27390072 capitalism democracy welfare

8/8/2019 27390072 Capitalism Democracy Welfare

1/336
http://www.cambridge.org/9780521848619


2/336

This page intentionally left blank


3/336

Capitalism, Democracy, and Welfare

This book builds on new institutionalist theory in both economics and polit-ical science to offer a general political economy framework for the study of welfare capitalism. Based on the key idea that social protection in a moderneconomy, both inside and outside the state, can be understood as protectionof specic investments in human capital, the book offers a systematic expla-nation of popular preferences for redistributive spending, the economic roleof political parties and electoral systems, and labor market stratication (in-cluding gender inequality). Contrary to the popular idea that competition inthe global economy undermines international differences in the level of socialprotection, the book argues that these differences are made possible by a highinternational division of labor. Such a division allows rms to specialize in pro-duction that requires an abundant supply of workers with specic skills, andhence high demand for protection. The rise of nontraded services underminesthis specialization and increases demands for more exible labor markets.

Torben Iversen is Professor of Government at Harvard University. He is theauthor of Contested Economic Institutions: The Politics of Macroeconomics and WagBargaining (Cambridge University Press, 1999) and coeditor of Unions, Em- ployers, and Central Bankers: Macroeconomic Coordination and Institutional Changein Social Market Economies (Cambridge University Press, 1999). He is also theauthor or coauthor of articles in such journals as the American Journal of Political Science, American Political Science Review, British Journal of Political Science, Com- parative Politics , Comparative Political Studies , International Organization, Oxford Review of Economic Policy, Public Choice, Quarterly Journal of Economics , and World Politics , as well as in numerous edited volumes.


4/336


5/336

Cambridge Studies in Comparative Politics

General Editor Margaret Levi University of Washington, Seattle

Assistant General Editor Stephen Hanson University of Washington, Seattle

Associate Editors Robert H. Bates Harvard UniversityPeter Hall Harvard UniversityPeter Lange Duke UniversityHelen Milner Columbia UniversityFrances Rosenbluth Yale UniversitySusan Stokes University of ChicagoSidney Tarrow Cornell University

Other Books in the Series Lisa Baldez,Why Women Protest: Womens Movements in ChileStefano Bartolini, The Political Mobilization of the European Left, 18601980:

The Class Cleavage Mark Beissinger,Nationalist Mobilization and the Collapse of the Soviet StateNancy Bermeo, ed., Unemployment in the New EuropeCarles Boix,Democracy and RedistributionCarles Boix,Political Parties, Growth and Equality: Conservative and Social

Democratic Economic Strategies in the World EconomyCatherine Boone, Merchant Capital and the Roots of State Power in Senegal,

19301985 Catherine Boone, Political Topographies of the African State: Territorial Authority

and Institutional Change Michael Bratton and Nicolas van de Walle,Democratic Experiments in Africa:

Regime Transition in Comparative Perspective Valerie Bunce, Leaving Socialism and Leaving the State: The End of Yugoslavia,the Soviet Union, and Czechoslovakia

Daniele Caramani, The Nationalization of Politics: The Formation of National Electorates and Party Systems in Western Europe

Kanchan Chandra, Why Ethnic Parties Succeed: Patronage and Ethnic Headcounts in India

(Continued after the index)


6/336


7/336


TORBEN IVERSEN Harvard University


8/336

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, So Paulo

Cambridge University PressThe Edinburgh Building, Cambridge , UK

First published in print format

- - - - -

- - - - -

- - - - -

Torben Iversen 2005

2005

Information on this title: www.cambridge.org/9780521848619

This publication is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.

- - - -

- - - -

- - - -

Cambridge University Press has no responsibility for the persistence or accuracy of sfor external or third-party internet websites referred to in this publication, and does notguarantee that any content on such websites is, or will remain, accurate or appropriate.

Published in the United States of America by Cambridge University Press, New York www.cambridge.org

hardback

paperback paperback

eBook (EBL)eBook (EBL)

hardback
http://www.cambridge.org/9780521848619http://www.cambridge.org/http://www.cambridge.org/9780521848619http://www.cambridge.org/


9/336

Til min mor og far


10/336


11/336

Contents

Tables pagex Figures xiii Acknowledgments xv

Part I: Welfare Production Regimes

1 A POLITICAL ECONOMY APPROACH TO THE WELFARE STATE 3

2 A BRIEF ANALYTICAL HISTORY OF MODERN WELFARE PRODUCTION REGIMES 30

Part II: Political Foundations of Social Policy3 EXPLAINING INDIVIDUAL SOCIAL POLICY

PREFERENCES 774 CREDIBLE COMMITMENT, POLITICAL

INSTITUTIONS, AND SOCIAL PROTECTION 122

Part III: Forces of Change5 COPING WITH RISK: THE EXPANSION OF SOCIAL

PROTECTION 183

6 NEW TRADEOFFS, NEW POLICIES: CHALLENGESOF THE SERVICE ECONOMY 217

Bibliography 279 Index 297

ix


12/336

Tables

1.1. Government Spending and Variation in Spending acrossSeventeen OECD Countries, 19601993 page16

1.2. Electoral System and the Number of Years with Left- andRight-Leaning Governments, 19451998 24

2.1. Structural Preconditions for Economic Policies andInstitutions, ca. 1950 42

2.2. Employment Protection in Eighteen OECD Countries 472.3. Unemployment Protection in Eighteen OECD Countries 502.4. Income Protection in Eighteen OECD Countries,

19731995 532.5. Percentage of Population over 25 with a Postsecondary

Education 542.6. Skill Systems in Eighteen OECD Countries 552.7. Scientic Citation Rates and Low-Wage Service

Employment in Eighteen OECD Countries 612.8. Unemployment in Selected Countries, 19501996 652.9. Average Annual Rates of Growth in Total Factor

Productivity for Fourteen OECD Countries, 19701994 713.1. Summary of Independent Skill Variables 963.2. Support for Social Spending among the Publics of Ten

OECD Countries, 1996 1003.3. Estimates of the Magnitude of the Effects of Independent Variables 102

3.4. Support for Spending in Four Areas of Social Protection 1043.5. Gender Effects on Preferences, Income, and Skills 1063.6. Income, Skills, and Support for Social Spending in Eleven

OECD Countries 108

x


13/336

Tables

3.7. Formal Education and Support for Two Types of Spending in Ten OECD Countries, 1996 110

4.1. Regression Results for Reduction in Inequality 1534.2. Key Indicators of Party and Electoral Systems 1564.3. Electoral System and the Number of Years with Left and

Right Government, 19451998 1574.4. Electoral System and the Number of Years with

Governments Farther to the Left or to the Right Thanthe Median Legislator, 19451998 160

4.5. Regression Results for Government Partisanship,

19501996 1624A.1. Electoral Systems and Incentives of Politicians toCampaign on the Party Platform 175

4A.2. Political Institutions and Capacity for Commitment 1764A.3. Country Means for Variables Used in Regression Analysis 1784A.4. Correlation Matrix 179

5.1. Regression Results for Government Spending 1995.2. Common Shocks, National Institutions, and Government

Spending 2025.3. Shocks and Spending in Two Subperiods 2055.4. Deindustrialization, National Institutions, and

Government Spending 2065.5. Deindustrialization, Partisanship, and Government

Spending 2085.6. Regression Results for Industrialization 2126.1. Real Exchange Rates for Eighteen OECD Countries,

19731997 2226.2. Productivity, Relative Earnings, Labor Shares, and

Private Employment in Four Sectors 2296.3. The Bivariate Relationship between Dispersion of

Earnings and Employment Growth in Three ServiceSectors 232

6.4. The Determinants of Private Service Sector EmploymentGrowth 2366.5. The Effect of Earnings Dispersion under Different

Model Specications 2406.6. The Effect of Taxation under Different Model

Specications 2426.7. Replication of Regression Results Using d5/d1 Ratios 244

xi


14/336

Tables

6.8. Wage Equality, Service Employment, and Institutions inEight OECD Countries, 19701996 251

6.9. Employment and Unemployment in Denmark, TheNetherlands, and Europe, 19901999 260

6.10. The Volume of Work in Twelve OECD Countries,19901999 264

6.11. Part-Time Employment in Twelve OECD Countries,19901999 266

xii


15/336

Figures

1.1. Vocational training and wage inequality page191.2. The percentage of adults with poor literacy scores

(bottom scale) and the percentage of adults with loweducation and high scores (top scale): thirteen OECDcountries, 19941998 21

