Before Norms: Institutions and Civic Culture

Before Norms

Before Norms Institutions

AND

Civic Culture

Robert W Jackman AND

Ross A. Miller

The University of Michigan Press Ann Arbor

Copyright 'Cl by the University of Michigan 2004 All rights reserved Published in the United States of America by The University of Michigan Press Manufactured in the United States of America @ Printed on acid-free paper 2007

2006

200 5

2004

4

2

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, or othenvise, without the written permission of the publisher.

A CIP catalog record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Jackman, Robert W. , 1946Refore norms : institutions and civic culture I Robert W. Jackman and Ross A. Miller. p. em. Includes bibliographical references and index. TSRN 0-472- T T 3 9 6- 8 (cloth : aJk. paper) I. Political culture . 2. Political sociology. 3. Social norms. I. Miller, Ross A., I96 s - II. Title. JA75 ·7·J3 2004 3 06. 2-dc22

for Rachael and Saul RJ for Kris and Mom RM

"An antipathy to labour, or an insensitivity to its advantages, is now, as in the days of Fynes Moryson and Bishop Berkeley, a distinguishing trait in the character of the Irish people. Their extreme poverty is principally a consequence of their extreme sloth." -J. R. McCulloch A Descriptive and Statistical Account of the British Empire, J:;xhibiting Its £xtent, Physical Capabilities, Population, Industry, and Civil and Religious Institutions (I84 7)

"The poor catholics in the south of Ireland spin wool very generally, but ...the professors of that religion are under such discouragements that they cannot engage in any trade which requires both industry and capital. If they succeed and make a fortune what are they to do with it? They can neither buy land, nor take a mortgage, nor even fine down the rent of a lease. Where is there a people in the world to be found industrious under such a circumstance?" -Arthur Young A Tour in Ireland, With General Observations on the Present State of That Kingdom, Made in the Years r776, 1777,

and r77R and Brought Down to the Hnd of 1779 (I78o)

"Had the Treaty of Limerick been faithfully kept; had the Irish Ro­ man Catholics, humbled by defeat, been treated with even-handed justice, as entitled to the protection of the laws equally with their Protestant fellow-subjects, the result of the one hundred and fifty years which have since elapsed might have been widely different. But almost the first act of the Irish legislature was to pass those penal laws, of which an eminent historian has declared, that 'to have exter­ minated the Catholics by the sword, or expelled them like the Moriscoes of Spain, would have been little more repugnant to justice and humanity, but incomparably more politic.' " -Jonathan Pim The Condition and Prospects of Ireland and the Evils Arising from the Present Distribution of Landed Property: With Suggestions for a Remedy (I 84 8)

Contents

x1

List of Tables List of Figures

xm

Preface and A cknowledgments The Issues

ONE.

Part One.

I

Cultural and Social Capital Approaches

The Protestant Ethic Thesis

TWO.

THREE.

33

Civic Virtue in Italy and the United States Civic Virtue, Growth, and Democratization

FOUR.

Part Two.

59 98

The Institutional Alternative

Institutions and Voter Turnout

FIVE. SIX.

xv

I3 7

Institutions and Support for Parties of the Extreme Right (with Karin Volpert) I63

SEVEN.

Culture, Institutions, and Political Behavior

References Index

225

20T

T88

Tables

I . I . Votes and Legislative Seats Received by the Congress Party of India in National Elections since Independence 2. I . High School Enrollments in Baden, 1 8 8 5 - 9 5 , by Type and Religious Affiliation 2 . 2 . Taxable Wealth i n Baden, 1 897, b y Type and Religious Affiliation 3 · T . Principal Component Analyses of Indicators of the Institutiona! Performance of Italian Regional Governments 3 . 2 . T-ratios from Regressions of the Institutional Performance of Italian Regional Governments and Its Components on ( a ) Current Civic Values and Current Economic Development and (b) Early Civic Values and Current Economic Development 3 - 3 - Average Levels of Trust, Helpfulness, and Advantage Regressed on Year, 1 9 74 - 9 4 3 -4 - Correlations among Trust, Helpfulness, and Advantage at the Individual and Aggregate Levels, 1 9 74- 94 3 · 5 · Average Levels of Membership i n Fraternal, Literary, Service, and " All " Groups 3 . 6. Regressions of Rates of Membership in Selected Groups on Survey Year, with and without a Control for Average Level of Education, 1 9 74 - 9 4 3 · 7· Rates of Church Attendance for Select Countries 4 · I . Regressions o f Growth o n Initial GDP, School Enrollment Ratios, Investment, and Need for Achievement Levels, T 9 5 0 4 . 2 . Regressions o f Need for Achievement, 1990, o n Prior Growth and Initial School Enrollment Ratios

24 s6 57 64

67 76 77 8o

81 87 T03 T05

Xll

TABL E S

4 - 3 - Principal Components Analyses o f Four Values Indicators, I99 0 4 + Regressions of Growth on Initial GDP, School Enrollment Ratios, Investment, and Other Sundry Characteristics 4 · 5 · Trust, Civic Virtue, and Growth, 1 9 8 0- 9 2 4 . 6. Product-Moment Correlations among Seven Components of Political Culture, circa 1 9 8 0 4 · 7 · Regressions o f Components of Political Culture, circa I 9 8o, on Basic Economic Conditions 4 . 8 . Estimates o f Muller and Seligson's Basic Model 4 · 9 · Estimates for a Modified Specification Using Muller and Seligson's Data 4 . IO. Regressions of Changes in Democracy, I 990-9 5 , on Cuitural and Economic Characteristics 5 - I . Turnout Rates, Electoral Disproportionality, and Number of Parties for 22 Established Democracies, I 9 so- 2000 ). 2 . Institutions and Voter Turnout, I 9 50- 2ooo 5 - 3 · Basic Information about the Swiss Experience with Referenda, I 8 79- 2ooo 5 + Regressions of Turnout in Swiss Referenda on the Number of Referendum Items in the Election and the Frequency of Elections in the Recent Past, I 8 79 - 2000 5 · 5 · Need for Achievement and Voter Turnout 5 . 6. Culture, Institutions, and Voter Turnout in the I 990S 6. r . Extreme Right Parties Included in the Analysis 6.2. Summary Statistics for A ll Variables 6. 3 . Sources of Support for Parties of the Extreme Right, I 9 70- 2000 6+ Immigration and Support for Parties of the Extreme Right, 1 9 70- 2000

