. 1
( 11)


This page intentionally left blank
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 modern
economy, both inside and outside the state, can be understood as protection
of speci¬c investments in human capital, the book offers a systematic expla-
nation of popular preferences for redistributive spending, the economic role
of political parties and electoral systems, and labor market strati¬cation (in-
cluding gender inequality). Contrary to the popular idea that competition in
the global economy undermines international differences in the level of social
protection, the book argues that these differences are made possible by a high
international division of labor. Such a division allows ¬rms to specialize in pro-
duction that requires an abundant supply of workers with speci¬c skills, and
hence high demand for protection. The rise of nontraded services undermines
this specialization and increases demands for more ¬‚exible labor markets.

Torben Iversen is Professor of Government at Harvard University. He is the
author of Contested Economic Institutions: The Politics of Macroeconomics and Wage
Bargaining (Cambridge University Press, 1999) and coeditor of Unions, Em-
ployers, and Central Bankers: Macroeconomic Coordination and Institutional Change
in Social Market Economies (Cambridge University Press, 1999). He is also the
author 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.
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 University
Peter Hall Harvard University
Peter Lange Duke University
Helen Milner Columbia University
Frances Rosenbluth Yale University
Susan Stokes University of Chicago
Sidney Tarrow Cornell University

Other Books in the Series
Lisa Baldez, Why Women Protest: Women™s Movements in Chile
Stefano Bartolini, The Political Mobilization of the European Left, 1860“1980:
The Class Cleavage
Mark Beissinger, Nationalist Mobilization and the Collapse of the Soviet State
Nancy Bermeo, ed., Unemployment in the New Europe
Carles Boix, Democracy and Redistribution
Carles Boix, Political Parties, Growth and Equality: Conservative and Social
Democratic Economic Strategies in the World Economy
Catherine Boone, Merchant Capital and the Roots of State Power in Senegal,
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)
Capitalism, Democracy,
and Welfare

Harvard University
©¤§ µ®©© °
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press
The Edinburgh Building, Cambridge  µ, UK
Published in the United States of America by Cambridge University Press, New York
Information on this title: www.cambridge.org/9780521848619

© Torben Iversen 2005

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.

First published in print format 2005

©®-± ·-°-µ±±-±°-± eBook (EBL)
©®-±° °-µ±±-±°-° eBook (EBL)

©®-± ·-°-µ±-±- hardback
©®-±° °-µ±-±- hardback

©®-± ·-°-µ±-±°·-± paperback
©®-±° °-µ±-±°·- paperback

Cambridge University Press has no responsibility for the persistence or accuracy of µ¬s
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Til min mor og far

page x

Part I: Welfare Production Regimes

Part II: Political Foundations of Social Policy

Part III: Forces of Change



1.1. Government Spending and Variation in Spending across
Seventeen OECD Countries, 1960“1993 page 16
1.2. Electoral System and the Number of Years with Left- and
Right-Leaning Governments, 1945“1998 24
2.1. Structural Preconditions for Economic Policies and
Institutions, ca. 1950 42
2.2. Employment Protection in Eighteen OECD Countries 47
2.3. Unemployment Protection in Eighteen OECD Countries 50
2.4. Income Protection in Eighteen OECD Countries,
1973“1995 53
2.5. Percentage of Population over 25 with a Postsecondary
Education 54
2.6. Skill Systems in Eighteen OECD Countries 55
2.7. Scienti¬c Citation Rates and Low-Wage Service
Employment in Eighteen OECD Countries 61
2.8. Unemployment in Selected Countries, 1950“1996 65
2.9. Average Annual Rates of Growth in Total Factor
Productivity for Fourteen OECD Countries, 1970“1994 71
3.1. Summary of Independent Skill Variables 96
3.2. Support for Social Spending among the Publics of Ten
OECD Countries, 1996 100
3.3. Estimates of the Magnitude of the Effects of Independent
Variables 102
3.4. Support for Spending in Four Areas of Social Protection 104
3.5. Gender Effects on Preferences, Income, and Skills 106
3.6. Income, Skills, and Support for Social Spending in Eleven
OECD Countries 108

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 153
4.2. Key Indicators of Party and Electoral Systems 156
4.3. Electoral System and the Number of Years with Left and
Right Government, 1945“1998 157
4.4. Electoral System and the Number of Years with
Governments Farther to the Left or to the Right Than
the Median Legislator, 1945“1998 160
4.5. Regression Results for Government Partisanship,
1950“1996 162
4A.1. Electoral Systems and Incentives of Politicians to
Campaign on the Party Platform 175
4A.2. Political Institutions and Capacity for Commitment 176
4A.3. Country Means for Variables Used in Regression Analysis 178
4A.4. Correlation Matrix 179
5.1. Regression Results for Government Spending 199
5.2. Common Shocks, National Institutions, and Government
Spending 202
5.3. Shocks and Spending in Two Subperiods 205
5.4. Deindustrialization, National Institutions, and
Government Spending 206
5.5. Deindustrialization, Partisanship, and Government
Spending 208
5.6. Regression Results for Industrialization 212
6.1. Real Exchange Rates for Eighteen OECD Countries,
1973“1997 222
6.2. Productivity, Relative Earnings, Labor Shares, and
Private Employment in Four Sectors 229
6.3. The Bivariate Relationship between Dispersion of
Earnings and Employment Growth in Three Service
Sectors 232
6.4. The Determinants of Private Service Sector Employment
Growth 236
6.5. The Effect of Earnings Dispersion under Different
Model Speci¬cations 240
6.6. The Effect of Taxation under Different Model
Speci¬cations 242
6.7. Replication of Regression Results Using d5/d1 Ratios 244