1.3. Redistribution as a function of taxes and transfers infourteen democracies 22

1.4. Taxes and transfers as a function of vocational trainingactivity 23

1.5. Skill specicity and occupational gender segregation,2000 27

2.1. Social protection and skill proles 582.2. Unemployment replacement rates, 19731995 642.3. The (un)employment gap between continental Europe

and the United States 692.4. The employment gap between continental Europe and

the United States, by sector 703.1. The three states of an individual in the labor market 803.2. Four models of social policy preferences 834.1. The relationship between institutionalization of the party

system, vocational training intensity, and governmentspending 1474A.1. The indifference curves forL and H and the empty LH

win-set of m* 1694A.2. The structure of the coalition game 171

5.1. Spending, trade, and deindustrialization in seventeenOECD countries, 19521995 184

xiii


16/336

Figures

5.2. Support for redistribution as a function of risk 1895.3. Support for redistribution as a function of shocks to

income 1905.4. Deindustrialization and change in welfare spending for

sixteen OECD countries, 19601993 1915.5. Trade openness and losses in traditional sectors 2105.6. Initial size and losses in traditional sectors 2115.7. Convergence toward the service economy, 19601995 2156.1. Wage dispersion and average currency overvaluation,

19931997 225

6.2. Wage dispersion and the relative price of Big Macs,19882000 2276.3. Dispersion of earnings and employment growth in four

economic sectors 2306.4. The service economy trilemma 2486.5. Stringency of employment regulation for regular

full-time employment, 1980s versus 1990s 2596.6. Stringency of employment regulation for temporary and

part-time employment, 1980s versus 1990s 2596.7. Relative wages in The Netherlands, 19801995 262

xiv


17/336

Acknowledgments

This manuscript is the result of more than half a decade of research intothe causes and consequences of social policy, and it is in large measure thefruit of collaboration with numerous outstanding scholars and colleagues.I am particularly grateful to David Soskice who worked with me on every part of this book, especially PartII, which is mainly the result of joint work.David has been a constant source of inspiration, and the book would nothave been possible without his ideas, insights, and encouragement evenas he remains a critic on parts of the argument in this book. In addition,parts of this book have beneted greatly from collaborative work with (in al-phabetical order) Tom Cusack,Barry Eichengreen,Margarita Estevez-Abe,Frances Rosenbluth, and Anne Wren. My primary claim to authorship isthat I am probably the only person prepared to assume responsibility forthe manuscript in its entirety, including all the errors, omissions, and ques-tionable arguments.

Three other people deserve special mention because they read and com-mented on the entire manuscript: Peter Lange, Margaret Levi, and CharlaRudisill. In addition to many helpful substantive comments, Peter Langesadvice on the organization and presentation of the materials, and especially his detailed comments on several iterations of Chapter1, made this a muchbetter book. Margaret Levi organized an extremely useful colloquium on a

preliminary version of the book in the Department of Political Science atthe University of Washington in May of 2002, where she and other partic-ipants (especially Terry Givens and Erik Wibbels) pushed me to be clearerabout the argument and to make signicant improvements to the analysis.I am grateful to two anonymous reviewers for the same reason. Last, butnot least, Charla Rudisill read every word in this book to make sure it made

xv


18/336

Acknowledgments

at least some sense to people who do not spend most of their waking hoursthinking about these issues. She also patiently put up with me while I did just that.

In addition, I have received many helpful comments on parts of thisbook from (in alphabetical order) Jim Alt, Uwe Becker, Pepper Culpepper,Robert Fannion, Jeff Frieden, Geoff Garrett, Robert Goodin, PeterHall, Bob Hanck e, Peter Katzenstein, Lane Kenworthy, Gary King,Herbert Kitschelt, Isabela Mares, Paul Pierson, Jonas Pontusson, NirmalaRavishankar, Philipp Rehn, Ron Rogowski, Fritz Scharpf, Ken Scheve, Michael Shalev, Ken Shepsle, and Wolfgang Streeck. Of these, Robert

Fannion and Nirmala Ravishankar also provided excellent researchassistance. Varioussegmentsofthemanuscripthavefoundtheirwayintoarticlesand

papers. Some of the arguments and data found in Chapters 1 and 2 originatein a paper I coauthored with Margarita Estevez-Abe and David Soskice,published in Peter Hall and David Soskice (eds.),Varieties of Capitalism:The Challenges Facing Contemporary Political Economies , Oxford University Press (2001). Chapter2 also draws on a paper with Barry Eichengreen that was previously published inOxford Review of Economic Policyunder the titleof Institutions and Economic Performance in the 20th Century: Evidencefromthe LaborMarket.Chapter 3, which presents and tests the individual-level implications of the argument, is an expanded version of a paper I wrote with David Soskice for the 2000 Annual Meeting of the American PoliticalScience Association (APSA) in Washington, DC, subsequently publishedin the American Political Science Reviewunder the title An Asset Theory of Social Policy Preferences. The parts in this chapter on gender preferencesappear in a somewhat different form as a section in a 2003 APSA paper with Frances Rosenbluth called The Political Economy of Gender: Ex-plaining Cross-National Variation in Household Bargaining, Divorce, andthe Gender Voting Gap. Chapter 4 is built on two unpublished papers withDavid Soskice.One was presented at the 2002 Annual Meeting of the PublicChoice Society, San Diego, and is entitled Political Parties and the Time

Inconsistency Problem in Social Welfare Provision; the other was rstpresented at the 2002 Annual Meeting of the American Political Science Association in Boston and is entitled Electoral Systems and the Politics of Coalitions:WhySome Democracies RedistributeMore Than Others.Thetwo sections in Chapter 5 on the effects of major shifts in the occupationalstructure build on a paper that I wrote with Tom Cusack and presented atthe 1998 Annual Meeting of the American Political Science Association in

xvi


19/336

Acknowledgments

Boston; it was later published inWorld Politics under the title The Causesof Welfare State Expansion: Deindustrialization or Globalization? (April2000). Finally, sections of Chapter6 draw from a paper written with Anne Wren that appeared in World Politics (July 1998) under the title Equal-ity, Employment, and Budgetary Restraint: The Trilemma of the ServiceEconomy.

xvii


20/336


21/336



22/336


23/336

PART I

Welfare Production Regimes


24/336


25/336

1

A Political Economy Approachto the Welfare State

Printing is one of the worlds oldest industries, and typography is one of the oldest occupations in the industrial economy. Typographers essentially transformed stacks of typed and handwritten manuscripts into a form thatpermitted the mass production of books, newspapers, and journals. Half technicians and half craftsmen, typographers were highly skilled, well paid,andproud harbingers of Gutenbergs revolutionary invention. However, thecraft wasradically transformed over time: rst from hot-metal typesettingto analog typesetting and then to digital CRT (cathode ray tube) andlaser image-setting. In the process of change, previous typesetting skills were swept aside in a matter of a decade or two, and large numbers of typesetters and other printing production workers lost their jobs many by an invention that the printed word helped set in motion: the computer.Lead molds, printing plates, and all the other paraphernalia that went intothe original printing processes were retired to the dusty shelves of industry museums.Butretirementwasnotanoptionforthemajorityoftypographers whose livelihood depended on using the skills they had acquired throughlong apprenticeships and years of learning by doing.

The depth of desperation these workers felt as their industry was trans-formed manifested in bitter strikes across the developed world can beloosely conceptualized as a product of the nontransferability of their skills

and the speed with which their skills were rendered obsolete by new tech-nology minus the availability of public policies such as unemployment in-surance, public health insurance, pensions, retraining programs, and public job creation that all cushion the effects of skill redundancy. And this for-mula for desperation can, of course, be applied not just to typographersbut to all workers past, present, and future who have skills that arelimited in application and can be made obsolete by new technology and

3


26/336


other forces of change. Social scientists are certainly no exception to thislogic. If it were not for the institution of tenure, many Kremlinologists would have had little marketable expertise following the collapse of theSoviet Union.

