Book Search:  


Google full text of our books:


Development, Democracy, and Welfare States:
Latin America, East Asia, and Eastern Europe
Stephan Haggard & Robert R. Kaufman

Book Description | Endorsements | Table of Contents

COPYRIGHT NOTICE: Published by Princeton University Press and copyrighted, © 2008, by Princeton University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. Follow links for Class Use and other Permissions. For more information, send e-mail to

This file is also available in Adobe Acrobat PDF format


Toward a Political Economy of Social Policy

The comparative study of social policy in developing countries is of recent vintage.1 Yet the middle-income countries of Latin America, East Asia, and Eastern Europe have welfare systems that originated in the early postwar period. The questions posed by these systems are the same as those that motivate the literature on the advanced welfare state: Why did governments undertake the provision of social insurance and services? How have welfare systems evolved over time, and how are benefits distributed?

The relevance of these questions has been heightened by epochal political and economic changes that occurred in the developing and formerly socialist world in the 1980s and 1990s. Most countries in the three regions democratized during this period, raising hopes that new governments would be more attentive to social issues. At the same time, most countries also experienced financial crises, recession, and associated fiscal constraints. These problems triggered wide-ranging reforms, including, but by no means limited to, liberalization and increased economic openness.

Economic crisis and market reforms entailed serious social dislocations and raised questions about the viability of existing welfare commitments. Could social spending be sustained in the face of severe fiscal constraints? Or would economic crises and reform force new democracies to limit, or retrench, social-policy commitments?

In this book, we analyze the development and reform of social policy among the middle-income countries of Latin America, East Asia, and Eastern Europe. By 1980, at the onset of the major economic and political changes of the late twentieth century, the three regions had developed distinctive social-welfare models. Eastern European welfare systems, though increasingly strained, provided comprehensive protections and services to almost all of their populations. East Asian welfare systems offered minimal social insurance, but a number placed a high priority on investment in education. In Latin America, the urban middle class and some blue-collar workers enjoyed access to relatively generous systems of public protection, but peasants and informal-sector workers were generally excluded or underserved.

These welfare legacies had a strong influence on both the politics and the economics of social policy as the countries of the three regions democratized. Past policies—or the absence of them—created constituencies and generated demands on incoming democratic governments. Prior welfare commitments also had important fiscal implications. In Eastern Europe and Latin America, entitlements placed heavy burdens on governments and generated strong pressures for reform and even retrenchment. In the high-growth Asian countries, by contrast, new democratic governments were relatively unencumbered by prior welfare commitments and had room to expand social insurance and services.

In exploring these distinctive welfare trajectories, we build on three lines of theoretical argument that have motivated the literature on the advanced welfare state. The first, and arguably the most basic, is the significance of distributive coalitions and economic interests. Following the power-resource approach, we consider the extent to which political elites incorporate or exclude organizations and political parties representing urban labor and the rural poor. We focus initially on critical political realignments that resulted in the long-term repression of these groups in some countries and allowed space for them to operate in others. Over time, as Pierson (1994) has argued with respect to advanced welfare states, the commitments established through these initial political choices created stakeholders and constituencies that influenced the subsequent course of social policy.

The second set of factors we consider are economic; this set includes both the performance of the economy and its organization. Economic performance exerts a crucial influence on social policy, particularly through its effect on the fiscal capacity of the state. High growth is at least a permissive condition for an expansion of entitlements and spending, slow growth, crises, and attendant fiscal constraints, by contrast, place political as well as economic limits on the ability to sustain welfare entitlements and services.

Our focus on the “organization” of the economy follows and modifies the varieties of capitalism literature.2 We show how the development strategies of governments, and the resultant production strategies of firms, are complementary to particular social policies and labor-market institutions. Over time, the sustainability of these different economic models also had a crucial impact on the path of social policy. The crisis and transformation of Latin America’s import-substitution model during the 1980s and the rapid implosion of state socialism in the early 1990s undermined import-substituting firms and state-owned enterprises that had been central pillars of social policy in both regions. Privatization, restructuring, and greater exposure to international competition had important implications for the social contract.

Political institutions constitute the third cluster of factors that influence social policy. Given the political heterogeneity of the countries in our sample, and the “third wave” of democratization that began in the 1970s, we are particularly interested in the influence of regime type. To what extent did democracy and democratization affect the responsiveness of governments to the interests of low-income groups?

Regime type is an important component of any explanation of social policy. Dictatorship and democracy determine the extent to which competing parties can enter the electoral arena and the freedom interest groups enjoy in organizing and exercising influence. However, we also emphasize the limitations of a purely institutional approach. Institutional rules of the game—the supply side of the political market—are not sufficient to account for the character of social policy without consideration of underlying interests and their organization—the demand side of the political market—and the economic context in which governments operate.


What do we mean by social or welfare policy? An expansive view of the social contract between states and citizens would arguably begin with the capacity of governments to deliver economic growth. However, the question of growth is analytically distinct from the question of how governments choose to redistribute income, either through insurance schemes that mitigate risk or through spending on basic social services that are of particular significance to the poor. We focus on these two broad areas of social policy and on the particular way they are combined in broader policy complexes.