6.8. Wage Equality, Service Employment, and Institutions in
Eight OECD Countries, 1970“1996 251
6.9. Employment and Unemployment in Denmark, The
Netherlands, and Europe, 1990“1999 260
6.10. The Volume of Work in Twelve OECD Countries,
1990“1999 264
6.11. Part-Time Employment in Twelve OECD Countries,
1990“1999 266


1.1. Vocational training and wage inequality page 19
1.2. The percentage of adults with poor literacy scores
(bottom scale) and the percentage of adults with low
education and high scores (top scale): thirteen OECD
countries, 1994“1998 21
1.3. Redistribution as a function of taxes and transfers in
fourteen democracies 22
1.4. Taxes and transfers as a function of vocational training
activity 23
1.5. Skill speci¬city and occupational gender segregation,
2000 27
2.1. Social protection and skill pro¬les 58
2.2. Unemployment replacement rates, 1973“1995 64
2.3. The (un)employment gap between continental Europe
and the United States 69
2.4. The employment gap between continental Europe and
the United States, by sector 70
3.1. The three states of an individual in the labor market 80
3.2. Four models of social policy preferences 83
4.1. The relationship between institutionalization of the party
system, vocational training intensity, and government
spending 147
4A.1. The indifference curves for L and H and the empty LH
win-set of m* 169
4A.2. The structure of the coalition game 171
5.1. Spending, trade, and deindustrialization in seventeen
OECD countries, 1952“1995 184

5.2. Support for redistribution as a function of risk 189
5.3. Support for redistribution as a function of shocks to
income 190
5.4. Deindustrialization and change in welfare spending for
sixteen OECD countries, 1960“1993 191
5.5. Trade openness and losses in traditional sectors 210
5.6. Initial size and losses in traditional sectors 211
5.7. Convergence toward the service economy, 1960“1995 215
6.1. Wage dispersion and average currency overvaluation,
1993“1997 225
6.2. Wage dispersion and the relative price of Big Macs,
1988“2000 227
6.3. Dispersion of earnings and employment growth in four
economic sectors 230
6.4. The service economy trilemma 248
6.5. Stringency of employment regulation for regular
full-time employment, 1980s versus 1990s 259
6.6. Stringency of employment regulation for temporary and
part-time employment, 1980s versus 1990s 259
6.7. Relative wages in The Netherlands, 1980“1995 262


This manuscript is the result of more than half a decade of research into
the causes and consequences of social policy, and it is in large measure the
fruit 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 Part II, which is mainly the result of joint work.
David has been a constant source of inspiration, and the book would not
have been possible without his ideas, insights, and encouragement “ even
as he remains a critic on parts of the argument in this book. In addition,
parts of this book have bene¬ted 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 is
that I am probably the only person prepared to assume responsibility for
the 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 Charla
Rudisill. In addition to many helpful substantive comments, Peter Lange™s
advice on the organization and presentation of the materials, and especially
his detailed comments on several iterations of Chapter 1, made this a much
better book. Margaret Levi organized an extremely useful colloquium on a
preliminary version of the book in the Department of Political Science at
the University of Washington in May of 2002, where she and other partic-
ipants (especially Terry Givens and Erik Wibbels) pushed me to be clearer
about the argument and to make signi¬cant improvements to the analysis.
I am grateful to two anonymous reviewers for the same reason. Last, but
not least, Charla Rudisill read every word in this book to make sure it made


at least some sense to people who do not spend most of their waking hours
thinking 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 this
book from (in alphabetical order) Jim Alt, Uwe Becker, Pepper Culpepper,
Robert Fannion, Jeff Frieden, Geoff Garrett, Robert Goodin, Peter
Hall, Bob Hanck´ , Peter Katzenstein, Lane Kenworthy, Gary King,
Herbert Kitschelt, Isabela Mares, Paul Pierson, Jonas Pontusson, Nirmala
Ravishankar, 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 research
Various segments of the manuscript have found their way into articles and
papers. Some of the arguments and data found in Chapters 1 and 2 originate
in 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). Chapter 2 also draws on a paper with Barry Eichengreen that
was previously published in Oxford Review of Economic Policy under the title
of “Institutions and Economic Performance in the 20th Century: Evidence
from the Labor Market.” 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 Political
Science Association (APSA) in Washington, DC, subsequently published
in the American Political Science Review under the title “An Asset Theory of
Social Policy Preferences.” The parts in this chapter on gender preferences
appear 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, and
the Gender Voting Gap.” Chapter 4 is built on two unpublished papers with
David Soskice. One was presented at the 2002 Annual Meeting of the Public
Choice Society, San Diego, and is entitled “Political Parties and the Time
Inconsistency Problem in Social Welfare Provision”; the other was ¬rst
presented at the 2002 Annual Meeting of the American Political Science
Association in Boston and is entitled “Electoral Systems and the Politics of
Coalitions: Why Some Democracies Redistribute More Than Others.” The
two sections in Chapter 5 on the effects of major shifts in the occupational
structure build on a paper that I wrote with Tom Cusack and presented at
the 1998 Annual Meeting of the American Political Science Association in