I06 I IO 113 II7 I I9 I22 125 128 I44 I47 I52

I53 I 57 I 6o I76 I78 I8I 183

Figures

I. I. 2. I. 2.2. 3·T. 3 .2. 3 -3-

Indian elections, 1 9 5 2-99 Growth pattern implied by the Weber thesis Growth a s a simple compounding process Membership trends ( T 9 74 - 9 4 ) by type of group The churching of America, 1 776- 1990 Membership i n Little League and United States Youth Soccer, 1 9 3 9-99 3 + Membership in American youth soccer, 1 9 64 - 9 9 4 · I . Democracy 1 9 8os against democracy 1 9 70s

26 43 45 79 84 93 95 123

Preface and Acknowledgments

I

iving conditions for most people in nineteenth-century Ireland were

l""'miserable. Was this the result of a flawed, indolent national character,

or did Irish poverty have other sources ? Since the collapse of the Soviet Union in 1 9 89, the pace of democratization has been much slower than many had anticipated. Does this slow pace stem from an ingrained au­ thoritarian streak among Russians or from other factors ? Italians are much more likely to vote than Americans. Does this signal a keener appreciation of electoral politics among ordinary citizens in Italy than in the United States ? Does the resurgence of parties of the extreme right in some countries of Western Europe reflect a distinctive set of values in the countries concerned, or does it arise from institutional and other struc­ tural factors ? These are instances of the broader question that motivates our book. On the one hand are those who argue that much political behavior, along with a host of other outcomes, can be traced back to distinctive sets of norms among different publics. Labeled the " value enactment" approach by Portes ( 1 9 7 6 ) , this account attaches great explanatory significance to political or social cultures, or, to use a more current term, social capital. The sets of values associated with different cultures are said to generate distinctive political and social outcomes. Hence the value enactment la­ bel. Further, the values involved are durable, changing slowly, if at all, across generations . The alternative view, which we find much more compelling, is that ordinary citizens everywhere optimize in light of institutional and other constraints that generate varieties of incentive structures. For example, potential voters and candidates for office in a country with a form of proportional representation face a different set of electoral choices than

XVI

PREfAC E A N D A CK N O WL E D G M E N T S

d o citizens i n a single-member district system. Peasants living i n a low­ income country with an authoritarian regime face a different (and much more constricted ) set of alternatives or choices than do the middle-class citizens of an established and prosperous democracy. These differences in available choices imply different sets of incentives that in turn affect pat­ terns of behavior. Further, and in contrast to the value enactment approach, modifica­ tions to institutional and other constraints alter the incentive structure and thereby lead to changed patterns of behavior. Instead of taken as fixed, value configurations become the product of institutional environ­ ments and therefore subject to change. To borrow a central example from recent discussions of social capital, this line of argument implies that levels of trust are created by the institutional environment, not a cause of it. In this account, norms retain a consequential role. At the same time, the institutional approach more readily allows for changed patterns of behavior as current generations are not cast as substantially burdened by the experiences of those who preceded them. The book is organized as follows. In chapter I, we frame the analysis by developing the distinctions that are ever so briefly sketched (and thereby oversimplified) here. In particular, we specify the cultural argument and elaborate the premises on which it rests. Along the way, we discuss the different versions of the value enactment approach that have been offered, ranging from earlier claims about national character to contemporary treatments of social capital. We then elaborate the institutional alternative and specify why it is, in fact, an alternative and preferable approach. The heart of the book has two parts . Part I (chaps. 2 through 4) ad­ dresses in detail the empirical bases for the cultural perspective. Here, we evaluate claims that have been made about the ways in which norms shape a variety of outcomes, including economic growth and democratization. Since so many recent studies take Max Weber's thesis about the Protestant ethic as a starting point, chapter 2 centers directly on that argument. Chapter 3 focuses on contemporary discussions of the political conse­ quences of cultural values or social capital in Italy and the United States, to determine whether southern Italy really is different and whether Ameri­ cans are increasingly prone to going it alone. We broaden the empirical focus across a wider range of cases in chapter 4 to address more general claims about the role of norms in bringing about economic growth and democratization. Throughout part I, we find limited evidence for the cultural account. At some points in the analysis there is little indication of distinctive cultural values that could be driving anything. At other points, the cultural account

Preface and Acknowledgments

xvu

only seems to work if we engage in the most egregious ex post reasoning. Indeed, at several stages in the analysis there are strong suggestions that any systematic patterns in the data that we do observe are more profitably interpreted by means of the alternative institutional framework. In part II ( chaps. 5 and 6), we show that key aspects of mass political behavior are more profitably explained in terms of the incentive patterns generated by different institutional configurations . Our focus on the po­ litical behavior of ordinary citizens (as opposed to political elites) is motivated by two considerations. First, the suggestion that rules affect the behavior of political elites such as legislators is hardly novel. While values and norms were once seen as the key element in determining legislators' behavior (e.g., Matthews 1 9 60; Huitt 1 9 6 1 ) , that day is long gone. Second, arguments about distinctive political cultures associated with the value enactment approach are, at their core, arguments about broad-based constellations of societal values. Hence our focus on the ways in which institutional constraints affect the behavior of ordinary citizens constitutes a stronger evaluation of the argument. Finally, the particular forms of political behavior considered in part II center on rates of electoral participation and patterns of support for parties of the extreme right. These two forms of political behavior have attracted considerable attention for some time. For example, understand­ ing the sources of political participation and patterns of support for extremist political movements were key components of Lipset's ( 1 9 60) classic synthesis of political sociology. More recently, Putnam ( 2ooo, chap. I ) begins his analysis of declining social capital in the United States with a discussion of patterns of electoral participation. And the unex­ pected electoral success of many extreme right parties over the past two decades in wealthy Western European democracies has rekindled debate about the cultural origins of antisystem political movements. We have been working on this project for a little longer than we had originally planned. Along the way, we have accumulated a number of debts. Much earlier versions of some of the material presented here appeared in "Political Institutions and Voter Turnout in the 1 9 8os," Comparative Political Studies 27 ( 1 9 9 5 ) : 4 67-92; " A Renaissance of Political Cul­ ture ? " and "The Poverty of Political Culture," A merican journal of Politi­ cal Science 40 ( 1 9 9 6 ) : 63 2- 5 9 , 697- 7 1 6; " Conditions Favouring Parties of the Extreme Right in Western Europe, " British Journal of Political Science 25 ( 1 9 9 6 ) : 5 0 1 - 2 1 (with Karin Volpert) ; and " Social Capital and Politics," Annual Review of Political Science 1 ( 1 99 8 ) : 4 7- 73 . While we