In principle, social insurance can offer protection against the full range of life-cycle and market risks, including sickness, work-related injury and disability, maternity and childbearing, unemployment, retirement, and death (through survivors’ benefits). However, pensions and health insurance are the most significant forms of social insurance in the countries in our sample, and we pay particular attention to them. In our consideration of more recent reforms in part 2, we also examine antipoverty and labor market policies, which are relative latecomers to the social-policy mix.

The second area of social policy is the provision of basic social services. Access to high-quality education and primary health services is widely viewed as critical for expanding human freedom and capabilities (Nussbaum and Sen 1993; Sen 1999) and is an underlying determinant of life chances and arguably of growth itself.3 Access to primary education and basic health care is especially important in labor-abundant economies, since it augments the human capital of the poor, their most basic resource (Lindert 2004).

Following Esping-Andersen (1990), we argue that specific social policies do not evolve in isolation but cluster together into identifiable complexes.4 Table I.1 provides a stylized summary of these models across the three regions during the early post–World War II decades. In Eastern Europe, social policy was anchored by an overarching employment guarantee, but also by a strong commitment to education and training, universal health care and pensions, and family allowances. These commitments began as occupational ones but were transformed over the postwar period into universal citizenship rights. In Latin America, most states established occupationally based social insurance and health systems that favored formal-sector workers but typically excluded informal urban workers and the rural sector. The provision of basic social services also showed a marked inequity in distribution, reenforcing rather than mitigating long-standing patterns of inequality in the region. In Asia, social insurance was limited and, where it did exist, was provided through mandated individual savings programs that had little or no redistributive component. Nonetheless, governments attached a high priority to the provision of primary and secondary education and, somewhat more unevenly, to public health and basic health services.

Contrasts between these systems constitute the pivot around which we organize the core arguments of this book. In part 1 (chapters 1 through 4) we examine the initiation and expansion of welfare commitments in the three regions from the early postwar period through the late 1970s, and, in the socialist cases, through the 1980s. We trace the origins and evolution of these systems to political realignments that occurred during the early and mid-twentieth century and the subsequent adoption of distinctive development models. Part 2 focuses on the political conflicts over social policy during and after the “third wave” of democratization (c. 1980–2005). We show that economic conditions and welfare legacies strongly influenced how new democracies dealt with these conflicts.

These arguments are outlined briefly in the first two sections of this introduction. In the third section, we take up the issue of institutions and regime type in more detail. Section four considers some methodological issues that arise in the comparative historical analysis that forms the empirical core of the book. We close this introduction by locating our work in the broader literature on the welfare state, a task we revisit in the conclusion to the volume.


In seeking to explain the origins and early development of welfare commitments, we must analyze why political elites have incentives to distribute and redistribute income in a particular way. But we must first make choices about the appropriate time frame for thinking about these issues. A number of recent studies of long-run growth have located the ultimate origins of key political and economic institutions in the colonial period (Kohli 2004; Acemoglu, Johnson, and Robinson 2001, 2005) or in wholly exogenous factors, such as resource endowments and geography (Engerman and Sokoloff 2000, 2002). These long-run constraints no doubt operate on social policy, and we consider the significance of colonial inheritance in particular. But this focus on very long-run forces seems misplaced for our purposes, given quite fundamental political discontinuities that occurred in the mid-twentieth century in the three regions, as well as important changes in the welfare policies we seek to explain.

A more plausible alternative is that the politics of social policy is determined by the more proximate processes of growth and structural economic change. The modernization approach to the welfare state traces the early origins of welfare commitments to the functional requirements of industrialization and the political demands unleashed by it.5 The differences in welfare models that we have highlighted in table I.1 might result, not from distinctive political factors, but from variation in the level of development and the extent of structural change across countries.

Again, we are skeptical. Industrial growth and social modernization certainly contributed to the emergence of the modern welfare state. But these are highly general processes, and if we have learned one thing about modern capitalism it is the absence of a single model. Countries experiencing “modernization” end up with very different market institutions and social-policy complexes. In chapter 1, we provide some simple, cross-national regressions to justify our skepticism. These models suggest that the relationship between level of development and social policy is by no means as robust as one might expect; even controlling for “modernization,” important cross-regional differences persist. These results are consistent with our claim that the course of social policy depends on political and economic factors that cannot be reduced to the modernization process alone.

Critical Realignments and the International Setting

In contrast to these “deep” historical and modernization arguments, we focus on discontinuities in patterns of political domination that occurred in each of the three regions during the first half of the twentieth century.6 We identify these discontinuities by the emergence of new ruling coalitions and political incorporation or exclusion of working-class and peasant organizations. The incorporation of urban working-class and rural-sector organizations influenced social policy through the basic mechanisms identified by the power-resource approach: by determining the constituencies to which politicians—whether democratic or authoritarian—responded. Exclusion of these groups gave elites greater leeway in pursuing their political and economic objectives.

In East Asia and Eastern Europe, critical realignments occurred in the aftermath of World War II and were strongly influenced by international political developments. In both regions, great powers provided crucial support for new political elites who brought with them novel political and economic projects. In both regions, these projects dramatically weakened organized labor, the left, and rural political movements.

However, they did so with fundamentally different aims. In Asia, the turbulent wave of decolonization set in train by the end of the Pacific War was followed quickly by the onset of the Cold War and the triumph of conservative, anti-Communist political leaderships. With external support, new governments beat back the challenge from the left in the cities and forestalled or defeated armed insurgencies in the countryside. But in varying degrees, and in contrast to Latin America, these new political elites also reached into the rural areas for political support.

