussion above, privatization of a majority stake has greater positive effects than the privatization of a stake below 50 percent. Conclusions and Policy Implications There is now a growing body of research on all aspects of privatization that uses detailed datasets and up-to-date methodology. This research provides solid evidence that privatization "generally" works, both for the firms that are privatized and for privatizing economies as a whole. While privatization usually results both in increased productivity and reduced employment in privatized firms, fears of negative overall effects at the economy level are not justified. An important caveat here is that the benefits of privatizations depend on market institutions being in place. The countries that manage to ensure property rights pro- tection and the rule of law, impose hard budget constraints, increase competition, and improve corporate governance reap the largest benefits. If appropriate institutions are not in place, privatization often fails to improve performance at the firm level and for the economy as a whole. Emprical research provides a strong case for openness in privatization. Virtually all studies point to a positive role of foreign investors. Firms privatized to foreign owners exhibit the highest productivity increases. Moreover, as foreign owners usually buy the assets in a more competitive bidding process, they are likely to pay a high price for the privatized assets--and the threat of competition from foreign bidders also tends to raise the bids of domestic investors. Receiving a high net privatization price is important, not only for fiscal reasons but also for the political legitimacy of emerging private property rights and the sustainability of reforms. How does one reconcile the privatization research with the two cases that seem to contradict the positive impact of privatization: China and Russia? Our understanding of the extant evidence is that China and Russia are not outliers. China's growth has come from private sector development, even as many SOEs are still destroying value. Moreover, much privatization that did occur in the last decade produced substantial benefits both for the firms and for the social welfare. Russia's privatization did not produce the expected benefits, but this is related to the lack of good institutions. Yet the privatizations did result--albeit more slowly than expected--in a demand for institutional change. Russia's transition has been bumpy, but it is not clear whether there was a better alternative. Russia lacked an outside anchor to provide a credible PRIVATIZATION: WHAT HAVE WE LEARNED? | 287 commitment to reforms, unlike Central and Eastern European countries. It could not privatize to foreigners, as this was politically unacceptable. Finally, as a resource-rich country, it faced substantial macroeconomic volatility because of huge terms-of-trade shocks in the 1990s. There are still many state-owned enterprises around the world (Nellis 2006), espe- cially in Asia, the Middle East, and Africa, and--after the recent re-nationalization wave--in Russia. What advice can the extant research provide to policy makers who are contemplating privatization of these firms? This paper suggests that the strength of the evidence varies. We know much more about the firm-level effects of privatiza- tion, but the evidence on the welfare effects of privatization is scant, and our under- standing of the specifics of complementarities between privatization and institutional change is limited. We start with the lessons for which the evidence is substantial and rather non- controversial: · Privatization can deliver substantial benefits for the privatized firm. In some cases productivity doubles; in other cases it increases by single percentage points. Capi- tal spending typically increases dramatically, and the companies become finan- cially healthier--in particular, less heavily leveraged. · Privatization is usually accompanied by either no change or a reduction in employment. Privatizers should be prepared to handle the increased unemploy- ment, and experience suggests that most privatizing countries manage this prob- lem reasonably well. · Mass privatization is usually inferior to the case-by-case approach. Noncash privatization is generally worse than trade sales and share-issue privatization. The choice between share-issue privatizations and trade sales is driven by several factors--including firm size, the need to develop national stock markets, and the trade-off between better governance under concentrated ownership versus the difficulty of finding a single buyer for a large company. · Policy trade-offs are resolved most effectively when privatization is transparent and open to foreign investors. However, insiders and domestic investors always lobby against allowing foreign participation and often stir up nationalistic senti- ment. Precluding foreign ownership always results in lower privatization prices and lower efficiency after privatization. · Share-issue privatization brings an important side benefit of contributing to the development of the national stock market. There is also preliminary evidence on the following important issues: · Privatization usually produces welfare gains beyond the increased productivity at the firm level. Privatization generally results in lower prices for consumers--espe- cially if competition is encouraged--and does not contribute to higher inequality. · Restructuring enterprises before privatization is unlikely to work. · Privatization works well wherever there are good institutions. 288 | SERGEI GURIEV AND WILLIAM MEGGINSON The relative lack of evidence on these questions is caused by methodological and data problems. In all three cases, one needs a convincing counterfactual, which is extremely difficult to create. In the case of the impact of restructuring before privatization, this seems possible in principle; a government could randomly choose to restructure some firms before privatization, yet leave others unrestruc- tured. It is much harder to construct a counterfactual for the other two issues, since their impact is related to economy-wide effects. In order to estimate the welfare and inequality effects, one has to collect economy-wide microeconomic data. In order to measure the complementarities between privatization and institutions, one has to measure the institutions and again have a convincing counterfactual. These issues certainly call for further research. While the data challenges are serious, they are not insurmountable. For example, to estimate the importance of the complementarities between privatization and institutions, one could compare the effects of privatization across industries (as in Rajan and Zingales 1998). As some industries differ in terms of their "institutions-intensity," one can expect to see similar effects of privatization in sectors that are low in institution-intensity in all countries; in the sectors where good institutions are crucial, the effect of privatization would be higher in countries with better institutions--provided one can measure institutions well. Another approach could rely on the emerging liter- ature on political connections (Faccio 2006). Do political connections matter for the choice of firms to be privatized, for the price received, or for their subsequent performance? What are the channels for political connections to affect privatiza- tion? These and other research avenues are well within reach and will promote understanding of privatization. Notes 1. For extensive reference lists, see Megginson and Netter (2001); Djankov and Murrell (2002); and Megginson (2005). 2. See also Shirley and Xu (1998); Shirley (1999). 3. See McKenzie and Mookherjee (2003) for four countries in Latin America; Fischer, Gutierrez, and Serra (2005) for Chile; Jones, Jammal, and Gokgur (1998) for Côte d'Ivoire; Appiah-Kubi (2001) for Ghana. 4. In the dataset in Faccio (2006), Russia ranks first in terms of political connections. 5. See Glaeser, Scheinkman, and Shleifer (2003); Sonin (2003); Acemoglu (2005). 6. In Serbia, the emphasis was on fairness of the auctions rather than on high price. In Russia, the sheer magnitude of privatization made both efficiency and optimality of the privatiza- tion auctions impossible. The reformers relied instead on the subsequent reallocation of ownership via the stock market; this eventually happened but has taken more time than was initially expected. References Acemoglu, Daron. 2005. "The Form of Property Rights: Oligarchic vs. Democratic Societies." Department of Economics, Massachusetts Institute of Technology. PRIVATIZATION: WHAT HAVE WE LEARNED? | 289 Anderson, James H., Young Lee, and Peter Murrell. 2000. "Competition and Privatization Amidst Weak Institutions; Evidence from Mongolia." Economic Inquiry 38: 527­49. Andreyeva, Tatiana. 2001. "Company Performance in Ukraine: What Governs its Success?" Working paper, RAND Graduate School of Policy Analysis. Angelucci, Mauella, Saul Estrin, Jozef Konings, and Zbigniew Zoliewski. 2001. "The Effect of Ownership and Competitive Pressure on firm Performance in Transition Countries: Micro Evidence from Bulgaria, Romania, and Poland," Discussion Paper 2985, Center for Economic Policy Research (CEPR), London. Appiah-Kubi, Kojo. 2001. "State-Owned Enterprises and Privatization in Ghana." Journal of Modern African Studies 39 (2): 197­229. Bai, Chong-En, Jiangyong Li, and Zhigang Tao. 2006. "How Does Privatization Work in China?" Tsinghua University, Beijing. Barone, Enrico. 1908. "The Ministry of Production in the Collectivist State." Reprinted in Collectivist Economic Planning: Critical Studies on the Possibilities of Socialism, ed. Friedrich A Hayek, 245­90. London: Routledge. Berg, Andrew, Eduardo Borensztein, Ratna Sahay, and Jeromin Zettelmeyer. 1999. "The Evolution of Output in Transition Economies: Explaining the Differences." Working paper, International Monetary Fund (IMF), Washington, D.C. Bertero, Elisabetta, and Laura Rondi. 2000. "Financial Pressure and the Behavior of Public Enterprises Under Soft and Hard Budget Constraints: Evidence from Italian Panel Data." Journal of Public Economics 75: 73­98. Biais, Bruno, and Enrico Perotti. 2002. "Machiavellian Privatization." American Economic Review 92: 240­58. Birch, Melissa, and Jerry Haar, eds. 2000. The Impact of Privatization in the Americas. Miami: North-South Center Press. Boardman, Anthony, and Aidan R. Vining. 1989. "Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises." Journal of Law and Economics 32: 1­33. Boardman, Anthony E., Claude Laurin, and Aidan Vining. 2000. "Privatization in Canada: Operating, Financial, and Stock Price Performance With International Comparisons." Working paper, University of British Columbia, Vancouver. Boone, Peter, and Denis Rodionov. 2002. "Rent Seeking in Russia and the CIS." Brunswick UBS Warburg, Moscow. Boubakri, Narjess, and Jean-Claude Cosset. 1998. "The Financial and Operating Performance of Newly Privatized Firms: Evidence From Developing Countries." Journal of Finance 53: 1081­1110. ------. Forthcoming. "Does Privatization Meet the Expectations? Evidence from African Coun- tries," Journal of African Economics. Boubakri, Narjess, Jean-Claude Cosset, and Omrane Guedhami. 2005. "Postprivatization Corporate Governance: The Role of Ownership Structure and Investor Protection." Journal of Financial Economics 76: 369­99. Boutchkova, Maria K., and William L. Megginson. 2000. "Privatization and the Rise of Global Capital Markets." Financial Management 29: 31­76. Boycko, Maxim, Andrei Shleifer, and Robert Vishny. 1995. Privatizing Russia. Cambridge, Mass.: MIT Press. Brown, David, John Earle, and Almos Telegdy. 2006. "The Productivity Effects of Privatiza- tion: Longitudinal Estimates from Hungary, Romania, Russia, and Ukraine." Journal of Political Economy 114: 61­99. 290 | SERGEI GURIEV AND WILLIAM MEGGINSON Carlin, Wendy, Steven Fries, Marc Schaffer, and Paul Seabright. 2001. "Competition and Enterprise Performance in Transition Economies: Evidence from a Cross-Country Survey." Discussion Paper 2840, Center for Economic Policy Research (CEPR), London. Chisari, Omar, Antonio Estache, and Carlos Romero. 1999. "Winners and Losers from the Privatization and Regulation of Utilities: Lessons from a General Equilibrium Model of Argentina." World Bank Economic Review 13(2): 357­78. Chong, Alberto, and Florencio López-de-Silanes. 2002. "Privatization and Labor Force Restructuring Around the World." Policy Research Working Paper 2884, World Bank, Washington, DC. ------. 2005. "The Truth about Privatization in Latin America." In Privatization in Latin America: Myths and Reality, ed. Alberto Chong and Florencio Lopez-de-Silanes, 1­66. Stanford University Press and the World Bank. Chong, Alberto, and Jose Miguel Sánchez, editors. 2003. Medios privados para fines públicos: Arreglos contractuales y participación privada en infrastructura en América Latina. Washington, DC: Inter-American Development Bank Press. Claessens, Stijn, and Simeon Djankov. 1999a. "Enterprise Performance and Management Turnover in the Czech Republic." European Economic Review 43: 1115­24. ------. 1999b. "Ownership Concentration and Corporate Performance in the Czech Republic." Journal of Comparative Economics 27: 498­513. ------. 2002. "Privatization Benefits in Eastern Europe." Journal of Public Economics 83: 307­24. Claessens, Stijn, Simeon Djankov, and Gerhard Pohl 1997. Ownership and Corporate Governance: Evidence from the Czech Republic. Policy Research Paper 1737, World Bank, Washington, DC. Clarke, George, and Robert Cull. 2001. "Bank Privatization in Argentina: A Model of Political Constraints and Differential Outcomes." Policy Research Working Paper 2633, World Bank, Washington, DC. Coricelli, Fabrizio, and Simeon Djankov. 2001. "Hardened Budgets and Enterprise Restruc- turing: Theory and an Application to Romania," Journal of Comparative Economics 29: 749­63. Cull, Robert, Jana Matesova, and Mary Shirley. 2002. "Ownership and the Temptation to Loot: Evidence from Privatized Firms in the Czech Republic." Journal of Comparative Economics 30: 1­25. Demsetz, Harold, and Kenneth Lehn 1985. "The Structure of Corporate Ownership: Causes and Consequences." Journal of Political Economy 93 (6): 1155­77. Dewenter, Kathryn, and Paul H. Malatesta. 2000. "State-Owned and Privately Owned Firms: An Empirical Analysis of Profitability, Leverage, and Labour Intensity," American Economic Review 91: 320­34. Djankov, Simeon. 1999a. "Ownership Structure and Enterprise Restructuring in Six Newly Independent States." Comparative Economic Studies 41: 75­95. ------. 1999b. "The Restructuring of Insider-Dominated Firms: A Comparative Analysis." Economics of Transition 7: 467­79. Djankov, Simeon, and Peter Murrell. 2002. "Enterprise Restructuring in Transition: A Quan- titative Survey." Journal of Economic Literature 40: 739­92. Djankov, Simeon, and Tatiana Nenova. 2000. "Why Did Privatization Fail in Kazakhstan?" Working paper, World Bank, Washington, DC. D'Souza, Juliet, and William L. Megginson. 1999. "The Financial and Operating Performance of Newly Privatized Firms in the 1990s," Journal of Finance 54: 1397­1438. PRIVATIZATION: WHAT HAVE WE LEARNED? | 291 Dumontier, Pascal, and Claude Laurin. 2002. "The Financial Impacts of the French Government Nationalization-Privatization Strategy." Working paper, University of Geneva. Dyck, I., and J. Alexander. 1997. "Privatization in Eastern Germany: Management Selection and Economic Transition." American Economic Review 87: 565­97. Earle, John S. 1998. "Post-Privatization Ownership Structure and Productivity in Russian Industrial Enterprises." SITE working paper, Stockholm School of Economics. Earle, John, and Saul Estrin 1998. "Privatization, Competition, and Budget Constraints: Disciplining Enterprises in Russia." SITE Working Paper 128, Stockholm School of Economics. Earle, John, and Almos Telgedy. 2001. "Productivity and Ownership Structure in Romania: Does Privatization Matter?" Working paper, Central European University, Budapest. Estache, Antonio. 2002. "Argentina's 1990s Utilities Privatization: A Cure or a Disease?" Working paper, Free University of Brussels. Estrin, Saul, and A. Rosevear. 1999. "Enterprise Restructuring and Ownership: The Case of Ukraine." European Economic Review 43: 1125­36. Faccio, Mara. 2006. "Politically Connected Firms." American Economic Review 96: 369­86. Fidrmuc, Jana P., and Jan Fidrmuc. 2001. "Firm Performance, CEO Turnover, and Managerial Labor Market in the Czech Republic." Working paper, Tilburg University, The Netherlands. Fischer, Ronald, Rodrigo Gutierrez, and Pablo Serra. 2005. "The Effects of Privatization on Firms and on Social Welfare." In Privatization in Latin America: Myths and Reality, ed. Alberto Chong and Florencio Lopez-de-Silanes, 197­274. Stanford University Press and the World Bank. Florio, Massimo. 2001. "The Welfare Impact of a Privatisation: The British Telecom Case- History." Working paper, University of Milan. Frydman, Roman, Marek Hessel, and Andrzej Rapaczynski. 2000. "Why Ownership Matters? Entrepreneurship and the Restructuring of Enterprises in Central Europe." Working paper, C.V. Starr Center for Applied Economics, New York University. Frydman, Roman, Cheryl W. Gray, Marek Hessel, and Andrzej Rapaczynski. 1999. "When Does Privatization Work? The Impact of Private Ownership on Corporate Performance in Transition Economies." Quarterly Journal of Economics 114: 1153­91. ------. 2000. "The Limits of Discipline: Ownership and Hard Budget Constraints in the Transition Economies." Working paper, C.V. Starr Center for Applied Economics, New York University. Galal, Ahmed, Leroy Jones, Pankaj Tandon, and Ingo Vogelsang. 1994. Welfare Conse- quences of Selling Public Enterprises. Oxford: Oxford University Press. Galiani, Sebastián, Paul Gertler, and Ernesto Schargrodsky. 2001. "Water for Life: The Impact of the Privatization of Water Services on Child Mortality." Working paper, University of California, Berkeley. Glaeser, Edward, Jose Scheinkman, and Andrei Shleifer. 2003. "The Injustice of Inequality." Journal of Monetary Economics, Carnegie-Rochester Series on Public Policy, 50 (1): 199­222. Glennerster, Rachel. 2003. "Evaluating Privatization in the Former Yugoslav Republic of Macedonia." Working paper, Kennedy School of Government, Harvard University. Green, Richard, and Ingo Vogelsang. 1994. "British Airways: A Turn-Around Anticipating Privatization." In Privatization and Economic Performance, ed. Matthew Bishop, John Kay, and Colin Mayer, chapter 4. Oxford: Oxford University Press. Grigorian, David. 2000. "Ownership and Performance of Lithuanian Enterprises." Policy Research Working Paper 2343, World Bank, Washington, DC. 292 | SERGEI GURIEV AND WILLIAM MEGGINSON Grosfeld, Irena, and Iraj Hashi. 2003. "Mass Privatisation, Corporate Governance, and Endo- genous Ownership Structure." Working Paper 596, William Davidson Institute, University of Michigan, Ann Arbor. Grosfeld, Irena, and Thierry Tressel. 2002. "Competition and Ownership Structure: Substitutes or Complements? Evidence from the Warsaw Stock Exchange." Economics of Transition 10 (3): 525­52. Groves, Theodore, Yongmiao Hong, John McMillan, and Barry Naughton. 1994. "Autonomy and Incentives in Chinese State Enterprises." Quarterly Journal of Economics 109: 183­209. Guriev, Sergei, and Andrei Rachinsky. 2005. "The Role of Oligarchs in Russian Capitalism." Journal of Economic Perspectives 19 (1): 131­50. Hachette, Dominique, and Rolf Luders. 1993. Privatization in Chile: An Economic Appraisal. San Francisco: International Center for Economic Growth. Harper, Joel T. 2001. "Short-Term Effects of Privatization on Operating Performance in the Czech Republic." Journal of Financial Research 24: 119­31. Hart, Oliver, Andrei Shleifer, and Robert Vishny. 1997. "The Proper Scope of Government: Theory and an Application to Prisons." Quarterly Journal of Economics 112 (4): 1126­61. Holmstrom, Bengt, and Paul Milgrom. 1991. "Multi-task Principal-agent Analyses: Linear Contracts, Asset Ownership, and Job Design." Journal of Law, Economics, and Organi- zation 7: 24­52. Jia, Jin, Qian Sun, and Wilson H. S. Tong. 2002. "Privatization via Overseas Listing: Evi- dence from China's H-Share Firms," Working paper, Nanyang Technological University, Singapore. Jones, Derek, and Niels Mygind. 2002. "Ownership and Productive Efficiency: Evidence from Estonian Panel Data," Review of Development Economics 6: 284­301. Jones, Leroy P., Yahya Jammal, and Nilgun Gokgur. 1998. Impact of Privatization in Côte d'Ivoire: Draft Final Report. Boston Institute for Developing Economies, Boston University. Kikeri, Sunita, and Aishetu Fatima Kolo. 2005. "Privatization: Trends and Recent Develop- ments." Policy Research Working Paper 3765, World Bank, Washington, DC. Kocenda, Evzen, and Jan Svejnar. 2002. "Ownership and Firm Performance after Large-Scale Privatization," Working paper, University of Michigan, Ann Arbor. Lange, Oscar. 1936. "On the Economic Theory of Socialism: Part One/Part Two." Review of Economic Studies 1936­37, 4(1): 53­71 and 4(2): 123­42. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "Benefits of Privatization--Evidence From Mexico," Quarterly Journal of Economics 114 (4): 1193­1242. Laurin, Claude, and Yves Bozec. 2000. "Privatization and Productivity Improvement: The Case of Canadian National (CN)." Working paper, Ecoles de HEC, Montreal. Lipton, David, and Jeffrey Sachs. 1990. "Creating a Market Economy in Eastern Europe: The Case of Poland." Brookings Papers on Economic Activity 1: 75­133. Lizal, Lubomir, and Jan Svejnar. 2001. "Investment, Credit Rationing, and the Soft Budget Con- straint: Evidence from Czech Panel Data," Review of Economics and Statistics 84: 353­70. ------. 2002. "Privatization Revisited: The Effects of Foreign and Domestic Owners on Corporate Performance." Working paper, University of Michigan, Ann Arbor. Lizal, Lubomir, Miroslav Singer, and Jan Svejnar. 2000, "Enterprise Break-ups and Per- formance during the Transition from Plan to Market," Review of Economic Statistics 83: 92­99. López-de-Silanes, Florencio. 1997. "Determinants of Privatization Prices." Quarterly Journal of Economics 112: 965­1025. PRIVATIZATION: WHAT HAVE WE LEARNED? | 293 ------. 2005. "Dos and Don'ts in Privatization: Evidence from Developing Countries." Pri- vatization Barometer Newsletter 2: 29­35. Macquieira, Carlos, and Salvador Zurita. 1996. "Privatizaciones en Chile: Eficiencia y Politi- cas Financieras." Estudios de Administracion 3: 1­36. Martin, Stephen, and David Parker. 1995. "Privatization and Economic Performance through- out the UK Business Cycle." Managerial and Decision Economics 16: 225­37. Maskin, Eric. 2000. "Auctions, Development, and Privatization: Efficient Auctions with Liquidity-Constrained Buyers." The Marshall Lecture to the European Economic Asso- ciation, European Economic Review 44 (4­6): 667­81. McKenzie, David, and Dilip Mookherjee. 2003. "The Distributive Impact of Privatization in Latin America: Evidence from Four Countries." Working paper, Stanford University. Megginson, William L. 2005. The Financial Economics of Privatization, 1st edition. New York: Oxford University Press. Megginson, William L., and Jeffry M. Netter. 2001. "From State to Market: A Survey of Empirical Studies on Privatization." Journal of Economic Literature 39: 321­89. Megginson, William L., Robert C. Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An International Empirical Analy- sis." Journal of Finance 49: 403­52. Megginson, William L., Robert C. Nash, Jeffry Netter, and Annette Poulsen. 2004. "The Choice of Public versus Private Markets: Evidence from Privatizations." Journal of Finance 59: 2835­70. Nellis, John. 1999. "Time to Rethink Privatization in Transition Economies?" Discussion Paper 38, International Finance Corporation, Washington, DC. ------. 2006. "Privatization in Developing Countries: A Summary Assessment." Working Paper 87, Center for Global Development, Washington, DC. Newbery, David, and Michael G. Pollitt. 1997. "The Restructuring and Privatization of Britain's CEGB--Was it Worth it?" Journal of Industrial Economics 45: 269­303. Okten, Cagla, and K. Peren Arin. 2001. "How Does Privatization Affect the Firm's Efficiency and Technology Choice? Evidence from Turkey." Working paper, Louisiana State Univer- sity, Baton Rouge. Omran, Mohammed. 2001. "The Performance of State-Owned Enterprises and Newly Privat- ized Firms: Empirical Evidence from Egypt," Working paper, Arab Academy for Science and Technology, Alexandria, Egypt. ------. 2002. "Initial and Aftermarket Performance of Egyptian Share Issue Privatization." Working paper, Arab Monetary Fund--Economic Policy Institute, Abu Dhabi, UAE. Petrazzini, Ben A., and Theodore H. Clark. 1996. "Costs and Benefits of Telecommunica- tions Liberalization in Developing Countries," Working paper, Hong Kong University of Science and Technology, Hong Kong. Pinheiro, Armando Castelar. 1996. "Impactos Microeconomicos da Privatizacao no Brasil." Pesquisa e Planejamento Economico 26: 357­98. Pivovarsky, Alexander. 2001. "How Does Privatization Work? Ownership Concentration and Enterprise Performance in Ukraine" Working Paper 01/42, International Monetary Fund, Washington, DC. Pohl, Gerhard, Robert E. Anderson, Stijn Claessens, and Simeon Djankov. 1997. "Privatization and Restructuring in Central and Eastern Europe: Evidence and Policy Options." Techni- cal Paper 368, World Bank, Washington, DC. Pombo, Carlos, and Manuel Ramírez. 2005. "Privatization in Columbia: A Plant Performance Analysis." In Privatization in Latin America: Myths and Reality, ed. Alberto Chong and Florencio Lopez-de-Silanes. Stanford University Press and the World Bank. 294 | SERGEI GURIEV AND WILLIAM MEGGINSON Price, Catherine Waddams, and Thomas Weyman-Jones. 1996. "Malmquist Indices of Produc- tivity Change in the UK Gas Industry before and after Privatization." Applied Economics 28: 29­39. Rajan, Raghuram, and Luigi Zingales 1998 "Financial Dependence and Growth." American Economic Review 88: 559­86. Ramamurti, Ravi. 1996. "The New Frontier of Privatization." In Privatizing Monopolies: Lessons From the Telecommunications and Transport Sectors in Latin America, Ravi Ramamurti, ed., 1­45. Baltimore: Johns Hopkins University Press. ------. 1997. "Testing the Limits of Privatization: Argentine Railroads." World Development 25: 1973­93. Ros, Augustin J. 1999. "Does Ownership or Competition Matter? The Effects of Telecommuni- cations Reform on Network Expansion and Efficiency," Journal of Regulatory Economics 15: 65­92. Saal, D. S., and David Parker. 2003. "The Impact of Privatization and Regulation on the Water and Sewerage Industry in England and Wales," Working paper, Aston University, Birmingham, UK. Sanchez, Manuel, and Rossana Corona. 1993. "Privatization in Latin America." Center for Research in Applied Economics, Inter-American Development Bank, Washington, DC. Sheshinski, Eytan, and Luis Lopez-Calva. 1999. "Privatization and its Benefits: Theory and Evidence." Discussion Paper 698, Harvard Institute for International Development (HIID), Harvard University. Shirley, Mary. 1999. "Bureaucrats in Business: The Role of Privatization in State Owned Enterprise Reform." World Development 27 (1): 115­36. Shirley, Mary, and Lixin Colin Xu. 1998. "Information, Incentives, and Commitment: An Empirical Analysis of Contracts between Government and State Enterprises." Journal of Law Economics and Organization 14 (2): 358­78. Shleifer, Andrei. 1998. "State versus Private Ownership." Journal of Economic Perspectives 12: 133­50. Smith, Stephen C., Beon-Cheol Cin, and Milan Vodopivec. 1997. "Privatization Incidence, Ownership Forms, and Firm performance: Evidence From Slovenia." Journal of Compar- ative Economics 25: 158­79. Sonin, Konstantin. 2003. "Why the Rich May Favor Poor Protection of Property Rights," Journal of Comparative Economics 31 (4): 715­31. Sun, Qian, and Wilson H. S. Tong. 2003. "China's Share Issue Privatization: The Extent of its Success." Journal of Financial Economics 70: 183­222. Sun, Qian, Feng Fang, and Wilson H. S. Tong. 2004 "Do Government-Linked Companies Necessarily Underperform?" Journal of Banking and Finance 28: 2461­92. Sun, Qian, Jin Jia, and Wilson H. S. Tong. 2002. "Malaysian Privatization: A Comprehensive Study." Financial Management 31: 5­31. Tian, George Lihui. 2003. "Government Shareholding and the Value of China's Modern Firms." Working paper, London Business School. Trujillo, Lourdes, Noelia Martín, Antonio Estache, and Javier Campos. "Macroeconomic Effects of Private Sector Participation in Latin America's Infrastructure." Policy Research Working Paper 2906, World Bank, Washington, DC. Verbrugge, James A., William L. Megginson, and Wanda Lee Owens. 1999. "The Financial Performance of Privatized Banks: An Empirical Analysis," Working paper, University of Georgia, Athens, Georgia. PRIVATIZATION: WHAT HAVE WE LEARNED? | 295 Villalonga, Belen. 2000. "Privatization and Efficiency: Differentiating Ownership Effects from Political, Organizational, and Dynamic Effects," Journal of Economic Behavior and Organization 42: 43­74. Wallsten, Scott J. 2001. "An Econometric Analysis of Telecom Competition, Privatization, and Regulation in Africa and Latin America." Journal of Industrial Economics 49: 1­19. Wei, Zuobao, Oscar Varela, Juliet D'Souza, and M. Kabir Hassan. 2003. "The Financial and Operating Performance of China's Newly Privatized Firms." Financial Management 32: 107­26. Weiss, Andrew, and Georgiy Nikitin. 1998. "Effects of Ownership by Investment Funds on the Performance of Czech Firms." Working paper, Boston University. Zinnes, Clifford, Yair Eilat, and Jeffrey Sachs. 