Boston; it was later published in World Politics under the title “The Causes
of Welfare State Expansion: Deindustrialization or Globalization?” (April
2000). Finally, sections of Chapter 6 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 Service

Capitalism, Democracy, and Welfare

Welfare Production Regimes

A Political Economy Approach
to the Welfare State

Printing is one of the world™s 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 that
permitted the mass production of books, newspapers, and journals. Half
technicians and half craftsmen, typographers were highly skilled, well paid,
and proud harbingers of Gutenberg™s revolutionary invention. However, the
craft was radically transformed over time: ¬rst from “hot-metal” typesetting
to “analog” typesetting and then to digital CRT (cathode ray tube) and
laser 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 into
the original printing processes were retired to the dusty shelves of industry
museums. But retirement was not an option for the majority of typographers
whose livelihood depended on using the skills they had acquired through
long 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 be
loosely 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 typographers
but to all workers “ past, present, and future “ who have skills that are
limited in application and can be made obsolete by new technology and
Welfare Production Regimes

other forces of change. Social scientists are certainly no exception to this
logic. If it were not for the institution of tenure, many Kremlinologists
would have had little marketable expertise following the collapse of the
Soviet Union.
The point of this story is to highlight a central theme of this book: the
importance of political and economic institutions for protecting the hu-
man capital in which people have invested. Job protection, unemployment
bene¬ts, income protection, and a host of related policies such as public re-
training programs and industry subsidies, all help to insure workers against
the loss in asset values when external shocks in technology and labor mar-
ket conditions shift the demand for skills. Indeed, having in place some
form of protection is a precondition for people making investments in spe-
ci¬c skills in the ¬rst place. High job security, wage protection backed by
union power, and guaranteed health and pension bene¬ts have encouraged
generations of young people to follow in their parents footsteps and choose
typography as their trade. And, needless to say, the health of the printing
industry depended on people willing to invest in speci¬c skills. Likewise,
the acquisition of specialized knowledge in academia, including that repre-
sented by Kremlinologists, would be very risky without some form of job
security, and specialized knowledge is the lifeblood of any major research
institution. Even if the institution of tenure was invented as a response to
the Red Scare in the 1920s, its persistence owes much to the fact that it is
functional to the production of new knowledge.
But social protection is clearly not only about insurance, it is also about
redistribution and political con¬‚ict. By this I mean that whereas insurance
is an institutional device for workers to consensually pool their risks and
reimburse each other for potential future losses, redistribution is a device
wherein money is taken from some workers and given to others in the
present, without prior consent to do so. When printers™ unions went on
strike across the industrialized world in the 1970s, it was to seek subsi-
dization of their own jobs and income, not to collect an already agreed
upon insurance or to guarantee the future reproduction of old typograph-
ical skills. Everyone understood that traditional typesetting as a trade was
doomed and that protection of current workers served largely distributive
purposes. 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 no
accident that the ¬rst publishing houses to introduce new technology, such
as LA Times and Oklahoma City Times, were ones with the weakest unions.
A Political Economy Approach to the Welfare State

As the printing example highlights, the political economy of insurance
and of redistribution are in fact closely intertwined. Policies adopted for
insurance purposes have redistributive consequences, and, as I will argue
in detail later in this chapter, redistribution can also sometimes serve in-
surance purposes. Indeed, a central contention of this book is that answers
to many of the most pressing questions about the relationship between
social protection and the economy can be found in the intersection of in-
surance and redistribution, or more speci¬cally 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 been
organized 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
Standard approaches to the welfare state fail to account for the rela-
tionship between production and social protection, and they leave behind
a number of key questions that any political economy approach to social
protection needs to answer. For example, if social protection undermines
markets, as commonly argued, why is there no apparent relationship be-
tween the generosity of such protection and economic growth? A related
question 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 intersection
between welfare production regimes and the creation of comparative ad-
vantages in the international economy. The same is true if we want to un-
derstand why employers are not universally opposed to generous social
protection, and why they continue to invest heavily in economies with high
social spending despite the widely held view that such spending is detri-
mental to business interests.
Even traditional distributive politics, I submit, is not well understood
in the existing literature. Though there is considerable evidence that class
politics matters, why is distributive politics played out so differently in dif-
ferent countries? The fact that partisan politics is systematically biased to
the 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
Welfare Production Regimes

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 the
most skewed distribution of income, where standard class arguments would
predict 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 in
the asset or welfare production regime argument (Section 1.1). I then spell
out the implications of the argument for understanding the role of elec-
toral politics (Section 1.2) and the relationship between the international
economy, economic change, and the rise of social protection (Section 1.3). I
¬nally explore how the approach can help explain cross-national variance in
some of the key dimensions of inequality and redistribution (Section 1.4).