In Eastern Europe, Soviet influence prevailed. As in East Asia, the postwar liberation initially unleashed a wide spectrum of new social forces. With the consolidation of power by Communist parties, however, came the destruction of independent social-democratic and peasant parties and the transformation of unions into arms of the party-state. The distinctive features of socialist social policy were not based on accommodation of labor and the left, as was the case in the European social democracies. Rather, labor and the peasantry were subordinated to the political, economic and organizational logic of the command economy.

In Latin America, we identify the critical realignments with reformist challenges to the oligarchic states of the nineteenth century. In contrast to Asia and Eastern Europe, these political changes predated the great-power rivalry of the Cold War era. New contenders for political power could not count on sustained support from powerful external patrons. Rather, they relied on the support of cross-class coalitions that offered legal status and influence to segments of organized labor and, in some instances, to popularly based parties. But antioligarchic coalitions also typically included some segments of the dominant landowning class itself, and urban-based political challengers found it difficult to penetrate the countryside to the same extent as in Eastern Europe and East Asia. Peasants and agricultural workers remained politically marginalized and the countryside relatively disadvantaged in the provision of social insurance and services.

The effects of these critical realignments cannot be reduced to differences in regime type. In Latin America, the incorporation of labor was accompanied by a widening of the franchise in some cases and various forms of authoritarian rule in others. In East Asia, decolonization was followed by democratic openings in some cases but not in others. In Eastern Europe, the shift to Stalinist political rule was uniform, but it is difficult to deduce the course of social policy from the nature of political institutions To the contrary, the socialist welfare state stands as a stark anomaly to the expectation that authoritarian regimes are likely to redistribute income in a more narrow way. In all three regions, critical realignments ran much deeper than formal institutions; they reflected fundamental shifts in the constellation of political interests.

Development Strategies

The middle of the twentieth century also marked the onset of industrialization and profound structural change in the countries of interest to us. Now, with the benefit of hindsight, we can see that the first three postwar decades were a period of unprecedented growth. The Eastern European countries were the first to take off, and they experienced rapid economic growth until the 1970s, when the constraints of the socialist model became manifest. The postwar period witnessed high growth across most of the middle-income countries of Latin America as well. Given subsequent crises, it is easy to forget that Brazil and Mexico were among the economic miracles of the period. The East Asian takeoff came somewhat later, but from the mid-1960s, the newly industrializing countries of Korea and Taiwan sustained unparalleled economic performance, followed by a number of Southeast Asian countries. Strong growth across the developing world did not necessarily lead to an expansion of social entitlements, as the East Asian cases show. However, it did provide the basis for increased social spending among those governments that chose to do so.

Growth occurred in the context of very different development strategies, however: import-substitution in Latin America and the early postwar period in East Asia, state socialism in Eastern Europe, and a more export-oriented model beginning in the 1960s in a number of the East Asian countries. In chapter 1, we extend the underlying logic of the varieties-of-capitalism approach by considering how differences in the development strategies of governments influenced labor-market and production strategies and the preferences of key groups with respect to social policy. We also show how the economic models adopted in the three regions reinforced and perpetuated the patterns of political incorporation and exclusion associated with the critical realignments we identify.

In the larger Latin American countries, the process of import-substitution industrialization (ISI) was accelerated by economic shocks that began as early as World War I in the larger countries and was subsequently pushed along by the Great Depression and World War II. The cross-class coalitions formed in the course of the critical realignments in the region were conducive to the adoption of ISI as a more self-conscious development strategy in the postwar period. ISI allowed state enterprises and private firms in the import-substituting sectors to accommodate welfare entitlements for the organized urban working class. However, such policies contributed to labor-market dualism and to well-known biases against agriculture and the rural sector. The social-insurance systems that developed in the region both reflected and reenforced these biases. Moreover, the structural characteristics of import-substituting economies also had adverse effects on the overall distribution of income and reduced incentives for governments, firms, and workers to invest in education.

In East Asia, the turn toward export-oriented growth occurred well after the conservative political realignments of the Cold War period. Nonetheless, outward-oriented strategies strongly influenced the incentives facing governments, firms, and workers with respect to social policy. On the one hand, strategies dependent on the export of labor-intensive manufactures put a premium on labor-market flexibility and made governments and firms highly resistant to social insurance schemes that would increase labor costs. The authoritarian regimes in the region, whether established early in the postwar period or after brief periods of semidemocratic rule, thus maintained far more limited systems of public protection than was the case in either Eastern Europe or Latin America. On the other hand, export-oriented growth strengthened incentives to expand access to primary, secondary, and vocational education. To a lesser extent, the policy emphasis on human capital as a core asset motivated the expansion of basic public health services as well.

In Eastern Europe, all Communist governments initiated centrally planned industrialization drives immediately after the consolidation of Communist power in the late 1940s. This strategy rested on high levels of investment in basic industry, the mobilization of labor, and a squeeze on the countryside. The commitment to full employment and government provision of social insurance and services emerged as side effects of the complete socialization of the economy. In the absence of any private sector, the government was perforce involved in both the financing and provision of health care, pensions, and even housing. Even though these entitlements originated in the state-owned enterprise sector, the collectivization of agriculture extended them into the countryside and provided the basis for the universalization of benefits. Similarly, government interest in the expansion of education, and its particular emphasis on vocational training, was a direct complement of the socialist system of manpower planning.