XVlll

PR E FA C E A N D ACK N O WL E DG M E N T S

have expanded our analyses considerably, w e are grateful t o the editors and referees of these journals for their constructive comments on those earlier reports from our project, and their comments have continued to inform our work. Since Karin Volpert contributed to the earlier analysis of patterns of support for the extreme right in Western Europe, she is listed as a coauthor of chapter 6 here. At various points in the book, we have employed materials from the Internet. While this is an impressive source of information, not all Web addresses last forever. The references that we have made to particular Web addresses on the Internet were all current as of November 2003. Several people played a vital role in ensuring that the transition from manuscript to book went smoothly. James Reische, Jeremy Shine, Charles Myers, Sarah Mann, and Kevin Rennells each contributed to this process, and we thank them for their patience and support. Many others have contributed to this book in a variety of ways, offering feedback and reactions to ideas when they were in more embryonic form, helping us locate data, and drawing our attention to studies of which we would otherwise have been unaware. These include Josephine Andrews, Bruce Bueno de Mesquita, Ethan Bueno de Mesquita, Alexander Field, Mary Jackman, Arend Lijphart, Bill McCarthy, Jeannette Money, Gabri­ ella Montinola, Stephen Nicholson, Randolph Siverson, James Spriggs, and Walter Stone. Randy Siverson also steered us away from an ungainly title, John Daniels offered crucial help on the figures, and Deborah Patton prepared the index. We were also fortunate to receive two detailed and constructive reviews (William Mishler subsequently identified himself as one of the referees ) . We appreciate the effort that went into these reviews and the fact that they closely complemented each other. The book is immea­ surably improved by these contributions, for which we are grateful.

ONE

The Issues Ever since Max Weber, many social scientists looked at the " right" cul­ tural attitudes and beliefs as necessary conditions ( " prerequisites " ) for economic progress, j ust as earlier theories had emphasized race, climate, or the presence of natural resources. In the 19 s os, newly fashioned cul­ tural theories of development competed strongly with the economic ones (which stressed capital formation ) , with Weber's Protestant Ethic being modernized into David McClelland's "achievement motivation " as a pre­ condition of progress and into Edward C. Banfield's "amoral familism " as an obstacle . According to my way of thinking, the very attitudes alleged to be preconditions of industrialization could be generated on the job and " on the way, " by certain characteristics of the industrialization process. Albert Hirschman ( 1 9 84, 9 9 )

re we prisoners of our distinctive pasts, governed by customs and habits that evolve, if they do so at all, at a glacial pace over several generations ? Such is the broad assertion of those who propose that cul­ tural differences are at the heart of the crucial variations in political behavior that we so often observe when comparing disparate social groups and states. Thus, the claim has recently been revived that there were distinctive traits to popular values in Germany that made Nazism possible, traits not shared, say, with Danes. Most notably, ordinary Ger­ mans had displayed singularly high levels of anti-Semitism for at least the preceding century and a half, which fact served as the foundation of the Nazi state. Given these traits, German behavior during the Nazi period was necessarily different from what Danish behavior could potentially have been, even had both groups faced similar political, social, and eco­ nomic conditions ( Goldhagen 1 9 9 6 ) .

2

BE F O RE N O RM S

Others counter such broad arguments by stressing that people every­ where optimize, given the known alternatives. That is, people process information in a similar manner to get as much as they can of what they want at the least possible cost. Observed differences in political behavior stem from the considerable variance across settings in the conditions under which people optimize. Some have an information advantage over others, and there are major differences in the political, social, and eco­ nomic circumstances within which individuals and groups find them­ selves. Hence, for example, we understand why people might be more sensitive to their short-run than to any possible long-term interests in a Hobbesian world where life is " solitary, poor, nasty, brutish, and short. " By the same token, we expect the same people to maximize their longer­ term interests in a different setting where their property rights are secure, where they can envisage encouraging economic prospects, and where they have a reasonable prospect of longevity. In other words, whether people are future-oriented is fundamentally a function of whether such an outlook makes sense, given their circumstances. These are the two alternatives sketched out by Albert Hirschman, and there is a world of practical difference between them. The first assigns causal priority to cultural values as driving other outcomes, while the second casts values as an outcome of the conditions within which people find themselves. Further, the first argument stresses the durability of cultural differences in values across generations, which implies that at any given time, the die is cast for any particular form of political behav­ ior. In other words, observed differences across cultural groups in politi­ cal and social behavior are predestined by shared collective experiences and group membership, and are therefore fundamentally impervious to change through intervention or changes in incentive structures. Thus, Goldhagen's ordinary Germans of the 1 9 3 0s and 1940s were basically wired to be genocidal by a distinctively virulent blend of anti-Semitism inherited from many previous generations of Germans. ' This has two immediate implications. If the die for particular behav­ ioral forms is indeed cast by waves of preceding generations, then the r. Goldhagen's analysis obviously struck a responsive chord and received an extraordi­ nary amount of press attention. However, the criticisms from specialists in the field have been devastating. Among the major problems identified are his selective use of both primary and secondary sources to fit his argument, his stereotyping of Germans, and his willingness to draw inferences about the allegedly distinctive behavior of Germans without reference to comparative data on the behavior of non-German individuals. On these issues, sec espe­ cially the two essays that comprise Finkelstein and Birns ( 1 9 9 8 ) and Jahoda ( 1 99 8 ) . More probing views than Goldhagen's of the motivation and behavior of those who implemented Nazi policy are offered in Browning ( 2ooo, esp. chaps. 5 and 6) and Rauer (2oo r ) .

The Issues

3

cultural argument is conservative in the sense that it minimizes the possi­ bilities for alternative forms of behavior. Those alternatives are simply precluded on the grounds that they are inconsistent with shared group values. There is an additional ethical implication. If group members are indeed all culturally wired to behave in distinctive ways, then individual group members cannot be held accountable for their own behavior. Thus, in what is admittedly an extreme case, if Goldhagen is correct that the behavior of ordinary Germans was a function of a shared set of anti­ Semitic values inherited through several previous generations, then Na­ zism was inevitable, and it is difficult to pass moral judgment on any particular individual Germans for their behavior during the Nazi period.2 The alternative view-that people optimize, given the alternatives of which they are aware-is subject to neither of these shortcomings. First, behavior is not cast as an irrevocable function of the values of the gener­ ations who have gone before. Instead, by emphasizing the selection of pat­ terns of behavior from a menu of the known alternatives, it is an argument that assigns special weight to incentive structures, that is, to the political, social, and economic conditions within which people find themselves. In this vein, the electoral fortunes of the Nazi Party from I9 2 5 through I9 3 3 would be taken to reflect its success in appealing to the material interests of a decisive number of German voters, given conditions after the First World War, the recency of the party system, and in light of the Great Depression (Brustein I 9 9 6; King et a!. 200 2 ) . Because changes in incentive structures lead to changes in behavior, the optimizing argument is not subject to the conservative implications that encumber the cultural argument. Further, from an ethical point of view, behavior is cast as the outcome of intentional choices. Even if conditions restrict the menu from which choices are se­ lected, they thus remain choices subject to ethical evaluation. Our purpose in this book is to adjudicate between these two general explanations of political behavior, that is, between the political culture approach and what we label the institutional approach. The task is impor­ tant given the marked discrepancies between the empirical and ethical implications of the two approaches . Our analyses include empirical eval­ uations of their relative merits, as they have found their way into contem­ porary political science. We conclude that the institutional account out­ performs the cultural account, and by a large margin. Let us begin at the beginning, however, by detailing the distinctive characteristics of each approach. 2 . For an important discussion of the issues in assigning responsibility in cases like these, see Kelman and Hamilton ( T 9 8 9, esp. chap . 8 ) .