1.1. Toward a New Approach to the Study of the Welfare State
As the printing industry example suggests, the ability of management to
introduce 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 among
neoclassical economists and political sociologists alike. According to those
who take this view, labor is an anonymous commodity, easily aggregated
into a single factor L, where each constituent unit (worker) is “replaceable,
easily redundant, and atomized” (Esping-Andersen 1990, p. 37). Logically,
the opposite of market (or “commodi¬cation”) is state (or “decommodi¬ca-
tion”). It means that “a person can maintain a livelihood without reliance on
the 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: “Decommodi¬cation strengthens the worker
and weakens the absolute authority of the employer. It is for exactly this rea-
son that employers have always opposed decommodi¬cation” (1990, p. 22).
The welfare state is “politics against markets” (Esping-Andersen 1985), and
the historical strength of the political left, mediated by alliances with the
middle classes, determine how much welfare state and how much market
A Political Economy Approach to the Welfare State

you end up with (Korpi 1983, 1989; Esping-Andersen 1990; Huber and
Stephens 2001).
The power resources model of the welfare state as it is known is the
dominant 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 a
central theme in the burgeoning literature on globalization where the “exit
option” for capital presents precisely such an opportunity. As Wolfgang
Streeck explains in the case of Germany: “Globalization, by increasing the
mobility of capital and labour across national borders, extricates the labour
supply from national control and enables the ¬nancial sector to refuse doing
service 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 generous
social programs” (Rodrik 1997).
Indeed, if welfare capitalism is primarily about decommodi¬cation and
exploitation 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 menu
options 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 relationship
among levels of public spending, investment, and national income in ad-
vanced democracies is that there is none (Lindert 1996). Or if there is one,
it is so weak that it does not appear to have imposed much of a constraint on
governments™ 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 resources
model, 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 take
social bene¬ts out of competition by creating a uniform, national social
insurance system. When labor markets are slack, on the other hand, high-
cost producers may feel compelled to impose costs on low-cost producers
through mandatory social insurance arrangements. Swenson argues that the
¬rst logic helps explain early welfare reforms in Sweden, while the latter
Welfare Production Regimes

helps explain salient features of the New Deal legislation in the United
In a similar vein, Isabela Mares (2003) has argued that companies and
industries that are highly exposed to risks will favor a social insurance sys-
tem where costs and risks are shared, leading these employers to push
universalistic unemployment and accident insurance. This is remarkable
because universalism is usually associated with strong unions and left gov-
ernments. Mares also suggests, and this idea is emphasized in this book, that
social 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 generous
unemployment insurance.
In a recent dissertation on the German welfare state, Philip Manow
(2002) has likewise advanced the thesis that the German system of social
protection, through a process of institutional coevolution, emerged as an
important complement to the collective bargaining system, which in turn
underwrote union wage restraint and international competitiveness. By del-
egating much of the responsibility for social policy to the social partners, the
institutional 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 the
event of unemployment and other shocks to income. In earlier work, Peter
Baldwin (1990) also challenges the notion that the welfare state was erected
by 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 nonworkers
at the turn of the century to be included in social programs that served as
instruments of insurance as much as tools of redistribution.
The evidence presented by Lindert, Mares, Swenson, Manow, and
Baldwin strongly suggests that social protection cannot be conceived exclu-
sively in terms of simple dichotomies between the state and the market, or
between commodi¬cation and decommodi¬cation. We need a “politics of
markets” rather than a “politics against 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), this
is 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
A Political Economy Approach to the Welfare State

links the study of the welfare state to recent work on the importance of
democratic institutions for social policy.
At the heart of the dif¬culties in the standard view of the welfare state
is a neoclassical conception of markets that largely ignores differentiated
skills. Although unskilled day workers can sensibly be analyzed as an undif-
ferentiated factor L, and although such labor can be exchanged ef¬ciently in
something that approximates spot markets, in Becker™s (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-speci¬c investments in skills “ Ls , where s = (1, 2, 3, . . . , n) refers to
differentiated skills “ coupled with nonmarket institutions that protect and
manage these investments.
Of course, workers are not the only ones who invest in speci¬c assets;
¬rms, merchants, and virtually any agent involved in economic exchange do
also. And because economic agents are engaged in exchange, and because
they sometimes own the assets jointly, most assets are cospeci¬c in the sense
that they tie together the welfare of people and make them dependent on
one another. For every worker whose livelihood is tied to a speci¬c skill,
there is an employer who depends on the worker with those skills. As argued
by Polanyi (1944), Williamson (1985), North (1990), and others, when an
economy is characterized by heavy investment in such cospeci¬c assets,
economic agents are exposed to risks that make ef¬cient market exchange
problematic. A precondition for such an economy to work ef¬ciently is
a dense network of institutions that provide information, offer insurance
against risk, and permit continuous and impartial enforcement of complex
contracts. In the absence of such institutions, exchange is possible only at a
small scale in local trading communities where people know each other well
and engage in repeated face-to-face interactions.1 At a larger scale, markets
left to their own devices either will fail to produce much exchange, will be
accompanied by costly and continuous haggling, or will involve only very
homogeneous types of assets (L as opposed to Ls ).
Nowhere is the importance of institutions more evident than in the labor
market where the welfare state plays a key mediating role. Social protection
is particularly important in solving market failures in the formation of skills.
Without implicit agreements for long-term employment and real wage
stability, investment in skills that are speci¬c to particular jobs, ¬rms, or

1 Marshall™s concept of industrial districts likewise emphasizes repeated interactions in local-
ized settings as a precondition for ef¬cient outcomes (Marshall 1922).