In sum, the welfare systems that had developed by the late 1970s and early 1980s in the three regions were deeply embedded in political interests and economic strategies that had evolved over decades. “Social contracts” were by no means the result of democratic politics, or even bargaining with affected interests; much more commonly, they were simply imposed from above. Nevertheless, distinctive social-policy complexes generated expectations about the benefits the state would provide. In turn, the interests, expectations, and even institutions formed through these earlier social-policy interventions strongly affected the political battles over social policy that unfolded in the new democracies that emerged in the 1980s and 1990s.


The 1980s and 1990s witnessed profound political and economic changes in the three regions of interest to us. Transitions to democratic rule strengthened incentives to expand public welfare commitments to vulnerable sectors of the population and to protect existing entitlements.

But the course of social policy was also deeply affected by economic developments and ideological trends that pushed in the direction of weakening the social responsibilities of the state.

In part 2, we show that responses to these cross-pressures varied substantially within and across geographic regions. Most countries instituted social-policy reforms, but the scope and direction of those changes depended on variations in economic performance and on the organizational and political legacy of past welfare entitlements.

Reversal of Fortune

The first three postwar decades can be considered a “golden era” of rapid growth. By contrast, economic performance since 1980 has shown much wider variance across regions and countries and much greater volatility within them. These changed economic circumstances not only directly affected the capacity of the government to maintain or expand existing entitlements but affected the politics of social policy as well. Good macroeconomic performance strengthened the hand of political actors arguing for an expansion of social commitments and weakened the force of technocratic arguments for reform or retrenchment. By contrast, crises—and particularly those accompanied by fiscal constraints and high inflation— increased the influence of technocrats and their allies in the international financial institutions. Liberal technocrats focused initially on macroeconomic stabilization and a variety of market-oriented reforms; but they came to press for liberalizing reforms of the social sector as well.

The Asian governments in our sample faced by far the most favorable economic environment. Growth remained robust in most of them until the financial crisis of 1997–98, providing the new democracies with the wherewithal to expand the state role in the provision of social insurance and services. The crisis of 1997–98 posed similar constraints to those seen in Latin America and Eastern Europe but did not pose the same fundamental challenge to the prior development model. Fiscal constraints were cyclical, rather than long-term and structural in nature, and as a result, new entitlements generally survived the crisis intact.

Latin America and Eastern Europe, by contrast, faced far more severe economic constraints. The Latin American economies experienced deep recessions during the debt crisis of the 1980s and a recurrence of financial crises in the 1990s and early 2000s. Eastern European countries faced a gradual slowdown in growth prior to the collapse of 1989, then deep “transitional” recessions in the first half of the 1990s.

Economic crises had contradictory implications for the politics of social policy. On the one hand, crises and the reforms that followed in their wake—liberalization of trade and capital markets, privatization, and a variety of other market-oriented reforms—were socially disruptive and exposed previously sheltered sectors to new market risks. These grievances provided the basis for electoral andinterest-group mobilization.

On the other hand, crises increased the influence of technocrats, the international financial institutions, and domestic policy networks that favored short-term fiscal adjustments and restructuring of long-term fiscal commitments. Social spending was implicated in these adjustments. Sheer fiscal constraint limited the capacity of governments to spend on social programs. The social policy models of the ISI and state-socialist period were also vulnerable to the criticism that they had contributed to the fiscal crises of the 1980s and 1990s and that they were not well suited to the new, market-oriented policy environment.

Distributive Politics: Defense of Entitlements, Pressures for Expansion

In addition to economic conditions, the welfare legacy shaped the distributive demands placed on the state. Welfare entitlements created electoral and interest-group constituencies; in general, the wider the coverage and the more effective the services provided, the more difficult it was for liberal reformers to initiate changes in the social-policy status quo. Constraints on government did not operate solely through the electoral connection, moreover. Past welfare policies created not only beneficiaries but also complex institutional and interest group constraints on welfare reform: from civil servants and public service providers; to unions with an institutional stake in the welfare system; to an array of private actors, from the financial sector and pharmaceutical companies to socially-oriented nongovernmental organizations (NGOs).

Again, the Asian countries stand apart from those in Latin America and Eastern Europe. Relatively minimalist welfare states provided politicians and those favoring an expansion of public commitments with a political tabula rasa. Operating under highly favorable economic conditions, they could offer major new social programs to voter-beneficiaries while facing little organized resistance from existing stakeholders. As we discuss below, these incentives to expand entitlements and services were significantly strengthened by the advent of democracy.

Latin America and Eastern Europe resembled one another both in the economic adversity they faced and in the inheritance of much more extensive public commitments. However, the regions had quite distinct welfare legacies. Most Latin American welfare states were deep but not wide. They involved substantial public expenditures on social insurance, but access to these resources was typically highly unequal. Where coverage was wider, as in Uruguay and Costa Rica, efforts to reform the system of social insurance and services faced greater political difficulties. Where coverage had been narrow and unequal, market-oriented reformers gained greater traction over the social-policy agenda.