2001. "The Gains from Privatization in Tran- sition Economies: Is `Change of Ownership' Enough?" IMF Staff Papers 48: 146­70. Comment on "The Institutional Determinants of State Capabilities in Latin America," by Ernesto Stein and Mariano Tommasi; "Lowering the Cost of Capital in Emerging Market Economies," by Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya; and "Privatization: What Have We Learned?" by Sergei Guriev and William Megginson IRENA GROSFELD Impediments to Growth What is common to the three papers presented in this session? All of them explore what characteristics of institutions encourage development and growth. They discuss the impediments to growth that contribute to the twin phenomenon of the high cost of capital and low return to economic activity. The recurrent themes of the three papers are corruption, expropriation of investors, capacity of special interests to capture rules, and the importance of enforcement. The political preconditions for successful reforms are clearly recognized. Quality of Policies Ernesto Stein and Mariano Tommasi focus on the impact of political institutions on the policy-making processes. They argue that the quality of the policy-making processes may matter more than the content of the policies. To be sure, if the state is unable Irena Grosfeld is professor of economics, Paris School of Economics (PSE), Paris. Annual World Bank Conference on Development Economics ©2007 The International Bank for Reconstruction and Development / The World Bank 297 298 | IRENA GROSFELD (or unwilling) to enforce laws and regulations, it is clear that the content of reform does not matter so much. But if reforms become credible, their content matters a lot. The authors suggest several features necessary to improve the quality of decisions: stability, adaptability, coordination and coherence, enforcement, public regarded- ness, and efficiency. The question is what makes policies stable but adaptable when necessary, consistent, and focused on the public interest. Stein and Tommasi identify several characteristics of political institutions that appear as strong determinants of the quality of policy making: a well- established party system, a capable legislature, independent judiciaries, and a strong bureaucracy. We can expect indeed that if politi- cal institutions incite political actors to reach and enforce intertemporal agreements, the policy making process will be more stable and adaptable. Sequencing The problem that remains unanswered is how to get there from where we are. Stein and Tommasi recognize the problem and in future work want to study the way in which desirable institutional characteristics are built over time. This is the problem of sequencing, which was extensively discussed during the initial period of transition and which gave rise to passionate debates. One aspect of those debates, on which Andrei Shleifer and Joseph Stiglitz disagreed, is discussed in the paper by Sergei Guriev and William Megginson: market institutions and appropriate regulation are needed in order to obtain the benefits of privatization, but the support for building such institu- tions requires a critical mass of private ownership. So what should come first? Guriev and Megginson remind us that the experience of transition provides exam- ples supporting both options. Let me name two such examples. The Czech Republic is a case of rapid privatization and delayed institutional change; Poland is one of careful institution building and slower privatization. Despite such opposed sequenc- ing, both countries succeeded in the medium term. The fact that Russia failed in this respect, and that the demand for good institutions has been slow to emerge, should not be attributed, however, to voucher privatization as such; we should rather look at the nature of the Russian political system. The existing political institutions do not provide a favorable framework for development-friendly institutional change. They did not stop vested interests created by the privatization process from blocking the necessary regulations. They do not stimulate the mobilization of social energies and ideas in order to make the institutional change growth-enhancing. The features of the political system described by Stein and Tommasi as desirable are lacking in Russia. They were, however, present in Poland and in the Czech Republic, two countries that have adopted two very different forms of voucher privatization. Relationship between Formal and Informal Institutions More generally, an important issue in this sequencing debate concerns the relation- ship between formal institutions (laws and formal rules) and informal institutions COMMENT | 299 (norms of behavior and conventions). Many observers of the transition experience argue that the change in formal institutions should have been slower because informal institutions inherited from the past were in conflict with the new formal ones, or at least not in line with them. To be sure, the ideal sequencing would be the one in which progressive evolution of social norms and ideas contributed to the emergence of formal laws and rules. However, this is not a one-way relationship. The formation of norms of behavior is strongly influenced by the existing laws and rules. The introduc- tion of laws protecting contracts can accelerate the emergence of behavior respecting contracts. Norms are cultivated and maintained by imitation and by sanctions. So the change of legal constraints can, very slowly, bring about the change of strategies, values, and eventually, of social norms. For instance, the speed of adjustment of the mental and behavioral constraints and legacies of the period of central planning, such as the level of understanding of the working of a market economy, the respect of contracts, or attitudes toward foreign investors, will be higher if the environment in which the economic agents operate is structured by well-designed formal rules. But it is clear that putting in place new laws is not enough. In order to trigger a virtuous dynamic of institutional change, it is necessary to reset expectations. This requires credible commitment to and actual implementation of reforms. Bankruptcy The way bankruptcy procedures are introduced and managed plays a key role in shaping expectations and in ensuring the credibility of economic transformation. The most important part of the paper by Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya is devoted to bankruptcy. The authors recognize that although the use of formal bankruptcy procedures is low in developing countries and most debt enforcement occurs outside formal bankruptcy, it is important to properly design bankruptcy laws. The paper gives a very careful and complete discussion of the intricacies of bankruptcy law in emerging market economies. Let me stress one difference between developing countries and transition economies. The key issue in the countries in transition is restructuring. These countries are not in the process of industrialization; they rather have to channel resources from the industrial sector to services, cut obsolete product lines, get rid of unpro- ductive assets, withdraw capital from some lines of production, and invest it in new ventures. The extent of the necessary disbanding of irrationally bundled assets distinguishes enterprise restructuring in transition countries from structural adjust- ment in developing countries. Bankruptcy procedures may have both direct and indirect impact on enterprise restructuring. Indirectly, the mere threat of bankruptcy may change the expecta- tions of managers and encourage them to adopt restructuring measures to avoid insolvency. Even a limited number of firms going under may convince managers to take bankruptcy seriously. The actual implementation of bankruptcy proce- dures directly affects enterprises restructuring. Here, the design of bankruptcy law appears particularly important. 300 | IRENA GROSFELD I strongly agree with the clear conclusion given in the paper concerning the right balance between reorganization and liquidation: bankruptcy law in emerging markets economies should have a liquidation bias. This is an important conclusion that goes against the usual argument in favor of reorganization, which is supposed to give nec- essary breathing space to enterprises facing unusually strong shocks. If an economy needs to profoundly reorient its productive structure, liquidation may be less damag- ing than is often assumed. Empirical evidence shows that reorganization plans in transition economies have usually included only measures of financial restructuring and have very rarely relied on well-elaborated business plans. Typically, the quality of reorganization submitted by the incumbent managers has been rather poor; moreover, these plans have rarely been implemented. Here, we touch upon the second question discussed in the paper. Who controls the assets during the bankruptcy process? An important advantage of liquidation-- straight sale of a firm or piecemeal sale of assets--is that it reassigns control over assets. An efficient matching of managers appears crucial for effective restructuring. A firm in bankruptcy should not be systematically left with the same managerial team that has brought it to insolvency. The quality of restructuring might be enhanced if some elements of contestability were introduced through, for instance, open bidding for the control of the firm. Lessons of Empirical Research on Privatization In their summary of the literature, Guriev and Megginson stress two conclusions: there are important complementarities between privatization and other reforms; and foreign ownership is beneficial. The first point is very important. There is an increasingly wide consensus that in order to be effective, privatization requires appropriate institutional reforms. However, if we move from such a general claim, the unanimity disappears. What is really important? Protection of property rights? Competition? Financial system? Corporate governance? The legal system? These questions bring us back to the issue of sequencing. It is clear that many institutions are strongly correlated, but in any system of governance there are also institutions that are substitutes and not complements. For instance, in the Czech Republic, the enforcement of the bank- ruptcy law was postponed for a couple of years, but it was partially compensated by the commitment to privatize rapidly. Concerning the benefits of foreign ownership, it should be stressed that several empirical studies failed to identify a positive impact of foreign shareholders on firm performance. More generally, the empirical results on the impact of privatization on performance should be assessed with caution. We should not draw conclusions putting together studies using poor data and high-quality data: given the experience of 15 years, we should not take too seriously studies covering the first two or three years. More importantly, we should carefully discriminate between studies using poor methodology and careful empirical strategy. It is increasingly recognized that studying the relationship between ownership and performance is a tricky question and we have at least to take into account endogeneity and selection bias. COMMENT | 301 Market Socialism As Guriev and Megginson refer to this old idea, let me say a word about it. Market socialism relied on an illusion that it was possible to eliminate the uncertainty plagu- ing capitalist economies. The central planner was supposed to have access to all necessary information to realize his objectives. He was supposed to be able to decide which enterprises and sectors should be developed and which should be contracted, what new projects should be realized, and so on. What we call transition indeed involves a change of the paradigm and could be defined as a process of abandoning the utopian view of predictable society and going back to the world full of risk and uncertainty. The uncertainty characterizing economies in transition is the one typical for any complex economic system that is confronted every day with new problems. But this "universal" uncertainty is magnified by the need for a profound redeployment of assets in the conditions of rapid changes in relative prices, profoundly reshaped macroeconomic policy, political uncertainty due to the slow emergence of democratic institutions, and to the new types of inequalities. This makes distinguishing potential winners from unquestionable losers in the sector of state-owned enterprises particularly difficult. Assessing the value of and the prospects for a firm comes up against the lack of adequate information, incentives, and competence. In such circumstances, the generation of information about various investment opportunities becomes the critical issue in enterprise restructuring. Viewing the process of transition as the switch to an economic system stimulating a lot of infor- mation generation may have important policy implications. It gives some clues for assessing reform measures and the emerging institutional and organizational order. It also makes blatant the inability of market socialism to respond to such challenge. Conclusions? We learned a lot from the papers about the intricate relationship between institutions and development. But we are still far from fully understanding the process of suc- cessful institutional change. We know that there are no ready-to-wear solutions. All three papers recognize great variability of rules and institutions across countries and over time and suggest caution in drawing general conclusions. The political system, culture, religion, legal origin, financial development: all of them may matter. Given the existing evidence and experience, let me venture a tentative conclusion: What appears most important is to create conditions for flexible adjustment of insti- tutional setting to country-, industry-, and firm-specific constraints. A democratic system seems to provide the best framework for such adjustment, which requires freely confronting different ideas about the organization of the society. Comment on "The Institutional Determinants of State Capabilities in Latin America," by Ernesto Stein and Mariano Tommasi; "Lowering the Cost of Capital in Emerging Market Economies," by Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya; and "Privatization: What Have We Learned?" by Sergei Guriev and William Megginson CHONG-EN BAI Governance is one of the most important factors behind economic development. The three papers in this session consider governance from three different but closely related aspects: namely, political governance, corporate governance, and ownership. It is very useful to consider these three topics together because they interact with one another, and the political economy of corporate governance and that of privatization are valuable case studies of political governance. Therefore, the three papers are excellent complements of one another. Ernesto Stein and Mariano Tommasi take the view that the policy-making process is more important than specific policies themselves. For tthe policy-making process to be effective, the state needs to have the capabilities to make policies in the interest of the general public rather than that of narrow interest groups, to commit to the intertemporal consistency of policies, to adjust policies when circumstances change, and to enforce policies effectively. The ability for political actors to cooperate over time is particularly important. These capabilities depend on the characteristics of the executive, legislative, and judicial branches of the government, on the party system, and on the quality of the bureaucracy. This framework for understanding policy is a Chong-En Bai is professor of economics, School of Economics and Management, Tsinghua University, Beijing. ©2007 The International Bank for Reconstruction and Development / The World Bank 303 304 | CHONG-EN BAI welcome change to the conventional approach of prescribing a one-size-fits-all menu of specific policies to developing countries. Erik Berglof, Patrick Bolton, Sergei Guriev, and Ekaterina Zhuravskaya emphasize the interaction between political governance and corporate governance. They argue that government failures as well as market failures affect corporate governance, and at the same time, corporations are potent political actors and have the capacity to influence government policy. As political institutions vary significantly across coun- tries, corporate governance practices should adapt to the local environment. Again, there is no one-size-fits-all solution to corporate governance problems. This political economy approach to understanding corporate governance is especially relevant for emerging markets. This paper makes a significant contribution by focusing one's attention to this important aspect. Sergei Guriev and William Megginson survey the empirical literature on the effect of privatization on the performance of the privatized firm and on society. Yet again, there is no simple conclusion. Although privatization usually has positive effects on firm productivity and social welfare, the effects depend on institutions. The experiences of China and Russia are used as case studies. Given the rapidly expanding literature on privatization, this survey is very timely. A major feature of public policy considered by Stein and Tommasi is public regardedness: that is, the degree to which policies pursue the public interest. This is one of the most difficult goals of public policy to achieve. In many policy areas, special interest groups benefit at the expense of the general public. Corporate governance and privatization policies are no exceptions. The next two sections of this discussion will present two examples from China to illustrate this point: one on the public interest and corporate governance, and the other on the public interest and privatization. But first, let me make a few observations about the two examples. In the example about corporate governance, one can see that simply protecting the interests of the investors is not enough. Even if the government does not own the firm or lend to it, public interests can still be damaged by the governance failures of the firm when the government's role is that of a regulator and it is pressured to carry out a regulator bailout. Therefore, corporate governance should help prevent regulator bailout. The example about privatization demonstrates that one should not just look at its effect on the financial performance of the firm but also consider the effects on the general public and other stakeholders. The examples also show that the specific policy of ownership of firms has signifi- cant effects on other policies in addition to the policy-making process. Furthermore, cooperation among political actors, which is emphasized by Stein and Tommasi, may be counterproductive when political actors in the policy arena do not fully represent the interests of the general public. Finally, the examples provide evidence of the importance of the capabilities for the bureaucrats to enforce and implement policies. The two examples illustrate how special interests can benefit from public policies at the expense of the general public, but they are not meant to belittle the importance of another issue: that is, how the interests of minority groups should be protected from "the tyranny of the majority." Stein and Tommasi do not spend much space on COMMENT | 305 this issue either. However, this is a very important issue, and more research should be conducted on it. The remainder of this note presents the two examples. The Public Interest and Corporate Governance To illustrate the role of government in corporate governance, consider the case of securities firms in China. There are about 130 securities firms in China. Their business lines include investment banking, asset management, and retail brokerage services, among others. Most of these firms are wholly or partially owned by various levels of governments, most by local governments. They are treasured by their government owners because they bring many benefits to state-owned enterprises under the juris- diction of the government owners and sometimes to government officials. These firms hold two types of accounts for investment in the securities markets: accounts owned by their brokerage clients, and accounts owned by the firms them- selves for managing their own assets. The retail clients often leave some cash in their accounts to facilitate future trading. To protect the interests of the retail clients, the government has issued regulations forbidding unauthorized transfers of funds between retail-client accounts and the securities firm's own accounts. In reality, however, the "firewall" between the two types of accounts required by the government regulation have turned out to be a "paper wall"; the regulation was written on paper but poorly enforced. As a result, it was a widespread practice for securities firms to divert funds from client accounts to make investment for their own accounts. If the investment paid off well, the diverted funds would be retuned to the client accounts when the clients needed to use the money. However, if the investment went badly, the securities firm would have difficulty paying back the clients. This problem got very severe when the whole market was in decline. In some cases, the securities firm would have gone bankrupt if there had been no cash infusion into the firm. If the securities firm had gone bankrupt, its brokerage clients would have lost their money not because they had made bad investment choice themselves but because their funds had been diverted from their accounts to make risky investment by the firm without their authorization. The poor enforcement of the firewall between the client accounts and the securities firm's own account was partly responsible for the problem. Therefore, the government faced considerable pressure from brokerage clients to bail out the securities firm. Meanwhile, the government owner of the secu- rities company also lobbied the central government to keep the securities firm afloat by injecting news funds into the firm. Usually, it was the central bank that was footing the bill for the bailout. The problem did not go away with the bailout. The securities firms formed the expectation of being bailed out when they ran into trouble and kept repeating the same practice. After all, when they gambled by investing clients' funds without authorization in risky assets, they would reap big profits if the investment happened to go well, and they did not have to bear the cost if the investment went bad. For 306 | CHONG-EN BAI example, one securities firm had only 100 million yuan in capital but owed its clients 2 billion yuan in 2002. After a round of bailouts, it lost more money in its gambles and its deficit increased to 4 billion yuan in 2004. In total, the government spent tens of billions of yuan to bail out securities firms in trouble. Several lessons related to governance and ownership can be learned from this example. First, the general public interest can be damaged when the government as the regulator is forced to bail out a firm for its corporate governance failure. A good corporate governance system should prevent this from happening. Corporate gover- nance is sometimes narrowly defined as the collection of mechanisms that protect the interests of investors so that they can obtain sufficient return to cover their costs of investment and therefore have continued incentives to invest in the future. An even narrower definition focuses on the interests of shareholders. However, creditors are also important investors, and their interests cannot be ignored either. To protect investor interests, one should find ways, including proper bankruptcy rules, to mitigate the potential conflicts between insiders and investors of a firm. The insiders whose expropriation activities one should protect the investors from include controlling shareholders as well as the management. In emerging markets where the rule of law is inadequate to control government expropriation, ways should also be found to protect the investors from the government. If we focus only on investor interests, the conventional corporate governance mechanisms seem useful in China. Bai and others (2004) find that publicly listed firms in China with better conventional corporate governance mechanisms tend to have higher market valuation. However, our example shows that merely protecting investor interests is not enough. Members of the general public are also stakeholders whose interests should be protected by the corporate governance mechanisms. In the literature, people have considered the interests of employees, customers, and other suppliers as stakeholders that should be protected by corporate governance mechanisms. The interests of the general public have also been considered when the government is an investor or when firms may cause environmental damages. However, except for the banking sector, the discussion seems to have ignored the possible damage to the public interest when the government, as the regulator, is forced to bail out firms due to their governance failure. This issue needs to be considered when discussing corporate governance. The second lesson is that in addition to the policy-making process that is empha- sized by Stein and Tommasi, the specific policy of firm ownership matters: not only for corporate governance but also for government behavior. In our example from China, the lobby of local government owners of the securities firms played an impor- tant role in pressing for the bailout. Furthermore, it seems that society is more ready to tolerate the bailout of a state-owned enterprise (SOE) than a private enterprise. Third, if some interests, especially the general public interest, are not represented in the policy-making arena, the close cooperation emphasized by Stein and Tommasi among players that are in the arena may damage the interests of those who are not represented. In the example, the general public is the silent majority in the matter. The securities firms, their owners, and their clients are all politically very vocal. The close cooperation between these vocal stakeholders and the policy makers made it more likely for the bail out to happen. COMMENT | 307 Fourth, the weakness of the bureaucracy in its regulation enforcement capabilities is another reason for the public interest to be sacrificed. This supports Stein and Tommasi's consideration of the development of the civil service as a determinant of policy outcomes. In our example, if the firewall between the clients' accounts and the securities firm's own accounts had been well enforced, the pressure for bailout would have been much weaker. The enforcement failure was partly due to strong influence of the government owners of the securities firms, but the weak capability of the bureaucracy was also responsible. Public Interest and Privatization The privatization process in China also illustrates the difficulty of making policies in the interest of the general public and the danger of special interests colluding with one another at the expense of the public interest. An SOE in China typically employs too many workers. If it is privatized, the new owner most likely will lay off many of these workers. If the prospect for the laid-off workers to be reemployed by other firms is poor and the social security system is so rudimentary that it is more efficient for the firm to provide the social safety net than for the government to do so, then privatization is costly. Privatization also brings some benefits to the government owner of the SOE. When an SOE is privatized, the government owner gets the proceeds from privatization and also gets rid of the burden of supporting the SOE from failing. SOEs in China are not all owned by the central government. Most are owned by local governments at the county, city, or provincial level. Different levels of the government trade off the cost and benefit of privatization differently. All levels of the government take full account of the benefits of privatization, but the local govern- ments consider only part of the cost of privatization because laid-off workers may migrate to other regions in the country. Therefore, local governments may be too aggressive in their privatization efforts and their action may not be in the interest of the whole country. This argument is consistent with the empirical findings in Bai, Lu, and Tao (2006a, 2006b, and 2006c). Bai, Lu, and Tao (2006a) find that privatization tends to reduce the number of employees in the firm on average. Bai, Lu, and Tao (2006b) further find that this is true only for SOEs affiliated with the lower two levels of the government. Bai, Lu, and Tao (2006c) find that for lower-level governments, an SOE affiliated with them is more likely to be privatized if there are more surplus workers in the firm; for higher-level governments, the opposite is true. The way debt is dealt with when an SOE is privatized suggests that local govern- ment owners of SOEs may collude with the buyers, and sometimes the management of the firm, to write off the debt owed by the firms at the expense of banks owned by the central government. Local governments often have close relations with the buyers of the privatized SOEs, but they do not have any stake in the banks to which privatized SOEs owe their money (except in the case of city commercial banks, which are very small compared to central government-owned banks). Therefore, local gov- ernments have incentives to help the buyers negotiate a debt write-off from the banks 308 | CHONG-EN BAI in the process of privatization. This benefits both the local governments and the new owners of the firms, but is costly to the bank and in turn to the interests of the general public. There is evidence that when a firm is privatized, the level of debt owed by the firm and the financial expenses of the firm both decrease on average (Bai, Lu, and Tao 2006a). Furthermore, this is true only for SOEs affiliated with lower-level governments, but not true for those affiliated with higher-level ones (Bai, Lu, and Tao 2006b). Finally, for lower-level governments, an SOE is more likely to be priva- tized if it owes more debt. For higher-level governments, the opposite is true (Bai, Lu, and Tao 2006c). Given that privatization is often used by the local government owners of SOEs to benefit themselves or people with close connections to them at the expense of laid- off workers and the banks owned by the central government, it is not surprising that the evidence does not suggest a significant real improvement in firm performance after privatization. Bai, Lu, and Tao (2006a) find that the profits of a firm tend to increase after privatization, but the profit increases come mainly from the reduction of management expenses and financial expenses. The former may be the result of laying off mid-level managers, and the latter may be due to the debt write-off. If so, there is not much real improvement. More research is needed if better data become available to further identify the concrete sources of the profitability improvement. Existing data does not allow us to identify other potential benefits of privatization, such as the reduction of bailout or other potential costs of privatiza- tion to the government, such as selling the firm at too low a price. Again, more work is needed. References Bai, Chong-En, Jiangyong Lu, and Zhigang Tao. 2006a. "How Does Privatization Work in China?" Tsinghua University. ------. 