Welfare Production Regimes

industries will be suboptimal. In the absence of insurance, workers shun
such investments because unanticipated shocks to the economy, whether as
a result of recessions or technological change, can prevent workers from
reaping the returns on their investments. Employers will also be reluctant
to invest in their employees™ skills, or in equipment that requires those skills,
unless they believe that institutions that prevent poaching and discourage
unions from exploiting the potential holdup power that speci¬c skills confer
are in place.
The importance of asset speci¬city 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 hold
their wealth in liquid assets that can be quickly moved from one jurisdiction
to another (Bates, Brock, and Tiefenthaler 1991). Even when basic prop-
erty rights are well protected, investments vary signi¬cantly in the degree of
their asset speci¬city. When investors cannot trust suppliers or employees
on whose cooperation they depend, they will shun investments in relation-
speci¬c assets and rely instead on anonymous market transactions where
one supplier or employee can easily be replaced by another. Conversely,
when investments in physical assets are speci¬c to a particular location,
supplier network, or employee relationship, ¬rms are more prone to lobby
the state for protection against uninsurable risks (Frieden 1991; 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 Soskice 2001).
A similar logic applies to human capital. When skills are speci¬c 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 bene¬ts. Skills that are portable,
by contrast, do not require extensive nonmarket protection, and when there
is little protection, investing in such skills is the best insurance against ad-
verse market conditions and technological change. Yet, despite its intuitive
appeal, asset speci¬city plays virtually no role in existing explanations of
the welfare state. Labor is L, and workers are “replacable, easily redundant,
and atomized.” Correspondingly, politics is labor against capital, L against
C. 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 speci¬city of
industries. See also Alt et al. (1996) for a more theoretical treatment of this and related
arguments concerning the importance of asset speci¬city.

A Political Economy Approach to the Welfare State

importance of the level and composition of human capital (Ls ) for explaining
the character of the welfare state “ the level because it determines income
and hence workers™ demand for redistribution; the composition because it
determines workers™ exposure to risk and hence their demand for insur-
ance. It is natural to label this an asset theory of the welfare state, although
political institutions are also important as we will see in a moment.
The link between assets and the welfare state explains the continued
and even growing importance of social policy in advanced, and therefore
human-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 school
degree and $16,900 earned by workers who had less than a high school de-
gree (U.S. Census Bureau 2000). Ignoring other group differences, having
a college degree is equivalent to a 3 percent real return on a net fortune
of 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 re¬‚ects accumulated past returns on
skills. Human capital is, thus, easily the most import asset for a majority of
Do ordinary people also worry about protecting the value of these assets?
The answer obviously varies from individual to individual according to the
level and mobility of her skills, but there is no question that many workers
face a substantial risk that their training can be made partially or entirely
redundant by new technology or other forces of change (as in the example of
typographers). Taken as a whole, manufacturing employment in the OECD
has 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 workers
and their unions to concern themselves with insurance against income losses
as 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 dif¬culties of quantifying the speci¬city of skills is that wage and social protection
systems are set up to reduce the riskiness of speci¬c skills. Skill certi¬cation and wage
standardization 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 an
indicator of asset speci¬city. Chapter 3 goes to considerable length in developing alternative
measures of skill speci¬city and to tie such speci¬city to social policy preferences.

Welfare Production Regimes

failures (Przeworski, 2003 Ch. 11). Writing complete contracts to cover all
contingencies in a labor market characterized by incomplete information
and highly speci¬c investments is precisely what transaction cost economics
rules out. Nearly all unemployment insurance, for example, is provided
through public insurance; health insurance markets always produce limited
coverage, and there is no effective private insurance against the adverse
effects of technological change on earnings capacity. For the vast major-
ity of people in advanced democracies, insurance against job and income
loss 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 strategies
that require speci¬c skills also have a vested interest in social policies that
reduce the risk of acquiring those skills (Mares 1998, 2001; Estevez-Abe
1999a; Swenson 2002). The focus on employers, however, tends to leave
the democratic state, electoral politics in particular, underexplored. The
power of employers is primarily “structural” in nature, but governments
must win elections to stay in power, and it cannot be assumed the electoral
incentives of politicians are perfectly aligned with their economic incentives
(Elkin 1985; Block 1994). In Swenson™s (1997) account of the New Deal,
for example, politicians are faithful representatives of employers™ long-tem
interests, yet they face an urgent need to accommodate popular demands
for political action. Indeed, Swenson acknowledges that business often op-
poses such action, yet it somehow ends up bene¬ting from it.5 But electoral
politics operates according to its own dynamic, and more often than not
this dynamic is powerfully affected by popular demands for redistribution.
As Stephen Elkin explains, “the democratic impulse is egalitarian, because
rule by all requires not only political equality but also economic equality
suf¬cient not to vitiate the premise of equal participation” (1985).
One of the great strengths of the power resources model is that it has
a credible account of electoral politics. But the role of democratic poli-
tics must be reconsidered in the context of a different understanding of the
economy and of employer interests. The emphasis on class interests ignores
the importance of insurance motives in people™s 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 on
the role of employers in the rise of the welfare state.

A Political Economy Approach to the Welfare State

with high inequality are not more redistributive than democracies with low
inequality. 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 the
loss of income not only is conducive to investment in risky assets but also
has the effect of redistributing income. This is obviously the case of the
unemployed, 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 behind
the 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 will
simultaneously serve insurance functions. Those who are unemployed, sick,
old, and have low pre-tax income more generally, will rationally press for
redistribution. But by doing so, many of those who are employed, healthy,
young, and enjoying a high incomes will enjoy some measure of insurance
against 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 struggles and a source
of comparative advantage. Those who see only the ¬rst face will tend to
conclude 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 who
see only the second face tend to reduce democratic politics, and electoral
politics in particular, to a symbolic game where the welfare state always
mirrors the needs of the capitalist economy (or employers), trumping the
pursuit of competing interests. To understand the welfare state, we must
understand how popular preferences for social insurance and redistribution
are rooted in peoples™ position in the economy, how these preferences are
aggregated 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 the
possible future loss of income from those investments. Modeling popular
preferences for social protection as a function of the assets people own in
the economy is the ¬rst departure from the power resources approach to
mass politics.
The second departure is my attention to the speci¬c design of democratic
institutions. Consider, for example, that because social insurance may only
Welfare Production Regimes