The point of this story is to highlight a central theme of this book: theimportance of political and economic institutions for protecting the hu-man capital in which people have invested. Job protection, unemploymentbenets, income protection, and a host of related policies such as public re-training programs and industry subsidies, all help to insure workers againstthe loss in asset values when external shocks in technology and labor mar-

ket conditions shift the demand for skills. Indeed, having in place someform of protection is a precondition for people making investments in spe-cic skills in the rst place. High job security, wage protection backed by union power, and guaranteed health and pension benets have encouragedgenerations of young people to follow in their parents footsteps and choosetypography as their trade. And, needless to say, the health of the printingindustry depended on people willing to invest in specic skills. Likewise,the acquisition of specialized knowledge in academia, including that repre-sented by Kremlinologists, would be very risky without some form of jobsecurity, and specialized knowledge is the lifeblood of any major researchinstitution. Even if the institution of tenure was invented as a response tothe Red Scare in the 1920s, its persistence owes much to the fact that it isfunctional to the production of new knowledge.

But social protection is clearly not only about insurance, it is also aboutredistribution and political conict. By this I mean that whereas insuranceis an institutional device for workers to consensually pool their risks andreimburse each other for potential future losses, redistribution is a device wherein money is taken from some workers and given to others in thepresent, without prior consent to do so. When printers unions went onstrike across the industrialized world in the 1970s, it was to seek subsi-dization of their own jobs and income, not to collect an already agreedupon insurance or to guarantee the future reproduction of old typograph-

ical skills. Everyone understood that traditional typesetting as a trade wasdoomed and that protection of current workers served largely distributivepurposes. For the unions, it was a matter of survival, and they fought bit-ter battles, sometimes violent, to delay the introduction of new technology and to force employers to retain their old typographical workers. It is noaccident that the rst publishing houses to introduce new technology, suchas LA Times and Oklahoma City Times , were ones with the weakest unions.

4


27/336

A Political Economy Approach to the Welfare State

As the printing example highlights, the political economy of insuranceand of redistribution are in fact closely intertwined. Policies adopted forinsurance purposes have redistributive consequences, and, as I will arguein detail later in this chapter, redistribution can also sometimes serve in-surance purposes. Indeed, a central contention of this book is that answersto many of the most pressing questions about the relationship betweensocial protection and the economy can be found in the intersection of in-surance and redistribution, or more specically in the interplay of income,skills, and democratic politics. Close linkages exist between workers invest-ment in skills, the international product market strategies of rms, electoral

politics, and social protection. As I have argued with Margarita Estevez- Abe and David Soskice (Estevez-Abe et al.2001), these linkages have beenorganized into distinct welfare production regimes in different coun-tries, each associated with its own political-economic dynamic and rein-forced, not undermined as often presumed, by the international division of labor.

Standard approaches to the welfare state fail to account for the rela-tionship between production and social protection, and they leave behinda number of key questions that any political economy approach to socialprotection needs to answer. For example, if social protection underminesmarkets, as commonly argued, why is there no apparent relationship be-tween the generosity of such protection and economic growth? A relatedquestion is why globalization has not led to a competitive race to the bot-tom as many feared. Indeed, it seems to be the rise of sheltered, nontraded,services that has prompted some governments to embark on labor mar-ket deregulation. To understand why, we need to examine the intersectionbetween welfare production regimes and the creation of comparative advantages in the international economy. The same is true if we want to un-derstand why employers are not universally opposed to generous socialprotection, and why they continue to invest heavily in economies with highsocial spending despite the widely held view that such spending is detri-mental to business interests.

Even traditional distributive politics, I submit, is not well understoodin the existing literature. Though there is considerable evidence that classpolitics matters, why is distributive politics played out so differently in dif-ferent countries? The fact that partisan politics is systematically biased tothe left in some countries but to the right in others is not in any straight-forward way related to the power of unions or the size of the traditional working class. For example, it is striking that the decline of the industrial

5


28/336


working class and their unions has been associated with a rise, not a col-lapse, in the political support for the welfare state. Also, countries with themost skewed distribution of income, where standard class arguments wouldpredict the most radical redistribution, are in fact the least redistributive. The solution to these puzzles, I argue later, is to be found in the interplay of insurance and redistributive incentives to support the welfare state, as well as in the political institutions that translate these motives into policy.In turn, preferences for social protection are explained by the key assets,especially skills, that economic agents have invested in.

In the rest of this chapter, I rst provide a more thorough critique of the

existing literature and introduce the key concepts and causal mechanisms inthe asset or welfare production regime argument (Section1.1). I then spellout the implications of the argument for understanding the role of elec-toral politics (Section1.2) and the relationship between the internationaleconomy, economic change, and the rise of social protection (Section1.3). Inally explore how the approach can help explain cross-national variance insome of the key dimensions of inequality and redistribution (Section1.4).

1.1. Toward a New Approach to the Study of the Welfare State

As the printing industry example suggests, the ability of management tointroduce radical new technology is undermined by strong unions and la-bor market regulation. Indeed, the notion that these institutions, and the welfare state more generally, erode the market is a central theme amongneoclassical economists and political sociologists alike. According to those who take this view, labor is an anonymous commodity, easily aggregatedinto a single factorL, where each constituent unit (worker) is replaceable,easily redundant, and atomized (Esping-Andersen1990, p. 37). Logically,the opposite of market (or commodication) is state (or decommodica-tion). It means that a person can maintain a livelihood without reliance onthe market (Esping-Andersen 1990, p. 22). The welfare state transforms L into not- L, and thereby set the worker free: free to organize, free to op-

pose capital, free to be an individual rather than a commodity. Again in the words of Esping-Andersen: Decommodication strengthens the workerand weakens the absolute authority of the employer. It is for exactly this rea-son that employers have always opposed decommodication (1990, p. 22). The welfare state is politics against markets (Esping-Andersen1985),andthe historical strength of the political left, mediated by alliances with themiddle classes, determine how much welfare state and how much market

6


29/336


you end up with (Korpi1983, 1989; Esping-Andersen 1990; Huber andStephens 2001).

The power resources model of the welfare state as it is known is thedominant approach to the study of the welfare state. It suggests that the welfare state is built on the shoulders of an unwilling capitalist class, who will be looking for any opportunities to unburden itself. This is also acentral theme in the burgeoning literature on globalization where the exitoption for capital presents precisely such an opportunity. As WolfgangStreeck explains in the case of Germany: Globalization, by increasing themobility of capital and labour across national borders, extricates the labour

supply from national control and enables the nancial sector to refuse doingservice as a national utility (Streeck 1997). In a similar vein, Dani Rodrik concludes that integration into the world economy reduces the ability of governments to undertake redistributive taxation or implement generoussocial programs (Rodrik 1997).

Indeed, if welfare capitalism is primarily about decommodication andexploitation of the rich, one should have expected capitalists to shun pro-ductive investment in large welfare states well in advance of the onset of globalization in the 1980s. Perhaps globalization has made the tradeoff be-tween redistribution and investment steeper because of expanded menuoptions for capital, but as argued by Lindblom (1980), Przeworski (1986),and others, economic performance has always depended on the cooper-ation of capital. Yet, the remarkable fact about the observed relationshipamong levels of public spending, investment, and national income in advanced democracies is that there is none (Lindert1996). Or if there is one,it is so weak that it does not appear to have imposed much of a constraint ongovernments ability to spend and regulate labor markets. Among democ-racies, the countries with the largest welfare states are no poorer, or richer,than countries that spend much less.

In recent years, an alternative approach to the welfare state has emerged, which emphasizes the role of employers. Contrary to the power resourcesmodel, Peter Swenson (2002) shows through careful archival research that

employers played a proactive role in the early formation of social policy.Swenson argues that in tight labor markets employers will seek to takesocial benets out of competition by creating a uniform, national socialinsurance system. When labor markets are slack, on the other hand, high-cost producers may feel compelled to impose costs on low-cost producersthroughmandatory social insurancearrangements. Swenson argues that therst logic helps explain early welfare reforms in Sweden, while the latter

7


30/336


helps explain salient features of the New Deal legislation in the UnitedStates.