2006b. "The Multitask Theory of State Enterprise Reform: Empirical Evidence from China." American Economic Review Papers and Proceedings 96 (2): 353­7. ------. 2006c. "Divergent Interests between Central and Local Governments: Testing Theo- ries of Public Ownership." Tsinghua University. Bai, Chong-En, Qiao Liu, Joe Lu, Frank M. Song, and Junxi Zhang. 2004. "Corporate Governance and Market Valuation in China." Journal of Comparative Economics 32 (4): 599­616. Judicial Foundations of a Market System Judicial Reform in Developing Economies: Constraints and Opportunities MATTHEW C. STEPHENSON This essay focuses on three recurring problems that bedevil efforts to design and implement effective legal and judicial reform projects. The first problem is a straight- forward resource constraint problem: Improving the capacity and quality of a judicial system requires material and human resources that are in short supply in developing economies. The second problem is an incentive compatibility problem: The ability of the judicial system to perform a positive role in promoting development depends on the willingness of affected parties to use the courts to resolve disputes and to abide by judicial decisions, and on the willingness of judges and other legal officers to behave in a manner that is consistent with the requirements of a well-functioning judicial system. The third problem is an institutional version of the General Theory of the Sec- ond Best: When a legal system is suboptimal in more than one respect, improving the law or the courts along one dimension may not improve overall institutional perform- ance, and may even worsen it. Scholars and practitioners should pay greater attention to the inherent tradeoffs induced by resource scarcity; the importance of making sure that individual incentives are aligned with institutional objectives; and the dangers that particular institutional reforms that appear to be welfare-improving when consid- ered in isolation may have counterproductive effects, if other institutional reforms are unachievable. Over the last decade, there has been an extraordinary increase in the attention paid to the role that public institutions play in promoting economic development.1 Indeed, the assertion that "institutions matter" has become commonplace, perhaps even cliché. This institutionalist revival in the development community has included a resurgence of interest in the role that legal and judicial institutions play, or ought to play, in promoting material improvements in the quality of life of the world's poor. Academics and policy analysts have sought to better understand the relationship between legal institutions and economic performance, while the development com- munity has promoted legal and judicial reform projects that range from modest Matthew C. Stephenson is assistant professor at Harvard Law School. The author is grateful to William Alford, Rick Messick, Ugo Panizza, Katharina Pistor, Andrei Shleifer, Stephen Voigt, and two anonymous referees for helpful comments on earlier versions of this paper. Annual World Bank Conference on Development Economics 2007 © 2007 The International Bank for Reconstruction and Development / The World Bank 311 312 | MATTHEW C. STEPHENSON efforts to improve court administration to ambitious attempts to eliminate judicial corruption, promote judicial independence, and craft better, more equitable, and more market-friendly legal systems. The diversity and complexity of the debate about legal and judicial reform, and of the myriad reform projects that have already been undertaken, put a comprehensive overview of the field beyond reach. My purpose here is a more modest one. First, I want to identify what I see as basic and recurring problems that bedevil efforts to design and implement effective legal and judicial reform projects. Second, I hope to suggest some conceptual tools that can be used to address these difficulties. I have three particular problems in mind. The first is a straightforward resource constraint problem. Improving the capacity and quality of a judicial system requires material and human resources that are in short supply in developing economies. The second problem is what one might think of as an incentive compatibility problem. The judiciary's capacity to perform the economic and other functions assigned to it by law-and-development theorists depends in large part on the willingness of affected parties to use the courts to resolve disputes and to abide by judicial decisions, and on the willingness of judges and other legal officers to behave in a manner that is con- sistent with the requirements of a well-functioning judicial system. But creating appropriate incentives often proves difficult. The third problem is an institutional version of the General Theory of the Second Best: When a legal system is suboptimal in more than one respect, improving the law or the courts along one dimension may not improve, and may even worsen, overall institutional performance. Understand- ing this principle is important to understanding, and attempting to avoid, the pitfalls associated with the necessarily incremental and partial nature of virtually all legal and judicial reform efforts. Why Reform Judiciaries? It may be useful to remind ourselves why reforming legal and judicial institutions in developing countries is thought to be important. In sketching an answer, I will glide over an even more basic set of conceptual questions. Simply defining "courts," "law," and "lawyers" in comparative or historical contexts can be a challenge, given the vari- ation in institutional arrangements and functions. There is also the vexed question of whether certain qualities of the legal system ought to be considered constitutive of, not merely causally connected to, "development" properly understood (Sen 2000). With- out disparaging the significance of these conceptual controversies, for reasons of brevity I will not engage them here. Instead, I use terms like "law," "courts," and "judi- cial" to refer to the set of institutional arrangements that conventionally carry those labels, even though I acknowledge that substantial institutional variation exists. As for "development," I will focus on the instrumental role of legal and judicial institutions in promoting social welfare, rather than on the intrinsic value of such institutions. With these definitional preliminaries out of the way, what can we say about the appropriate role of the judicial system in promoting economic development? Generally, the primary service provided by courts is thought to be reliable and efficient dispute JUDICIAL REFORM IN DEVELOPING ECONOMIES | 313 resolution. This service is important to development for at least three reasons. First, courts enforce contract and property rights, and secure property and con- tract rights are important for fostering productive investment and arms' length economic transactions (North 1990; World Bank 2005, ch. 4). Second, state-funded courts may improve economic performance by correcting various market failures. For example, judicial imposition of legal liability for certain types of harm may induce private parties to internalize what would otherwise be negative externalities associated with their conduct. To put the same point in more Coasian terms, a well-functioning judicial system may allocate liability in such a way that total social costs (including the transaction costs associated with bargaining around the initial allocation of legal rights) are minimized (Coase 1960). Third, judicial enforcement can make commitments--particularly commitments by government--more credible. The basic credible commitment problem identified by Finn Kydland and Edward Prescott (1977) has particular salience for the govern- ments of developing economies, which need to convince both their citizens and international investors to invest in the long term without fear that the government will expropriate the value of these investments (Brunetti and Weder 1994; Henisz 2000). Because courts are supposed to resolve disputes according to preexisting legal commitments--whether contained in contracts, statutes, or constitutions-- judicial dispute resolution by independent, effective courts helps enable parties, including government, to bind themselves to take or forgo certain actions under specified circumstances. To be sure, at least some of these functions can be performed by other institutions, or even by private parties. Thus the American Arbitration Association, the Interna- tional Chamber of Commerce, the World Bank's International Center for the Settle- ment of Investment Disputes, and a host of other providers offer conflict resolution services that compete with state-backed courts. But, while competitive private provi- sion of dispute resolution services can be both healthy and desirable (Benson 1990; Landes and Posner 1979), there are several reasons why public provision of dispute resolution services, in the form of effective courts, is superior to exclusive reliance on the private market. First, many forms of private dispute resolution are inherently limited in size or scope (Bueno de Mesquita and Stephenson 2006; Greif 1993). Second, many nongovernmental substitutes for judicial dispute resolution produce significant negative externalities. For example, Curtis Milhaupt and Mark West (2000) show that in the absence of effective state dispute resolution and contract enforcement in Japan, the Yakuza (the Japanese mafia) provides an unsavory substitute. Likewise, Diego Gambetta (1993) found that the Sicilian mafia arose to supply landowners with protection from predatory attacks in an environment where state-supplied law enforcement and dispute resolution was unavailable. Third, courts develop rules, doctrines, and principles that offer guidance for the resolution of future disputes. This is particularly so in common law countries, but it is increasingly the case in civil law countries as well (MacCormick, Summers, and Goodhart 1997). This body of judge-made (or judge-"discovered") law is a public good that benefits individuals other than the parties to the dispute. It would therefore tend to be undersupplied in a private market for dispute resolution services (Landes and Posner 1979). 314 | MATTHEW C. STEPHENSON The preceding summary of the role of the judiciary in economic development is both abstract and general. Specifying the optimal set of judicial and legal institu- tions for any given country is a much more difficult and context-specific task, one that is well beyond the scope of this essay. The point I want to emphasize is that even if we could specify the optimal judicial and legal institutions for any given developing country, reformers who wanted to bring about progress toward that ideal could not escape three challenging problems: resource constraints, incentive compatibility, and the second best problem. It is to these three issues that I now turn. Three Dilemmas for Judicial Reformers Resource Constraints The first important limitation on the ability of legal and judicial reform to improve overall economic well-being in developing countries is the simple fact that material and human resources are limited. This observation is not especially interesting analytically, but it has great practical significance. After all, every dollar spent on judicial reform is a dollar that cannot be spent on other public goods or put toward economically productive private investment. Every hour spent by government officials drafting judicial reform legislation or investigating methods for improving judicial performance is an hour that could have been spent on other legislative activities. And every talented young man or woman in a developing country who decides to become a lawyer or a judge generally forgoes the possibility of becoming an engineer or a doctor or an entrepreneur (Murphy, Shleifer, and Vishney 1991). (For that matter, every academic paper about legal reform diverts time and attention from papers on other aspects of the development project.) This is not to disparage the importance of legal and judicial reform as part of the larger project of economic reform. Clearly, legal and judicial reform has some role in the overall development project. The question, from a practical standpoint, is how much of a role it should have when resources are scarce. This is not a question that admits of easy or generic answers. My point, which may be obvious but is nonetheless worth restating, is that devoting development resources to judicial reform projects, and allocating those resources among various judicial reform projects, entails difficult trade-offs. It is therefore important to think more critically about the role of judicial reform as part of a larger development strategy, and about how to set institutional reform priorities. The prioritization issue relates to a more general set of debates in the academic and policy communities about the degree to which high-quality institutions--including but not limited to judicial institutions--are primarily a cause or a consequence of economic growth (Acemoglu, Johnson, and Robinson 2001; Chong and Calderon 2000). The short, simple, and not very helpful answer to this question is "both." But we need to know more about the nature of the causal relationships in order to make intelligent decisions about how to allocate scarce human and material resources in developing countries. If, for example, a well-functioning judicial system is a necessary precondition for large-scale economic activity, then it might make sense to devote JUDICIAL REFORM IN DEVELOPING ECONOMIES | 315 substantial resources up front to improving the court system. If, on the other hand, a great deal of economic progress and social welfare can be generated with a more modest court system, then it may make sense to devote relatively fewer government resources to the court system early on, targeting these resources instead at other things--such as health care, basic education, and infrastructure--thought to be more important for priming the pump of economic growth. Of course, the productivity of these other sorts of reforms may depend on a well-functioning system for regulating service delivery and resolving disputes, which may require a reasonably effective judicial system. The point is not that judicial reform should be postponed entirely, but rather that resource constraints mean that the allocation of scarce resources to different types of reform efforts involves difficult questions of prioritization. A similar resource constraint problem, and a similar set of hard choices, appears when we think about how to allocate resources among different types of judicial reform projects. There are, to be sure, some low-hanging fruit: straightforward, inexpensive reforms that yield a very high payoff. Thus, for example, the simple introduction of a computerized list of the jail population can significantly reduce the time those suspected of a crime are held before trial (Hammergren forthcoming). In most instances, however, such low-cost, easy reforms were embraced long ago, leaving the more complex, expensive, and politically controversial ones to be taken up. But usually it is simply impossible, in light of limited resources, for developing country governments or the donor community to tackle all of these at once, and it therefore becomes necessary to pick and choose among different projects. In that situation, how should priorities be set? Is it more important to train judges or to computerize the case filing and tracking system? Is it more important to invest in fighting judicial corruption or in educating the poor about their legal rights? Does it make more sense to concentrate resources on creating a few highly capable specialized tribunals--say, to deal with disputes involving foreign investors or major business transactions--or to spread resources more widely to improve the average local court? Again, all these things may be valuable, and these choices are "more- less" choices, not "either-or" choices. But they are choices nonetheless, and as Linn Hammergren's forthcoming review of the Latin American experience with judicial reform over the past 25 years shows, there has not been sufficient attention in the field to issues of prioritization and sequencing of judicial reform efforts (ch. 7). The lack of attention to prioritization and sequencing reflects what Thomas Carothers (2006) has dubbed "the problem of knowledge." Despite more than a decade of experience with programs of all types, knowledge about what factors are conducive to success, and why, remains scarce. This knowledge gap itself reflects a resource problem: the unwillingness of donor agencies and developing country governments to invest in better up-front analysis and more thorough post-reform evaluation (Carothers 2006; Hammergren 2002; Messick 2000). Yet without more robust data on judicial systems and reform experiences, analyzed in the context of the ongoing research on the role of institutions in development, the ability to set appropriate reform priorities is unlikely to improve. 316 | MATTHEW C. STEPHENSON Incentive Compatibility In order for the judiciary to perform the functions generally assigned to it by law and development theorists, the relevant parties must have appropriate incentives. Individ- uals must have an incentive to rely on the courts to adjudicate their disputes rather than relying on alternative, socially undesirable dispute resolution mechanisms or forgoing certain transactions altogether. Those with the power to disregard judicial decisions or to subvert judicial independence must have an incentive to refrain from such activities. And the judges themselves must have an incentive to carry out the functions assigned to them. Consider first the private parties who we would like to encourage to use the courts rather than other mechanisms to resolve their disputes. Of course, it is manifestly not the case that a society or an economy is better off if all potentially justiciable contro- versies are litigated, as that would entail an enormous social cost (Shavell 1997). Many nonjudicial dispute resolution mechanisms may often be more efficient (from both a private and a social perspective) than the court system, and therefore decisions to forgo judicial adjudication may often reflect a market success rather than a market failure (Bueno de Mesquita and Stephenson 2006). But, we might reasonably suppose that, in an ideal world, the judiciary would be the best forum for the resolution of some nontrivial subset of private disputes, either because alternatives are unavailable or because they are too socially costly. In this subset of cases, the private parties to a dispute must have incentives to rely on the court system. There are a number of reasons, however, why such incentives may not be present. The first and most obvious reason is that the court system may fail to provide dispute resolution services of acceptable quality. Judges or court administrators may be incompetent, venal, or corrupt, and the law itself may be inefficient or unfair. Or, the private costs to litigants of using the court system--attorneys' fees, filing fees, and other court costs--may be inefficiently high (Shavell 1997). Perhaps there are too few courts or is too much delay in hearing or deciding cases. Redressing these and similar failings is the bread and butter of most judicial reform efforts. There are also more subtle disincentives to reliance on judicial dispute resolution. Katharina Pistor's (1996) examination of Russian businesses' use of the courts to resolve commercial arbitration in the early 1990s provides an interesting illustration. Numerous observers claimed that during this period Russian businesses tended not to use the courts to resolve commercial disputes; instead, firms relied on nonlegal (or illegal) enforcement mechanisms. Contrary to this prevailing conventional wisdom, Pistor suggests that the reluctance of many Russian businesses to use the courts to resolve contractual disputes was not because the courts were inefficient or because enforcement was unreliable. Rather, the reason had to do with the gross inefficiencies of the Russian legal system, particularly the tax system, which induced most busi- nesses to engage in numerous illegal or semilegal transactions. Going to court would risk disclosing these transactions--even if they were only peripherally related to the dispute at issue--which would often result in undesirable consequences for the firm. The upshot, for legal and judicial reformers, is that bad law, or other bad collat- eral effects of invoking the judicial process, can deter use of otherwise efficient and effective judicial institutions. The point may seem obvious in the abstract, but the JUDICIAL REFORM IN DEVELOPING ECONOMIES | 317 more significant lesson here is that, when private parties are not using the courts, it is important to understand why. Improving the law may be of little relevance if judicial institutions are dysfunctional, but improving judicial institutions may likewise prove futile if private parties have strong incentives to avoid the courts for other reasons. Another hypothesized deterrent to socially efficient use of the court system is "legal culture." It is often claimed, particularly but not exclusively in the context of devel- oping countries, that the use of courts to resolve disputes is considered culturally taboo or otherwise inappropriate, and therefore the use of the courts is inefficiently low and reliance on nonjudicial dispute resolution mechanisms is inefficiently high (Bierbrauer 1994). There are at least two problems with this hypothesis, however. First, it is difficult to specify in advance those cultural norms that pose a socially undesirable impediment to judicial adjudication. Indeed there are several counterex- amples in which reliance on judicial dispute resolution became widespread, despite what one might have supposed were adverse cultural norms.2 Second, many of the examples cited in support of the proposition that cultural predispositions deter use of the courts may actually be examples of the more mundane--though important--type of problem described earlier: the courts simply are not working well, and therefore they are not an efficient alternative to nonjudicial dispute resolution mechanisms (Bueno de Mesquita and Stephenson 2006). A cultural aversion to relying on the formal court system may be an effect, rather than a cause, of a poorly functioning judicial system. Despite these concerns, however, the question whether specific cultural character- istics deter (or encourage) reliance on judicial dispute resolution is clearly important. If such effects do exist and can be identified, then cultural norms may be both an important constraint on reform and themselves an object of reform. If not, then judi- cial reformers should be wary of cultural determinist arguments about what sorts of judicial systems will or will not "take" in a particular cultural context. Up to now, the discussion has focused on the incentives of private parties to rely on the judiciary for the resolution of disputes. An equally significant incentive com- patibility question concerns the government. The government must have incentives both to abide by adverse judicial decisions to which it is a party and to enforce against other parties judicial decisions with which the government disagrees. The problem of creating adequate incentives for the government to respect judicial inde- pendence and authority is particularly salient if we believe that one of the most important functions of the judiciary is to enable the government to make credible commitments. Appropriate government incentives are important for private dispute resolution as well, since most judicial rulings rely on the government's willingness to enforce them in order to be effective. There is now a sizable literature on the political and economic factors that may induce governments to respect the rulings of an independent judiciary even when the government dislikes a given decision.3 One of the most interesting and persuasive explanations for why governments may have an incentive to respect adverse judicial decisions involves the role of long-term political competition between rivals for legislative and executive power. The hypothesis is that, when political competition is robust and the competitors tend to be long-lived political parties, then holders of 318 | MATTHEW C. STEPHENSON political power may prefer to respect the limits imposed by independent courts so long as their rivals do the same when they are in power. Mark Ramseyer (1994) developed this hypothesis to explain why courts in the United States exhibited so much more independence than courts in Japan under LDP rule. I have formalized Ramseyer's hypothesis and provided cross-country statistical evidence of a correlation between stable political competition and judicial independence (Stephenson 2003). The hypothesis finds further support in Thomas Ginsburg's (2003) study of judicial politics in East Asia and Andrew Hanssen's (2004) analysis of the variation in judi- cial independence across U.S. states. If the findings of this research prove robust, it will put to rest the claim that rule of law reforms can and should precede democratic reform (Zakaria 2003). If an independent judicial constraint on the government depends on robust and stable democratic competition, then attempts to promote judicial independence, or to implement legal or institutional reforms that presume an independent judiciary, are likely to founder in the absence of such competition. The scathing critique of the World Bank's efforts to promote judicial reform in Peru under President Fujimori may lend some case-specific support to this general con- clusion (Lawyers' Committee on Human Rights 2000). Another political mechanism that is sometimes thought to provide the government in power with sufficient incentives to respect independent courts is public support for the judiciary. On this account, because judicial independence improves social welfare-- for instance, by ensuring that the government respects welfare-enhancing limits on its own power--attempts by the government to subvert judicial independence or to defy judicial decrees would be detected by public watchdogs and punished by public opinion (Sutter 1997; Vanberg 2001). Anecdotal support for this view is occasion- ally drawn from instances in U.S. political history in which perceived attempts to defy the courts have triggered political punishment. The negative public reaction to President Roosevelt's plan to "pack" the Supreme Court and to President Nixon's hints that he might defy a Supreme Court order to turn over incriminating evidence during the Watergate investigation are the most prominent examples. President Johnson's willingness to call out the National Guard to enforce the Supreme Court's school desegregation decisions and the negative public reaction to congressional attempts to meddle in the recent Terry Schiavo fiasco in Florida may also illustrate the political support for insulating judicial decisions from government interference. The implication of this hypothesis, if it proves correct, may be that it is important for judicial reformers to promote a "rule of law culture" in which defiance or manip- ulation of courts engenders political opposition. But figuring out exactly what that entails is not easy. Moreover, there are reasons to doubt whether a "rule of law culture" is really an independent factor that causes the public to rise to the defense of independ- ent courts. Indeed, there are a number of troubling cases in which an apparent "rule of law culture" proved transient or powerless in the face of determined government hostility to the courts. For instance, Malaysia in the late 1970s and the early to mid- 1980s was generally seen as having a greater public commitment to judicial inde- pendence and the rule of law than most developing countries. Malaysia also boasted a relatively sophisticated and organized bench and bar. Despite this, after a series of politically controversial rulings provoked a constitutional crisis in 1988, Prime Minister JUDICIAL REFORM IN DEVELOPING ECONOMIES | 319 Mahatir forced out the Lord President of the Supreme Court and cowed the Court into submission (Harding 1990). It may be that public willingness to defend the courts from political interference arises not because of some general, abstract political commitment to judicial inde- pendence, but rather because in certain circumstances, the public is rationally distrustful of government decisions that fail to obtain judicial approval (Stephenson 2004). If that is so, it is not at all clear that the general public would have sufficient incentives to protect the courts from government action that clearly benefited large and powerful political constituencies. The policy implications of this perspective on public support for the courts may differ from those of the "rule of law culture" per- spective, in that the key to ensuring an effective judicial check in this