4

BE F O RE N O RM S

The Political Culture Approach

The belief that cultural differences underlie many of the observed varia­ tions across political units has a long pedigree, especially among students of comparative politics. The classic exemplar of the argument in the social sciences remains Max Weber's analysis of linkages between the Protestant ethic and the rise of capitalism in Europe ( [ 1 9 0 5 ] 19 5 S a ) , although as J. R. McCulloch's mid-nineteenth-century remarks about the Irish reproduced in the front of this book remind us, the claim was hardly original with Weber. And one of the benchmark studies of the behavioral revolution in the study of politics is Almond and Verba's ( 1 9 6 3 ) exploration of linkages between civic cultures and democracy across five different societies . While the popularity o f this general perspective h a s waxed and waned over the past three decades (e.g., the essays in Almond and Verba 1 9 80; Barry 1 9 70, esp. chaps. 3, 4), the presumption that cultural differences drive significant elements of political and economic life enjoys wide cur­ rency. Indeed, the perspective has experienced a substantial resurgence in the past decade ( Street 1994 ) . 3 Consider Fukuyama's ( 1 9 9 5 ) analysis o f the role o f trust i n economic performance. Comparing three " high-trust" economies (the United States, Germany, and Japan) with three of their " low-trust" counterparts, Fuku­ yama concludes that while such factors as technology, markets, and human capital contribute to economic growth, the key ingredient in generating growth is social capital in the form of a supporting culture of trust, or, as he terms it, " spontaneous sociability. " This claim is very similar to Harrison's ( 1 9 8 5 , 1 992, 1 9 9 7 ) declaration that political, social, and economic devel­ opment all hinge critically on values involving trust, ethical codes, the exercise of authority, and orientations to work and risk-taking. In a similar manner, Putnam ( 1 99 3 ) compares the performance of regional governments in contemporary Italy to conclude that effective governance hinges critically on durable traditions of civic engagement that go back at least a century, and possibly a millennium. His study parallels Inglehart's ( 1 990, 1 9 9 7 ) analyses of a broader range of societies, which argue that cultural values fundamentally drive economic perfor­ mance and democratic stability. Such claims are of more than academic interest. For example, it is often asserted that democratization will continue to face severe handicaps in southern Europe, much of Latin America, and most strikingly in East 3· It also remains a fundamental theme in cultural anthropology (for a critical survey, see Kuper T 9 99 ) .

The Issues

5

Europe and Russia, given the strong legacy of authoritarianism in the recent past. Attempts to link values to economic performance have simi­ larly clear policy implications, as evidenced by the many discussions about possible links between cultural values and the economic growth rates of the East Asian tigers. Moreover, these claims have already re­ ceived a good deal of attention. The Economist, for example, waxed rhapsodically about Putnam's book: "a great work of social science, worthy to rank alongside de Tocqueville, Pareto, and Weber. " 4 Laitin was only slightly less enthusiastic in his appraisal of the book as a " stunning breakthrough in political culture research" ( 1 9 9 5 , 1 7 1 ) . Origins o f the Argument The intellectual godfather to work on political culture is Weber's (r 1 9 0 5l 1 9 5 8 a ) thesis linking Protestantism with the spirit of capitalism.s Weber, of course, sought to identify the peculiar characteristics of Protestantism that he believed gave rise to a distinctive entrepreneurial spirit conducive to economic growth. This argument is at the heart of contemporary claims about political culture, and we examine it more fully in chapter 2 . Subsequent studies o f political culture frequently offered a unique exegesis of political behavior within a given state, often cast in terms of national character. Perhaps the most systematic of these involved a series of country studies on the organization of civic training put together by Charles Merriam in the 1 9 20s and summarized in The Making of Citi­ zens ([19 3 1 ] 1 9 6 6 ) . However, studies explicitly cast in the national­ character mold became increasingly rare after the mid- 1 9 5 os, a fact Pye ( 1 99 1 a) attributes to the difficulties such studies encountered in meeting the standards proposed by Inkeles and Levinson ( 1 9 54 ) . 6 4 · " Pro Bono Publico, " Economist, February 6 , ' 99 3 , 9 6 . 5 . We a r e not, of course, assigning exclusive paternity to Weber. Indeed, versions of the argument can also be found in such writers as de Tocqueville. Consider the following. [Without] common belief no society can prosper; say, rather, no society can exist; for without ideas held in common there is no common action, and without common action there may still be men, but there is no society. In order that society should exist and, a fortiori, that a society should prosper, it is necessary that the minds of all the citizens should be rallied and held together by certain predominant ideas; and this cannot be the case unless each of them sometimes draws his opinions from the common source and consents to accept certain matters of belief already formed. ( de Tocqueville [ ' 8 3 5 - 40] ' 94 5 , 2: 8 ) Primary paternity remains Weber's, however, since he most fully articulated the argument. 6. According to Pye, "the national-character approach was . . . brought to an almost complete stop in T9 5 4 " because of the lnkeles and Levinson paper, "which was supposed to