In a similar vein, Isabela Mares (2003) has argued that companies andindustries that are highly exposed to risks will favor a social insurance sys-tem where costs and risks are shared, leading these employers to pushuniversalistic unemployment and accident insurance. This is remarkablebecause universalism is usually associated with strong unions and left gov-ernments. Mares also suggests, and this idea is emphasized in this book, thatsocial protection may encourage the acquisition of skills in the labor force, which in turn enhances the ability of some rms to compete in international

markets. Consequently, for example, some high-skill rms favor generousunemployment insurance.In a recent dissertation on the German welfare state, Philip Manow

(2002) has likewise advanced the thesis that the German system of socialprotection, through a process of institutional coevolution, emerged as animportant complement to the collective bargaining system, which in turnunderwroteunion wage restraint and international competitiveness. By del-egating much of the responsibility forsocial policy to the social partners, theinstitutional incapacity of the German state to guarantee full employment(as a result of federalism, an independent central bank, etc.) was compen-sated for by a social system that provided very high levels of insurance in theevent of unemployment and other shocks to income. In earlier work, PeterBaldwin (1990) also challenges the notion that the welfare state was erectedby the industrial working class alone, against the will of the middle classes. Much universalism in the social democratic welfare states of Scandinavia,for example, was the result of pressure by farmers and other nonworkersat the turn of the century to be included in social programs that served asinstruments of insurance as much as tools of redistribution.

The evidence presented by Lindert, Mares, Swenson, Manow, andBaldwin strongly suggests that social protection cannot be conceived exclu-sively in terms of simple dichotomies between the state and the market, orbetween commodication and decommodication. We need a politicsof

markets rather thana politicsagainst markets,or, more precisely, a theory that acknowledges that social protection can improve the operation of mar-kets as well as undermine them. Building on Estevez-Abe et al. (2001), thisis precisely what this book aims to provide. It develops an approach to pro-duction and labor markets in which the role of social protection is explicitly modeled. The theory reconciles the controversy between the power re-sources perspective and the new employer-focused approaches, and it also

8


31/336


links the study of the welfare state to recent work on the importance of democratic institutions for social policy.

At the heart of the difculties in the standard view of the welfare stateis a neoclassical conception of markets that largely ignores differentiatedskills. Although unskilled day workers can sensibly be analyzed as an undif-ferentiated factor L, and although such labor can be exchanged efciently insomething that approximates spot markets, in Beckers (1964) seminal work and in new institutional economics, these conditions are considered the ex-treme of a continuum. At the other extreme, you nd workers with highly asset-specic investments in skills Ls, where s = (1, 2, 3, . . . , n) refers to

differentiated skills coupled with nonmarket institutions that protect andmanage these investments.Of course, workers are not the only ones who invest in specic assets;

rms, merchants, and virtually any agent involved in economic exchange doalso. And because economic agents are engaged in exchange, and becausethey sometimes own the assets jointly, most assets are cospecic in the sensethat they tie together the welfare of people and make them dependent onone another. For every worker whose livelihood is tied to a specic skill,there is an employer who depends on the worker with those skills. As arguedby Polanyi (1944), Williamson (1985), North (1990), and others, when aneconomy is characterized by heavy investment in such cospecic assets,economic agents are exposed to risks that make efcient market exchangeproblematic. A precondition for such an economy to work efciently isa dense network of institutions that provide information, offer insuranceagainst risk, and permit continuous and impartial enforcement of complexcontracts. In the absence of such institutions, exchange is possible only at asmall scale in local trading communities where people know each other welland engage in repeated face-to-face interactions.1 At a larger scale, marketsleft to their own devices either will fail to produce much exchange, will beaccompanied by costly and continuous haggling, or will involve only very homogeneous types of assets ( L as opposed to L s ).

Nowhere is the importance of institutions more evident than in the labor

market where the welfare state plays a key mediating role. Social protectionis particularly important in solving market failures in the formation of skills. Without implicit agreements for long-term employment and real wagestability, investment in skills that are specic to particular jobs, rms, or

1 Marshalls concept of industrial districts likewise emphasizes repeated interactions in local-ized settings as a precondition for efcient outcomes (Marshall1922).

9


32/336


industries will be suboptimal. In the absence of insurance, workers shunsuch investments because unanticipated shocks to the economy, whether asa result of recessions or technological change, can prevent workers fromreaping the returns on their investments. Employers will also be reluctantto invest in their employees skills, or in equipment that requires those skills,unless they believe that institutions that prevent poaching and discourageunions from exploiting the potential holdup power that specic skills conferare in place.

The importance of asset specicity is already well understood for ex-plaining other policy areas. For example, when there is little credible pro-

tection of property rights, property owners will be more inclined to holdtheir wealth in liquid assets that can be quickly moved from one jurisdictionto another (Bates, Brock, and Tiefenthaler 1991). Even when basic prop-erty rights are well protected, investments vary signicantly in the degree of their asset specicity. When investors cannot trust suppliers or employeeson whose cooperation they depend, they will shun investments in relation-specic assets and rely instead on anonymous market transactions whereone supplier or employee can easily be replaced by another. Conversely, when investments in physical assets are specic to a particular location,supplier network, or employee relationship, rms are more prone to lobby the state for protection against uninsurable risks (Frieden1991; Alt et al.1999).2 In the most general varieties of capitalism (VoC) formulation, na-tional or regional institutions act as complements to the strategies of rms,allowing them to make better use of their assets (Hall and Soskice2001).

A similar logic applies to human capital. When skills are specic to a par-ticular rm, industry, or occupation, their owners are exposed to risks for which they will seek nonmarket protection such as protection of jobs, stan-dardization of wages, or state-guaranteed benets. Skills that are portable,by contrast, do not require extensive nonmarket protection, and when thereis little protection, investing in such skills is the best insurance against adverse market conditions and technological change. Yet, despite its intuitiveappeal, asset specicity plays virtually no role in existing explanations of

the welfare state. Labor isL, and workers are replacable, easily redundant,and atomized. Correspondingly, politics is labor against capital,L againstC . By contrast, the approach offered in this book emphasizes the critical

2 Alt et al. (1999) shows empirical evidence that lobbying rises with the asset specicity of industries. See also Alt et al. (1996) for a more theoretical treatment of this and relatedarguments concerning the importance of asset specicity.

10


33/336


importance of the level and compositionof human capital ( L s ) for explainingthe character of the welfare state thelevel because it determines incomeand hence workers demand for redistribution; thecompositionbecause itdetermines workers exposure to risk and hence their demand for insur-ance. It is natural to label this an asset theory of the welfare state, althoughpolitical institutions are also important as we will see in a moment.

The link between assets and the welfare state explains the continuedand even growing importance of social policy in advanced, and thereforehuman-capital-intensive, economies. In 1999, for example, American work-ers over the age of 25 with a four-year college degree earned an average

of $47,400 compared to $26,500 earned by workers with a high schooldegree and $16,900 earned by workers who had less than a high school de-gree (U.S. Census Bureau2000). Ignoring other group differences, havinga college degree is equivalent to a 3 percent real return on a net fortuneof about $925,000 (compared to someone with less than a high school de-gree). For comparison, the median net worth of an American household,most of which is tied up in real estate wealth, is $53,000 (Wolff 1998).3 And, of course, some of this wealth reects accumulated past returns onskills. Human capital is, thus, easily the most import asset for a majority of people.

Do ordinary people also worry about protecting the value of these assets? The answer obviously varies from individual to individual according to thelevel and mobility of her skills, but there is no question that many workersface a substantial risk that their training can be made partially or entirely redundant by new technology or other forcesof change (as in the example of typographers). Taken as a whole, manufacturing employment in the OECDhas been cut in half since the 1960s, and a large portion of the jobs that re-main require substantially different skills. There is every reason for workersand their unions to concern themselveswith insurance against income lossesas a result of redundant skills, although it is hard to quantify.4 And such in-surance cannot be provided exclusively through the market as a result of well-known problems of moral hazard, adverse selection, and other market

3 These are 1995 numbers expressed in 1999 dollars.4 One of the difculties of quantifying the specicity of skills is that wage and social protection

systems are set up to reduce the riskiness of specic skills. Skill certication and wagestandardization by skill categories, for example, are ways for unions to prevent individual workers from experiencing large drops in income. Variability of wages is therefore not anindicator of asset specicity. Chapter3 goes to considerable length in developing alternativemeasures of skill specicity and to tie such specicity to social policy preferences.

11


34/336


failures (Przeworski,2003 Ch. 11). Writing complete contracts to cover allcontingencies in a labor market characterized by incomplete informationand highly specic investments is precisely what transaction cost economicsrules out. Nearly all unemployment insurance, for example, is providedthrough public insurance; health insurance markets always produce limitedcoverage, and there is no effective private insurance against the adverseeffects of technological change on earnings capacity. For the vast major-ity of people in advanced democracies, insurance against job and incomeloss comes from the state and, to a lesser extent, from individual savings.