view may be developing institutions and practices in which politically relevant constituencies rationally place a high value on the signal sent by judicial decisions, rather than attempting to promote a more general cultural change in affective attitudes toward the courts. A final incentive compatibility issue concerns the incentives of the judges them- selves. If the judges do not have incentives to decide cases in an appropriate manner, the judicial system will cease to function effectively as a forum for dispute resolution or as a source of new or improved law. One source of bad judicial incentives, already discussed, are threats and promises offered by the government in power. But even if the government has incentives to respect judicial decisions, the judges themselves may lack appropriate incentives. The most obvious problems here are the various forms of improper influence brought to bear by interested parties, either in the form of threats (the problem of judicial coercion) or promises (the problem of judicial cor- ruption). This sort of problem is easy to define but hard to combat. Another concern regarding judicial incentives is that even if judges are not corrupt, their interests may not align with social interests. For instance, judicial decisions, especially in controversial cases, may reflect judges' preexisting political or ideological commitments. Political scientists who study U.S. legal institutions have documented ideological voting on the U.S. Supreme Court and elsewhere, though the extent and significance of ideology in this context is a matter of considerable controversy (Segal and Spaeth 2002). Others have suggested that, in addition to advancing ideological or political goals, judges are concerned with their reputations (Posner 1993; Schauer 2000). This can be a good thing, if judges benefit from a reputation for probity and impartiality. But it can be a bad thing if the judge cares about his or her reputation for loyalty to a particular cause, faction, or ethnic group--a concern that may be particularly acute in deeply divided societies. The more general take-away point here for those interested in promoting eco- nomic development through judicial reform is the importance of thinking about the judges (and other legal professionals) who must carry out the business of the judicial system not as generic idealized arbiters but as flesh-and-blood human beings who are both rational and fallible. If judicial reform is to achieve its intended goals, it must succeed in aligning judicial incentives with social incentives. One particularly nettle- some problem with efforts to address this issue is that many of the mechanisms that would facilitate the government's ability to monitor judges and punish those who are 320 | MATTHEW C. STEPHENSON biased or corrupt may also make it easier for the government to undermine judicial independence. That is, there is a well-known tension between promoting judicial accountability and promoting judicial independence. Furthermore, any attempt to address an incentive problem by relying on external monitors merely shifts the incen- tive compatibility problem up one level. The Institutional Version of the General Theory of the Second Best There is yet a third problem that judicial reformers often encounter. This problem is sometimes described as the problem of "partial" or "incremental" reform, or as the problem of the "interdependence" of legal rules and institutions. The basic idea is that individual reforms that look like good ideas when considered in isolation can sometimes have unintended negative consequences. This problem can be conceptualized as an institutional version of the General Theory of the Second Best described by R.G. Lipsey and Kelvin Lancaster (1956). Though the General Theory of the Second Best may be familiar to many readers, let me offer a quick and informal summary. When a market contains multiple imper- fections, correcting or redressing a subset of those imperfections does not always lead to overall improvements in social welfare. In fact, in some cases the correction of some but not all market failures can lead to an overall reduction in social welfare. Although the "first best" world may be the one in which all market distortions have been corrected, if the elimination of certain market failures is not possible for some reason, then the "second best" world is not necessarily the one in which other market distortions are minimized. The correction of some market distortions can worsen others. Consider a stylized illustration involving a monopolist that produces a good with some negative externality, such as environmental pollution. Both the monopoly and the externality are market failures. If an otherwise well-functioning market is domi- nated by a single firm, the lack of perfect competition means that, relative to the social optimum, the market price will be too high and the quantity will be too low. In an otherwise efficient market for a good that imposes a significant negative exter- nality, the quantity produced will be too high and the market price will be too low. Now imagine a market in which both market failures are present. If both failures could be eliminated--say, through the combination of effective antitrust policy and an optimal tax on the externality--then we would be in the world of the first best. But if one of the market failures is uncorrectable, or simply uncorrected, fixing the other one might make matters worse. Suppose, for example, that the market becomes competitive but the negative externality problem is left unaddressed. The market price will drop and the quantity produced and consumed will increase. This exacer- bates the costs imposed by the negative externality, and if this externality is sufficiently costly, then the social welfare loss may exceed the social welfare gain. Though the General Theory of the Second Best has typically been applied to classic market failures like the ones just described, the basic insight also applies to legal and judicial reform, and to institutional reform more generally. Of course, it may be more difficult to specify the "first best" conditions for complex public institutions than to JUDICIAL REFORM IN DEVELOPING ECONOMIES | 321 do so for markets, but let us assume for the moment that we could adequately characterize a particular constellation of institutions as first best. When actual insti- tutional arrangements deviate from this institutional optimum in multiple respects, reforms that "improve" institutions along some but not all of these dimensions may not improve--and may in some cases worsen--overall social welfare. If certain insti- tutional reforms are simply off the table, at least for the time being, then would-be legal and judicial reformers in the developing world are likely to find themselves confronting a version of the second best problem. A failure to appreciate the fact that movements toward first best legal and judicial institutions do not necessarily lead to better overall performance can lead the most well-intentioned reformers astray. A simple, generic example of the legal-judicial version of the second best problem concerns the optimal complexity of legal rules. Many legal rules may be thought, with justification, to be too crude. Such rules may be inferior, from a social welfare perspective, to either more complex rules or more discretionary standards. There are ongoing debates in the legal and economic literature about the optimal precision of legal rules and about the relative virtues of rules and more discretionary standards,4 but we can assume for purposes of illustration that in the first-best world, legal rules would entail a reasonable amount of complexity and/or room for judicial discretion. But the first best world also involves sophisticated judges subject to the right incen- tives. In the real world, certain legal systems may be characterized by crude legal rules and unsophisticated judges. "Improving" the legal rules to make them more complex and nuanced might be a move toward the first-best world along that dimen- sion; however, the overall effect might be negative if unsophisticated judges make more welfare-reducing errors when attempting to implement complex legal rules than would be the case if these unsophisticated judges implemented cruder but simpler legal rules (Posner 1998; Hay, Shleifer, and Vishney 1996). We can illustrate this general problem with a more concrete example. Although it is generally believed that private agreements between suppliers and customers or manufacturers can be welfare-enhancing, it is also possible that such vertical arrangements may facilitate monopolistic pricing (Posner 2001). It is often difficult to distinguish between good and bad vertical contracts. A sophisticated judiciary, aided by high-priced advocates, may be reasonably good at applying a general "reasonableness" standard without making too many errors. But if the bench and bar are unsophisticated, a country might do better to adopt a simple rule--either banning or permitting all vertical integration contracts--than to allow unsophisti- cated courts to try to evaluate particular arrangements on a case-by-case basis. In this simple example, the policy implication appears to be that improvements in the quality and sophistication of the judiciary must precede improvements in the quality of the law. It is also possible to imagine a different example in which improvements in the sophistication of the judiciary, without improvements in the quality of the law, can make matters worse rather than better. Suppose that the law on the books is a bad, welfare-reducing law, but that sophisticated parties have fig- ured out how to get around it, and the unsophisticated judiciary is generally unable to detect such subterfuge. In the first-best world, we might have both efficient law and sophisticated judges. But suppose that we have both bad law of the sort just 322 | MATTHEW C. STEPHENSON described and also unsophisticated judges, and that it is not possible to improve the law. Will improving the sophistication of the judiciary in this situation improve over- all welfare? Not necessarily: The improvement in judicial sophistication may make it impossible to avoid the application of the bad, welfare-reducing legal rules. This example is a close cousin of the hypothesis that, at least in some contexts, corruption can be efficiency-enhancing because it allows parties to avoid excessively cumber- some regulatory requirements, and therefore efforts to combat corruption may be counterproductive if unaccompanied by regulatory reform (Huntington 1968). Another example of how the second best problem can affect the pursuit of judicial reform goals involves the credible commitment problem discussed earlier. In the first- best world, we might want governments to enact welfare-enhancing legal rules that are enforced by independent courts with the power to constrain the government. This constraint is important because firms' willingness to commit assets to long-term projects may be contingent on their confidence that the government will honor its promises not to expropriate these firms' profits. But suppose we are in a world where the law is growth-retarding rather than growth-promoting, and the judiciary is under the government's thumb. Reforms that strengthen the independence of the judiciary without altering the legal rules to which the state has committed itself may make matters worse because the courts may impede efforts by the government to adopt socially desirable legal reforms. Possible, though controversial, real-world illustra- tions of this problem might be the decisions by a number of courts in Latin America and Eastern Europe to block, usually on constitutional grounds, neoliberal economic reforms thought by domestic governments and outside advisors to be important for economic growth. For instance, a recently created chamber of the Costa Rican constitutional court, Sala IV, was designed to safeguard individual rights, but has had the unintended consequence of obstructing economic liberalization (Handberg and Wilson 2000).5 The lesson here is not necessarily that welfare-improving changes in the law must always precede credibility-enhancing improvements in judicial power. Imagine, for example, that law is bad and the judiciary is weak. In this institutional environment, the government's only source of credibility might be reputational. That is, the govern- ment might have to demonstrate its credibility by sticking to its announced policy no matter what. If there were a truly independent court and sufficiently cumbersome impediments to policy change, then the government might be able to change its economic policies without a significant loss of credibility, since the government would be as credibly committed to the new policy as to the old. But if we are in a second-best world where no serious constraints on policy change exist, attempts to revise the legal rules might lead to a net loss of social welfare if the loss of government credibility due to the unexpected change in policy outweighs the welfare gain from the improvement in the content of the law. In contrast to the earlier example, here improvements in the credibility-enhancing mechanisms, such as judicial independence, would need to precede improvements in the substantive content of the law. Let me give another, more concrete example of how the institutions that affect dis- cretion and credibility can give rise to a second-best problem in the context of legal JUDICIAL REFORM IN DEVELOPING ECONOMIES | 323 and institutional reform. This example is drawn from Brian Levy and Pablo Spiller's (1994) analysis of telecommunications regulation in Jamaica. Prior to 1966, Jamaica's telecommunications regulation strategy involved detailed licensing agreements between the government and Jamaica's domestic telecommunications company. These contractual agreements specified relatively precise rates of return over a long period of time, typically 25 years; they could not be modified without the company's consent; and they were enforceable in Jamaica's independent courts. According to Levy and Spiller, this system provided a relatively high degree of credibility and engendered high rates of investment in the telecommunications sector. The problem with a contractual, license-based system of regulation, however, is its lack of flexibility. Even when dramatic technological, economic, or political changes made a modification to the rate of return desirable, such a change could not be implemented under the Jamaican license-based system without the licensee's consent. Though the design of optimal regulatory institutions is a complicated and con- troversial topic in its own right, one might reasonably suppose that in a world of first- best institutions, telecommunications regulation would ensure credibility but would also entail more flexibility--perhaps through the use of an independent public utility commission (PUC) subject to appropriate institutional incentives and constraints-- than the contractual scheme in pre-1966 Jamaica allowed. In apparent response to the perceived rigidity of the license-based approach to regulation, in 1966 Jamaica decided to move to a different system. The Jamaican Public Utilities Act of 1966 established an independent regulatory commission, the Jamaican Public Utilities Commission, to regulate domestic telecommunications services. The statute directed the Commission to set a "fair" rate of return. The result, according to Levy and Spiller, was disastrous: The relationship between the Commission and the Jamaican Telephone Company quickly deteriorated, rate increases lagged behind inflation, and investment and network expansion ground to a virtual halt. The reason, Levy and Spiller explain, was that Jamaica in that period lacked institutions that could impose substantive restraints on the Commission's decisions. In contrast to the United States, which has had a reasonably successful (albeit imperfect) experience with rate-setting by independent utility commissions, Jamaica lacked the cluster of formal and informal institutional constraints on bureaucratic discretion necessary to maintain credibility. Also, although the Jamaican courts had proven adept at independently enforcing license contracts, their approach to what we would think of as administrative law--in particular, their review of bureaucratic discretion--was deferential to the point of being ineffectual. The problems in the Jamaican telecommunications sector lasted until the Commission was abolished and replaced with a new license-based scheme (albeit one that differed in many respects from the pre-1966 system). This example provides a nice illustration of the second best problem. Even if we stipulate that a first best world involves the use of administrative law rather than contract law to determine public utility rates, this does not necessarily mean that changing a system from one that relies on contract law to one that relies on admin- istrative law will necessarily improve the performance of the regulated sector. In a 324 | MATTHEW C. STEPHENSON country like Jamaica that lacks the political and legal institutional endowments necessary to make an administrative law system credible and workable, a regulatory system based on long-term contracts, while imperfect, may well be second best. Let me give one more historical example of the second best problem in a context that may have particular relevance for present-day legal and judicial reformers. This example is drawn from Rachel Kranton and Anand Swamy's (1999) account of the introduction by the British colonial government of civil courts into the Bombay Deccan region of India in the nineteenth century. Prior to the British introduction of civil courts, agricultural credit markets in this region were dominated by local moneylenders who relied on their own resources and other nonlegal mechanisms to recover loans. Reliance on such costly, personalized enforcement mechanisms meant that the scope of operation for any given moneylender was geographically limited, which led to a segmented market characterized by local oligopolies and high interest rates. The British colonial authorities believed that the introduction of well-functioning civil courts would facilitate arm's-length credit transactions, thereby introducing more competition into the rural credit market and lowering interest rates to competi- tive levels. Therefore, they introduced civil courts capable of enforcing simple debt contracts--but, importantly, incapable of enforcing more complex exclusive dealing or state-contingent contracts. If the British civil courts had simply failed to work--if they turned out to be corrupt and unreliable, or if nobody had used them--then this case might simply be another illustration of the importance of incentive compatibility, or an example invoked by those who advance the claim that legal and institutional "transplants" generally do not work. But what makes Kranton and Swamy's account of the British introduction of civil courts in the Bombay Deccan so interesting is that the courts did have their intended effect of increasing the feasibility of arms' length transactions, thereby stimulating competition and lowering interest rates. At least in the short term, the introduction of effective judicial contract enforcement in the Bom- bay Deccan appeared to be a great success. There was a problem, though. The problem had to do with the impact of effective enforcement of simple debt contracts on the ability of borrowers to insure against financial risk. Prior to the introduction of effective civil contract enforcement, if a borrower found himself unable to repay his debt due to natural disaster or some other misfortune, the local moneylender had an incentive to forgive or roll over the debt rather than to seize and sell the farmer's land. If the defaulting farmer retained his land and remained financially viable, then the moneylender could be confident that this farmer would provide a stream of super-competitive interest payments on future loans. Once the civil courts made arms' length transactions feasible, however, no moneylender could be confident of extracting a future stream of monopolistic profits from any individual borrower, and so the incentive to forgive or roll over debt instead of seizing assets dropped considerably. The result, as Kranton and Swamy persuasively argue, was economic disaster and widespread rioting when exogenous economic shocks, including a sudden drop in the international price of cotton, led to widespread defaults and asset seizures by creditors. The situation in central India prior to the British reform of the civil court system was not first best. Instead, it was characterized by at least two relevant legal/institutional JUDICIAL REFORM IN DEVELOPING ECONOMIES | 325 failures. First, the absence of effective civil contract enforcement led to monopolistic interest rates and an undersupply of credit. Second, the region lacked both an adequate public social safety net and the institutional infrastructure for a well-functioning private insurance market. The introduction of judicial institutions that could effec- tively enforce simple debt contracts alleviated the first market failure but, in so doing, it exacerbated the second one. This example may be especially salient for modern legal and judicial reformers, given the fact that many rural and poor urban commu- nities may still rely primarily on informal insurance mechanisms. Legal and judicial reforms--even, and perhaps especially, successful ones--may sometimes disrupt these risk-spreading mechanisms. I have dwelled on these examples of the institutional version of the second best problem because it is, in my view, underappreciated in the literature on institutional reform. It should not, however, be interpreted as a counsel of despair. Incremental institutional reform can, and often does, lead to improvements in overall welfare: The theory of the second best shows that correcting some but not all market imper- fections may lead to social welfare reductions, not that it necessarily will do so. And, even when partial reform does have counterproductive effects, these problems may be short lived if the initial incremental reform efforts are followed by more extensive reform of other institutions. The important lesson is that individual reforms cannot be considered in isolation, and that we can and should draw on the tools of economic analysis, applied in a particular context, to try to identify situations in which certain institutional reforms that appear to be movements toward an unachievable first-best world will actually move us away from an achievable second best. Conclusion In this short essay, I have attempted to provide a summary of some of the difficult problems that confront reformers who hope to address the problem of global poverty through the reform of institutions, particularly legal and judicial institutions. The goal is to encourage both scholars and practitioners to pay greater attention to the inherent trade-offs induced by resource scarcity; the importance of making sure that individual incentives are properly aligned with institutional objectives; and the dangers that particular institutional reforms that appear to be welfare-improving when considered in isolation may have counterproductive effects, if other institu- tional reforms are unachievable. The more general lesson, it seems to me, is the importance of greater cooperation between those in the policy and scholarly communities who specialize in more abstract and general economic theory and those who possess detailed, country-specific knowledge of particular institutional environments. The need for such cooperation seems self-evident, yet for some reason the communication between technically minded general theorists and context-sensitive country experts has sometimes been characterized by misunderstanding and mutual skepticism. My hope in elaborating on some of the more difficult and recurring generic problems in the field of legal and judicial reform is that the exercise will make an incremental contribution to thinking 326 | MATTHEW C. STEPHENSON more seriously and collaboratively about ways to identify and avoid these pitfalls in the context of specific legal and judicial reform efforts. Notes 1. For two recent examples, see Acemoglu and Robinson (2006) and Greif (2006). 2. See, for example, Kranton and Swamy (1999); Lee (1993). 3. For a summary of various theories, see Stephenson (2003). 4. See, for example, Kaplow (1992). 5. To be clear, I am agnostic as to the desirability of the particular economic reforms at issue. I proceed under the assumption that these reforms would have been welfare- enhancing in order to illustrate the nature of the problem. References Acemoglu, Daron, and James Robinson. 2006. Economic Origins of Dictatorship and Democracy. Cambridge: Cambridge University Press. Acemoglu, Daron, Simon Johnson, and James Robinson. 2001. "The Colonial Origins of Comparative Development: An Empirical Investigation." American Economic Review 91: 1369­1401. Benson, Bruce. 1990. The Enterprise of Law: Justice Without the State. San Francisco: Pacific Research Institute for Public Policy. Bierbrauer, Gunter. 1994. "Toward an Understanding of Legal Culture: Variations in Individualism and Collectivism Between Kurds, Lebanese, and Germans." Law and Society Review 28 (2): 243­64. Brunetti, Aymo, and Beatrice Weder. 1994. "Political Credibility and Economic Growth in Less Developed Countries." Constitutional Political Economy 5 (1): 23­43. Bueno de Mesquita, Ethan, and Matthew Stephenson. 2006. "Legal Institutions and Informal Networks." Journal of Theoretical Politics 18: 40­67. Carothers, Thomas C. 2006. Promoting the Rule of Law Abroad: In Search of Knowledge. Washington, DC: Carnegie Endowment. Coase, Ronald. 1960. "The Problem of Social Cost." Journal of Law and Economics 3: 1­44. Chong, Alberto, and Cesar Calderon. 2000. "On the Causality and Feedback between Insti- tutional Measures and Economic Growth." Economics and Politics 12 (1): 69­81. Gambetta, Diego. 1993. The Sicilian Mafia: The Business of Private Protection. Cambridge: Harvard University Press. Ginsburg, Thomas. 2003. Judicial Review in New Democracies: Constitutional Courts in Asian Cases. Cambridge: Cambridge University Press. Greif, Avner. 2006. Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge: Cambridge University Press. Hammergren, Linn. 2002. Reforming Courts: The Role of Empirical Research. Washington, DC: World Bank (PREM Note 65). ------. Forthcoming. Envisioning Reform: Practical and Conceptual Obstacles to Improving Judicial Performance in Latin America. The Pennsylvania State University. Handberg, Roger, and Bruce Wilson. 2000. "Costa Rica's New Constitutional Court: A Loose Political Cannon?" South Eastern Latin Americanist 43 (4): 58­79. JUDICIAL REFORM IN DEVELOPING ECONOMIES | 327 Hanssen, Andrew. 2004. "Is There a Politically Optimal Level of Judicial Independence?" American Economic Review 94: 712­29. Harding, Andrew. 1990. "The 1988 Constitutional Crisis in Malaysia." International and Comparative Law Quarterly 39: 57­81. Hay, Jonathan, Andrei Shleifer, and Robert Vishny. 1996. "Toward a Theory of Legal Reform." European Economic Review 40: 559­67. Henisz, Witold. 2000. "The Institutional Environment for Economic Growth." Economics and Politics 12 (1): 1­31. Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven: Yale University Press. Kaplow, Louis. 1992. "Rules Versus Standards: An Economic Analysis." Duke Law Journal 42 (3): 557­629. Kranton, Rachel, and Anand Swamy. 1999. "The Hazards of Piecemeal Reform: British Civil Courts and the Credit Market in Colonial India." Journal of Development Economics 58: 1­24. Kydland, Finn, and Edward Prescott. 1977. "Rules Rather than Discretion: The Inconsistency of Optimal Plans." Journal of Political Economy 85: 473­90. Landes, William, and Richard Posner. 1979. "Adjudication as a Private Good." Journal of Legal Studies 5 : 235­84. Lawyers' Committee on Human Rights. 2000. Building on Quicksand: The Collapse of the World Bank's Judicial Reform Project in Peru. New York. Lee, Tahirih. 1993. "Risky Business: Courts, Culture, and the Marketplace" University of Miami Law Review 47: 1335­1414. Levy, Brian, and Pablo Spiller. 1994. "The Institutional Foundations of Regulatory Commit- ment: A Comparative Analysis of Telecommunications Regulation." Journal of Law, Economics, and Organization 10 (2): 201­46. Lipsey, R.G., and Kelvin Lancaster. 1956. "The General Theory of the Second Best." Review of Economic Studies 24 (1): 11­32. MacCormick, Neil D., Robert S. Summers, and Arthur L. Goodhart, eds. 1997. Interpreting Precedents: A Comparative Study. Brookfield: Ashgate/Dartmouth. Messick, Richard E. 2000. International Support for Civil Justice Reform in Developing and Transition Countries: An Overview and Evaluation. http://www1.worldbank.org/ publicsector/legal/donorsupported.doc. Milhaupt, Curtis, and Mark West. 2000. "The Dark Side of Private Ordering: An Institu- tional and Empirical Analysis of Organized Crime." University of Chicago Law Review 67: 41­98. Murphy, Kevin, Andrei Shleifer, and Robert Vishney. 1991. "The Allocation of Talent: Impli- cations for Growth." Quarterly Journal of Economics 106: 503­30. North, Douglass. 1990. Institutions, Institutional Change, and Economic Performance. Cambridge: Cambridge University Press. Pistor, Katharina. 1996. "Supply and Demand for Contract Enforcement in Russia: Courts, Arbitration, and Private Enforcement." Review of Central and East European Law 22 (1): 55­87. Posner, Richard. 1993. "What Do Judges and Justices Maximize? (The Same Thing Everybody Else Does)." Supreme Court Economic Review 3: 1­41. ------. 1998. "Creating a Legal Framework for Economic Development." World Bank Research Observer 13 (1): 1­11. ------. 2001. Antitrust Law, 2nd Edition. Chicago: University of Chicago Press. 328 | MATTHEW C. STEPHENSON Ramseyer, Mark. 1994. "The Puzzling (In)dependence of Courts: A Comparative Approach." Journal of Legal Studies 23: 721­47. Schauer, Frederick. 