In the Eastern European cases, the transition to the market required a fundamental shift of resources out of the state sector. Yet the socialist welfare legacy had offsetting political consequences. Citizens had been incorporated into a dense network of social entitlements, and even where the value of these protections and the quality of services had deteriorated, scaling them back posed serious political risks. New democratic governments in Eastern Europe did undertake reforms. Yet they also strove to maintain universal coverage of a number of important life-cycle risks, most notably with respect to health and pensions. They also attached a relatively high priority to providing social safety nets for formal sector workers.


How do political institutions affect the political economy of welfare policy? More specifically, does the turn toward democracy result in more generous and progressive social contracts? The literature on the advanced welfare state has paid increasing attention to institutional differences within democratic systems, and these factors can be important to new developing-country democracies as well.7 However, the tremendous heterogeneity in basic political institutions across the three regions in the postwar period and their more recent convergence on democratic forms of rule compel attention to the prior question of the effects of regime type. Does democracy result in a more generous and progressive social contract, ceteris paribus? Does authoritarian rule preclude or limit the scope of redistributive politics?

Democratic politics has long been associated with pressures for redistribution, through mechanisms that are basic, constitutive features of democratic rule itself. These include, most notably, the electoral connection and the freedom of association that allows interest groups to organize and press their claims on the state.8 The coming of democracy is associated with an expansion of the franchise and competition for office. Politicians must appeal to broader constituencies in order to win and retain office. They do so by offering competing packages of public and private goods to voters, including those that redistribute income.9 Similar expectations follow from an interest-group approach to the policy process (see Grossman and Helpman 2001 for an overview). Interest groups do many things: they mobilize votes; supply money and information to politicians; and engage in contentious politics (McAdam, Tarrow, and Tilly 2001). As Mancur Olson (1982) pointed out in The Rise and Decline of Nations, however, these various forms of collective political action are all undertaken with one fundamental aim: to redistribute resources toward the members of the given group.

Nondemocratic regimes also redistribute income, in some cases quite dramatically (Bueno de Mesquita et. al. 2003). Nevertheless, the presence of these two closely related mechanisms—electoral competition and independent interest-group activity—would lead us to expect democracies to be more generous in the provision of social insurance and services than their authoritarian counterparts.

The focus on these institutional differences finds justification in both the broader literature and in our own research. The relationship between the expansion of the franchise and the redistribution associated with the welfare state was noted by the early-twentieth-century European social democrats (Przeworski 1985), in the classic work of Marshall (1965) and more recently in Lindert’s (2004) magisterial overview of the role of political “voice” in the expansion of social entitlements in the advanced industrial states. In appendix 1, we survey a broader range of empirical studies that test for the effects of regime type on social policy. A majority of these studies—but by no means all10—find that democracy has positive effects on government effort and on actual welfare outcomes, such as health.

We also find regime type consequential. In Latin America, welfare commitments in long-standing democracies such as Costa Rica, Uruguay, and Chile were more generous than in short-lived democracies and in those political systems that oscillated between authoritarian and democratic rule. The incidence of democracy in Asia prior to the 1980s was substantially less than in Latin America, but periods of democratic or semidemocratic rule were more likely to be accompanied by an expansion of social commitments than were periods of authoritarian governance. In all three regions, finally, the “third wave” transitions of the 1980s and 1990s saw renewed attention to the social question and pressures both to protect existing entitlements and to expand social insurance and services to new groups.

If electoral competition and freedom of interest groups to mobilize are the key mechanisms through which regime type operates on social policy, however, it is very clear that neither of these factors is strictly dichotomous. Some autocracies do not allow elections at all or completely monopolize the electoral process and interest-group activity. But a surprising number are “competitive autocracies” (Linz 2000, 34; Levitsky and Way 2002, 52). They subject themselves to controlled electoral tests and tolerate a degree of interest-group pluralism. Even controlled elections and limited ability to organize may create incentives for authoritarian rulers to seek support through redistribution.

We find vindication for the effort to distinguish among democratic and authoritarian regimes in a more nuanced way. Intermediate regimes— what we call semidemocratic and semiauthoritarian governments (chapter 1)—exhibit greater attention to social policy than do “hard” authoritarian regimes, at least in the nonsocialist cases.

Yet, as we have argued, it is unlikely that the institutional “rules of the game” that define regime type can fully explain the origins or reform of the particular complexes of social-welfare policies that are of interest to us. The analysis of institutional mechanisms provides insight into the supply side of the political market. But as formal models of redistribution show, the effects of institutions are conditional on the distribution of underlying preferences over the policy in question and the strength of contending social groups in the political process.11 Purely institutional models are underspecified; the effects of democracy and authoritarianism depend on the underlying coalitional alignments and economic interests that we have outlined above.

This point is most evident in the wide array of social policies adopted by the authoritarian regimes that we examine in part 1—variations that proved highly consequential for subsequent reform efforts. Restrictions on the franchise allow rulers to censor the groups that can formally influence the political process through electoral or interest-group mechanisms. But even authoritarian rulers court support and vary in the extent to which they repress or coopt low-income groups. Examples of this variance in authoritarian strategies abound. In Latin America, military dictatorships were generally less inclined than democracies to extend coverage of social security to new sectors of the population, but most of them used existing schemes to deflect protest from workers. In Asia, authoritarian regimes expanded the provision of health and education services in the countryside both as a counterweight to urban political forces and to dampen insurgencies; indeed, they reached into the countryside more aggressively than most Latin American democracies.