6

BE F O RE N O RM S

Against this backdrop, Almond and Verba's classic study ( 1 9 63 ) repre­ sented a major breakthrough because it replaced the idiosyncratic, case­ by-case explanations typical of the national-character studies with an ex­ planation couched in terms of general features of political culture that vary systematically from one setting to another and that foster democratic per­ formance. In this sense, their analysis embodied a return to the concerns that motivated Weber in a format that was also sensitive to the analytic and empirical issues raised by Inkeles and Levinson ( 19 5 4 ) . Almond and Verba paid special attention to the contrasts among participant, subj ect, and parochial cultures, arguing that democratic outcomes are more likely where participatory norms are widespread, and less likely where trust is low and where values assume a predominantly passive, subject form. At about the same time, McClelland ( 1 9 6 1 ) drew explicitly on Weber to suggest that high concentrations of values emphasizing need for achieve­ ment are the engine that drives economic growth. Since need for achieve­ ment is a syndrome that emphasizes entrepreneurial skills, the parallels with Weber are straightforward.7 McClelland further suggested that au­ thoritarian regimes were the likely outcome of cultures that stressed high levels of need for power and low levels of need for affiliation. Banfield made the similar argument that economic and political " backwardness" is a function of high levels of amoral familism according to which the norm is to "maximize the material, short-run advantage of the nuclear family; assume that all others will do likewise" ( 1 9 5 8 , 8 5 ) . Thus defined, amoral familism involves low levels of trust and resembles both Almond and Verba's subj ect culture and McClelland's views on low levels of need for achievement. By the late 1 9 6os, the argument had thoroughly permeated analyses of development. For example, in his influential Political Order in Changing Societies, Huntington wrote: The absence of trust in the culture of the society provides formidable obstacles to the creation of public institutions . Those societies deficient have been written in the spirit of constructive criticism but which called for impossibly high scientific standards. Inkeles never rejected the idea of national character; he only made it impractical to usc it" (Pyc I99 Ia, 4 9 6 ) . 7 · I n a companion paper, l'vlcCielland complained that " national character is not taken very seriously in discussing economic and political development" ( T 9 6 3 ) and proposed need for achievement as a key ingredient of national character. A subsequent book addressed the ways in which levels of achievement motivation might be manipulated in order to enhance economic performance (McClelland and Winter r 9 6 9 ) . McClelland's work inspired a num­ ber of other specific analyses of the impact of achievement norms on economic perfor­ mance; see, for example, Bradburn and Rerlew ( T 9 6 T ) , Cortes ( T 9 6 T ), and Morgan ( r 9 64 ) .

The Issues

7

stable and effective government are also deficient in mutual trust among their citizens, in national and public loyalties, and in organiza­ tion skills and capacity. Their political cultures are often said to be marked by suspicion, jealousy, and latent or actual hostility toward everyone who is not a member of the family, the village, or, perhaps, the tribe. These characteristics are found in many cultures, their most exten­ sive manifestation perhaps being in the Arab world and in Latin Amer­ ica. ( 1 9 6 8 , 2 8 ) m

The theme i s echoed repeatedly i n the more recent analyses. Thus, Harrison acknowledges that resource endowments, policy choices, and sheer luck may be among the operative factors explaining variations in performance across nations and ethic groups: " But it is values and atti­ tudes-culture-that differentiate ethnic groups and are mainly respon­ sible for such phenomena as Latin America's persistent instability and inequity, Taiwan's and Korea's economic 'miracles,' and the achievements of the Japanese-in Japan, in Brazil, and in America " ( 199 2, r ) . Fukuyama arrives at the same conclusion, casting culture as social capital. A society's endowment of social capital is critical to understanding its industrial structure, and hence its position in the global division of labor. . . . [But] social capital has implications that go well beyond the economy. Sociability is also a vital support for self-governing political institutions, and is, in many respects, an end in itself. Social capital, which is practiced as a matter of arational habit and has its origins in " irrational " phenomena like religion and traditional ethics, would ap­ pear to be necessary to permit the proper functioning of rational mod­ ern economic and political institutions-a fact that has interesting im­ plications for the nature of the modernization process as a whole. ( 19 9 5 , 3 2 5 )

Among the more rigorous of the current analyses, Putnam's emphasis on sense of civic community matches Banfield's emphasis on trust and Almond and Verba's case for the importance of a civic political culture. The absence of civic virtue is exemplified in the " amoral familism" that Edward Banfield reported as the dominant ethos in Montegrano . . . . Participation in a civic community is more public-spirited than that, more oriented to shared benefits . Citizens in a civic community, though not selfless saints, regard the public domain as more than a battle­ ground for pursuing personal interest. ( 19 9 3 , 8 8 )

8

BE F O RE N O RM S

In like manner, Inglehart places primary emphasis o n variations over time and space in the configuration of mass value priorities. Data from roughly two dozen nations reveal a consistent cultural­ economic syndrome. The wealthier countries and those with highly developed tertiary sectors are most likely to be long-established democ­ racies, and the publics of these nations tend to show relatively high rates of political discussion, have less Materialist value priorities, and tend to be Protestant in religion . . . . These nations, furthermore, tend to have publics that are characterized by high levels of life satisfaction and interpersonal trust, low levels of support for revolutionary change, high levels of satisfaction with the way democracy is working, and high rates of political discussion. Conversely, the less wealthy, less demo­ cratic, and less Protestant nations tend to be characterized by political cultures that show low levels of trust and satisfaction, high levels of support for revolutionary change, and low rates of political discussion. ( 1 990, 57)

The political culture perspective clearly has a long pedigree. Equally clear is the central role it continues to play in current explanations of political behavior and economic performance. Distinctive Features of the Argument Central to all of these claims is the idea that political cultures stem from particular configurations of attitudes across a broad group of individuals. In other words, cultures reflect relatively coherent clusters of attitudes. On some occasions, the syndrome or modal personality type is taken to reflect the prevalence of entrepreneurial sentiments (e.g., need for achieve­ ment), while at other times it is said to reveal a sense of civic community or civic virtue (as in Banfield, Almond and Verba, Inglehart, and Put­ nam ) . The important point, however, is that the syndrome is coherent. A second point follows immediately. Arguments about political culture are fundamentally concerned with the prevalence of such value clusters within societies. While the cultural pattern reflects the attitudes of individu­ als, it assumes political and social significance to the extent that it is widely endorsed. This means that at their heart, cultural arguments are concerned with aggregate properties of societies. The point was made with great clarity by Inkeles and Levinson, who insisted that a culture be equated with " modal personality structure: that is, it should refer to the mode or modes of the distribution of [individual] personality variants within a given soci-

The Issues

9

ety" ( I 9 69, 4 2 5 ) . By Weber's original account, for example, it was the widespread diffusion of entrepreneurial attitudes fostered by Protestant­ ism that gave rise to economic growth. Third, these cultural syndromes are durable. Even if slightly modified by short-term forces, their fundamental effects persist over the long haul. Again, as Inkeles and Levinson observed: Whatever their specific nature, [the components of culture] are rela­ tively enduring personality characteristics, for example, character traits, modes of dealing with impulses and affects, conceptions of the self, and the like. These are not phenotypic, behavior-descriptive terms . Rather, they are higher-level abstractions that refer to stable, general­ ized dispositions or modes of functioning and may take a great variety of concrete behavioral terms . (Inkeles and Levinson r 9 69, 4 2 6; em­ phasis in original)