2000. "Incentives, Reputation, and the Inglorious Determinants of Judicial Behavior." University of Cincinnati Law Review 6: 615­36. Segal, Jeffrey A., and Harold J. Spaeth. 2002. The Supreme Court and the Attitudinal Model Revisited. Cambridge: Cambridge University Press. Sen, Amartya. 2000. "What Is the Role of Legal and Judicial Reform in the Development Process?" Speech delivered at the World Bank Conference on Legal and Judicial Reform. http://www1.worldbank.org/publicsector/legal/legalandjudicail.pdf. Shavell, Steven. 1997. "On the Fundamental Divergence between the Private and Social Motive to Use the Legal System." Journal of Legal Studies 26 (2): 575­612. Stephenson, Matthew. 2003. "`When the Devil Turns...": The Political Foundations of Inde- pendent Judicial Review." Journal of Legal Studies 32: 59­89. ------. 2004. "Court of Public Opinion: Government Accountability and Judicial Indepen- dence." Journal of Law, Economics, and Organization 20: 379­99. Sutter, Daniel. 1997. "Enforcing Constitutional Constraints." Constitutional Political Economy 8: 139­50. Vanberg, Georg. 2001. "Legislative-Judicial Relations: A Game-Theoretic Approach to Constitutional Review." American Journal of Political Science 45: 346­61. World Bank. 2005. A Better Investment Climate for Everyone: World Development Report 2005. Washington, DC: World Bank. Zakaria, Fareed. 2003. The Future of Freedom: Illiberal Democracy at Home and Abroad. New York: W.W. Norton Co. Transforming Judicial Systems in Europe and Central Asia JAMES H. ANDERSON AND CHERYL W. GRAY This paper updates a recent World Bank report, Judicial Systems in Transition Economies: Assessing the Past, Looking to the Future (Anderson, Bernstein, and Gray 2005) by incorporating the findings of a large survey of enterprises throughout the region undertaken in spring 2005, the third EBRD-World Bank Business Environment and Enterprise Performance Survey, or BEEPS (EBRD and World Bank 2005). The study emphasizes that judicial reform is a critical challenge for most transition countries. The majority of these countries have made progress in establishing independence in their judiciaries, but accountability, transparency, and efficiency have lagged behind. Many transition countries need to focus now on strengthening the fairness and honesty of their courts--which requires broad actions along many fronts to select the right judges and support staff, train, remunerate, and evaluate them adequately, and provide infrastruc- ture and IT systems to promote efficiency and transparency. More generally, transition countries share many of the same priorities and concerns as other countries, whether developed or developing, notably strengthening judicial accountability. The judicial systems in the transition countries of Central and Eastern Europe and the former Soviet Union are under heightened scrutiny these days, 17 years after tran- sition began. In Central and Eastern Europe, the European Union is exerting strong pressure on new members and candidate countries to root out corruption and improve the functioning of their judiciaries. Further east, judicial systems in Russia and other countries in the former Soviet Union have been increasingly in the spot- light due to high-profile roles in controversial cases, such as the Yukos case in Russia and the dispute surrounding the presidential elections in Ukraine. As economic reforms mature and these countries become increasingly interconnected with the out- side world, the need for good governance and the constraints imposed by weak judi- cial systems are rising in visibility and importance.1 James H. Anderson is senior economist in the Europe and Central Asia Region of the World Bank. Cheryl W. Gray is sector director in the Europe and Central Asia Region of the World Bank. The authors would like to thank David Bernstein for detailed comments and country examples and Laura Lanteri for helpful assistance with data analysis. Two anonymous referees provided helpful comments, as did Ugo Panizza and Stefan Voigt, the discussants, and other participants at the ABCDE in St. Petersburg. Annual World Bank Conference on Economic Development © 2007 The International Bank for Reconstruction and Development / The World Bank 329 330 | JAMES H. ANDERSON AND CHERYL W. GRAY A recent World Bank report, Judicial Systems in Transition Economies: Assessing the Past, Looking to the Future (Anderson, Bernstein, and Gray 2005, hereinafter Judicial Systems in Transition Economies), reviewed the transition countries' experi- ence with judicial reform since 1990 and drew on numerous data sources to compile a snapshot of the state of their judiciaries in the first few years of the twenty-first century. This paper updates that report by incorporating the findings of a large survey of enterprises throughout the region undertaken in spring 2005, the third EBRD-World Bank Business Environment and Enterprise Performance Survey, known as BEEPS (EBRD and World Bank 2005), described in detail in the annex. This paper goes into further detail on the judicial reform programs underway in transition countries and addresses three broad questions:2 · What kinds of judicial reforms are needed for successful transition from socialism to market-based economies, and in what sequence are they likely to occur? · How much progress has been made in this transition, both by individual countries and by subregion, and what factors may explain the extent of progress to date? · How do firms' evaluations of judicial systems in transition countries, and by implication the priorities and challenges that these systems face, compare with those in more advanced countries? To what extent do transition countries share common concerns and priorities with countries in Western Europe? From Plan to Market: Judicial Systems and the Sequencing of Reforms When looking from the vantage point of 1990, the magnitude of the changes needed to adapt the judicial systems of transition countries to the needs of a market eco- nomy seemed daunting. While on their face they had many of the elements of West- ern judicial systems--such as courts, judges, lawyers, prosecutors, and bailiffs--the roles, capacities, and expectations of each set of actors were fundamentally different. The entire purpose of the legal system under communism was to enforce the inter- ests of the working class, as represented by the communist party. Courts and judges were part of the executive branch and fully subordinated to the political leadership of the communist party. There was no idea of limited government, checks and bal- ances, or individual or corporate rights vis-á-vis the state. Laws in the commercial sphere dealt primarily with relationships between administrative agencies and the regulation of production by state-owned entities to meet centrally coordinated out- put targets. Most commercial disputes were handled through state-sponsored arbi- tration, while formal courts and judges handled criminal and civil matters (such as family law and minor personal property issues). The position of judge was not partic- ularly prestigious and was often staffed on a part-time basis. Courthouses were drab and unwelcoming, designed for an inquisitorial system of criminal prosecution where the defendant was almost always found guilty. Far-reaching changes would clearly be needed in the transition from socialism to capitalism. The existing legal framework--constitutions as well as civil, criminal, and commercial legislation--would need to be rewritten to recognize and respect TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 331 individual rights and limitations on state power, and the public would need to be educated about its new rights and how to enforce them. Judiciaries would need to be made independent of the executive branch to enable them to safeguard these rights and limitations. Many new laws--from property to evidence to banking to securities to bankruptcy laws--would need to be drafted and put in force to meet the needs of a private market economy, and the number of qualified judges and their training and knowledge base would need to be significantly expanded in order to understand and enforce these laws. Existing courthouses, often in dilapidated condition, would need to be renovated to improve public access and serve new due process requirements, and many new ones would need to be built to meet the rapidly expanding demand for dispute resolution. Finally, in the absence of heretofore strong executive control, new mechanisms would be needed to ensure capacity, accountability, and profession- alism not only of judges but also of the many related professions--such as lawyers, bailiffs, notaries, trustees, and court clerks--that make a judicial system work. Judicial Systems in Transition Economies describes the path of legal and judicial reform and the progress made in the 1990s. It documents how changes in the legal framework ("legal extensiveness") went much faster than institutional reforms ("legal effectiveness") and how, among institutional reforms, establishing independ- ence took precedence over building capacity or ensuring accountability.3 Overall, judicial reform tended to take a back seat to fundamental political and economic reforms, as reformers dealt with the pressures of declining output, rising inflation, and the scramble by some to appropriate state property--whether through state- sponsored programs of privatization or less legitimate means--that arose immediately after the collapse of communism. There is some logic to this sequencing. Institutions do not change in a vacuum; rather they change in response to pressure from within or without. Privatization of state assets, the creation of property rights and a private business class, and the increase in foreign trade and foreign investment that resulted from economic liber- alization have led to an increasing demand for more objective dispute resolution mechanisms and better-functioning regulatory and judicial systems in many transi- tion economies. Many countries are seeing a flood of new cases entering their judicial systems as a result of liberalization. In Russia, for example, the total number of cases filed with the commercial courts nearly doubled between 1995 and 2000, with tax and bankruptcy cases rising particularly quickly. In Ukraine some 6 million new cases enter the courts each year, to be handled by about 6,500 judges. Increasing demand has spurred training and investment in judicial systems that have slowly increased their capacity, as well as broader economic growth that helps to increase the resources available to the legal system as a whole. Figure 1 places countries on a continuum along two dimensions: the demand for judicial services (dependent in part on the extent of economic reform) and the resources available to the country's legal system to deliver judicial services (approxi- mated by a country's per capita GDP).4 The proxy used to measure demand is the average of the percentage of firms that have used the courts and the mean EBRD transition indicator for 2005. The proxy used for resource availability is the log of grossdomesticproduct(GDP)percapita,basedontheviewthatgreaterGDPpercapita translates into greater resource availability, which is itself expected to translate over 332 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 1. Resource Availability and Demand for Judicial Services, 2005 Pol Est Hun Lit Cze Cro 2 Slk Sln Bul Lat 1.8 Rom vices ser Mac Kyr 1.6 GeoUkr Rus Kaz Arm Alb judicialrof Mol SAM BiH 1.4 Aze Demand 1.2 Taj Uzk Bel 1 6 7 8 9 10 Availability of resources to deliver judicial services Sources: BEEPS, EBRD Transition Report, World Development Indicators. Note: Resource availability is the log of GDP per capita (2004); demand is based on court usage (2005) and the EBRD transition indicators (2005). Alb = Albania; Arm = Armenia; Aze = Azerbaijan; Bel = Belarus; BH = Bosnia and Herzegovina; Bul = Bulgaria; Cro = Croatia; Cze = Czech Republic; Esp = Spain; Est = Estonia; Geo = Georgia; Ger = Germany; Gre = Greece; Hun = Hungary; Ire = Ireland; Kaz = Kazakhstan; Kyr = Kyrgyz Republic; Lat = Latvia; Lit = Lithuania; Mac = FYR Macedonia; Mol = Moldova; Pol = Poland; Por = Portugal; Rom = Romania; Rus = Russian Federation; SAM = Serbia and Montenegro; Slk = Slovak Republic; Sln = Slovenia; Taj = Tajikistan; Tur = Turkey; Ukr = Ukraine; Uzb = Uzbekistan. time into stronger capacity. The intention here is not to prove an exact relationship (for which more refined proxies may be required) but rather to illustrate a key insight: that "demand" is as important as "supply" in pushing the judicial reform process for- ward, andcountriesatthepoorestandleasteconomicallyreformedendofthespectrum are unlikely to be the ones where immediate prospects for change are greatest. For those countries in the bottom left corner, the priorities should be to build basic demand for impartial dispute resolution through continued market reforms and to take initial steps to create or reinforce the independence and accountability of the judi- ciary. As countries move toward the upper right, the demand for more extensive and far-ranging judicial reform strengthens and there is a greater likelihood that efforts at reform will succeed, given greater resource availability. Three clear examples now are Bulgaria, the Former Yugoslav Republic of Macedonia, and Romania, where the demands for reform--both internally from the business community and externally from the European Union--are very strong and the likelihood of improvement high. Progress in Building Judicial Systems, 1990­2005 The first five or so years of transition, until the mid-1990s, saw little real change in the judiciaries in transition countries. As noted above, other priories--most notably TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 333 economic liberalization, privatization, and stabilization--took center stage, and little attention and few resources were devoted to longer-term institution building. The efforts that were made during this early period focused on constitutional change to lock in political reforms and judicial independence (as described further below), as well as the rapid preparation and adoption of commercial legislation. By the late 1990s, it became increasingly clear that weak capacity in the legal and judicial system was imped- ing investment and growth and contributing to corruption and poor governance. Citizen feedback mechanisms highlighted a growing distrust of legal institutions (Rose and Haerpfer 1994, 1996, 1998), and inability to implement or enforce new legislation led donors to focus more on the need for resources and capacity building (Anderson, Bernstein, and Gray 2005). In many countries in the region, strong and concerted efforts at change began in earnest only at the close of the decade. In some--primarily the countries that are also less advanced in economic and political reforms--those efforts are only just beginning; in a few, they have not yet begun. Progress along various dimensions of judicial reform and capacity building is outlined below. Judicial Independence and Accountability Independence of the judiciary is fundamental to a democratic political system and a free market economy, and most former socialist countries began their judicial reform efforts by moving to make their judiciaries independent from the executive branch of government. They were often assisted by foreign donors and democracy-promoting nongovernmental organizations (NGOs), which also focused primarily on judicial independence (rather than judicial capacity building) in the early years. New consti- tutions enshrined the principle of judicial independence, and new institutions--typically some type of judge-controlled judicial council for overall governance and a related judicial department for day-to-day court administration--were set up to oversee the selection and oversight of judges (often in conjunction with parliament and the minister of justice) and the day-to-day management of the courts. The process of establishing judicial independence was closely intertwined with the deepening of democratic processes in the overall political system; in general, the more democratic the political system, the more independent the judiciary has become. Judiciaries are now legally independent in virtually all European transition countries and are moving strongly in that direction in many Commonwealth of Independent States (CIS) coun- tries (with the exception of the few regimes where democracy has not yet taken hold). Indeed, judiciaries zealously promote and guard their independence, and there are often tense relationships between them and ministries of justice. The principle issue at present in most transition countries is not ensuring greater judicial independence--although admittedly de jure independence (typically sup- ported by judiciaries and government leaders alike) may not always be fully matched by de facto independence. The most pressing issue in many transition countries is ensuring judicial accountability, given newfound independence. As judiciaries have gained independence, their ability to ensure accountability has not kept pace. Most observers think that judicial corruption has increased during the 1990s along with the increased role and discretion of judges in the market economy. The paradox 334 | JAMES H. ANDERSON AND CHERYL W. GRAY is that judicial independence is necessary for true economic and political reform, but lack of judicial accountability is a major obstacle to economic development. Reform-minded ministers of justice want to push for greater accountability, but independence has taken away most of their levers of influence. Some chief justices are also pushing for greater accountability but face an uphill struggle to change entrenched and dysfunctional norms and practices. Evidence from the 2005 BEEPS survey throws light on the issue of accountability. Firms asked about honesty in the judiciary reported improvements in some countries from 2002 but deterioration in others (figure 2).5 It is particularly striking how poorly most countries fare. The only transition country where a majority of firms saw courts as honest in mid-2005 was Estonia. Perceptions of honesty improved in a number of countries--one of the most notable being Georgia,6 where a strongly FIGURE 2. Firms' Assessments of Courts as Honest and Uncorrupted, 2002 and 2005 Estonia Slovenia Hungary Belarus Romania level in 2005/04 Latvia increase between Armenia 2002 and 2005 Azerbaijan Tajikistan Georgia Bulgaria Poland y Croatia Countr Kazakhstan decrease between Slovak Republic 2002 and 2005 Czech Republic Lithuania Uzbekistan Serbia and Montenegro Russian Federation Albania FYR Macedonia Ukraine Bosnia and Herzegovina Kyrgyz Republic Moldova 0 25 50 75 100 Percent of firms indicating the courts are honest and uncorrupted Source: BEEPS 2002, 2005. Note: The chart shows the percent of firms indicating the courts were frequently, usually, or always "honest and uncorrupted" (4, 5, or 6 on a six-point scale). The sample includes all firms with nonmissing data. TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 335 reformist government is trying hard to tackle corruption--but worsened in Bosnia and Herzegovina, Hungary, FYR Macedonia, Moldova, Serbia and Montenegro, and Uzbekistan. On average about one-third of business managers viewed courts as honest, and even fewer in some of the new EU members such as the Czech Republic, Lithuania, Poland, and the Slovak Republic. Overall the change from 2002 to 2005 in the region as a whole was not substantial, and as of now there is little evidence that judicial corruption has been tackled successfully in most transition economies (Anderson and Gray 2006). It is interesting to compare perceptions of firms in the BEEPS sample that have been to court and those who have not, as these two groups often have different perspectives. Studies in the U.S. state of Wisconsin, for example, found that the general public has a different and often more pessimistic view of the courts than recent court users (Kritzer and Voelker 1998). Similarly, firms in the BEEPS sample that have actually used courts provided somewhat better assessments of honesty than those that have not, although the assessments of the former group have not changed significantly over the past three years, while assessment of the latter have improved (figure 3). However, firms that had actually been to court reported that unofficial payments are more frequent at courts than did firms that have not used the courts (figure 4).7 These two findings appear contradictory, in that one would typically equate higher levels of bribery with lower perceptions of honesty. One possible explanation is that some of the bribes might be paid to court functionaries to speed up the judicial process and may not be perceived as undermining the honesty of the judges themselves. Moreover, the patterns evident in figures 3 and 4 are regional patterns. For some individual countries, firms that use courts provide better assess- ments of the extent of bribery (Hungary and Poland), and for others firms that use courts provide worse assessments of honesty (Serbia and Montenegro). Trends may also be somewhat contradictory. While the overall assessment of the honesty of FIGURE 3. Assessments of Honesty in Courts 50 2002 2005 are ts 40 cour upted 30 ying sa uncorr ms and 20 fir of honest10 ercentP 0 Firms that have Firms that have not been to court been to court Source: BEEPS 2002, 2005. Note: The chart shows the percent of firms indicating the courts were frequently, usually, of always "honest and uncorrupted" (4, 5, or 6 on a six-point scale). The sample includes all firms with nonmissing data. Each country was given an equal weight. 336 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 4. Frequency of Bribery at Courts 15 2002 y 2005 iber br 10 ying sa frequent is ms ts fir of cour 5 at ercentP 0 Firms that have Firms that have not been to court been to court Source: BEEPS 2002, 2005. Note: The chart shows the percent of firms indicating that firms frequently, usually, or always use unofficial payments when dealing with courts (4, 5, or 6 on a six-point scale). The sample includes all firms with nonmissing data. Each country was given an equal weight. courts in the Slovak Republic improved only slightly (figure 2), the assessments of firms that had actually used the courts improved considerably.8 What would it take to establish true accountability in the judiciary? A myriad of individual steps are needed, including ensuring merit-based systems for judicial appointment, promotion, and disciplinary proceedings; providing adequate judicial salaries; and prosecution of some high-profile corruption cases, whether related to the judiciary or to government more broadly. Only through the "carrot" of profes- sional stature and remuneration and the "stick" of potential punishment for wrong- doing--together with the incentives and self-enforcement mechanisms that arise from transparency (see below)--can corruption be successfully tackled in the judiciary or any other branch of the public sector. Judiciaries and governments are aware of the dismal stigma of corruption, and sig- nificant steps are being taken to address it in many countries. In Romania, Russia, and Ukraine, for example, judicial salaries have been raised substantially to a level that compares reasonably to average private sector salaries.9 This move has raised the status of the profession, its "value" to incumbents, and its attractiveness to potential candidates. The process of judicial selection is also being tightened. Georgia, for example, was one of the first countries to introduce examinations for judges, and other transition countries have followed suit. While the examination process itself is not without difficulties,10 it is a step in the right direction compared to selection processes of old. As a complement to merit-based selection of judges, the Slovak Republic has put major efforts into strengthening government's capacity to prosecute cases of judicial corruption, including setting up a special court and prosecution office to deal with cases of corruption and organized crime. Public Information and Transparency Transition countries are also taking important steps to teach citizens about their rights and to increase the transparency of the legal system. In Armenia, for example, TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 337 a television show called "My Rights," in which a government official played the role of a judge hearing cases, became the most popular show on television in its two years of production. Much to his surprise, the government official, formerly a deputy minister of justice, became a national star and was recently appointed as a judge. In Russia, the government set up a network of "Legal Information Centers" in public libraries and other locations in the late 1990s, where the public can access informa- tion on laws and the justice system. In Croatia (and many other transition countries), the courts are adopting an automated case management system that will not only improve efficiency but also produce better statistical data to monitor performance. Countries' judiciaries and ministries of justice throughout the region are establishing Web sites to publicize laws, judicial calendars, and decisions in individual cases. As with other areas of reform, there is still a long way to go, and public informa- tion and transparency remains an area fraught with resistance. Judiciaries were not at all open and transparent in Soviet times, and there remains a concern for confi- dentiality that clouds many judges' views of the issue (and may serve to protect more corrupt or less competent judges). In speaking with judicial leaders in the region, one often hears the view that case decisions should not or need not be made public, either because litigants' privacy needs to be protected or because the public "would not be interested" in most routine decisions. It is also true that even in Western Europe, not all decisions of lower-level courts are necessarily published.11 While privacy rights are a concern, however, problems of accountability and corruption are serious enough in transition economies to justify strong measures to promote transparency. Most privacy concerns can be handled through special rules, such as the use of generic names, such as "John Doe," in lieu of actual names. Judicial Infrastructure and Management Enormous needs for infrastructure faced the courts in the transition economies in the 1990s. Courthouses were typically run-down and dreary places (particularly in loca- tions outside of capital cities), a legacy of the relatively low status and minor role of communist judiciaries. They often shared space in a building with other government agencies or even private businesses or apartments. Courtrooms were small and limited in number. As the number of cases rose with the expansion of market economies in the 1990s, it became increasingly difficult to find premises to hold trials and to accommodate the increasing number of citizens who wanted to observe them. Without sufficient trial space, litigants and judges were sometimes forced to meet in closed offices, raising further suspicions of impropriety. Furthermore, the trial venues that did exist were not well outfitted. They did not give the public a sense of confidence in the independence and impartiality of the system. They often did not have space to accommodate juries where needed12 or to allow defendants to confront accusers or cross-examine witnesses. Indeed, criminal defendants often sat in cages in the middle of the courtrooms, as in communist times--hardly a reflection of the concept of "innocent until proven guilty." The equipment needed to run courts efficiently was also lacking. Few judges had access to computers, and few of the computers that did exist had access to the Internet. Paper-based case files were bulky, difficult to manage, and easy to "lose" 338 | JAMES H. ANDERSON AND CHERYL W. GRAY or tamper with. Courts in far-off locations had difficulty keeping up-to-date on recent legislation or changes in judicial policy. Months could sometimes go by after parliamentary adoption of new legislation before all judges were aware of the changes. For example, until recently Albanian judges had limited or no access to new laws due to a lack of funds needed to provide each judge with a copy of the Official Gazette. The release in early 2006 of the Armenian Legal Information System, which contains all Armenian legislation in a database accessible and searchable from the Internet, has improved the situation in that country. Economic downturns in the 1990s contributed to these problems by severely limiting the resources available to judiciaries or ministries of justice to update their facilities. Few international donors focused on judicial capacity building, and in any case most donors were not allowed to fund building construction or renovation. Yet without access to resources, how could such ill-equipped judiciaries hope to meet the rising demand for their services? Fortunately this situation is now changing. The economic upturns since 2000 have provided more resources to government budgets, and some of those resources are going to judicial systems (both for infrastructure and for increases in judicial salaries, as noted above). The World Bank and other donors are providing substantial funding to upgrade existing courthouses and information technology (IT) infrastructure,13 supplementing substantial renovation programs financed by government budgets. Courts are increasingly installing computer equipment, modern case management software, and procedures and support staff to enhance court management. They are connecting district and higher-level courts together through wide-area networks and developing Web sites for information sharing (as noted above). In Russia, for example, almost all courts now have computers, and a new federally funded program will help support full connectivity among all of the courts in each of the