The highly comprehensive socialist welfare states provide the most obvious anomaly. Despite the most rigid authoritarian rule in our sample, socialist governments committed to universal and quite extensive social protections. They did so not because of constraints operating through formal political institutions but as a component of a larger socialist economic project.

The distribution of social-policy interests is also important for understanding variation in the behavior of democratic regimes. In all three regions, democracy created a new politics of welfare reform. However, the policy outcomes were strongly affected by differences in economic conditions and in the distribution and organization of social-policy interests that had emerged in the earlier period.

The high-growth democracies of East Asia correspond most neatly to the expectation that democratization is associated with an expansion of social entitlements. Outside of education, these countries had the narrowest social-insurance coverage and had relied more extensively on government-mandated private savings, self-insurance, and private delivery. With democratization, NGOs, unions, and civil-society organizations burst onto the political scene and pressed for a broader public role in the provision of social insurance and services. Politicians saw significant opportunities to attract support through the extension of new social protections. By contrast, the two semiauthoritarian systems in the region—Singapore and Malaysia—show much greater continuity in social policy and even a bias toward a liberalizing agenda.

The effects of democracy were much more complex in Latin America’s relatively costly but unequal welfare systems. The turn toward democracy expanded incentives for political entrepreneurs to appeal to marginalized voters but also provided opportunities for organized stakeholders to defend entitlements and institutional prerogatives. At the same time, the collapse of the old development model and deep recessions created especially severe fiscal constraints, and new governments faced strong pressures for liberalizing reforms of social insurance and even outright retrenchment of prior commitments. Given these economic conditions, expansion of the comprehensive social insurance initiatives visible in the East Asian democracies was largely off the table. Instead, reformist technocrats sought to combine far-reaching liberalization of existing social insurance schemes, compensation for some of the most powerful stakeholders, and targeted antipoverty programs aimed at the very poor.

In Eastern Europe, the socialist welfare state included broad segments of the population in a system of entitlements. With democratization, wide coverage created strong electoral and interest group constraints on liberalization and retrenchment, even in the face of strong fiscal pressures. Democracy also provided incentives for parties across the political spectrum to attend to the interests of labor and other groups disadvantaged by the transition to the market.


Our analysis focuses on twenty-one middle-income countries drawn from three regions. In Latin America, we consider Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico, Peru, Venezuela, and Uruguay. In Asia, we examine developments in Korea, Taiwan, Singapore, Malaysia, Thailand, and the Philippines. In Eastern Europe, we cover Hungary, Poland, Romania, Bulgaria, and Czechoslovakia with its successor states, the Czech and Slovak Republics. This choice of countries encompasses all or most of the middle-income countries of the three regions and allows us to consider not only differences across the regions but also variations within them.12 The number of effective cases is much larger, however, because we consider several policy areas within each country and exploit changes in the principal causal variables over time.

As we noted at the outset, our study was strongly motivated by quite striking cross-regional differences in the evolution of welfare systems; we outline these differences in more detail in chapter 1. Of course, there is an ample number of candidates for explaining this variation, and in both chapters 1 and 5 we introduce our comparative case studies with some cross-national statistical analysis. In chapter 1, this analysis considers some correlates of social spending and other measures of social-service delivery, controlling for a number of possible structural determinants. The purpose is to motivate our supposition that regionally distinct patterns of social policy do in fact exist. In chapter 5, we develop a more dynamic panel model that permits a consideration of political and economic changes on the path of social spending.

We believe, however, that the utility of cross-national statistical analysis faces much more severe limitations than is typically acknowledged.13 Not only are the data on social policy characterized by fundamental limitations, but cross-national statistical designs also face a host of problems even when the data are of adequate quality. The tremendous heterogeneity that exists across developing countries places exacting demands on the modeling exercise; more cases can be a minus as well as a plus, because each dimension of heterogeneity must be correctly modeled. As these dimensions multiply, it becomes more and more difficult to arrive at the appropriate specification of the causal model, given a plethora of plausible models and the problem of modeling interactions among the covariates (Gerring 2006, 3). Identification problems, selection problems, and endogeneity are also serious constraints, although they are of concern to qualitative designs as well. Despite their increasing popularity, it is hard to solve these problems through instrumental variable techniques or selection models. Comparative case studies by no means resolve these difficulties, but they do provide a way to gauge the plausibility of contending conjectures.

Finally, comparative case studies provide the opportunity to demonstrate causal relationships in a way that statistical analysis cannot. The advantages of a larger sample and the ability to introduce controls in statistical work are offset by the difficulty of showing that the postulated mechanisms work as proposed—for example, that electoral competition or interest group pressures are the means through which democracy operates on policy outcomes.

Once we established that regionally distinct patterns of policy existed, we faced the question of how to choose cases for more detailed analysis. One proposal from James Fearon and David Laitin (2005) is to randomly select cases from the larger sample. This method may be appropriate if the primary analytic objective is to test theories believed to have a general applicability and validity. As we will argue in more detail below, however, we are more concerned with issues of internal rather than external validity, and as a result, Fearon and Laitin’s random sampling technique struck us as inappropriate. An alternative is to choose a representative or typical case from each region.14 This method requires the identification of the distribution of cases within each regional subsample and the use of typical or modal cases or even ideal-typical analysis, to explore the differences across types (or regions, in our case); well-known studies by Esping-Andersen (1990) and Swank (2002) employ this strategy, as does the varieties-of-capitalism (VOC) literature (Hall and Soskice 2001). However, this method is always subject to the concern that the selected case may not be fully representative—even in the context of a broader statistical analysis—or that it may exhibit some idiosyncratic features.