The durability theme appears repeatedly in studies of national cul­ tures, which cast distinctive values as the product of socialization. For example, Banfield claimed bluntly that " the Montegranesi act like selfish children because they are brought up as selfish children" ( I9 5 8 , I 5 I ) , describing an iterative process repeated across many generations. Simi­ larly, in his wide-ranging essay on Russian political culture, Keenan ( I 9 8 6) asserts that Russian political culture is distinctly centralized, bu­ reaucratic, and authoritarian, and that Russians display low levels of interpersonal trust and prefer to act in a shroud of secrecy. Keenan traces these traditions from sixteenth-century Muscovy directly through the Brezhnev years, and presumably the period since. The claimed continuity is thus extensive, covering as it does fully five hundred years.s Durability is also fundamental to more general cultural accounts. Thus, the " basic message " that Harrison seeks to convey is: Human development is frustrated in most Hispanic-American coun­ tries-and most Third World countries-by a way of seeing the world that impedes the achievement of political pluralism, social equity, and dynamic economic progress. And that way of the world has been driven, without significant deviation, by the momentum of centuries. ( r 9 8 5 , r 6 8 ; emphasis added) 8. An extensive discussion of Keenan's analysis is provided by Crummey et al. ( r 9 8 7 ) . Keenan's basic argument has been echoed repeatedly (e.g., Joyce r 9 84; White r 9 84; Hed­ lund T 9 9 9 ) .

IO

BE F O RE N O RM S

Putnam ( 1 9 9 3 ) goes even further to trace differences i n performance across Italian regional governments to legacies of civic engagement reach­ ing back a thousand years to the Middle Ages. The point is also under­ scored by Inglehart. The political culture approach is distinctive in arguing that ( r ) people's responses to their situations are shaped by subj ective orientations, which vary cross-culturally and within subcultures; and ( 2 ) these varia­ tions in subj ective orientations reflect differences in one's socialization experience, with early learning conditioning later learning, making the former more difficult to undo. Consequently, action can not be inter­ preted as simply the result of external situations : Enduring differences in cultural learning also play an important part in shaping what people think and do. ( 1990, 19; emphasis in original )

He further adds: " People live in the past much more than they realize. We interpret reality in terms of concepts and world views based on past experi­ ences" ( 4 2 2 ) . Thus described, the distinguishing elements of cultural ac­ counts are their emphases on the durability of norms arising from early socialization and on the limited impact of " external situations . " Indeed, Inglehart ( 1 9 9 7, 1 84 - 8 8 ) subsequently contends that norms and values are sufficiently durable that estimates obtained in 1990 can usefully be used as surrogates for unobtainable estimates for 1 9 20, fully seventy years earlier. While this represents a shorter span than Putnam's millen­ nium, it remains a strong claim about the durability of cultural values. In an interesting twist, Rice and Feldman ( 1 99 7 ) compared the civic values of current residents in eleven West European countries with those of Americans of European descent. Thus, they compared Italians with Italian-Americans, Danes with Danish-Americans, and so on. They re­ ported that, in the aggregate, Americans whose predecessors emigrated from populations that are now highly civic themselves hold civic values, while descendants of emigrants from less civic populations themselves are less civic. They further reported that timing of immigration has no bear­ ing on civic values in the sense that those values do not hinge on whether all, some, or none of the grandparents of European-origin Americans were born in the United States. This suggests one of two things. Perhaps civic values are remarkably long-lived and insensitive to any diverging conditions, having basically been frozen over several generations in both origin and destination countries. Alternatively, such values have evolved in exactly the same manner in both origin and destination settings, so that Danish-Americans continue to resemble Danes, Italian-Americans con­ tinue to resemble Italians, while Americans of Danish descent are quite

The Issues

II

unlike their Italian-origin compatriots. Either way, Rice and Feldman emphasize the durability of these values. Durability was, of course, also central to earlier accounts of political culture. Verba, for example, stressed the role of socialization and direct experience in the formation of political culture but continued: One must look beyond the direct political experiences of the individual. The political memories passed from generation to generation and the way these memories are formed are crucial. One is forced to consider the historical experiences of a nation from the point of view of their impact on political beliefs . ( 1 9 6 5 , 5 5 4 )

Although somewhat more critical of cultural explanations on the grounds that they are inordinately conservative, Moore made the same point. Common observation is enough to show that human beings individu­ ally and collectively do not react to an " obj ective " situation in the same way as one chemical reacts to another when they are put together in a test tube . . . . There is always an intervening variable, a filter, one might say, between people and an " obj ective " situation, made up from all sorts of wants, expectations, and other ideas derived from the past. This intervening variable, which it is convenient to call culture, screens out certain parts of the obj ective situation and emphasizes other parts . There are limits to the amount of variations in perception and human behavior that can come from this source. Still the residue of truth in the cultural explanation is that what looks like an opportunity or a tempta­ tion to one group of people will not necessarily seem so to another group with a different historical experience and living in a different form of society. ( 1 9 66, 4 8 5 )

Note the claim that deep-seated configurations of norms inhibit people from adapting to changes in broader political conditions, labeled " objec­ tive " and " external " situations by Moore and Inglehart, respectively. Cultural factors are thus said to constrain how individuals adapt to changes in incentive structures embodied in political institutions and the like. Consequently, as Inglehart and many others have pointed out, argu­ ments couched in terms of political culture are at variance with social choice arguments, which imply that different individuals will respond to the same change in incentive structures in a similar manner.9 We return to this point later. 9· This is the case even though lnglehart's suggestion that social choice arguments are concerned solely with the optimization of economic utility is wrong.