two court systems (the commercial, or "Arbitrazh," courts and the courts of general jurisdiction). The needs in Europe and Central Asia are enormous, however, and there is still a long way to go to equip these judicial systems with adequate infrastructure and IT systems to serve the public efficiently and effectively. Judicial Education and Training Judicial reforms would be incomplete without also addressing judicial education and training. Not only were communist-era judges ill prepared for the kinds of cases that arise in a market economy, but the sheer volume of cases has expanded dramatically, meaning that both more well-trained judges and more efficient ways of handling the caseload are needed. With regard to basic legal education, the demand for places in law schools has expanded tremendously. Many new private law schools have opened to serve this demand, although quality varies widely. The best law graduates typically seek lucra- tive positions in private law firms or international companies, but recent increases in judicial salaries in many transition countries have made the judicial profession more attractive than it was in the 1990s.14 Unfortunately, endemic corruption can be a major problem in higher education, as in other areas in transition economies. Some TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 339 countries--such as Albania and Georgia--have responded to this concern by institut- ing new written entrance examinations to allocate university positions. This does not always make the problem disappear. The 2005 round of testing for entrance to the Tirana law school was itself marred by corruption, as it was discovered by the authorities that the answers had been sold in advance. On the positive side, however, the corruption was detected and announced nationwide, and the exams were read- ministered in full. Most transition countries are expanding their judicial training programs for new and in-situ judges. Judicial academies are being equipped and expanded, often with donor support, and several international groups are also sponsoring independent multicountry judicial training. New opportunities are arising for e-training programs, taking advantage of the increasing computer networking described above. Given the enormous changes in laws and judicial norms, however, the need for effective and timely judicial training typically far outpaces its availability. While more advanced in some new EU members, judicial training systems are still in relative infancy in most transition countries, particularly in South-East Europe and the CIS. Supporting Professions: Lawyers, Notaries, and Bailiffs Judges do not operate in a vacuum, and judicial systems cannot be effective unless the many supporting professions, including attorneys and bailiffs, also function effec- tively. Yet all of these professions were in the same position as the judiciaries at the start of the 1990s: that is, either nonexistent or totally ill prepared for the needs of a market economy. Of these supporting professions, private attorneys have arguably developed the furthest, thanks to strong market incentives and significant investments from abroad. Most transition economies have a large and growing number of law firms, both domestic and foreign, with significant competition among them. Quality is not always assured, however, and prices can be high (in part because bar associations have, as elsewhere, sometimes functioned more as cartels than quality assurers), but overall the profession has grown rapidly in most transition countries. The notary profession--also populated by private attorneys--has similarly flour- ished in some settings, albeit with mixed economic impact.15 In some cases, the mix of complex legislation and the heavy regulatory role of notaries have added to the duration of judicial proceedings, although in other countries the notary process offers a way to circumvent court proceedings altogether. In Poland, for example, parties can proceed directly to execution of a judgment if certain documents are notarized (World Bank 2006). The role of bailiffs is to enforce judicial decisions, and this is a particularly problem- atic area in transition economies. As can be seen in figure 5, only about 40 percent of firms surveyed in the BEEPS believed that courts could enforce judicial decisions. Interestingly, the problem seems to be worse in the new EU members than in the former Soviet Union (and worst of all in South-East Europe), a pattern that could be partially explained by the much larger demand for courts and number of judgments to be enforced in the higher-income transition countries. 340 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 5. Firms' Assessments of Courts' Ability to Enforce Decisions, 2002­05 55 le 50 ab Increase between Level in 2005 2002 and 2005 are ts 45 cour 40 decisions ying sa orce 35 ms Decrease between fir enf of to 2002 and 2005 30 ercentP 25 20 EU EU EU8 Other SEE Northern Southern accession candidate CIS CIS Source: BEEPS 2002, 2005. Note: EU accession includes Bulgaria and Romania; EU candidate is Croatia; EU8 includes Poland, Czech Republic, Slovak Republic, Slovenia, Hungary, Latvia, Lithuania, and Estonia; other SEE includes Albania, FYR Macedonia, Bosnia and Herzegovina, and Serbia and Montenegro; northern CIS includes Belarus, Russian Federation, Ukraine, and Kazakhstan; southern CIS includes Armenia, Azerbaijan, Georgia, Kyrgyz Republic, Tajikistan, Moldova, and Uzbekistan. Regional averages allow each country in the region the same weight. Regulating bailiffs appropriately involves combining incentives for vigorous col- lection with supervision to be sure that even the smallest case receives attention. Some countries have moved toward private incentives for bailiffs, but not always with a governance framework to ensure accountability. Russia adopted new legisla- tion for bailiffs in the late 1990s, giving notaries the right to a 5 percent incentive payment when enforcing judgments. The general view, however, is that this has not been well implemented in practice and that judgments in Russia are still very difficult to enforce. FYR Macedonia has decided to follow an emerging trend in Western Europe by creating a private profession of enforcement agents or bailiffs. The bailiffs will be licensed and regulated by the ministry of justice but will be a private profes- sion working in the market to enforce judicial decisions. Poland continues to have a mixed system. Bailiffs are court officials, with rights and immunities commensurate with public office and with the number of bailiffs fixed by law. In all other respects, however, they operate no differently than private business, funded entirely by a fixed 15 percent of successful collections and hiring staff and outfitting their offices from these proceeds exactly as a private firm would (World Bank 2006). Access to Justice Finally, there has been insufficient progress in promoting access to justice in transi- tion countries. The high cost of both lawyers and notaries are no doubt a significant reason why judicial proceedings are considered by many firms to be unaffordable: most notably, again, firms in South-East Europe (figure 6). Providing legal aid ser- vices is beyond the reach of many public budgets and has not been given significant emphasis by most transition governments. The former socialist countries in Europe TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 341 FIGURE 6. Firms' Assessments of the Affordability of Courts, 2002 and 2005 50 Level in 2005 ts 40 cour le ying 30 sa ordab ms Decrease between aff fir 20 Increase between 2002 and 2005 of are 2002 and 2005 ercentP 10 0 EU EU EU8 Other SEE Northern Southern accession candidate CIS CIS Source: BEEPS 2002, 2005. Note: See figure 5 for country categories. tend to be quite legalistic; indeed, in the 1980s Yugoslavia had more lawyers per capita than any other country in the world. Unlike Asia, Latin America, or Africa, for example, transition countries do not have widely accepted systems of indigenous "customary" legal processes that the poor can turn to for the resolution of disputes, nor are they particularly enthusiastic about alternative methods of dispute resolution such as formal mediation and arbitration. Thus access to justice for the broad swath of the population is likely to grow only slowly, as the economies and the judicial sys- tems continue to grow and develop. Is There a Standard? Comparisons with Selected Nontransition Countries Most transition countries in the Europe and Central Asia region--including new EU members, actual or potential EU candidate countries in South-East Europe, EU "neighbors" such as Ukraine and the south Caucasus, and countries further east-- look toward higher-income West European countries as models for the future. They envision societies based on respect for rule of law and well-functioning judicial sys- tems, but believe there is still a long way to go for them to "catch up" with the West. Yet efficiency, honesty, and affordability are still challenges for judicial systems in Western Europe, as well. For the first time in 2004 and 2005, the BEEPS survey was also conducted in a number of nontransition European countries, including Ireland, Germany (eastern and western), Greece, Portugal, and Spain.16 While this sample of countries is not necessarily representative of all nontransition countries in Europe, comparisons between these two groups of countries is illuminating. Figures 7­10 show comparisons of the evaluations of courts by firms along four dimensions--honesty, quickness, 342 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 7. Firms' Assessments of Courts as Honest and Uncorrupted in 2005-- Transition Countries versus European Comparator Countries Germany Greece Ireland Estonia Spain Turkey Slovenia Hungary Belarus Romania Latvia Armenia Portugal Azerbaijan Tajikistan y Georgia Bulgaria Countr Poland Croatia Kazakhstan Slovak Republic Czech Republic Lithuania Uzbekistan Serbia and Montenegro Russian Federation Albania FYR Macedonia Ukraine Transition countries Bosnia and Herzegovina Kyrgyz Republic Comparator countries Moldova 0 25 50 75 100 Percent of firms saying courts are honest and uncorrupted Source: BEEPS 2005. Note: Comparator countries include Germany, Greece, Ireland, Portugal, Spain, and Turkey. Transition countries and Ireland, Spain, and Turkey were surveyed in 2005. Germany, Greece, and Portugal were surveyed in late 2004. ability to enforce decisions, and affordability--in the six nontransition European countries covered by BEEPS and in the transition countries.17 The most notable differences between transition countries and Western European countries are in perceptions of honesty and fairness. Germany scores much higher than any other country on honesty (figure 7), followed in order by Greece, Ireland, Estonia (the highest-scoring transition country), Spain, and Turkey. In all other coun- tries fewer than 50 percent of firms viewed courts as honest, with Portugal scoring below a number of transition countries. Firms in all countries have major concerns about speed (figure 8). Fewer than half the firms in any country evaluate courts as quick. In Turkey and several transition countries that score slightly better on quickness (Armenia, Azerbaijan, and Tajikistan), demand for judicial services by firms is still relatively small, which may help to explain this outcome (figure 11). Whether from the perspective of firms responding to BEEPS or of lawyers providing assessments for the Doing Business study on the business climate (World Bank 2005), courts appear to be the slowest in the new and TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 343 Figure 8. Firms' Assessments of Courts as Quick in 2005--Transition Countries versus European Comparator Countries Turkey Germany Azerbaijan Belarus Armenia Tajikistan Uzbekistan Greece Georgia Hungary Kazakhstan Romania Moldova Kyrgyz Republic y Albania Lithuania Estonia Countr Spain Ireland Russian Federation FYR Macedonia Poland Ukraine Latvia Bosnia and Herzegovina Bulgaria Slovak Republic Serbia and Montenegro Czech Republic Croatia Transition countries Portugal Comparator countries Slovenia 0 25 50 75 100 Percent of firms saying courts are quick Source: BEEPS 2005. Note: Comparator countries include Germany, Greece, Ireland, Portugal, Spain, and Turkey. Transition countries and Ireland, Spain, and Turkey were surveyed in 2005. Germany, Greece, and Portugal were surveyed in late 2004. prospective EU members in Central, Eastern, and South-East Europe, and the situation may be getting worse rather than better (figure 12). It is critical that these countries unclog and speed up court proceedings through legal reforms to eliminate unnecessary procedures, institutional reforms to create stronger incentives for efficiency, and additional resources to increase judicial capacity where clearly warranted. Transition countries can take some comfort, however, from the fact that Ireland, Spain, and Portugal face similar challenges. Transition countries also fare poorly relative to Greece, Turkey, and Germany in their ability to enforce decisions (figure 9), although Belarus scores better than other transition countries, perhaps reflecting the fact that it is still a centrally controlled economy.18 Ireland and Spain fall behind a number of transition countries on this indicator, and Portugal's scores are among the lowest of all countries surveyed. The one area where transition countries fare relatively well in comparison with Western Europe appears to be affordability, although on average only about one- third of all firms surveyed agreed that courts are affordable (figure 10). There is 344 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 9. Firms' Assessments of Courts as Able to Enforce Decisions in 2005-- Transition Countries versus European Comparator Countries Greece Belarus Turkey Germany Estonia Georgia Spain Slovenia Tajikistan Albania Bulgaria Ireland Slovak Republic Azerbaijan y Kyrgyz Republic Russian Federation Romania Countr Armenia Uzbekistan Kazakhstan Hungary Croatia Lithuania Ukraine Latvia Poland Serbia and Montenegro Portugal Czech Republic Bosnia and Herzegovina Transition countries Moldova Comparator countries FYR Macedonia 0 25 50 75 100 Percent of firms saying courts are able to enforce decisions Source: BEEPS 2005. Note: Comparator countries include Germany, Greece, Ireland, Portugal, Spain, and Turkey. Transition countries and Ireland, Spain, and Turkey were surveyed in 2005. Germany, Greece, and Portugal were surveyed in late 2004. significant variation among countries, with the highest marks (as well as a clear improving trend from 2002 to 2005) for Estonia, Belarus, and Latvia. Again, most Central and South-East European countries--including Bosnia and Herzegovina, the Czech Republic, FYR Macedonia, and Serbia and Montenegro--trailed behind the others and appeared to have deteriorated even further from 2002 to 2005. The two countries where firms considered courts to be least affordable were Ireland and Portugal. It is likely that the reasons for this are rooted as much if not more in the structure and regulation of related professions (which affects, for example, lawyers' fees) as in the extent of demand or capacity in the judicial system as a whole. Finally, variations in assessments by firms in eastern and western Germany pro- vide a glimpse into what happens when institutions are adopted wholesale with plenty of financial and technical support, while also illustrating the tenacious grip of history. Before the transition, the country that most resembled the former German Democratic Republic (GDR) in economic structure was Czechoslovakia, while Slovenia was the closest in per capita income. Figures 13­16 show how the Czech Republic, the TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 345 FIGURE 10. Firms' Assessments of Courts as Affordable in 2005--Transition Countries versus European Comparator Countries Estonia Belarus Latvia Poland Turkey Tajikistan Kazakhstan Bulgaria Germany Azerbaijan Greece Georgia Ukraine Albania y Romania Armenia Russian Federation Countr Kyrgyz Republic Lithuania Croatia Hungary Spain Slovenia Slovak Republic Uzbekistan Moldova Serbia and Montenegro FYR Macedonia Czech Republic Bosnia and Herzegovina Transition countries Portugal Comparator countries Ireland 0 25 50 75 100 Percent of firms saying courts are affordable Source: BEEPS 2005. Note: Comparator countries include Germany, Greece, Ireland, Portugal, Spain, and Turkey. Transition countries and Ireland, Spain, and Turkey were surveyed in 2005. Germany, Greece, and Portugal were surveyed in late 2004. FIGURE 11. Pressure on Court Slows Them Down k 45 Tur quic 40 Aze Ger are 35 Arm Bel ts Taj 30 cour Uzb 25 Geo Gre Hun ying Kaz Rom sa 20 Mol Kyr Est Lit ms 15 Alb Esp fir Ire Lat Rus of BiH 10 Slk Por Cze MacPol UkrSAM Bul Cro Sln 5 ercentP 0 0 10 20 30 40 50 60 70 Percent of firms that have been to court Comparator countries Transition countries Source: BEEPS 2005/2004. Note: Data for transition countries. Esp, Ire, and Tur refer to 2005; data for Ger, Gre, and Por refer to late 2004. 346 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 12. Firms' Assessments of the Courts as "Quick," 2002­05 30 Increase between k 2002 and 2005 25 quic Decrease between are ts 2002 and 2005 20 cour ying 15 sa ms fir 10 of Level in 2005 ercentP 5 0 EU EU EU8 Other SEE Northern Southern accession candidate CIS CIS Source: BEEPS 2002, 2005. Note: See figure 5 for country categories. FIGURE 13. Firms' Assessments of Court Honesty--Former German Democratic Republic, Former Federal Republic of Germany, and Transition Comparators 100 2002 2005 75 saying honest firms 50 are of courts 25 Percent 0 Czech Slovak Slovenia former former Rep. Rep. GDR FRG Source: BEEPS 2002, 2004, 2005. Note: Germany was surveyed in late 2004. The Czech Republic, the Slovak Republic, and Slovenia were surveyed in spring 2005. Slovak Republic, and Slovenia compare to both eastern and western Germany today. Although firms in eastern Germany continue to provide worse assessments than those in western German along every dimension of court performance except ability to enforce decisions, their assessments are nevertheless better than those provided by firms in the other transition countries. German unification clearly had a strong positive effect on institutions in eastern Germany, but the influence of history lingers even after 15 years. TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 347 FIGURE 14. Firms' Assessments of Court Speed--Former German Democratic Republic, Former Federal Republic of Germany, and Transition Comparators 100 2002 2005 75 saying quick firms are 50 of courts 25 Percent 0 Czech Slovak Slovenia former former Rep. Rep. GDR FRG Source: BEEPS 2002, 2004, 2005. Note: Germany was surveyed in late 2004. The Czech Republic, the Slovak Republic, and Slovenia were surveyed in spring 2005. FIGURE 15. Firms' Assessments of Court Affordability--Former German Democratic Republic, Former Federal Republic of Germany, and Transition Comparators 100 2002 2005 75 saying affordable firms 50 of are courts 25 Percent 0 Czech Slovak Slovenia former former Rep. Rep. GDR FRG Source: BEEPS 2002, 2004, 2005. Note: Germany was surveyed in late 2004. The Czech Republic, the Slovak Republic, and Slovenia were surveyed in spring 2005. Conclusion What does all this tell us? First, judicial reform is a critical challenge for most tran- sition countries. The majority of these countries have made progress in establishing independence in their judiciaries, but accountability, transparency, and efficiency have lagged behind. Many transition countries need to focus now on strengthening the fairness and honesty of their courts--which requires broad actions along many 348 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE 16. Firms' Assessments of Courts' Ability to Enforce Decisions--Former German Democratic Republic, Former Federal Republic of Germany, and Transition Comparators 100 2002 2005 courts 75 decisions saying 50 firms enforce of to able 25 Percent are 0 Czech Slovak Slovenia former former Rep. Rep. GDR FRG Source: BEEPS 2002, 2004, 2005. Note: Germany was surveyed in late 2004. The Czech Republic, the Slovak Republic, and Slovenia were surveyed in spring 2005. fronts to select the right judges and support staff; train, remunerate, and evaluate them adequately; and provide infrastructure and IT systems to promote efficiency and transparency. Countries at the very early stages of transition may not feel the pinch, but as economic reforms proceed and private business grows, the public's demand for more capable and efficient judiciaries is likely to become stronger and stronger, creating ever greater pressures for reform (as are now evident in countries such as Bulgaria and Romania). More generally, transition countries share many of these same priorities and con- cerns as other countries, whether developed or developing. Strengthening judicial accountability is also a critical challenge for some OECD countries, not to mention most countries in the developing world. And even those more advanced countries in which citizens trust the honesty and competence of their judges must grapple with problems of judicial delay, affordability, and ability to enforce decisions. Justice sys- tems that are so slow or expensive as to be out of reach or impractical for most cit- izens to use, or that cannot enforce judges' decisions, are unlikely to ensure rule of law. Judicial strengthening may not be perceived by businesses as the highest prior- ity in all societies (figure 17), but it will be a continuing challenge almost everywhere for years to come. Annex: The Business Environment and Enterprise Performance Survey The EBRD-World Bank Business Environment and Enterprise Performance Survey (BEEPS),developedjointlybytheEuropeanBankforReconstructionandDevelopment and the World Bank (2005), is a survey of managers and owners of more than 20,000 firms across the countries of central and eastern Europe, the former Soviet FIGURE 17. Problems Doing Business, 2005 a. EU8 b. European comparators 4.0 4.0 TRANSFORMING obstacle 3.5 obstacle 3.5 3.0 3.0 4=major 4=major 2.5 2.5 2.0 2.0 obstacle obstacle JUDICIAL an 1.5 an 1.5 1=not1.0 1=not1.0 Taxes Finance Crime Corruption Labor Customs Judiciary Business Macro- Taxes Finance Crime Corruption Labor Customs Judiciary Business Macro- regulations license economic regulations license economic SYSTEMS stability stability c. EU accession, EU candidate, and other SEE d. CIS 4.0 4.0 IN obstacle 3.5 obstacle 3.5 EUROPE 3.0 3.0 4=major 4=major 2.5 2.5 AND 2.0 2.0 obstacle obstacle CENTRAL an 1.5 an 1.5 1=not 1.0 1=not 1.0 Taxes Finance Crime Corruption Labor Customs Judiciary Business Macro- Taxes Finance Crime Corruption Labor Customs Judiciary Business Macro- regulations license economic regulations license economic ASIA stability stability Source: BEEPS 2005. Note: EU8 includes the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, the Slovak Republic, and Slovenia. European comparators include Germany, Greece, Ireland, | Portugal, Spain, and Turkey. EU accession, EU candidate, and other SEE include Albania, Bosnia and Herzegovina, Bulgaria, Croatia, the former Yugaslav Republic of Macedonia, 349 Romania, and Serbia and Montenegro. CIS includes Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyz Republic, Moldova, Tajikistan, Russia, Ukraine, and Uzbekistan. These groupings are for convenience. Croatia and Turkey are both EU candidate countries, but Turkey is included with the European comparators to highlight differences between former socialist and other countries. All transition countries and Ireland, Spain, and Turkey were surveyed in 2005. Germany, Greece, and Portugal were surveyed in late 2004. 350 | JAMES H. ANDERSON AND CHERYL W. GRAY Union, and Turkey. (It has not been possible to implement this survey in Turkmenistan.) The survey has been carried out in three rounds: 1999, 2002, and 2005. The BEEPS is designed to examine the quality of the business environment as deter- mined by a wide range of interactions between firms and the state, including in the following areas: problems doing business, unofficial payments and corruption, crime, regulations and red tape, customs and taxes, labor issues, firm financing, legal and judicial issues, and infrastructure. All questionnaires in every country in every round of the BEEPS were implemented the same way, through face-to-face interviews. The BEEPS sample was drawn from the universe of firms in a broad range of economic activities. In each country, the sectoral composition of the sample in terms of manufacturing (including agro-processing) versus services (including commerce) was determined by their relative contribution to GDP. The BEEPS sampling approach was the same in all three rounds of the BEEPS, and was implemented nationwide in all countries. The BEEPS sample in all three years included quotas related to size, ownership, export orientation, and geographical location to ensure sufficient numbers of firms to conduct analysis of firms with certain characteristics. From a practical perspective, the quotas that had an actual impact on the sample, compared to what would have arisen from a wholly random sample, were the ones for state ownership, for foreign ownership, and for large size. As ownership and size are highly correlated, the quo- tas ultimately affected a relatively small proportion of the sample. · The 2005 round of the BEEPS consisted of 9,655 interviews.19 Sample sizes ranged from 200 in smaller countries to about 600 in Russia. The survey was carried out in 27 countries in the World Bank's Europe and Central Asia Region (ECA). This group of countries includes Turkey and all of the former socialist countries of Europe and the former Soviet Union, except for Turkmenistan. · The BEEPS in Comparator Countries consisted of 4,453 firms in seven countries: Germany, Greece, the Republic of Korea, Portugal, and Vietnam were surveyed in late 2004, and Ireland and Spain were surveyed in 2005. Samples ranged from 500 to 1,197 per country. · The 2002 round of the BEEPS consisted of 6,667 interviews, covering a range of 170 to 514 firms per country. The survey was carried out in every ECA country except Turkmenistan. · The 1999 round of the BEEPS consisted of 4,104 interviews, covering a range of 112 to 552 firms per country. The survey was carried out in every ECA country except Serbia and Montenegro, Tajikistan, and Turkmenistan. The BEEPS is unique as a tool that allows monitoring of how firms experience and perceive their environments over a large number of countries and across time. The BEEPS is an original source of data that offers several useful features not found in aggregate indicators, including a common yardstick for country comparisons; the possibility to examine changes over time; the possibility to examine changes in more narrowly defined areas, such as the speed, affordability, credibility, and honesty of courts, as opposed to a generic "rule of law." TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 351 The BEEPS also provides a strong complement to the World Bank Group's Doing Business (DB) indicators. The two use different methodologies and answer related, but different, questions. Most of the DB indicators are generated by asking lawyers, accountants, and other professionals in each country about the details of the laws, rules, and procedures that govern various aspects of doing business. In order to compare apples to apples, the DB methodology presents hypothetical cases or situations that are the same for each country. The BEEPS, in contrast, asks 200 to 600 firms in each country questions about their business environment and their interactions with the state. The samples are chosen in a uniform way in each country, with sector composition divided according to contribution to GDP. Whereas DB can be thought of as a compilation of indicators about the content of various government policies, rules, and procedures, the BEEPS can be thought of as a compilation of indicators about what firms are saying about the ways that these government policies, rules, and procedures affect their everyday business. The DB indicators and BEEPS usually point in the same direction. Doing Business in 2006--Creating Jobs (World Bank 2005) highlighted Europe and Central Asia as the leading reformer in 2005, and the BEEPS 2005 results also suggest improvement from 2002 to 2005 in many areas. In cases where the two diverge, there are often valid explanations. Firms may have found ways to work around problematic regulations so that they are less burdensome; conversely, the formal rules and procedures may appear benign, while nontransparent implementation may cause firms considerable difficulty. In addition, improvements captured in the Doing Business indicators may take time to be recognized by the business community. For example, reductions in minimum capital requirements to start a company will not help firms that already exist. The DB indicators and assessments by firms in BEEPS tell the same broad story about judicial systems in transition countries. The two DB indicators that are most closely related to the performance of the courts are those for enforcing contracts FIGURE A1. Contract Enforcement and Court Speed 3.5 Ger Aze Tur 3.0 Bel Arm speed Gre court Uzb Geo Kaz of 2.5 Alb Rom Lit Hun Kyr Est Mol Esp Lat Ire Ukr Rus Slk BiH 2.0 SAM Mac Pol Sln Assessments Cze Bul Por Cro 1.5 5.0 5.5 6.0 6.5 7.0 Days to enforce a contract (log) Source: The assessment of court speed measure is from BEEPS (2005); the measure of days to enforce a contract is the log of the Doing Business indicator (2005). Note: See figure 1 for country names. 