1.2. Bringing Electoral Politics Back in

As implied earlier, employers who are pursuing product market strategiesthat require specic skills also have a vested interest in social policies thatreduce the risk of acquiring those skills (Mares1998, 2001; Estevez-Abe1999a; Swenson 2002). The focus on employers, however, tends to leavethe democratic state, electoral politics in particular, underexplored. The

power of employers is primarily structural in nature, but governmentsmust win elections to stay in power, and it cannot be assumed the electoralincentives of politicians areperfectly aligned with their economic incentives(Elkin 1985; Block 1994). In Swensons (1997) account of the New Deal,for example, politicians are faithful representatives of employers long-teminterests, yet they face an urgent need to accommodate popular demandsfor political action. Indeed, Swenson acknowledges that business often op-poses such action, yet it somehow ends up beneting from it.5 But electoralpolitics operates according to its own dynamic, and more often than notthis dynamic is powerfully affected by popular demands for redistribution. As Stephen Elkin explains, the democratic impulse is egalitarian, becauserule by all requires not only political equality but also economic equality sufcient not to vitiate the premise of equal participation (1985).

One of the great strengths of the power resources model is that it hasa credible account of electoral politics. But the role of democratic poli-

tics must be reconsidered in the context of a different understanding of theeconomy and of employer interests. The emphasis on class interests ignoresthe importance of insurance motives in peoples demand for social protec-tion, and, as noted earlier, it leaves us with the puzzle of why democracies

5 See Hacker and Pierson (2002) for an extensive critique of Swenson and related work onthe role of employers in the rise of the welfare state.

12


35/336


with high inequality are not more redistributive than democracies with lowinequality. The key here is to understand that redistribution (the focus of the power resources model) and insurance (the focus of institutional eco-nomics and VoC approaches) are intimately related. Insurance against theloss of income not only is conducive to investment in risky assets but alsohas the effect of redistributing income. This is obviously the case of theunemployed, who have no income, but it applies much more widely to any social protection, such as health insurance or pensions, that is not com-pletely dependent on current employment and income. Ex ante, or behindthe veil of ignorance as Rawls would say, people may support policies for

purely insurance reasons, which, ex post (after the veil is raised), will redis-tribute income. Conversely, policies that are deliberately redistributive willsimultaneously serve insurance functions. Those who are unemployed, sick,old, and have low pre-tax income more generally, will rationally press forredistribution. But by doing so, many of those who are employed, healthy, young, and enjoying a high incomes will enjoy some measure of insuranceagainst losing these goods.

This Janus-face of the welfare state means that it is unlikely to be under-stood simply as a tool of power as a complement to the economy. The wel-fare state is simultaneously an arena for distributive strugglesand a sourceof comparative advantage. Those who see only the rst face will tend toconclude that it is an impediment to market capitalism and that it can sur- vive only if capital is held captive and labor is politically strong. Those whosee only the second face tend to reduce democratic politics, and electoralpolitics in particular, to a symbolic game where the welfare state alwaysmirrors the needs of the capitalist economy (or employers), trumping thepursuit of competing interests. To understand the welfare state, we mustunderstand how popular preferences for social insurance and redistributionare rooted in peoples position in the economy, how these preferences areaggregated into social policies, and how policies in turn affect individu-als investments into assets that shape economic performance and interests.Chapter 3 presents a theory of social policy preferences in which individu-

als who have made risky investments in skills demand insurance against thepossible future loss of income from those investments. Modeling popularpreferences for social protection as a function of the assets people own inthe economy is the rst departure from the power resources approach tomass politics.

The second departure ismyattention to the specic designof democraticinstitutions. Consider, for example, that because social insurance may only

13


36/336


be enjoyed by the current median votersome time in the future, the median voter has an incentive to support such insuranceonlyif future median votersdo the same. The current median voter, therefore, faces a problem of howto commit future median voters. This translates into a time-inconsistency problemfor the government because it has an incentive to renege on itspromise to the current median voter when it seeks to attract the supportof the future median voter. Again, the reason is that the median voter atany given time, when choosing a policy for the present, does not have aninterest in high transfers. This problem is addressed in Chapter4.

One solution points to the role of institutions that can hold the gov-

ernment to its promises about future policy. The organization of politicalparties and their relation to private groups is particularly important in thisrespect. Another solution builds on the close relationship between redistri-butionandinsurance. Because redistributionalso servesinsurance purposes,institutions that promote redistribution serve as (imperfect) solutions to thetime-inconsistency problem. In Chapter 4, I use recent work on the eco-nomic effects of political institutions by Persson and Tabellini (2000, 2003)and others to show that redistribution is intimately related to the electoralsystem (and that the electoral system is also closely associated with thepresence of responsible and programmatic parties).

1.3. Globalization, Deindustrialization, and the Expansionof Social Protection

As noted previously, it is a puzzle that globalization has not led to convergence in social protection. The coupling of social protection and skillsystems helps us understand the puzzle by pointing to their effect on theinternational product market strategies of companies and the creation of comparative advantages in the global economy. Specically, where there isa large pool of workers with advanced and highly portable skills and wheresocial protection is low, companies enjoy considerable exibility in attract-ing new workers, laying off old ones, or starting new production lines. This

exibility allows for high responsiveness to new business opportunities andfacilitates the use of rapid product innovation strategies. Such capacities arelower for rms in economies that rely heavily on nontransferable skills andthat protect these skills through restrictions on the ability of rms to hireand re workers. On the other hand, the latter types of welfare-productionregimes give a comparative advantage to companies that compete in mar-kets where there is a premium on the ability to develop deep competencies

14


37/336


within established technologies and to upgrade and diversify existing prod-uct lines continuously what Wolfgang Streeck in a seminal article hasdubbed diversied quality production (Streeck 1991).

The international division of labor not only perpetuates particular prod-uct market strategies but is also likely to feed back into political supportfor existing social protection regimes. As countries specialize in productionthat uses abundant factors intensely, demand by the owners of those factorsfor protection of their value rises. Contrary to the popular notion of a so-cial race to the bottom, differences across countries, therefore, need notdisappear with a deepening of the international division of labor a propo-

sition implied by Hall and Soskices concept of comparative institutionaladvantage (Hall and Soskice2001). Social spending in continental Europecontinues to be much higher than in Ireland and the Anglo-Saxoncountries,and in many areas the gap has increased. Moreover, whereas labor marketshave become even more deregulated in the latter countries, employmentprotection for full-time employees has stayed high and largely unchangedin the former (OECD 1999b).6

The asset theory of social protection also suggests a different expla-nation of the expansion of the welfare state than is offered by either thepower resources model or theories emphasizing the role of the interna-tional economy. One of the most remarkable facts about the welfare stateis that public spending did not vastly differ between the United States,continental Europe, and Scandinavia in the early 1960s (Rothstein1998).In the 1960s, writes Rothstein, the difference between these countriesin total public spending was much smaller [than today] the level in theUnited States was about 28 percent compared to a mean of 29 percent forthe Scandinavian countries. This does not mean that basic differences inunemployment, employment, and wage protection through labor marketinstitutions did not exist at that time. They did, but the role of the statein the social insurance system through taxes and transfers was not terribly dissimilar.

The tremendous expansion of social spending since then, and the in-

creased variation across countries, can be gleaned from Table1.1. It showstotal government spending as a percentage of gross domestic product(GDP) across seventeen OECD countries, the standard deviation of spend-ing in these countries, and the difference in spending between Sweden and

6 Therearehowever signicant reforms in the regulation of part-time andtemporaryemploy-ment, as well as in a range of social transfer programs, which will be discussed in Chapter6.

15


38/336


Table 1.1. Government Spending and Variation in Spending across Seventeen OECDCountries, 19601993

Government Difference betweenSpending as Standard Deviation Swedish and U.S.Percentage of GDP of Spending Spending

1960 28.7 4.9 3.01965 31.8 5.2 8.11970 35.2 6.4 9.11975 42.2 6.9 13.11980 45.3 8.9 25.61985 48.0 9.1 27.11990 46.3 8.4 24.61995 49.0 9.0 31.0Ratio of 1995 level 1.71 1.85 10.38to 1960 level

Notes:Government spending includes government consumption, includes government trans-fers, plus interest payments and subsidies.Source:OECD, National Accounts , Part II: Detailed Tables (various years).

the United States from 1960 to 1995. Note that spending rose by about70 percent during this period, from 29 percent in 1960 to 49 percent in1995, but the variation in spending grew even faster. Thus, the standarddeviation in spending increased by about 85 percent in this period, and thedifference between Sweden and the United States ballooned from 3 percentto 26 percent of GDP a tenfold increase.