We opted for a strategy of choosing virtually all the significant cases in the regions of interest to us. Despite the criticisms of such a “medium-n” strategy (including those of our earlier work; see Brady, Collier and Sea-right 2004, 92–94, 100), we believe that this approach is justified for at least three reasons. First, the comparative case analysis is nested in a wider cross-national statistical analysis that allows us to control for a range of other possible determinants of the observed variation. Second, the multiplication of cases increases the confidence that the findings with respect to interregional variations are not driven by unrepresentative or outlier cases and permits intraregional comparison; we return to this issue below.

The third justification for this approach requires somewhat more extended discussion. Our approach is unapologetically historical and configurative. We believe strongly that the phenomena that interest us are best explained as the result of long-run historical processes that vary across regions. We are struck by the fact that numerous cross-national statistical studies implicitly confirm this approach—although without adequate additional exploration—by finding that regional dummies prove to be statistically significant in cross-national models. We are therefore less interested than standard statistical analyses in questions of external validity and whether findings are portable to out-of-sample cases and time periods; to consider these questions, we believe, would require an extended historical consideration of how critical realignments, development strategies, and regime type interacted in altogether different regional settings. We are more preoccupied with questions of internal validity: whether the causal arguments we have outlined do in fact explain the cases in question.

Although we are interested in cross-regional variations, it is also clear that the cases within any given region also exhibit important variations as well, and that these differences provide additional opportunities for testing. Democracy provides an important example. We have already noted how we use differences between long-and short-lived democracies in Latin America to bolster our contention that regime type matters. Similarly, in part 2 we have a number of cases in which authoritarian rule persisted into the period of interest to us here, including Singapore, Malaysia, and Mexico, as well as cases that reverted to authoritarian rule following an initial transition, notably Peru and Venezuela. We also have differences in the political transition path. Although most countries in our sample moved relatively swiftly from authoritarian to democratic rule, we also have cases that passed through somewhat more prolonged semi-authoritarian or semidemocratic phases, including Mexico, Thailand, Taiwan, the Slovak Republic, and Romania. These variations all provide opportunities for testing conjectures about the effects of regime type.

Although we emphasize regional similarities in the economic environment facing new democracies in each region, there are variations, both across cases and over time, on this dimension as well. For example, in part 2 we exploit a comparison between three democracies that transited during periods of high growth—Korea, Taiwan, and Thailand—with the Philippines, which experienced a debt crisis similar to Latin America’s. In Latin America, we sort cases on the basis of the severity of the crises they experienced in the 1980s as well as on regime type and legacies. Similarly in Eastern Europe, we consider differences between early reformers, which generally experienced more rapid recovery from the transitional recessions of the early 1990s, and late reformers that relapsed into crisis.

Narrative accounts also allow us to confirm the results of our cross-sectional time-series models by tracking the effect of changing economic conditions over time. Although we identify the East Asian cases as high-growth ones, all the countries of interest to us experienced the effects of the region-wide crisis of 1997–98. Conversely, the Latin American and Eastern European countries experienced severe crises at the outset of their transitions to democratic rule, but some experienced periods of relatively sustained recovery thereafter (such as Poland) and others fell back into crisis at some later point (Argentina, Brazil, Bulgaria, Romania). Again, these differences provide opportunities for more refined comparisons within the regions, and allow us to test some of our expectations regarding the effects of economic circumstances on the politics of reform.


Our study is framed within the context of an extensive and rich literature on the origins, development, and reform of the advanced industrial welfare states.15 This subfield has moved forward through intense theoretical debate, which has, in turn, spurred the collection of new data and outstanding empirical work across disciplines. We now have a much greater understanding of the evolution and distributive impact of social policy in the advanced welfare states (see Huber and Stephens 2005 for a compact review).

Despite ongoing controversy, the literature has converged around a relatively limited set of core metatheoretical preoccupations, including the role of distributive interests and their partisan representatives; the changing relationship between underlying economic structure, shorter-run economic performance, and welfare commitments; and the mediating role of political institutions. We, too, have framed our work in the context of these broad intellectual preoccupations, and on a number of issues we have come to conclusions that confirm consensus findings in the literature.

However, the literature on the advanced industrial states has been surprisingly insular, geographically confined to the advanced industrial states, and more particularly to an understanding of the European experience. An underlying motivation of this study is not only to extend and adapt this literature to new geographic and historical contexts, but also to place the advanced industrial-state experience in a broader comparative context.

We signal here a few of our findings that may be of particular interest to students of the advanced industrial states, revisiting them in more depth in the conclusion. These findings concern the merits of the power-resource approach; the effects of economic factors on welfare commitments, including both globalization and varieties of market coordination; and the effects of formal political institutions.