352 | JAMES H. ANDERSON AND CHERYL W. GRAY FIGURE A2. Property Registration and Court Speed 3.5 Ger Tur Aze 3.0 Bel Arm speed Gre court Uzb Geo of Kaz 2.5 Alb Rom Hun Lit Kyr MolEst Esp Rus Slk IreLat Ukr SAM 2.0 BiH Mac Pol Sln Assessments Bul Cze Por Cro 1.5 0 2 4 6 8 Days to register property (log) Source: The assessment of court speed measure is from BEEPS (2005); The measure of days to register property is the log of the Doing Business indicator (2005). Note: See figure 1 for country names. through the courts and for registering property, a function handled by the courts in many transition countries. The DB indicators include assessments by a small number of lawyers, and in some cases judges, on how long each of these processes may take for a given hypothetical situation. Both are significantly correlated with BEEPS meas- ures of court speed (see figures A-1 and A-2). For further information on the BEEPS, see www.worldbank.org/eca/econ. Notes 1. For a discussion of the role of institutions in economic development, see North (1990), Bardhan (1997), Williamson (1985), and World Bank (1996, 2002, 2004). For a collection of empirical essays on the use of law, see Murrell (2001). For a recent compendium of articles on law and economic development, see Schafer and Raja (2006). 2. This is of course not an exhaustive treatment of the many interesting issues surrounding legal and judicial reform in transition economies. For example, the paper does not address the substantive content of laws and regulations and how it has changed during transition; the structure of court systems, including jurisdictional issues or the role of specialized courts; or arbitration, mediation, and other nonjudicial means of enforcing contracts and resolving disputes, which are important in every society. The survey, on which much of the paper is based, focused on private commercial law--interactions between firms and courts--rather than administrative, constitutional, civil, or criminal law. 3. "Legal extensiveness" is intended to measure the content of laws and regulations and the extent to which the legal framework addresses critical issues in a market economy, while "legal effectiveness" is intended to capture how those laws and regulations are actually being implemented in practice in the country concerned. Measures of legal extensiveness and legal effectiveness were constructed based on responses to the EBRD Legal Indicator Survey, available at http://www.ebrd.com/country/sector/law/about/assess/main.htm. TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 353 4. The dotted lines are clearly arbitrary and are used for illustrative purposes only. 5. Many of the figures in this paper refer to changes in indicators between the BEEPS rounds in 2002 and 2005. For country-specific analysis of the changes from 1999 to 2002, see Anderson, Bernstein, and Gray (2005). All figures in this paper depict simple averages of nonmissing observations. Further information on the BEEPS is available in the annex. 6. While the findings for Georgia are positive, it should be noted that they come in the context of generalized improvements with regard to corruption in the country. Indeed, firms' assessments of corruption in other sectors improved even more. The Transparency International (TI) Global Corruption Barometer 2005 (Transparency International 2005) suggests that citizens' assessments are similar. Although perceptions of corruption in the judiciary have not changed much in the TI survey compared to one year earlier, perceptions of corruption in other areas (police, tax, and customs) have improved markedly, shifting the judiciary to the top of the list in terms of citizens' perceptions of corruption. 7. For both of these figures, the change between 2002 and 2005 is significant for firms that have not used courts, but not for those that have. All of the differences between court users and nonusers in figures 3 and 4 are significant at the 1 percent level. 8. There may also be an issue of selection bias, if firms with a better perception of courts are more likely to use them (which appears to be the case in some but not all countries). 9. In Russia the level of salaries is not differentiated across localities, and thus it is still considered to be too low in Moscow and St. Petersburg, where the cost of living is much higher than in surrounding regions. 10. In some countries the examinations are still oral and thus thought by some observers to be open to manipulation. 11. This reflects in part the fact that cases are not "law" in civil law systems but are merely applications of law. All court decisions are published in common law systems in part because these decisions take the status of law and are binding as precedent. 12. Juries are not a common element of civil law criminal systems but have been adopted in some countries, including Russia. 13. Recent World Bank loans in Albania, Armenia, Azerbaijan, Croatia, Georgia, FYR Mace- donia, Romania, and Russia and loans under preparation in Poland and Ukraine, devote a share of their resources to upgrading court houses and/or providing computers, case management software, and sound recording or other equipment (always accompanied by resources to enhance capacity, accountability, and transparency in other ways). 14. In civil law systems, a judicial career is typically chosen right out of law school and normally begins with a few years of internship. In common law systems, in contrast, lawyers typically enter judicial positions mid-career, after decades of law practice in the private sector or in government. 15. Many economists question the highly regulated and interventionist role of notaries in some European settings, such as Germany. Some transition economies, such as Russia, are moving to reduce this role. 16. The Republic of Korea and Vietnam were also surveyed but are not discussed in this paper. Germany, Greece, and Portugal were surveyed in late 2004. Ireland, Spain, and Turkey were surveyed in 2005. 17. Fairness was also measured by the BEEPS, but is not shown separately here because firms' assessments of fairness are highly correlated with their assessments of honesty. 18. One of the first major studies to employ the BEEPS data, the World Bank's 2000 report Anticorruption in Transition--A Contribution to the Policy Debate, describes the chal- lenges in interpreting data from a survey oriented for private business in countries where the private sector is in its infancy, and those observations continue to hold six years later. See World Bank (2000); Gray, Hellman, and Ryterman (2004); and Anderson and Gray (2006). 354 | JAMES H. ANDERSON AND CHERYL W. GRAY 19. The 2005 round of the BEEPS included 1,715 firms (included in the 9,655 number cited in the text) in a special "manufacturing overlay" in seven countries. These observations were not used in this study so that the sector proportions would be comparable across the whole sample. References Anderson, James H., and Cheryl W. Gray. 2006. Anticorruption in Transition 3­­Who Is Succeeding . . . And Why? Washington, DC: World Bank. Anderson, James H., David S. Bernstein, and Cheryl W. Gray. 2005. Judicial Systems in Transition Economies: Assessing the Past, Looking to the Future. Washington, DC: World Bank. Bardhan, Pranab. 1997. The Role of Governance in Economic Development: A Political Economy. Paris: Organisation for Economic Co-operation and Development (OECD). EBRD (European Bank for Reconstruction and Development). Various years. Legal Indicator Survey. http://www.ebrd.com/country/sector/law/about/assess/main.htm. ------. 2005. Transition Report 2005--Business in Transition. London: EBRD. EBRD (European Bank for Reconstruction and Development) and the World Bank. 2005. EBRD-World Bank Business Environment and Enterprise Performance Survey (BEEPS). http://www.worldbank.org/eca/governance/. Gray, Cheryl W., Joel Hellman, and Randi Ryterman. 2004. Anticorruption in Transition 2: Corruption in Enterprise-State Interactions in Europe and Central Asia 1999­2002. Washington, DC: World Bank. Kritzer, Herbert M., and John Voelker. 1998. "Familiarity Breeds Respect: How Wisconsin Citizens View their Courts." Judicature 82: 58­64. Murrell, Peter, ed. 2001. Assessing the Value of Law in Transition Countries. Ann Arbor: University of Michigan Press. North, Douglas. 1990. Institutions, Institutional Change, and Economic Performance. The Political Economy of Institutions and Decisions Series. New York and Melbourne: Cambridge University Press. Rose, Richard, and Christian Haerpfer. 1994. "New Democracies Barometer III: Learning from What is Happening." Studies in Public Policy Number 230. Centre for the Study of Public Policy, University of Strathclyde. ------. 1996. "New Democracies Barometer IV: A 10-Nation Survey." Studies in Public Policy 262, Centre for the Study of Public Policy, University of Strathclyde. ------. 1998. "New Democracies Barometer V: A 12-Nation Survey." Studies in Public Policy 306, Centre for the Study of Public Policy, University of Strathclyde. Schafer, Hans-Bernd, and Angara V. Raja, eds. 2006. Law and Economic Development. Northampton: Edward Elgar. Transparency International. 2005. Report on the Transparency International Global Corrup- tion Barometer 2005. Berlin: Transparency International. Williamson, Oliver. 1985. The Economic Institutions of Capitalism. New York: Macmillan. World Bank. 1996. From Plan to Market. World Development Report. Washington, DC: World Bank. ------. 2000. Anticorruption in Transition--A Contribution to the Policy Debate. Washington, DC: World Bank. TRANSFORMING JUDICIAL SYSTEMS IN EUROPE AND CENTRAL ASIA | 355 ------. 2002. Building Institutions for Markets. World Development Report. Washington, DC: World Bank. ------. 2004. Economic Growth in the 1990s: Learning from a Decade of Reform. Washington, DC: World Bank. ------. 2005. Doing Business in 2006--Creating Jobs. Washington, DC: World Bank and International Finance Corporation. http://www.doingbusiness.org. ------. 2006. Poland: Legal Barriers to Contract Enforcement. Washington, DC: World Bank. Comment on "Judicial Reform in Developing Economies: Constraints and Opportunities," by Matthew C. Stephenson, and "Transforming Judicial Systems in Europe and Central Asia," by James H. Anderson and Cheryl W. Gray UGO PANIZZA Comment on "Judicial Reform in Developing Economies: Opportunities and Constraints" The objective of Matthew Stephenson's paper is to highlight the main problems that arise when countries try to devise and implement judicial reforms. On the one hand, this paper is refreshing because it cautions against the "just do it" approach to reforms and suggests that countries should stop and think before implementing deep reforms of their judicial systems (and, probably, in any other field). On the other hand, this paper could give ammunition to those who oppose reforms because they extract rents from a potentially inefficient status quo. The paper highlights three possible problems with reforms of the judicial system: resource constraints, incentive compatibility, and second best optimality. As Stephenson makes a good case for the importance of these problems, I will play the role of the devil's advocate and try to deconstruct some of his arguments and suggest that, even with the caveats mentioned by Stephenson, reforming a poorly working judicial system is often a worthy endeavor. Before discussing these points in detail, I would like to mention that it would be useful to have a better definition of judicial reform. This definition should at least specify the type of the reform under study and the type of court that is affected by the reform process (from more details, see my discussion of the paper by James Anderson and Cheryl Gray in the second half of this comment). Ugo Panizza is senior economist in the Debt and Development Finance Branch of the Division of Globalization and Development Strategies at UNCTAD. Annual World Bank Conference on Development Economics © 2007 The International Bank for Reconstruction and Development / The World Bank 357 358 | UGO PANIZZA Resource Constraint All economists are well aware of the importance of the budget constraint and of the fact that, when implementing a given reform, policy makers need to make sure that the marginal dollar spent in a given activity must have a return that is not smaller than the return of any alternative public expenditure program and of the welfare cost of the distortion created by the mechanism used to finance the reform (either current or future taxation). However, this reasoning assumes that countries are on their efficiency frontier and hence that it is impossible to reform and improve a country's legal system without incurring more expenditure. A testable implication of this assumption is that one should observe a positive relationship between judicial expenditure and judicial efficiency. Figure 1 plots the result of a regression of the efficiency of the judiciary (the BERI index of judicial efficiency) over total public expenditure in the judicial system (measured in PPP U.S. dollar per 1 million inhabitants) and shows that there seems to be no relationship between these two variables (the regression's coefficient is basically zero and so is the regression's R2). As the relationship between these two variables may be nonlinear, I also tried to take the log of justice expenditure. Figure 2 shows a positive relationship between this variable and judicial efficiency, but the coefficient is small and far from being statisti- cally significant (the t-statistic is 0.4) and the regression's R2 remains close to zero. As a last experiment, I regressed judicial efficiency on the ratio between justice expenditure and total public expenditure (see figure 3). Again, I find no significant relationship between these two variables; if anything, the relationship is negative. While these are very crude exercises that use a very small set of countries (these were the only developing countries for which I could find data) and do not control for country characteristics that may affect judicial efficiency, they all yield the consistent message that there seem to be no relationship between public expenditure in justice FIGURE 1. Efficiency of the Judiciary and Judicial Expenditure 3 MYS X)| 2 TTO judiciary 1 CHL COL the LKA PER of 0 ECU KOR ZAF MEX NIC ARG BRA ­1 e(efficiency PHL GHA ­2 ­20 ­10 0 10 20 30 e(total judicial expenditure) coef = .00103082, se = .02281707, t = .05 Source: Author's calculations. Note: Judicial expenditure is total judicial expenditure per 1,000,0000 inhabitants. COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 359 FIGURE 2. Efficiency of the Judiciary and Judicial Expenditure (log specification) 3 MYS X)| 2 TTO judiciary 1 CHL COL the LKA PER of 0 NIC ECU KOR ZAF MEX ARG BRA ­1 e(efficiency PHL GHA ­2 ­2 ­1 0 1 2 e(log of total judicial expenditure | X) coef = .11600096, se = .31780314, t = .37 Source: Author's calculations. Note: Judicial expenditure is the log of total judicial expenditure per 1,000,0000 inhabitants. FIGURE 3. Efficiency of the Judiciary and Share of Judicial Expenditure over Total Public Expenditure 3 MYS X)| 2 TTO judiciary 1 CHL COL the LKA PER of 0 ECU KOR ZAF MEX ARG NIC BRA ­1 e(efficiency PHL GHA ­2 ­.01 0 .01 .02 e(share of judicial expenditure/total public expenditure | X) coef = ­15.757422, se = 37.215183, t = ­.42 Source: Author's calculations. and efficiency of the judiciary.1 This indicates that in most countries the resource constraint is not really binding and that, at the margin, countries could improve the working of their judicial systems without dedicating more resources to this sector. Incentive Compatibility The paper discusses incentive compatibility for three types of actors: private parties; thestate,politicians,orboth; andjudges. Itarguesthatprivatepartiesmaynotbeinter- ested in using the judicial system for two types of reasons: the judiciary system does not work well, and there is another distortion in the system that discourages private parties 360 | UGO PANIZZA fromusingcourts(forinstance,taxevasion,asinPistor's(1996)example.Iamlesscon- vincedbythe"cultural"reason).Arethesegoodreasonsfornottryingtoreformthesys- tem? In the first case, a reform is clearly optimal because an improvement of the judicial system will induce private actors to make a greater use of the system. In the second case, a judicial reform that improves the court system may have a small impact but could still be Pareto improving (especially if the cost of the reform is low).2 With respect to politicians, it is certainly true that several political systems may not give politicians the incentive to have a well-working judiciary and, more generally, to implement policies aimed at increasing the welfare of the country's citizens. However, this is a deeper problem; politicians that do not have the right incentives are likely to implement bad public policies in all fields (see IDB 2005). This is a fundamental issue, but it goes well beyond the scope of this paper. Finally, if judges do not have the right incentives, there is clearly a problem with the working of the judicial system: yet one more reason to embark in a reform program. Second Best At the risk of trivializing Stephenson's argument (and second best theory), one can summarize the point made in this part of the paper as follows: "If the law is bad, then it is optimal to have a bad judicial system and judges that do not apply the law." While this statement looks trivially true, I have three concerns with it. The first has to do with endogeneity and dynamics, the second has to do with uncertainty, and the third concerns the evaluation of the "optimality" of the law. Endogeneity and dynamics My first concern relates to the possibility of a feedback between distortions and the quality of the judicial system. Consider, for instance, the example of complex and superior legislation that becomes inferior in the presence of a bad judicial system. What if judges are bad because they cannot learn by working on more complex legal cases? Keeping the law simple might be the static second best, but in a dynamic sett- ing it would end up in a vicious circle in which there is no improvement in judicial capacity and hence no ability to adopt more efficient laws. Clearly one should not jump from a very simple legal system to a very complex one, but one could adopt gradual reforms and generate a positive feedback between the complexity of the legal system and the ability of judges. Uncertainty My second concern is that a bad law applied in a consistent fashion may be superior to the uncertainty generated by a system with a poor working judiciary and an erratic application of the law. In this sense, I tend to disagree with the statement that, under certain conditions, corruption can be optimal. How do we know what a good law is? My final concern has to do with the definition of "good law." In democracies, laws reflect a series of compromises aimed at maintaining some sort of political equi- librium, which should reflect society's preference; with the exception of a few cases, COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 361 it is very hard to judge whether a law is purely good or bad. This is especially the case if the person giving advice is a member of a foreign development agency with limited knowledge of the local situation, who may not realize that a law that looks bad from the outside is just a second best response to yet another distortion.3 Finally, it is unlikely that all the laws in a given country are bad. It is hard to think that a bad judiciary would enforce good laws and not enforce bad laws (if anything, this should be the characteristics of a good judiciary). Conclusions It is worth concluding by restating that the objective of my discussion was to be the devil's advocate, and in this role I sometimes forced Stephenson's arguments. It is clear to me that Stephenson's objective is not to discourage countries from embarking on judicial reforms, but only to say: "Be careful when you do it." My point is that the rule of law is a necessary condition for economic development. My advice to a country with a poorly working judicial system that would like to embark on a reform process is: "Be careful. Do not imitate others, but make sure to find something that is appro- priate for you. But definitely do it!" Comment on "Transforming Judicial Systems in Europe and Central Asia" The objective of James Anderson and Cheryl Gray's paper is to address three main issues: the types and sequencing of judicial reforms that are necessary for a successful transition from socialism to a market-based economy; the state of judicial reforms in the Europe and Central Asia region and an analysis of the factors that explain the extent of progress to date; and a cross-country comparison of the way in which firms evaluate the various judicial systems. The authors have a deep and detailed knowledge of the judicial systems of the various countries covered in the paper, but it is extremely difficult to address these three points in a single paper and the reader ends up being overwhelmed by the amount of information included. Hence the richness of the paper also ends up being its main weakness. In fact, the paper provides several interesting intuitions and facts but it cannot develop any of these themes in greater detail, as it tries to compress so much information into a relatively small number of pages. On the basis of these considerations, rather than providing specific detailed com- ments and minor criticism to Anderson and Gray's paper, I will try to comment on their work by producing the outline of the paper I would like to write if I were asked to rewrite their paper. Clearly, my "shadow paper" is a rhetorical device that has the benefit of hindsight because it internalizes what I learned by reading Anderson and Gray's work.4 What would be the research question of a shadow paper inspired by Anderson and Gray? There are three things I found particularly interesting. The first is the tension between independence and accountability. While Anderson and Gray are not the first to raise this issue (see, for instance, Gloppen, Gargarella, and Skaar 2004), 362 | UGO PANIZZA they make a convincing case that transition countries were more successful in guaranteeing independence than in making judges accountable, and that this is an obstacle to economic development. The second is the remark that judicial corruption can be fought only by providing both carrots (higher wages) and sticks (more severe punishments for corrupted judges). Again, this is not a new idea (in Panizza 2001, I survey the literature on the relationship between public sector wages and public sector performance), but there are many people in the development community who still think that higher public sector pay is the solution to corruption. Hence it would be useful to have a research piece offering clear evidence that higher pay for judges is not a sufficient condition to combat judicial corruption. The third is the richness of the survey (BEEPS) used in the paper. Given that there is extensive work on the characteristics of judicial reforms in tran- sition countries (most of it done by the authors; see, for instance, Anderson, Bernstein, and Gray 2005) and that the innovation of Anderson and Gray's paper is in the use of BEEPS data, I will focus the outline of my shadow paper on this latter point. The Shadow Paper My shadow paper would consist of three parts. The first part would include a precise definition of judicial reform and use a taxonomy to classify judicial reforms imple- mented in transition countries. This section would also try to formulate a series of hypotheses as to which actors would likely be the main beneficiaries of the status quo or of the reform process. The second part would be very close to Anderson and Gray's paper and provide a brief description of cross-country data based on BEEPS. The third and last part would describe the data using firm-level observations and use firm-level data to test some of the hypotheses formulated in the first part of the paper. Part I: Defining Judicial Reforms and Formulating Hypotheses The term judicial reform is rather vague. It would be good to have a framework to classify the various types of reforms. Such a framework could classify reforms along two dimensions: the institution (or part of the judicial system) that is object of the reform process, and the type of reform to be implemented. With respect to the first dimension, three types of courts are generally observed: the supreme (or constitutional) court; penal courts; and civil courts. Clearly, there are interactions between the three types of courts, but it is fair to think that reforms of the supreme court are much more linked with the democratization process, and reforms of the civil courts are much more linked with the business environment; penal courts fall somewhere in between.5 From reading Anderson and Gray's paper, it is never clear whether their analysis is focused on a specific type of court or on the judicial system in general. Given the availability of data on firms' attitudes toward the judiciary, I would definitely focus on reforms of civil courts. With respect to the second dimension, I would use a classification of the various reforms based on one of the taxonomies suggested in the literature. In particular, I would probably adopt the taxonomy proposed by Thomas Carothers (1998). This taxonomy divides judicial reforms into three types. Type I reforms focus on the COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 363 TABLE 1. A Matrix of Judicial Reforms Objective of the reform Type Supreme court Penal courts Civil courts Type of reform Type I Type II Type III Source: Author, based on Carothers (1998). ongoing process of cumulative changes in the law. These reforms include important modifications to legislations or procedural codes. Type II reforms focus on the insti- tutions that interpret and enforce the law. They involve efforts to strengthen the courts and various law enforcement agencies. Type III reforms are those that focus on the independency of the judiciary. They include changes to the process of nomination, promotion, tenure, and evaluation of judges. They also have to do with the budgetary autonomy of the judiciary.6 I would use the following matrix to classify judicial reforms in transition countries.7 As a second step, I would describe both the status quo and the reform process and try to formulate a set of hypotheses as to what types of firm benefit or suffer from the status quo or from the reform process. For instance, one could divide firms according to size (large, medium, and small), sector of operation (manufacturing, financial servi- ces, nonfinancial services, construction, agriculture), location (rural versus urban, close to or far away from the capital), and need for financial resources (Rajan and Zingales 1998 provide a useful classification). These hypotheses could then be tested in part 3 of the paper. Part II: Cross-Country Analysis In the second part of the paper, I would follow Anderson and Gray and describe the data at the country level. However, besides using simple averages, I would also look at second moments to get an idea of the dispersion of opinion and try to correct for firm characteristics by running the following regression: Fi,j Xi,j j i,j ' (1) where Fi,j is the assessment of firm i, located in country j, Xi is a matrix of firm char- ,j acteristics, is a vector of parameters, and j is a set of country fixed effects. Then I would use the fixed effects to compare countries. The advantage of this methodology is that it corrects for the fact that the characteristics of the firms included in the survey may vary across countries. As a next step, I would look at the correlation between these country-level averages (both the simple averages and the ones obtained with the correction outlined above) and country characteristics (GDP per capita, level of democracy, corruption, rule of law, index of competitiveness, and the like). While this is not very different from what Anderson and Gray do, I do not fully agree with their interpretation of some of these correlations. 