be enjoyed by the current median voter some time in the future, the median
voter has an incentive to support such insurance only if future median voters
do the same. The current median voter, therefore, faces a problem of how
to commit future median voters. This translates into a time-inconsistency
problem for the government because it has an incentive to renege on its
promise to the current median voter when it seeks to attract the support
of the future median voter. Again, the reason is that the median voter at
any given time, when choosing a policy for the present, does not have an
interest in high transfers. This problem is addressed in Chapter 4.
One solution points to the role of institutions that can hold the gov-
ernment to its promises about future policy. The organization of political
parties and their relation to private groups is particularly important in this
respect. Another solution builds on the close relationship between redistri-
bution and insurance. Because redistribution also serves insurance purposes,
institutions that promote redistribution serve as (imperfect) solutions to the
time-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 electoral
system (and that the electoral system is also closely associated with the
presence of responsible and programmatic parties).

1.3. Globalization, Deindustrialization, and the Expansion
of Social Protection
As noted previously, it is a puzzle that globalization has not led to con-
vergence in social protection. The coupling of social protection and skill
systems helps us understand the puzzle by pointing to their effect on the
international product market strategies of companies and the creation of
comparative advantages in the global economy. Speci¬cally, where there is
a large pool of workers with advanced and highly portable skills and where
social 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 and
facilitates the use of rapid product innovation strategies. Such capacities are
lower for ¬rms in economies that rely heavily on nontransferable skills and
that protect these skills through restrictions on the ability of ¬rms to hire
and ¬re workers. On the other hand, the latter types of welfare-production
regimes give a comparative advantage to companies that compete in mar-
kets where there is a premium on the ability to develop deep competencies
A Political Economy Approach to the Welfare State

within established technologies and to upgrade and diversify existing prod-
uct lines continuously “ what Wolfgang Streeck in a seminal article has
dubbed “diversi¬ed 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 support
for existing social protection regimes. As countries specialize in production
that uses abundant factors intensely, demand by the owners of those factors
for protection of their value rises. Contrary to the popular notion of a so-
cial “race to the bottom,” differences across countries, therefore, need not
disappear with a deepening of the international division of labor “ a propo-
sition implied by Hall and Soskice™s concept of comparative institutional
advantage (Hall and Soskice 2001). Social spending in continental Europe
continues to be much higher than in Ireland and the Anglo-Saxon countries,
and in many areas the gap has increased. Moreover, whereas labor markets
have become even more deregulated in the latter countries, employment
protection for full-time employees has stayed high and largely unchanged
in 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 the
power resources model or theories emphasizing the role of the interna-
tional economy. One of the most remarkable facts about the welfare state
is that public spending did not vastly differ between the United States,
continental Europe, and Scandinavia in the early 1960s (Rothstein 1998).
“In the 1960s,” writes Rothstein, “the difference between these countries
in total public spending was much smaller [than today] “ the level in the
United States was about 28 percent compared to a mean of 29 percent for
the Scandinavian countries.” This does not mean that basic differences in
unemployment, employment, and wage protection through labor market
institutions did not exist at that time. They did, but the role of the state
in the social insurance system through taxes and transfers was not terribly
The tremendous expansion of social spending since then, and the in-
creased variation across countries, can be gleaned from Table 1.1. It shows
total 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 There are however signi¬cant reforms in the regulation of part-time and temporary employ-
ment, as well as in a range of social transfer programs, which will be discussed in Chapter 6.

Welfare Production Regimes

Table 1.1. Government Spending and Variation in Spending across Seventeen OECD
Countries, 1960“1993

Government Difference between
Spending as Standard Deviation Swedish and U.S.
Percentage of GDP of Spending Spending
1960 28.7 4.9 3.0
1965 31.8 5.2 8.1
1970 35.2 6.4 9.1
1975 42.2 6.9 13.1
1980 45.3 8.9 25.6
1985 48.0 9.1 27.1
1990 46.3 8.4 24.6
1995 49.0 9.0 31.0
Ratio of 1995 level 1.71 1.85 10.38
to 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 about
70 percent during this period, from 29 percent in 1960 to 49 percent in
1995, but the variation in spending grew even faster. Thus, the standard
deviation in spending increased by about 85 percent in this period, and the
difference between Sweden and the United States ballooned from 3 percent
to 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 is
awkward to emphasize the role of the industrial working class in the postwar
rise of the welfare state because it has been on the decline everywhere. In
Chapter 5, I show that there is also little empirical evidence for the other
prominent argument that the growth of the welfare state is the result of
increased exposure to risks in the international economy (Cameron 1978;
Garrett 1998; Rodrik 1998). As I have argued with Thomas Cusack (Iversen
and 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 deindustrialization
represents a serious threat to those workers who have made signi¬cant in-
vestments in ¬rm- or industry-speci¬c skills “ a threat that cannot easily be
addressed within the “private” system of protection in the labor market “ it
A Political Economy Approach to the Welfare State