A central point of departure for the literature on the origins of the advanced welfare states is the observation that the balance of class power, and particularly the relative strength of encompassing unions and left parties, was an important determinant of welfare commitments. This power resource perspective has been challenged and modified in recent years. The effects of partisanship have been modified by a consideration of the effects of Christian Democratic and liberal traditions of welfare provision. Paul Pierson (1994) has argued that the importance of unions and left parties is likely to decline over time as welfare systems themselves generate broader constituencies of stakeholders and beneficiaries. And most recently, the focus on labor and the left has been supplemented, if not challenged outright, by research showing that private sector actors have their own distinctive social policy interests depending on their underlying production strategies (Estevez-Abe, Iversen, and Soskice 2001; Mares 2001, 2004). Despite these very important challenges and amendments, there is nonetheless ample quantitative and case study evidence that unions and left parties were, in fact, central determinants of the nature and scope of the welfare state (Huber and Stephens 2001, 113).

At the most general level, our study is anchored in an approach that emphasizes the causal significance of organized interests. Our explanation for early welfare commitments is rooted in the concept of critical political realignments, and we find some support for the general proposition that organized-labor and left parties are important actors in the extension of social insurance and services. We also follow Pierson in considering the subsequent influence of what we call the welfare legacy on the maintenance of welfare entitlements.

At the same time, we found ourselves pulled back to discussions of the origins of the European welfare state that emphasized not simply the urban working class but also the political position of the rural sector. Given the weight of this sector in the developing countries of interest to us, this emphasis should not be surprising. Both the comprehensive social entitlements of the socialist welfare state and the relatively egalitarian distribution of social services in a number of the East Asian cases ultimately hinged on the inclusion of the rural sector. This inclusion was not political in the same sense that we talk about the inclusion of the urban working class in Latin America. Nonetheless, the shape of the overall social contract could only be understood by grasping why governments had incentives to reach the countryside.

These considerations of the significance of the rural sector are not unrelated to how we think about the distributive consequences of partisanship and labor power. In Western Europe, the strength and scope of the labor movement seemed almost linearly related to the scope and progressivity of the welfare state. In the developing countries, by contrast, strong labor movements representing a narrow segment of the work force often had regressive implications. Benefits were financed not only by payroll taxes from workers and employers but also by general taxes and inflation levied primarily on low-income groups. Paradoxically, in Eastern Europe and East Asia, repression of the left and labor provided the conditions for policies that encouraged much more egalitarian distribution of social insurance and services and arguably more egalitarian distributions of income. These findings raise important comparative questions about the circumstances under which the interests of labor have progressive or regressive effects for social policy as a whole.

The structure and performance of the economy have also played a major role in the development of the welfare literature. We focus on two economic issues that have received particular attention: the potentially negative effects of globalization, and the consequences of different forms of market coordination on social-policy commitments.

Research on the contemporary effects of “globalization” has yielded highly mixed results. A number of studies have found that the impact of economic openness on the volatility of growth, income, or consumption is much less than thought and might work in the exactly opposite direction;more open economies may be better positioned to manage external shocks (Iversen 2001; Kim 2007). Moreover, this literature finds that the fate of the advanced welfare states has been affected much more substantially by technological and demographic changes or overall economic performance than by globalization, at least as narrowly conceived (Iversen and Cusack 2000).

We argue as well that a narrow focus on economic openness does not capture the most direct and significant economic determinants of social policy. In our sample, the sustainability of different development models, growth, and fiscal capacity all appear as much more important proximate causes of changes in social policy. Although these factors are no doubt influenced by the external environment, our results do not line up with common expectations about openness per se. The more closed economies of Eastern Europe and Latin America ultimately experienced the most profound shocks and the most severe crises, while the high-growth Asian cases combined high levels of economic openness with some of the more pronounced expansions of social entitlements in our entire sample. The effects of globalization on social policy are by no means resolved, but analysis along these lines should not squeeze out equally, if not more, promising lines of economic analysis, most importantly on the long-run fiscal capacity of developing-country governments and the effects of the tax base on the capacity to address the social question.

In addition to the effects of globalization, theories of the welfare state have increasingly focused on variations in the form of market capitalism (Hall and Soskice 2001). Our discussion of alternative development models is inspired by the VOC literature, but it is far from a simple extension of that perspective. The middle-income countries of the three regions pursued a far broader range of economic alternatives than captured in the typologies of the advanced industrial states. Such experiences indicate a need both for a broader understanding of the range of coordination mechanisms and a better understanding about how closed-economy models are being transformed.

The mediating effect of political institutions, finally, is a third important element in the theoretical scaffolding of the welfare-state literature. We have operated from the theoretical premise that democratic systems have institutional mechanisms that encourage redistribution and thus broader and more generous welfare systems. However, these expectations can be undercut both by economic constraints and by pressure from interest groups. Conversely, authoritarian regimes can, under specified circumstances, produce positive distributive effects. The research program on welfare and social policy needs to broaden its horizon beyond the democratic cases, to consider whether the benefits of democracy hold, if not why not, and under which conditions authoritarian regimes may be responsive to the interests of the poor. Such a research project would join the historical work on the European welfare state with the consideration of the developing and socialist systems we consider here.

Return to Book Description

File created: 7/10/2008

Questions and comments to:
Princeton University Press

New Book E-mails
New In Print
PUP Blog
Princeton APPS
Sample Chapters
Princeton Legacy Library
Exam/Desk Copy
Recent Awards
Princeton Shorts
Freshman Reading
PUP Europe
About Us
Contact Us
PUP Home

Bookmark and Share