12

BE F O RE N O RM S

Fourth, the significance o f these enduring cultural syndromes stems from the way in which they drive other outcomes. Weber's analysis con­ tinues to attract attention because of the effects on economic growth he attributed to Protestantism, and it is hard to imagine this continuing attention without that imputation. Weber was, of course, quite explicit on the point. His fundamental goal was to undermine any form of eco­ nomic determinism by showing that values are not epiphenomenal, Marx to the contrary ( see on this point, e.g., Zeitlin 1 9 9 0 ) . Banfield ( 1 9 5 8 ) was equally explicit: Why was Montegrano so backward ? Because amoral familism was endemic. Indeed, the emphasis on values as the generating force led Portes ( 1 9 7 6 ) to label the cultural account as the " value enact­ ment" approach to development. The same spirit motivates the current analyses. Thus, both Harrison ( 1 99 2, chap. 1 ) and Fukuyama ( 1 9 9 5 , chap. 5 ) argue at some length that Marx was wrong and Weber was right, precisely on these grounds. Similarly, Putnam weighs in against any " simple economic determin­ ism" ( 1 9 9 3 , 1 5 2 ) , insisting instead that cultural norms as reflected in sense of civic community are responsible for both economic and politi­ cal performance: The predictive power of the civic community is greater than the power of economic development . . . . In other words, economically advanced regions appear to have more successful regional governments merely because they are more civic. ( 9 8-9 9 )

Inglehart also casts the cultural account a s a n alternative t o " economic determinism" ( 1 990, chap. 8) or " institutional determinism, " where the latter term refers to " an extreme claim . . . that institutions alone deter­ mine a society's values" ( 1 9 9 7, 9 9 ) . On the basis of this contrast, he concludes: The available evidence indicates that the values and cultural norms held by given peoples are a major influence on whether or not democratic institutions are viable Culture not only responds to changes in the .

.

.

.

environment; it also helps shape the social, economic, and political world. ( 1 990, 4 32; emphasis added) The political culture account thus identifies distinctive clusters of atti­ tudes that are widely held across the relevant set of individuals. These durable clusters form subjective orientations to the world that are highly resistant to change and are seen as the fundamental generator of eco-

The Issues

I3

nomic and political performance. They are, in this sense, more important than objective conditions, and they persist in the face of changes in objec­ tive conditions. Despite the radically different twentieth-century political histories of Italy and the United States, Americans of Italian descent therefore share more political values with contemporary Italians than they do, say, with Americans of Danish lineage (Rice and Feldman 1 9 9 7 ) . We are hardly the first t o observe the conservative implications of this general claim. As Moore ( 1 9 6 6 ) , Partes ( 1 9 76), and many others have pointed out, the argument is one that allows for slow change at most, since it casts subjective orientations as more important than, and substan­ tially independent of, objective conditions. This implies that changes in the latter will have minimal effects on the former. Thus, Banfield's ( 1 9 5 8 ) discussion of amoral familism in Montegrano led him quite naturally to his subsequent well-known declaration that present-orientedness is en­ demic to lower-class culture in the urban United States, that this lower­ class culture is highly resistant to change, and that " so long as the city contains a sizable lower class, nothing basic can be done about its most serious problems" ( 1 974, 2 3 4 ) . Putnam's more recent conclusion that regional differences i n civic­ mindedness across contemporary Italy date back to at least the Middle Ages has the same implications. Poor economic and political perfor­ mance stem from cultural pathologies that, in the critical words of an­ other scholar, are " self-generating in the double sense that socialization perpetuates both the cultural patterns of the group and consequent indi­ vidual psycho-social inadequacies blocking escape from rtheml " (Valen­ tine 1 9 6 8 , 1 4 1 ) . For these reasons, LaPalombara j udged the analysis to have " dismal " implications. Civic traditions not only have remarkable staying power; they are also almost impossi ble to change, even in the long run . . . . This will be small solace to southern Italians, j ust as it will, unfortunately, confirm the stereotype held by so many northern Italians that " Italy, from Rome south, is really part of Africa. " ( 1 9 9 3 , 5 3 0 )

A southern Italian regional president had a similar reaction: "This i s a counsel of despair! You're telling me that nothing I can do will improve our prospects for success. The fate of reform was sealed centuries ago " (quoted in Putnam 1 99 3 , 1 8 3 ) . The conclusion is inescapable. The cultural account thus constitutes a coherent claim that has major implications for the ways in which we envision political life. We turn now to the principal alternative explanation of the same outcomes.

14

B E F O RE N O RM S

The I n s titution al Approach

Unlike the cultural perspective, the institutional view specifies that people optimize, choosing from the menu of available alternatives. Observed behavioral variations across groups do not stem from distinctive values or from group-specific ways of processing information but instead reflect the different menus available to different actors and groups. The institu­ tional approach thus assigns special weight to the incentive structures generated by the political, social, and economic conditions within which people find themselves. Let us digress briefly to explain the institutional label, since our usage may not be self-evident.10 Until the 1 9 6os, political science was dominated by an approach that emphasized formal rules and procedures, embodied in constitutions and the like, implicitly assuming that political life could be understood as an almost perfect function of those rules. The " behavioral revolution " that occurred in the 1 9 6os, but that began much earlier with the work of Charles Merriam in particular, can be seen as a reaction to this emphasis on formal rules. Drawing heavily on related social sciences, such as anthropology and social psychology, scholars in the behavioral mode drew our attention to less formal patterns of political behavior, emphasiz­ ing the importance of understanding how citizens actually behave rather than the ways they are supposed to behave, given the rules (Dahl r 9 6 r b ) . We have collectively learned much from the behavioral approach: for example, we have an enhanced sense of the limited levels of political infor­ mation exhibited by citizens across the industrial democracies. As we have already indicated, Almond and Verba's ( r 9 6 3 ) analysis is widely regarded as a benchmark study in this mode. ' ' Over the last twenty years, students of politics have come increasingly to adopt a "new" form of institutionalism. Unlike the older form of analysis that predominated until the 19 s os, this more recent focus ac­ cepts the behavioral premise that we focus on observed patterns of politi­ cal behavior, as opposed to those prescribed by rules. At the same time, the newer form of institutionalism emphasizes the contexts within which these forms of behavior are concerned, contexts defined by institutions and rules. These contexts, in turn, are seen as generating distinctive incen­ tive structures that constrain the choices of political actors and hence ro. Parallel treatments of the issues involved in this and the following paragraph can be found, for example, in Shepsle ( r 9 8 9 ) and Calvert ( r 99 5 ) . r r . Other key behavioral studies include The American Voter ( Campbell e t al. r 9 6o), Political Man (Lipset r 9 6o), and Who Governs? (Dahl r 9 6 r a ) .