is associated with a rise in electoral demands for public compensation and
risk sharing. Additionally, even though deindustrialization has occurred ev-
erywhere, the speed at which it has taken hold has varied considerably across
More importantly, the effects of deindustrialization have been mediated
by the skill regime, as well as by the institutional capacity of the political sys-
tem for credible commitment. Building on recent work on unemployment
by Blanchard and Wolfers (2000), I demonstrate this institutional condi-
tioning of common shocks with a variety of empirical tests in Chapter 5.
The growing electoral pressure for government spending has also pro-
vided politicians and political parties with an opportunity to shape the
structure of social protection according to ideological preferences. In this
respect, I am entirely in line with scholars such as John Stephens and
Geoffrey 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 state
should 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 the
potential for job growth (especially for women) is greatest, social demo-
cratic parties have favored an expansion of jobs in public services while
Christian democratic parties have emphasized transfers and social services
provided through the family. Liberal parties, by contrast, have advocated
A critical issue examined in Chapter 6 is the relationship between social
protection, especially a relatively ¬‚at wage structure, and employment. Al-
though high-protection countries have been very successful in international
markets, 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 performers
such as the United States have paid a heavy price in the form of greater
inequality. The underlying problem, I argue, is that lack of international
trade in services has undermined the ability of countries to take advantage of
their comparative advantage. High-protection countries, for example, have
squeezed 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 for
part-time and temporary employment) is coupled with new tax and trans-
fer policies to shield the inequalizing consequences. I assess the limits and
Welfare Production Regimes

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 af¬nity between the insurance and
distributive aspects of the approach. Building on Estevez-Abe et al. (2001)
in this section, I suggest how the asset argument can be extended to unravel
three sets of previously neglected causes logics by which welfare production
regimes affect distribution. These propositions help explain several of the
remaining puzzles noted in the opening to this chapter. They are elaborated
and 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 skill
acquisition at the low end of the academic ability distribution. The skill
system is also related to the wage-setting system, which strongly affects the
earnings distribution. This helps explain why welfare production regimes
are linked to wage dispersion. Second, demand for insurance against social
risks leads to signi¬cant redistribution of income through the welfare state,
and redistributive pressures are accommodated by their insurance bene¬ts.
This helps explain why the welfare state is so broadly supported in some
countries, despite modest levels of inequality. Finally, gender inequality in
the labor market is intimately related to skill and social protection regimes,
and such inequality, unlike wage inequality, tends to be higher in speci¬c
skills systems. The skill argument helps us understand why that is the case.
All in all, speci¬c skills systems tend to be notably more egalitarian and re-
distributive than general skills systems, but labor markets in these countries
tend 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-speci¬c skills have developed effective wage coordination
at the industry level. Conversely, general skills countries, especially coun-
tries with a strong emphasis on ¬rm-speci¬c skills ( Japan in particular),
lack such coordination. Very extensive evidence has in turn been accu-
mulated and veri¬es that the structure of the wage bargaining system has
important consequences for the wage structure (see especially Rowthorn
1992; Freeman and Katz 1994; Iversen 1999; Wallerstein 1999; and Rueda
A Political Economy Approach to the Welfare State

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 earnings
distribution 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, intraoccupational
compression 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 higher
levels also gives low-income groups in¬‚uence over the distribution of wages
that they lack in the market. Such in¬‚uence tends to promote equality.7
But the skill system itself is also important as suggested in Figure 1.1.
This graph uses the incidence of vocational training as an indicator of the
extent to which workers are acquiring speci¬c vocational skills as opposed

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

Welfare Production Regimes

to general academic skills.8 Note the strong empirical association between
skills and earnings equality. Because speci¬c skills systems generate high de-
mand for workers with good vocational training, young people who are not
academically inclined enjoy career opportunities that are largely missing in
general skills systems. Whereas a large proportion of early school leavers in
the former acquire valuable skills through the vocational training system,
in the latter most early school leavers end up as low-paid unskilled workers
for most or all of their working lives.
This pattern also implies that young school goers in speci¬c 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 go
on to college.9 Although there are clearly alternative interpretations, data
from standardized international literacy tests are consistent with this idea
as suggested in Figure 1.2. Countries are ranked by their share of poor
performers, and the rough division between general and speci¬c skills coun-
tries has been indicated with brackets.10 Whereas the percentage with the
lowest score averages 20 in Ireland and the Anglo-Saxon countries, the
comparable ¬gure in countries emphasizing more speci¬c skills is 10. The
correlation between vocational training intensity and the percentage with
low test scores is .73. Likewise, those who leave school without an upper
secondary education tend to have much higher test scores on job-relevant
skills (here measured by document literacy) when they are in speci¬c skills
rather than in general skills systems: 46 percent of early school leavers in
the former have high literacy scores compared to only 28 percent in the
In combination, the wage bargaining system (i.e., whether it is industry
coordinated or not) and the skill system (i.e., whether it is speci¬c skills or
general skills biased) provide a powerful explanation of earnings inequality.
It points to the importance of paying attention to factors outside the welfare
state that affect distribution. Much of the welfare state literature fails to
do this, notwithstanding its almost exclusive focus on distribution. In the
welfare production regime argument, nonstate institutions are integral parts

8 Measures of skills will be discussed extensively in Chapters 2 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 because
the test was administered in different areas using only one of the three of¬cial languages;
consequently, many people took the test in a nonnative language. This practice appears to
have signi¬cantly affected test scores (OECD and Statistics Canada 2000, 56). If included,
Switzerland is positioned below Belgium, where a similar issue may be at play.