364 | UGO PANIZZA Take, for instance, their figure 1, where they assume that the log of GDP per capita is a good proxy of the capacity of delivering judicial services. This statement requires three strong assumptions: that all countries have the same capacity to tax; that the level of GDP does not depend on the quality of judicial services; and that improving judicial services requires higher expenditure. These are problematic assumptions. With respect to the first assumption, it is well known that, when compared with industrial countries, developing countries tend to have much smaller public sectors (table 2) and that this probably due to an inferior capacity to tax. The second assumption is even more problematic. In fact, the authors are the first to recognize that a well-working judiciary is a necessary condition for economic development. With respect to the third assumption, I am not sure that budget restrictions are a key obstacle to having a well-working judiciary. It is my impression that most countries are far away from the efficiency frontier and hence they could substantially improve their judiciary without requiring a larger budget (for anecdotal evidence, see my discussion on Matthew Stephenson's paper, in the first part of this comment). I also have some problems with the interpretation of the y-axis of figure 1 in Anderson and Gray's paper. While it is trivially true that demand of judicial services can be proxied by the number of firms that use the court, it must also be true that this depends by how well courts work. More firms will use courts in countries with courts that work well and fewer firms will use courts in countries with courts that do not work well. As a consequence, both the x- and y-axis of figure 1 are partly driven by how well courts work. Hence it is not surprising that one finds a strong positive rela- tionship between these two variables. As I said before, I still think that reporting this correlation is valuable, but one must be careful in assigning causal interpretations. Part III: Exploiting Firm-level Data BEEPS data are a gold mine waiting to be exploited (Recanatini, Prati, and Tabellini 2005 is an example of what one can do with this sort of data). As a first pass, I would perform three exercises. As a first exercise, I would re-run equation (1) country by country (of course without the country fixed effects) and compare the vector of parameters, , across countries and across different types of assessments within a country. By comparing across countries, one could answer questions of the following kind: do large (rural, sector XYZ, . . .) firms in country j have a better assessment (relative to small, urban, sector WTF, . . .) of the judicial system than similar firms in country z? By comparing within countries, one could answer questions of the following kind: do firms with a TABLE 2. Size of the Public Sector, 1990­2004 No. of Country group Mean Median countries No. of obs. Total revenues/GDP Industrial 30.24 31.72 24 199 Developing 21.82 20.86 119 1,101 Total expenditure/GDP Industrial 33.59 34.90 23 195 Developing 25.45 24.80 118 1,096 Source: IMF International Financial Statistics. COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 365 given set of characteristics tend to give a better assessment of a given aspect of the judicial sector but a worse assessment of another aspect? As a second exercise, I would compare the results described above with the hypothesis discussed in part I.8 This would allow me to check whether judicial sys- tems that seem to favor, say, large firms receive a positive assessment by large firms (relative to small firms). While the previous two exercises are static analyses (focusing on the status quo before or after reform), my third exercise would exploit the panel dimension of BEEPS. Again, I would use the hypotheses formulated in part I and check whether reforms that one would expect to favor a specific type of firm have a positive effect on the assessment given by this type of firm. Conclusions Anderson and Gray do an excellent job of describing and comparing the main charac- teristics of the judicial systems in transition countries and in describing the main challenges for the reform agenda. The next step is to fully exploit the potential of the firm-level information contained in the BEEPS data. The purpose of my shadow paper is to discuss a possible way to do this. Of course some of the things I suggest are not easy to implement,9 and perhaps are not even the best way to use this rich dataset. However, I do think that we can learn much more from going beyond the cross-country analysis of BEEPS data.10 Notes 1. I also found data for six industrial countries. Including these countries in the sample does not affect the results described here. 2. As a better working of judicial system will increase the benefits of using the system, some marginal firms may decide to start using the system and pay its cost in terms of disclosure. 3. Consider, for example, a country divided into two regions (call them East and West) popu- lated by a large number of firms that have one plant in each region. Assume that there are plant-specific productivity shocks. Given these assumptions, moving some workers from a plant that receives a negative shock to a plant that receives a positive shock would clearly lead to an increase in total productivity. Hence an external observer may be tempted to judge as bad a law that prevents firms from reallocating workers across plants and advise the country that this law should be either abolished or, at the very least, not enforced. But what if this law is a second best response to another distortion (maybe in the housing or education market)? Then not applying (or eliminating) the law may lead to a reduction in total social welfare. 4. Furthermore, it allows to me to say what I would like to do without the need of actually doing it. 5. While it is hard to think of a dictatorship with an independent and well-working supreme court, there have been examples of dictatorships with well-working civil courts. 6. For more details and an application to Latin America, see Sousa (2005). 7. Anderson and Gray do a good job in classifying reforms that focus on: public information and transparency; judicial infrastructure and management; judicial education and training; and supporting professions. However, using a well-established taxonomy would probably increase the paper's readability. 366 | UGO PANIZZA 8. This could be done by either simple comparisons or by pooling all countries and interacting a set of coefficients representing the hypotheses described in part I with firm characteristics. 9. For instance, formulating the hypotheses discussed in part I is a very difficult task that requires deep institutional knowledge of the judicial systems of several countries. 10. Anderson and Gray use the firm-level dimension of the data when they compare firms that have used the courts with firms that have not used the courts. References Anderson, James H., David S. Bernstein, and Cheryl W. Gray. 2005. Judicial Systems in Transition Economies. Washington, DC: World Bank. Carothers, Thomas. 1998. "The Rule of Law Revival." Foreign Affairs 77 (2): 95­106. Gloppen, Siri, Roberto Gargarella, and Elin Skaar, eds. 2004. Democratization and the Judi- ciary. The Accountability Functions of Courts in New Democracies. London: Routledge. IDB (Inter-American Development Bank). 2005. The Politics of Policies. Economic and Social Progress in Latin America 2006 Report. Washington, DC and Cambridge, Mass.: Inter- American Development Bank and David Rockefeller Center for Latin American Studies, Harvard University. Panizza, Ugo. 2001. "Public-Private Wage Differentials and Bureaucratic Quality. Evidence from Latin America." Economia (Fall): 97­151. Rajan, Raghuram, and Luigi Zingales. 1998. "Financial Dependence and Growth." The American Economic Review 88: 559­86. Pistor, Katharina. 1996. "Supply and Demand for Contract Enforcement in Russia: Courts, Arbitrations, and Private Enforcement." Review of Central and Eastern European Law 22 (1): 55­87. Recanatini, Francesca, Alessandro Prati, and Guido Tabellini. 2005. "Why Are some Public Agencies less Corrupt? Lesson for Institutional Reforms from Survey Data." World Bank, Washington, DC. Sousa, Mariana. 2005. "Judicial Reforms, the PMP, and Public Policy." Department of Political Science, Notre Dame University. Background paper for Politics of Policies (IDB 2005). Comment on "Judicial Reform in Developing Economies: Constraints and Opportunities," by Matthew C. Stephenson, and "Transforming Judicial Systems in Europe and Central Asia," by James H. Anderson and Cheryl W. Gray STEFAN VOIGT Today, there seems to be widespread consensus that it is hard to exaggerate the impor- tance of well-functioning judicial systems for the beneficial development of entire economies. This consensus, however, has emerged only recently, along with the more general consensus that institutions are crucial for economic development. The transition processes that started in Central and Eastern Europe some 15 years ago have been instrumental in bringing this consensus about. Likewise, the transition processes also helped bring another insight to the fore: namely, that institutions cannot be entirely designed "from above," and that the implementation of market-friendly institutions often takes years, if not decades. Some people had long argued that a functioning judiciary would be crucial for the beneficial functioning of market systems, including--notably--Charles Montesquieu (The Spirit of the Laws, 1748) and Alexander Hamilton, James Madison, and John Jay (The Federalist Papers, 1788). However, empirical knowledge concerning the economic effects of judicial institutions, as well as the variables that determine them in the first place, is still scarce, although it has developed substantially over the last number of years. Both papers to be discussed in this comment are important for improving under- standing of the issues involved. The paper by James Anderson and Cheryl Gray hypothesizes that the transition countries in Central and Eastern Europe currently have a problem of judicial accountability rather than judicial independence. The paper does a great job in pointing to the problems of honesty, the absence of corrup- tion, and fairness by drawing on the BEEPS survey carried out by the World Bank in three waves among almost 20,000 firms in the region. Stefan Voigt is professor of economics at Philipps University, Marburg, Germany. Annual World Bank Conference on Development Economics © 2007 The International Bank for Reconstruction and Development / The World Bank 367 368 | STEFAN VOIGT In his paper, Matthew Stephenson is interested in the difficulties of implementing newly gained insights concerning the judiciary into real-world judicial systems. He deals with three kinds of constraints that judicial reforms systematically face: the resource constraint; the incentive compatibility of all groups of actors involved (that is, private parties, government representatives, and judges); and the sequencing of reforms, here described in terms of the familiar second best theorem. His main message is that the interdependence of various institutions needs to be explicitly taken into account; otherwise we are in for some unpleasant surprises. Both papers are highly stimulating. Instead of discussing them in any detail here, I propose to ask three questions. First, what are the functions of the judiciary in a market system? Second, what do we know about the effects of the judiciary--and what do we not know? And third, what can be done to improve the functioning of the judiciary? In discussing these three questions, I will return to the two papers. What are the Functions of the Judiciary in Market Systems? Transactions between private actors are coordinated via contracts that are based on property rights in market systems. Contracts are mutual promises to do certain things and to refrain from doing other things. In developed economies, many transactions, optimally, will not be executed simultaneously but sequentially. The sequential exe- cution of contracts suffers from a time inconsistency problem. That is to say, given that the other party has already delivered and my term comes, I have little incentive to carry out my original promise (for example, to pay) unless there is some third party that can credibly threaten to make me even worse off than if I pay: the judiciary. Hence the judiciary enables private parties to make credible commitments toward other parties. This can greatly increase the number of transactions taking place, prolong the planning horizon of the actors, increase the aggregate amount of investment and, at the end of the day, improve economic growth and incomes. This welfare-enhancing function of the judiciary will materialize, however, only if a number of conditions are met. First, legislation must be favorable to market transactions. If private property rights are not even promised, taxes are so high that private parties have little incentive to make profits, and so forth, a judiciary will not be sufficient to increase welfare. Secondly, none of the conflicting parties can buy a favorable decision by bribing the judges. In other words, judicial accountability is present. Thirdly, government representatives do not put any pressure on judges, and enforce the court decisions even if they are not in accord with their own preferences. This can be referred to as judicial independence. If these conditions are met, the judicial system will not only enable private actors to make credible commitments, but the government, as well. This is a crucial precon- dition for economic development, since the government also suffers from the problem of time-inconsistency. A simple promise to honor private property rights will not be credible. Once private actors are invested, government has incentives to renege on its promises. If, in such a situation, an independent judiciary is able to make the govern- ment stick to its promises, this will foster investment and growth. COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 369 Having dealt with the functions of the judiciary, we now turn to describe what is included in the notion of "the judiciary." It is important to note that the notion is clearly a lot broader than "the courts." Given that criminal law is involved, we would have to add "the police," "the prosecution agency," and "the prison system." Given that private law is involved, we would have to add "notaries" and "bailiffs."1 These components can be called "the value chain of the judiciary." This use of words sug- gests a hypothesis: namely, that the judiciary can only be as good as the weakest element of the chain: that is, as good as its worst component. The degree to which judicial systems will be able to generate welfare-enhancing effects will also depend on a host of other actors such as audit institutions (increasing the incentives to spend the judicial budget effectively) and anticorruption agencies (reducing the incentives to accept bribes). What Do We Know about the Effects of the Judiciary--and What Do We Not Know? Until recently, precious few empirical results concerning the hypotheses spelled out in the last section were available. The single most important impediment was per- ceived to be the difficulty--or even impossibility--of coming up with measures for both judicial independence and judicial accountability. This has changed, however, and I want to present four recent results here. First, Feld and Voigt (2003) introduce two indicators for judicial independence: one measuring de jure independence (based on twelve different variables) and the other one measuring de facto independence (based on eight different variables). In a cross-section of some 80 countries, Feld and Voigt (2006) find that their de facto indicator has a strong and significant positive impact on economic growth, while the de jure indicator does not. Second, judicial independence is not only a necessary condition for the impartiality of judges, it can also endanger it: judges who are independent could have incentives to remain uninformed, be lazy, or even corrupt. It is therefore often argued that judicial independence and judicial accountability are competing ends. Stephenson, in his paper, refers to this view. Anderson and Gray can be read as being optimistic concerning the compatibility of both concepts. Voigt (2005) argues that the two concepts need not exclude each other: judges could be independent from undue influences by other actors but still be accountable to the law. Voigt (2005) introduces two simple proxies for judicial accountability. One is the absence of perceived corruption among business people. The other is an indicator for the fairness of trials based on nine different components, such as the right to an appeal, timeliness of the court's decision, whether charges are pre- sented before trial, and whether the trial is public. After controlling for standard explanatory variables, the study finds that both indicators are very significantly and robustly linked to per capita income on a basis of 75 countries. Third, as noted in the last section, "the judiciary" encompasses many more actors than just judges. There are very few empirical studies available that deal with the value chain explicitly. In one such study (Voigt, Feld, and van Aaken 2005), it is conjectured 370 | STEFAN VOIGT that prosecution agencies that are not independent from the executive will lead to higher levels of corruption, as it will be more attractive for government members to accept bribes if they can count on, say, the minister of justice to stop prosecution of their crimes. The authors present two indicators, one for de jure and one for de facto prosecutorial independence. In a cross-section of 62 countries, they find that higher degrees of de facto prosecutorial independence are indeed robustly and signif- icantly correlated with lower degrees of perceived corruption. The fourth and last empirical result I want to present in this section deals with the interrelationship of de facto judicial independence and the factual independence of other government agencies, such as the central bank. Various writers have observed that higher degrees of formal central bank independence are correlated with lower degrees of inflation only among the OECD countries. In less developed countries, it is, rather, the turnover rate of central bank governors that is a good predictor for inflation rates. This could mean that de jure central bank independence is only a good proxy for factual independence in OECD countries. But what are the determi- nants of factual central bank independence? Hayo and Voigt (2005) recently argued that high degrees of judicial independence can influence inflation rates both indirectly (by increasing average tenure of the central bank governor) and directly (by lowering transaction costs, which would lead to a lower rate of natural unemployment and an increase in potential output). There is evidence that both transmission channels are empirically relevant. This is one case where institutions are highly interdependent: an independent judiciary is an important component for creating a stable currency. It can be hypoth- esized (but needs to be shown empirically) that this is also true for a host of other independent government bureaus, such as competition agencies and network indus- try regulators. This leads directly to the effects of the judiciary about which we do not yet know very much. Again, I would like to name four issues. First, there seems to be a significant and highly robust correlation between judicial independence and income. Yet we do not know very much about the transmission channels through which one affects the other. We could, for example, ask whether a factually independent judiciary induces additional investment, or whether there are differential effects on human versus physical capital, or whether there are different effects concerning the origin of investment: that is, between domestic and foreign investment. Could it also be the case that a high degree of factual judicial independence enhances (total) factor productivity? This leads directly to the second open question. The judiciary can be hypothesized to have important effects in two altogether different interaction situations. The first is in cases of conflict between private parties. As long as both sides expect the judiciary to be impartial and independent, the propensity to enter into such contracts in the first place can be assumed to be higher, which will lead to more welfare-enhancing trans- actions taking place, and hence to higher economic growth. The second is in cases of conflict between government and the citizens. Citizens are in need of an organization that has the power to adjudicate even against the government in case it has not fol- lowed the law. We are, in other words, dealing with the distinction between private COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 371 and public law. A second set of questions arises with regard to this distinction: Is it possible to evaluate the relative importance of either of the two channels just outlined? Does an independent judiciary increase overall government efficiency? Third, it could be argued that the ineffectiveness of the judiciary run by the state would be comparatively less relevant with regard to private law: if private parties believe that they could make a mutually attractive deal, they might be able to agree on non-state third parties as "judges." When one deals with the state, this possibility seems to be close to impossible. This hypothesis leads directly to a third set of unanswered questions: to what extent can non-state arbitration make up for an ineffective state-run judiciary? If non-state arbitration is a substitute for state-run arbitration, this might be policy relevant: if it is less costly and time-consuming to implement the preconditions for non-state arbitration, this could be an intermedi- ate fix for states that suffer from ineffective judicial systems. The fourth issue deals with the question, "correlation or causation?" As of yet, it cannot be excluded that it is wealthy societies that can afford to buy a well-functioning judiciary: that is, that the causation runs from being wealthy to the judiciary, and not the other way around. To shed more light on this question, time-series data that are just emerging would be helpful. Alternatively, instrumental variables that are more truly exogenous than judicial independence could do the job. In sum, we can be fairly certain that judicial independence as well as judicial accountability are robustly linked to income and growth, that prosecutorial indepen- dence is robustly linked to the absence of corruption, and that more independent courts also increase the factual independence of other agencies, with potentially wide-ranging effects. On the other hand, our knowledge concerning transmission channels, the effects of non-state dispute resolution, and the question of causality is still insufficient. What Can Be Done to Improve the Functioning of the Judiciary? If the judiciary has wide-ranging effects on the development of the economy, what can be done to improve its functioning? Is there "the" optimal way or are there different roads to an effective system? Or, more pessimistically, could countries that do not have specific conditions at their disposition be doomed to live with an inef- fective judiciary? In his paper, Stephenson emphasizes the need to allocate resources among various judicial reform options competing for scarce resources. But the ques- tion is not confined to the sequence of reforms within the judiciary, but also to the kinds of steps taken within the judiciary, as well as interrelationships with other insti- tutional arrangements that can extend well beyond the judiciary. A number of factors promise to be relevant. First, the indicators introduced by Feld and Voigt (2003) are based on twelve and eight different variables. The authors did ask what single variables were most impor- tant for the high correlation with economic growth, and found that anchoring the basics of the judiciary within the constitution was highly relevant with regard to de jure 372 | STEFAN VOIGT independence. Among the de facto independence variables, factual implementation of legal terms of the judges and paying them adequately proved to be most important. The policy advice that can be drawn from these insights seems straightforward. Second, what about other characteristics of the judiciary, like having a supreme court (the U.S. model) or a constitutional court (the Austrian model)? Feld and Voigt (2006) find that this component of the court structure does not have significant effects on economic growth. With regard to legal families, only countries with socialist legal origins performed clearly worse than the benchmark legal origin (Scandinavian). This means that no matter whether a country has a French, an English, or a German legal origin, this will not have important consequences for its growth prospects. One question for which we do not have good answers yet is what are the effects of lay participation (either juries or lay assessors) on economic performance.2 Third, the effectiveness of the judiciary might also depend on institutional choices that are made with regard to other branches. Here, Feld and Voigt (2006) find that the existence of strong checks and balances enforces the impact of de facto judicial independence on economic growth. Further, the growth-enhancing effects are particu- larly strong in presidential systems. However, policy recommendations ought not to be drawn too hastily: presidential systems are significantly less likely to realize high degrees of de facto judicial independence in the first place. Lastly, it could be the case that realizing high degrees of de facto judicial inde- pendence is possible only under certain conditions that are beyond the immediate reach of policy makers. Stephenson, for example, mentions the possibility that courts might only be factually independent if they enjoy broad support by the public. Hayo and Voigt (2006) have tested two closely related hypotheses: namely, that the realized degree of press freedom and the realized degree of civil society determine the degree of factual judicial independence. Their results show that press freedom is indeed an important determinant, but the degree of civil society does not survive their rigorous empirical model reduction process. Their results can even make policy makers somewhat optimistic: although the correlation between de jure and de facto judicial independence is rather low, de jure independence turns out to be the single most important predictor for the realized level of factual independence. This is, clearly, a variable under the control of policy makers. The other important determinants they found were the degree of legal trust that the population has in the judiciary and the extent of democratization of a country. These variables are much less subject to explicit policy measures. Conclusions and Outlook Our knowledge concerning the effects of the judiciary on economic development has substantially increased over the last number of years, but important gaps remain. De facto independence is far more important than the formal independence of the judiciary. Yet improving the functioning of the judiciary is not beyond the reach of governments. COMMENT ON STEPHENSON AND ON ANDERSON AND GRAY | 373 Notes 1. In both kinds of law, private lawyers will regularly play an important role. The terms on which they work are, to a large extent, defined by the respective legislation­­and hence by the state. 2. Voigt (2006) is a conceptual paper, not an empirical one. References Feld, Lars, and Stefan Voigt. 2003. "Economic Growth and Judicial Independence: Cross Country Evidence Using a New Set of Indicators." European Journal of Political Economy 19 (3): 497­527. ------. 2006. "Making Judges Independent--Some Proposals Regarding the Judiciary." In Democratic Constitutional Design and Public Policy--Analysis and Evidence, ed. R. Congleton and B. Swedenborg, pp. 251­88. Cambridge, Mass.: MIT Press. http://papers. ssrn.com/sol3/papers.cfm?abstract_id=597721. Hamilton, Alexander, James Madison, and John Jay. 1788, lst edition. Reprinted in 1961. The Federalist Papers, with an introduction by C. Rossiter. New York: Mentor. Hayo, Bernd, and Stefan Voigt. 2005. "Inflation, Central Bank Independence, and the Legal System." http://papers.ssrn.com/sol3/papers.cfm?abstract_id=724661. ------. 2006. "Explaining de facto Judicial Independence." Forthcoming in International Review of Law and Economics. Montesquieu, Charles L. de. 1748, 1st ed. Reprinted in 1989. The Spirit of the Laws (Cambridge Texts in the History of Political Thought). Cambridge: Cambridge University Press. Voigt, Stefan. 2005. "The Economic Effects of Judicial Accountability." http://papers.ssrn.com/ sol3/papers.cfm?abstract_id=732723. ------. 2006. "Civil Society Elements in European Court Systems--Towards a Comparative Analysis." http://www.asp-research.com/Papers%20CiSoNet/MadridPres.pdf. Voigt, Stefan, Lars Feld, and Anne van Aaken. 2005. "Power over Prosecutors Corrupts Politi- cians: Cross-country Evidence Using a New Indicator." Paper presented at the Annual Meeting of the Public Choice Society, New Orleans, March. http://www.pubchoicesoc.org/ papers2005/Voigt_Feld_van_Aaken.pdf. ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserving Saved: endangered forests and natural resources. · 17 trees The Office of the Publisher has chosen to · 12 million BTUs of print The Annual World Bank Conference total energy on Development Economics 2007: Beyond · 1,518 lbs. of net green- Transition on recycled paper with 30 percent house gases post-consumer waste, in accordance with · 6,300 gallons of the recommended standards for paper usage wastewater set by the Green Press Initiative, a nonprofit · 809 lbs. of solid waste program supporting publishers in using fiber that is not sourced from endangered forests. For more information, visit www. greenpressinitiative.org. Themes and Participants for the ANNUAL WORLD BANK CONFERENCE ON DEVELOPMENT ECONOMICS--REGIONAL Beijing, China "HIGHER EDUCATION AND DEVELOPMENT" January 16­17, 2007 Higher Education and Migration: Brain Drain and Sharing Skills in the Region Private-Public Provision of Higher Education Financing Higher Education Technological Innovation: Linkages between Universities and Industry Higher Education and Labour Markets in Asia François Bourguignon · Pawan Agarwal · Nicholas Barr · Satro Soematri Brodjonegoro · Kai-Ming Cheng · Frederic Docquier · Mary Ann Feldman · Fumio Kodama · Norman LaRocque · Daniel Levy · Bengt-Ake Lundvall · Wu Qidi · Mark Rosenzweig · Sharifah Hapsah Syed Hasan Shahabudin · Patricia A. Sto Tomas · Rong Wang · Min Weifang · L. Alan Winters · Cheonsik Woo · Shahid Yusuf Look for publication of the proceedings from this conference in Spring 2008 under the title Annual World Bank Conference on Development Economics 2008, Regional: Higher Education and Development (ISBN 978-0-8213-7123-7). T he Annual World Bank Conference on Development Economics (ABCDE) brings together the world's leading scholars and development practitioners for a lively debate on state-of-the-art thinking in development policy and the implications for the global economy. The 18th conference was held in St. Petersburg, Russia, from January 18­19, 2006. The theme of the conference was beyond transition, which was divided into four topics: growth after transition, economic space, governance, and judicial foundations of a market system. IN THIS VOLUME Introduction by François Bourguignon and Boris Pleskovic; keynote addresses by François Bourguignon, Anders Åslund and Nazgul Jenish, Gur Ofer, and Yegor Gaidar; papers by James H. Anderson, Erik Berglof, Patrick Bolton, Kiran Gajwani, Cheryl W. Gray, Sergei Guriev, Ravi Kanbur, William Megginson, Pradeep Mitra, Marcelo Olarreaga, Guillermo Perry, Ernesto Stein, Matthew C. Stephenson, Mariano Tommasi, Ruslan Yemtsov, Xiaobo Zhang, and Ekaterina Zhuravskaya; and comments by Chong-En Bai, Alan Gelb, Irena Grosfeld, Thierry Mayer, Ugo Panizza, Andrés Solimano, Jan Svejnar, and Stefan Voigt. ISBN 0-8213-6843-5