The power resources model attributes this growing gap to differences in working class power. But as noted in the introduction to this chapter, it isawkward to emphasize the role of the industrial working class in the postwarrise of the welfare state because it has been on the decline everywhere. InChapter 5, I show that there is also little empirical evidence for the otherprominent argument that the growth of the welfare state is the result of increased exposure to risks in the international economy (Cameron1978;

Garrett 1998; Rodrik 1998). As I have argued with Thomas Cusack (Iversenand Cusack 2000), the asset theory points to a quite different force of change, one that is in some respects the opposite of globalization: the tran-sition to a largely sheltered service economy. Because deindustrializationrepresents a serious threat to those workers who have made signicant investments in rm- or industry-specic skills a threat that cannot easily beaddressed within the private system of protection in the labor market it

16


39/336


is associated with a rise in electoral demands for public compensation andrisk sharing. Additionally, even though deindustrialization has occurred ev-erywhere, the speed at which ithas taken hold has variedconsiderablyacrosscountries.

More importantly, the effects of deindustrialization have been mediatedby the skill regime, as well as by the institutional capacity of the political sys-tem for credible commitment. Building on recent work on unemploymentby Blanchard and Wolfers (2000), I demonstrate this institutional condi-tioning of common shocks with a variety of empirical tests in Chapter5.

The growing electoral pressure for government spending has also pro-

vided politicians and political parties with an opportunity to shape thestructure of social protection according to ideological preferences. In thisrespect, I am entirely in line with scholars such as John Stephens andGeoffrey Garrett who underscore the importance of partisan politics. As Anne Wren and I have argued (Iversen and Wren 1998), a particularly contentious partisan issue has concerned the extent to which the stateshould expand publicly provided services. Because high-protection coun-tries with extensive wage and employment regulation have created rela-tively few jobs in low-productivity services, and because this is where thepotential for job growth (especially for women) is greatest, social demo-cratic parties have favored an expansion of jobs in public services whileChristian democratic parties have emphasized transfers and social servicesprovided through the family. Liberal parties, by contrast, have advocatedderegulation.

A critical issue examined in Chapter6 is the relationship between socialprotection, especially a relatively at wage structure, and employment. Al-though high-protection countries have been very successful in internationalmarkets, belying the notion that high protection reduce competitiveness,they have been poor employment performers in nontraded private services(Iversen and Wren 1998). At the same time, good employment performerssuch as the United States have paid a heavy price in the form of greaterinequality. The underlying problem, I argue, is that lack of international

trade inserviceshas undermined the ability of countries to takeadvantage of their comparative advantage. High-protection countries, for example, havesqueezed out low-skill jobs without an offsetting expansion of high-skill jobs. I call the emerging response selective and shielded deregulation, which means that greater exibility in parts the labor market (especially forpart-time and temporary employment) is coupled with new tax and trans-fer policies to shield the inequalizing consequences. I assess the limits and

17


40/336


possibilities of this strategy and compare it to the welfare reforms charac-terizing Anglo-Saxon countries.

1.4. Implications for Inequality and Redistribution

As pointed out earlier, there is a close afnity between the insurance anddistributive aspects of the approach. Building on Estevez-Abe et al. (2001)in this section, I suggest how the asset argument can be extended to unravelthree sets of previously neglected causes logics by which welfare productionregimes affect distribution. These propositions help explain several of the

remaining puzzles noted in the opening to this chapter. They are elaboratedand tested more extensively in subsequent chapters.First, general skill systems are more likely to generate wage inequality

and poverty traps because they limit opportunities and incentives for skillacquisition at the low end of the academic ability distribution. The skillsystem is also related to the wage-setting system, which strongly affects theearnings distribution. This helps explain why welfare production regimesare linked to wage dispersion. Second, demand for insurance against socialrisks leads to signicant redistribution of income through the welfare state,and redistributive pressures are accommodated by their insurance benets. This helps explain why the welfare state is so broadly supported in somecountries, despite modest levels of inequality. Finally, gender inequality inthe labor market is intimately related to skill and social protection regimes,and such inequality, unlike wage inequality, tends to be higher in specicskills systems. The skill argument helps us understand why that is the case. All in all, specic skills systems tend to be notably more egalitarian and re-distributive than general skills systems, but labor markets in these countriestend to be more gender segregated.

1.4.1. Skills and Wage Inequality

It is striking, though not surprising, that all countries with a strong empha-

sis on industry-specic skills have developed effective wage coordinationat the industry level. Conversely, general skills countries, especially coun-tries with a strong emphasis on rm-specic skills ( Japan in particular),lack such coordination. Very extensive evidence has in turn been accu-mulated and veries that the structure of the wage bargaining system hasimportant consequences for the wage structure (see especially Rowthorn1992; Freeman and Katz 1994; Iversen 1999; Wallerstein 1999; and Rueda

18


41/336


Figure 1.1 Vocational training and wage inequality. Notes:The d1/d9 wage ratios are the earnings of workers in the top decile of the earningsdistribution relative to a workers in the bottom decile of the earnings distribution. The inci-dence of vocational training is the share of an age cohort in either secondary or postsecondary (ISCED5) vocational training.Sources:UNESCO ( 1999); OECD, Electronic Data Base on Wage Dispersion(undated).

and Pontusson 2000). As implied by the skill argument, intraoccupationalcompression of wages serves as a complement to employment and unem-ployment protection because it helps ensure against a big drop in income if a worker loses his or her job. Collective bargaining at the industry or higherlevels also gives low-income groups inuence over the distribution of wagesthat they lack in the market. Such inuence tends to promote equality.7

But the skill system itself is also important as suggested in Figure1.1. This graph uses the incidence of vocational training as an indicator of the

extent to which workers are acquiring specic vocational skills as opposed

7 This inuence also reduces wage differentials between skills categories, which is contrary tothe goal of maintaining high wage protection for the employed (i.e., maintaining stable wagedifferentials across occupations). The problem was particularly great in Sweden and led toa revolt against centralized (though not coordinated) bargaining among skilled workers andtheir employers (Iversen 1996; Pontusson and Swenson 1996).

19


42/336


to general academic skills.8 Note the strong empirical association betweenskills and earnings equality. Because specic skills systems generate high de-mand for workers with good vocational training, young people who are notacademically inclined enjoy career opportunities that are largely missing ingeneral skills systems. Whereas a large proportion of early school leavers inthe former acquire valuable skills through the vocational training system,in the latter most early school leavers end up as low-paid unskilled workersfor most or all of their working lives.

This pattern also implies that young school goers in specic skills sys-tems have strong incentive to work hard in school, whereas the same is

not true for students in general skills systems who do not expect to goon to college.9 Although there are clearly alternative interpretations, datafrom standardized international literacy tests are consistent with this ideaas suggested in Figure1.2. Countries are ranked by their share of poorperformers, and the roughdivision between general and specic skills coun-tries has been indicated with brackets.10 Whereas the percentage with thelowest score averages 20 in Ireland and the Anglo-Saxon countries, thecomparable gure in countries emphasizing more specic skills is 10. Thecorrelation between vocational training intensity and the percentage withlow test scores is .73. Likewise, those who leave school without an uppersecondary education tend to have much higher test scores on job-relevantskills (here measured by document literacy) when they are in specic skillsrather than in general skills systems: 46 percent of early school leavers inthe former have high literacy scores compared to only 28 percent in thelatter.

In combination, the wage bargaining system (i.e., whether it is industry coordinated or not) and the skill system (i.e., whether it is specic skills orgeneral skills biased) provide a powerful explanation of earnings inequality.It points to the importance of paying attention to factors outside the welfarestate that affect distribution. Much of the welfare state literature fails todo this, notwithstanding its almost exclusive focus on distribution. In the welfareproductionregimeargument,nonstate institutionsare integral parts

8 Measures of skills will be discussed extensively in Chapters2 and 3.9 This idea was suggested to me by David Soskice.

10 Country coverage is limited. One country, Switzerland is excluded from Figure 1.2 becausethe test was administered in different areas using only one of the three ofcial languages;consequently, many people took the test in a nonnative language. This practice appears tohave signicantly affected test scores (OECD and Statistics Canada2000, 56). If included,Switzerland is positioned below Belgium, where a similar issue may be at play.