A Political Economy Approach to the Welfare State

of the story even though the main functions of these institutions may be
insurance and ef¬cient management of cospeci¬c assets.

1.4.2. Social Insurance and Redistribution
As I noted previously, workers who are behind Rawl™s veil of ignorance do
not know with certainty how they will fare in terms of future employment
and income. In this situation, risk-averse people will demand insurance
against loss of employment and income. If these preferences are translated
into policy, when the 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 result
in 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) and
the percentage of adults with low education and high scores (top scale): thirteen
OECD countries, 1994“1998.
Notes: The top bars (using top scale) show the percentage of adults who have not completed
an 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).

Welfare Production Regimes

40 Sweden
Finland Denmark



10 Switzerland

-3 -2 -1 0 1 2 3 4
Taxes and transfers
Figure 1.3 Redistribution as a function of taxes and transfers in fourteen democ-
Notes: Redistribution is the percentage reduction in the Gini coef¬cient from pre- to posttax
and transfer for households with working-age adults. Taxes and transfers is total taxes as
a percent of GDP plus total transfers as percent of GDP, after both measures have been
standardized. Both measures were developed by Bradley et al. (2003) based on LIS data and
OECD 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 this
basic idea can be illustrated with some data on pre- and posttax and transfer
income from the Luxembourg Income Study (LIS; see Bradley et al. 2003
for details).11 The inequality measure is the Gini coef¬cient, and redistri-
bution is the percentage reduction in the Gini from the pre- to posttax

11 I™m using the Bradley et al. (2003) data where pretax and transfer income consists of income
from wages and salaries, self-employment income, property income, and private pension
income, while posttax and transfer income is disposable personal income, including all
market income, social transfers, and taxes.

A Political Economy Approach to the Welfare State

and transfer distribution of income. To ensure that the measure is related
to the concept of wage protection, the data only include the working age
population. The data are available for fourteen advanced democracies and
represent averages for a period starting in the late 1970s and ending in the
mid-1990s (time coverage varies by country).
Figure 1.3 shows the reduction in pre- and posttax and transfer inequality
as a function of the level of taxes and transfers. As expected, there is a positive
relationship (r = .68), and, although we do not know from this relationship
how much is the result of a deliberate attempt to produce redistribution as
opposed to insurance, Huber, Stephens, and their associates have found a
strong positive relationship between tax and transfers and redistribution
after controlling for the partisan preferences of governments and a host of
other factors (Bradley et al. 2003). In fact, their ¬ndings indicate that the
level of taxes and transfers (what they term “welfare state generosity”) is
one of the most important determinants of both redistribution and poverty
reduction. They do not, however, explain welfare state generosity itself. The
asset theory implies that generosity is strongly affected by the structure of
skills and the demand for insurance to which they give rise.


Taxes and transfers

Sweden Belgium

-1 UK

0 10 20 30 40 50 60
Vocational training
Figure 1.4 Taxes and transfers as a function of vocational training activity.
Notes: Same as in Figures 1.1 and 1.3.
Sources: Same as in Table 1.1 and Figure 1.3.

Welfare Production Regimes

The second step in the argument, therefore, is to relate skill structure
to the level of taxes and transfers. Again using vocational training rates as a
rough indicator for national skill structure, Figure 1.4 shows that skills are
indeed closely related to the magnitude of taxes and transfers (r = .86) and
indirectly to redistribution (r = .50). Despite the emphasis on insurance
over distribution, skill structure is, thus, important for explaining not only
pretax/transfer income equality as shown previously) but also welfare state
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 speci¬c skills is a function of the availability of in-
come insurance, redistribution via the tax and transfer system will tend to
produce, in equilibrium, a skill pro¬le that is more speci¬c. Without re-
distributive insurance, investment in general skills is the best defense again
adverse changes in the labor market. And there is strong empirical evidence
that 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-leaning governments while others are dominated by right governments.
If the left and right took turns in government, it would provide no institu-
tionalized support for investment in risky assets. But Chapter 4 argues that
partisanship is determined by the differences in coalitional dynamics result-
ing from differences in electoral systems. Table 1.2 shows the strong empir-
ical relationship between electoral system and partisanship using a new data
set 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, 1945“1998

Proportion of
Left Right Governments
Proportional 342 120 0.26
(8) (1)
Electoral system
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).

A Political Economy Approach to the Welfare State

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 percent
of 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 the
level of countries, indicating the number of countries that have an over-
weight (more than 50 percent) of center-left or center-right governments
during the 1945“98 period.
The importance of the pattern revealed in Table 1.2 for the argument in
this book is that the electoral system is, in fact, related to the production sys-
tem. Peter Katzenstein (1985) pointed out this association many years ago
by linking social corporatism to PR. More recently, Hall and Soskice (2001)
have argued that “coordinated market economies” are much more likely to
have PR institutions than “liberal market economies.” Here the association
is explained by the equilibrium relationship between electoral institutions,
redistribution, insurance, and investment in speci¬c skills. In terms of the ar-
gument of this book, if the government is induced by the electoral system to
engage in redistributive spending, the latter serves as insurance against the
loss of income when speci¬c skills are rendered obsolete by technological
and other forms of change. Chapter 4 argues that PR is a key commitment
mechanism in political economies that depend on workers making heavy
investments in highly speci¬c 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 ascribe
this disadvantage to “statistical discrimination” by employers: If women
are more likely to interrupt their careers for child birth and child rear-

. 1
( 11)