20


43/336


of the story even though the main functions of these institutions may beinsurance and efcient management of cospecic assets.

1.4.2. Social Insurance and Redistribution

As I noted previously, workers who are behind Rawls veil of ignorance donot know with certainty how they will fare in terms of future employmentand income. In this situation, risk-averse people will demand insuranceagainst loss of employment and income. If these preferences are translatedinto policy, whenthe future arrives, and some workers have experienced a

drop in income, the distribution of income will be more egalitarian than without insurance. A generous tax and transfer system will therefore resultin redistribution of income even if the system is solely intended for social in-surance purposes. A related logic works from redistribution to insurance. If pressure for redistribution produces a more egalitarian after tax and transfer

Figure 1.2 The percentage of adults with poor literacy scores (bottom scale) andthe percentage of adults with low education and high scores (top scale): thirteenOECD countries, 19941998. Notes:The top bars (using top scale) show the percentage of adults who have not completedan upper secondary education but have high scores on document literacy. The bottom bars(using bottom scale) show the percentage of adults taking the test who get the lowest score,averaged across three test categories.Source:OECD and Statistics Canada (2000).

21


44/336


5

10

15

20

25

30

35

40

-3 -2 -1 0 1 2 3 4

Taxes and transfers

R e d

i s t r i b u t

i o n

Denmark

Sweden

Australia

Netherlands

USA

UK

Belgium

Canada

Finland

France

SwitzerlandItaly

Germany

Norway

Figure 1.3 Redistribution as a function of taxes and transfers in fourteen democ-racies. Notes:Redistribution is the percentage reduction in the Gini coefcient from pre- to posttaxand transfer for households with working-age adults. Taxes and transfers is total taxes asa percent of GDP plus total transfers as percent of GDP, after both measures have beenstandardized. Both measures were developed by Bradley et al. (2003) based on LIS data andOECD national accounts data.

distribution of income, such redistribution will serve an insurance function. This is particularly important to understand because redistribution, unlike

provision of social insurance, does not imply a time-inconsistency problem. Also it helps us understand why countries that redistribute income regularly perform better than we would expect from a standard neoclassical analysis.

Chapter 4 explores this interaction between redistribution, insurance,and the economy in detail. A key proposition is that much observed redis-tribution can be attributed to the political support for insurance. Here thisbasic idea can be illustrated with some data on pre- and posttax and transferincome from the Luxembourg Income Study (LIS; see Bradley et al.2003for details).11 The inequality measure is the Gini coefcient, and redistri-bution is the percentage reduction in the Gini from the pre- to posttax

11 Im using the Bradley et al.(2003) data where pretax and transfer income consists of incomefrom wages and salaries, self-employment income, property income, and private pensionincome, while posttax and transfer income is disposable personal income, including allmarket income, social transfers, and taxes.

22


45/336


and transfer distribution of income. To ensure that the measure is relatedto the concept of wage protection, the data only include the working agepopulation. The data are available for fourteen advanced democracies andrepresent averages for a period starting in the late 1970s and ending in themid-1990s (time coverage varies by country).

Figure 1.3 shows the reduction inpre- and posttax and transfer inequality asafunctionoftheleveloftaxesandtransfers.Asexpected,thereisapositiverelationship (r = .68), and, although we do not know from this relationshiphow much is the result of a deliberate attempt to produce redistribution asopposed to insurance, Huber, Stephens, and their associates have found a

strong positive relationship between tax and transfers and redistributionafter controlling for the partisan preferences of governments and a host of other factors (Bradley et al.2003). In fact, their ndings indicate that thelevel of taxes and transfers (what they term welfare state generosity) isone of the most important determinants of both redistribution and poverty reduction. Theydo not,however, explain welfare state generosity itself. Theasset theory implies that generosity is strongly affected by the structure of skills and the demand for insurance to which they give rise.

-3

-2

-1

0

1

2

3

4

0 10 20 30 40 50 60

Vocational training

T a x e s a n

d t r a n s

f e r s Denmark

Sweden

Australia

Netherlands

USA

UK

Belgium

Canada

FinlandFrance

Switzerland

ItalyGermany

Norway

Figure 1.4 Taxes and transfers as a function of vocational training activity. Notes:Same as in Figures1.1 and 1.3.Sources:Same as in Table 1.1 and Figure 1.3.

23


46/336


The second step in the argument, therefore, is to relate skill structureto the level of taxes and transfers. Again using vocational training rates as arough indicator for national skill structure, Figure1.4 shows that skills areindeed closely related to the magnitude of taxes and transfers (r = .86) andindirectly to redistribution (r = .50). Despite the emphasis on insuranceover distribution, skill structure is, thus, important for explaining not only pretax/transfer income equality as shown previously) but also welfare stateredistribution.

But this does not mean that the power resource story is irrelevant be-cause, again, the causality also runs in the opposite direction. That is to

say, if investment in specic skills is a function of the availability of in-come insurance, redistribution via the tax and transfer system will tend toproduce, in equilibrium, a skill prole that is more specic. Without re-distributive insurance, investment in general skills is the best defense againadverse changes in the labor market. And there is strong empirical evidencethat countries dominated by politically left-leaning governments also re-distribute more (Hibbs 1977; Korpi 1983, 1989; Boix 1998; Bradley et al.2003; Pontusson and Kwon 2004).

This raises the question, however, why some countries are dominated by left-leaninggovernments while othersaredominatedby right governments.If the left and right took turns in government, it would provide no institu-tionalized support for investment in risky assets. But Chapter4 argues thatpartisanship is determined by the differences in coalitional dynamics result-ing from differences in electoral systems. Table1.2 shows the strong empir-ical relationship between electoral system and partisanship using a new dataset on parties and legislatures assembled by Cusack and Engelhardt (2002)

Table 1.2. Electoral System and the Number of Years with Left- and Right-Leaning Governments, 19451998

GovernmentPartisanship Proportion of Right-Leaning

Left Right GovernmentsProportional 342 120 0.26

Electoral system (8) (1) Majoritarian 86 256 0.75(0) (8)

Note: Excludes centrist governments and PR cases with single-party majority governments.Source:Cusack and Engelhardt (2002) and Cusack and Fuchs (2002).

24


47/336


and Cusack and Fuchs (2002). The gures are the total number of years with right- and left-leaning governments in seventeen advanced democ-racies between 1945 and 1998. Among majoritarian systems, 75 percentof governments were center-right, whereas in proportional representation(PR) systems 70 percent were center-left (excluding pure center govern-ments). The numbers in parentheses convey a sense of the evidence at thelevel of countries, indicating the number of countries that have an over- weight (more than 50 percent) of center-left or center-right governmentsduring the 194598 period.

The importance of the pattern revealed in Table 1.2 f or the argument in

this book is that the electoral systemis, in fact, related to the production sys-tem. Peter Katzenstein (1985) pointed out this association many years agoby linking social corporatism to PR. More recently, Hall and Soskice (2001)have argued that coordinated market economies are much more likely tohave PR institutions than liberal market economies. Here the associationis explained by the equilibrium relationship between electoral institutions,redistribution, insurance,and investment inspecic skills. In terms of the ar-gument of this book, if the government is induced by the electoral system toengage in redistributive spending, the latter serves as insurance against theloss of income when specic skills are rendered obsolete by technologicaland other forms of change. Chapter 4 argues that PR is a key commitmentmechanism in political economies that depend on workers making heavy investments in highly specic skills.

1.4.3. Skills and Gender Inequality When we compare access to high-skilled and high-paid jobs, it is well doc-umented that women are at a disadvantage. Economists usually ascribethis disadvantage to statistical discrimination by employers: If womenare more likely to interrupt their careers for child birth and child rear-ing, and if individual women do not have access to effective commitmentmechanisms, all women will be treated as a less valuable source of labor for

employers. As argued by Estevez-Abe (1999b, 2002), however, this disadvantage clearly depends on the skills involved. If employers can easily nd workers with the skills they need in the external labor market, career inter-ruptions are of less consequence. Indeed, in a competitive neoclassical labormarket, every employed worker has a perfect substitute willing to work forthe same wage (it is labor L). Any employee is therefore readily replaceable. When employees bring skills that cannot be easily substituted ( L s ), career

25


48/336


interruptions become critical to the employer. In terms of providing train-ing and offering promotions, employers in this setting have an obvious in-centive to discriminate against women. This is reinforced by womens owninvestment decisions because the returns on specic skills will be negatively related to the prospect of interrupted careers. This implies a heavily gen-dered structure of educational choices that is probably reinforced throughsocialization (Estevez-Abe2002).

Let us look at the outcomes. Figure1.5 relates a measure of the special-ization of skills in different occupations to the percentage of women in theseoccupations using the International Labour Organizations (ILOs) Interna-

tional Standard Classication of Occupations (ISCO-88).12

The numbersare averages for the thirteen countries where comparable ISCO-88 data areavailable. Bolded occupations are those that have disproportionately largenumbers of low-skilled and low-paying jobs.

As expected, there is a strong negative relationship with men dominat-ing occupations that, as implied by this measure, require highly specializedskills.Thisbasicpatternisrepeatedinallthirteencases.Notsurprisinglyoc-cupations with specialized skills are found in agriculture and manufacturingrather than in services. Also note that because men on average participatemore in the labor market than women do, most occupations have an un-derrepresentation of women. Women are also overrepresented in jobs thatrequire relatively low general skills (and that tend to pay low wages), but women have at the same time achieved near-parity with men in professionaloccupations (in the United States there is complete gender parity amongtechnicians and associate professionals).13

An important part of the gender story that is not captured by Figure1.5is the cross-national variance in the labor market position of women.Numerous studies document such cross-national variance (e.g., Melkas and

12 Ignoring military personnel, ISCO-88 contains nine broad occupational groups, whichare subdivided into numerous subgroups depending on the diversity of skills represented within each major group. At the most detailed level, there are 390 groups, each supposedly

characterized by a high degree of skill homogeneity. By dividing the share of unit groups ina particular major group by the share of the labor force in that groups, we can get a roughmeasure of the degreeof specicity of skills represented by each major group. See Chapter3for details.

13 Within the professions, however, it is still the case that women tend to be in jobs andpositions with lower pay and skills for example, teachers in primary and secondary ed-ucation rather than in higher education, nurses rather than doctors, and junior ratherthan senior associates and they are notably underrepresented among senior ofcials andmanagers.

26


49/336


Anker 1997; Chang 2000; Breen and Garcia-Penalosa 2002; Porter 2002). The evidence suggests that specic skill systems exhibit high levels of occu-pational gender segregation (Estevez-Abe2002). Comparative survey datafrom the International Social Survey Program (ISSP), for example, showthat the correlation between vocational training activity and womens shareof private sector employment is .71 for twelve Organization forEconomicCo-ordination and Development (OECD) countries in 1997 (ISSP 2000).

The weak position of women in the private labor market in specic skillscountries is modied by public policies deliberately designed to counterthem, especially public provision of social services and hence employment

opportunities for women (Orloff 1993). These public policies enable theScandinavian countries to have high female participation rates despite hav-inga male-dominatedprivate sector (Esping-Andersen1990, 1999a;Iversenand Wren 1998). At the same time, however, women in countries with largeservice-oriented welfare states become ghettoized into the public sectorinstead of competing equally with men for the best private sector jobs.

10

20

30

40

50

60

70

P e r c e n

t a g e o

f w o m e n

0 0.5 1 1.5 2 2.5Skill specificity

Clerks

Service andsales workers

Professionals

Techniciansand associateprofessionals

Legislators,senior officials,and managers

Elementaryoccupations

Craft andrelatedworkers

Plant andmachineoperators

Skilledagriculturalworkers

r = .74

Figure 1.5 Skill specicity and occupational gender segregation, 2000. Note: Percentage of women refers to the average percentage share of women in differentmajor categories of ILOs International Standard Classication of Occupations across thirteencountries. The skill specicity measure is explained in footnote 12 and in Chapter 3.Source:International Labour Organization, Labour Statistics Database. ILO 19982004.

27


50/336


The implication of this logic is that conict over public sector serviceprovision, parental leave policies, and education is likely to be gender re-lated. Especially in specic skills countries, women will have a strong in-terest in an active state that provides daycare services and counters thedisadvantages of women in the private job market. Also, women dependon generous leave policies to balance family and work; however, for suchpolicies not to put women in a serious disadvantage in the competition forthe best jobs, men must shoulder some of the burden of caring for infants(essentially forms of mandatory parental leave). Finally, because womenhave a comparative advantage in general skills, we should expect them to

be more supportive of public spending on general education than men. Iexplore these hypotheses in Chapter 3.

1.5. Recap

This book builds on new institutionalist theory in both economics andpolitical science (especially Hall and Soskices VoC perspective) to offer ageneral political economy approach to the study of welfare capitalism. Thebook demonstrates that to a substantial extent social protection in a moderneconomy, both inside and outside the state, can be understood as protec-tion of specic investments by both workers and rms in human capital. Itthen shows that such an understanding provides a systematic explanationfor popular preferences for redistributive spending, the economic role of political parties and electoral systems, and labor market stratication (in-cluding gender inequality). In doing so, it helps resolve the debate betweenpower resources theory and recent work on employers and the welfare stateby systemically linking demand for redistribution and demand for socialinsurance and by tying social protection to the way rms compete in theinternational economy. Contrary to the popular idea that global competi-tion undermines cross-national differences in the level of social protection,the book argues that these differences are made possible by a high inter-national division of labor. Such a division is what allows rms to specialize

in production that requires an abundant supply of workers with specic (orgeneral) skills, and hence high (low) demand for protection. The rise of nontraded services, however, undermines this specialization

and leads to pressure for more exible labor markets. The reason isthat when specic skills countries can no longer achieve a completespecialization in production that uses such skills intensely, governmentsconcerned about employment must create regulatory and institutional

28


51/336


conditions that are conducive to the expansion of general skills jobs, es-pecially at the low end of the wage scale. In turn, this raises the critical issueof how governments can deregulate a portion of the labor market withoutundermining thecomparative advantages of generous social protection, and without causing a signicant rise in inequality.

1.6. Outline of Book

The book is divided into three parts. This rst part is an overview, whichincludes this chapter. Chapter 2 presents a comparative-historical analysis

that traces the rise of the postwar social protection system to compromisesbetween interests rooted in the industrial economy. It also provides muchof the comparative data used throughout the book. Those who are familiar with these data and economic history can jump to the subsequent chapters.Part II focuses on the comparative statistics of the argument and explorescross-national differences and the microfoundations of these differences.Chapter 3 develops an asset-based theory of the sources of individual socialpolicy preferences and shows how skills and gender are key in explainingthese preferences. Chapter 4 considers the translation of these preferencesinto policies through the political system, emphasizing the role of politicalparties and electoral systems. PartIII f ocuses on the dynamics of the argu-ment. Chapter 5 explains the rise of the welfare state since the 1960s as apolitically mediated outcome of the shift from industry to services. Chapter6 explores the distributive and political consequences of this shift, as wellas the recent reforms that can be attributed to it.

29


52/336

2

A Brief Analytical History of ModernWelfare Production Regimes

The introductory chapteroutlined a broad theoreticalapproach to the study of social protection and hypothesized close linkages between social pro-tection, skills, rm strategies, and the political-institutional foundationsof the welfare state. This chapter ties national and international develop-ments to the emergence of a few national institutional equilibria or idealtypes. Although the chapter will discuss causal mechanisms, the main pur-pose is to put some empirical meat on the conceptual bones presentedin Chapter 1. I will provide the reader with a range of background in-formation, and I will show some striking interconnections between skillsystems, social protection, and political institutions that cry out for ex-planation. Building on a joint paper with Barry Eichengreen, I argue thatthese interconnections metamorphosed into very distinct regime clustersin response to the challenges of postwar reconstruction and internationaleconomic integration, and each cluster is associated with distinct economicadvantages that are reinforced through the international division of labor.Relative advantages have been shifting over time, however, and I discussthese forces of change with a view to developing the themes that are ex-plored in greater depth, and with a sharper analytical knife, in subsequentchapters.

Much of the political-institutional divergence that occurred during thepostwar period, I argue, reects the structural-institutional potential forpostwar growth in particular countries, the strength of organized laborand capital, and the inherited capacity for centralized collective action.In countries where the rewards

27390072 capitalism democracy welfare

Documents