The Issues

rs

affect the forms that their political behavior assumes. In this sense, ob­ served differences in political behavior across groups are seen as a func­ tion in part of variations in the institutional arrangements to which they are exposed. We use the term institutional to refer to this newer form of analysis. While the approach is not exclusively associated with any one study, well­ known papers by Shepsle ( 1 9 79 ) and March and Olsen ( 1 9 84 ) are often taken as general benchmarks.U We drop new from the label on two grounds. First, there is an issue of brevity. Second, there has to be a statute of limitations issue: the approach is surely much less novel now than it was two decades ago ! In broad terms, the institutional approach that we adopt casts behav­ ior as the outcome of ( r ) the activities of optimizing actors ( 2 ) who operate under institutional constraints. We briefly elaborate our usage of these two terms. The Optimizing-Actors Premise The institutional approach hinges on the conventional premise that hu­ man actors seek to optimize their utility. We substitute the term optimize for maximize here solely to underscore the elementary point that, across individuals and groups, there are major differences in terms of the avail­ ability of both information and choices. Some actors have much more information about alternatives and their likely consequences than others, and some have access to a broader menu of choices than others. Decision making is thus surrounded by considerable uncertainty for all actors, although some face much more uncertainty than others. Saying that ac­ tors seek to optimize their utility does not mean that all actors have the same information nor that they are faced with the same set of choices or incentives. Despite the plausibility of this premise, casting political behavior as rational continues to generate considerable resistance among students of r 2 . Dating shifts in disciplinary emphasis is always difficult. For example, while we have dated the behavioral approach to the r 9 6os, this period simply reflects the time at which that emphasis peaked. As we have noted, the behavioral approach to politics can be traced back to Merriam's work of the T 9 20s and T 9 J OS, and key subsequent scholars like Gabriel Almond and Robert Dahl had earlier been students of Merriam. Similarly, the March and Olsen paper b uilt on a considerable body of research that had already been completed, and many of the concerns of the new institutionalism can be traced back to key analyses like Arrow ( r9 5 1 ) and Black ( r 9 5 8 ) that underscored the difficulties of aggregating preferences into a coherent outcome in the absence of an institutional framework.

r6

B E F O RE N O RM S

political life (e.g., Hindess I 9 8 8 ; Lowi I 9 9 2; Green and Shapiro I 9 94 ) . ' 3 Since at least Lasswell ( I 9 3 0 ) , many have preferred t o see political behav­ ior as motivated by subconscious nonrational drives that entail no design. Indeed, part of the purpose of many cultural accounts of political life is to emphasize supposedly nonrational and subconsciously driven elements of behavior (e.g., Almond I 990; Pye I 9 9 I a ) . The impact on the field of Lasswell's Psychopathology and Politics ( 1 9 3 0) has thus been monumen­ tal, and confusion too often reigns over the meaning of the term rational. Some of this confusion can be traced back to those behavioral studies that examined electoral participation. Classics in this genre like Berelson, Lazarsfeld, and McPhee ( I 9 54 ) and Campbell et a!. ( I 9 6o) concluded that voters (and nonvoters) are fundamentally irrational in the sense that they exhibit very low levels of information about and apparent compre­ hension of politics. This picture of voters seemed to square poorly with notions of how ideal democratic citizens might be supposed to behave. Berelson was especially emphatic on this point, suggesting that demo­ cratic theory was in need of wholesale revision as a result of his empirical results. Obviously, such conclusions hinge on a substantive view of rationality, in the Weberian sense. This view is often claimed to represent a " broader " perspective on rationality, but its empirical referents are seldom defined with great precision. Thus, Smelser and Swedberg claim that substantive rationality "refers to allocation within the guidelines of other principles, such as communal loyalties or sacred values, " and contrast this with utility maximization ( 1 994, 5 ) . The point they overlook is that both communal loyalties and " sacred values " can fruitfully be analyzed in an optimizing framework (e.g., Hechter 1 9 8 7; Bates, de Figueiredo, and Weingast 1 99 8 ; Iannaccone I 99 5; Hardin I 99 5; Stark and Finke 2000 ). They further gloss over the fact that substantive rationality too often implies j udgments about goals deemed desirable by the analyst rather than goals selected by the obj ects of analysis. In the case of voter turnout, substantive rationality specifies goals and characteristics ( including an active, informed citizenry) that the analyst believes potential voters should adopt. In contrast, the institutional approach is predicated on a procedural r 3· Of these, Green and Shapiro ( r 9 9 4 ) is perhaps the most wide-ranging critique. Even so, it is ultimately an unsatisfactory appraisal. Among other things, Green and Shapiro offer no coherent alternative against which to j udge rational choice arguments, and their criteria for theory evaluations arc idiosyncratic. On these issues, sec the commentaries in hicdman ( r 99 5 ), especially the chapters by Dennis Chong, Daniel Diermeier, John Ferejohn and Debra Satz, l\1orris Fiorina, Stanley Kelley, Susanne Lohmann, Norman Schofield, and Kenneth Shepsle, along with the commentary in Mueller ( 2003 , chap. 2 8 ) .

The Issues

17

view o f rationality. Riker succinctly stated the elements o f the rational choice model as follows. Actors are able to order their alternative goals, values, tastes, and strategies. This means that the relation of preference and indiffer­ ence among the alternatives is transitive . . . . 2 . Actors choose from available alternatives so as to maximize their satisfaction. ( 1990, 1 7 2; emphasis added) r.

Observe that this definition does not speak to the content of actors' goals. Instead, all that is specified is that, given a set of goals, actors will behave efficiently to optimize their utility. Obviously, different actors will often have distinctive goals, a fundamental source of political conflict. The basic point, however, is that goals are chosen from the alternatives that various actors (as opposed to the analyst playing God) believe to be available or feasible. Observe further that this definition does not preclude miscalculations on the part of actors. Much of social choice theory addresses the ways in which the aggregation of individual preferences, rationally generated, can lead to collectively irrational outcomes (e.g., Arrow 1 9 5 1 ) . Thus, Downs ( 1 9 5 7, chaps. 8, 9) examined how rational parties attempt to induce irrationality among individual voters by adopting ambiguous policy posi­ tions. Similarly, Andrews ( 2002 ) has shown the ways in which cycling over proposals for constitutional reform prevented legislative majorities in the Russian Duma from adopting a new constitution in the early 1 990s. The emphasis then is on optimizing expected or anticipated satis­ faction. Miscalculations are also to be expected, given uncertainty and imperfect information. Riker pointed out: It is quite possible for people to choose alternative actions that frustrate their primary goals. It is also quite possible that, lacking information about others' choices, people choose actions ( even ones with undesired consequences) that would be different from those they would choose with full information. In short, this definition requires only that, within the limits of available information about circumstances and conse­ quences, actors choose so as to maximize their satisfaction. ( 1 990, 1 73 )

The emphases on uncertainty and incomplete information are charac­ teristics of this framework that have given it a decidedly political bent since at least Downs ( 1 9 5 7) . Given these emphases, the idea of proce­ dural rationality has proven to be an important premise that has led to

r8

B E F O RE N O RM S

the analysis of such obviously political phenomena a s agenda setting, ambiguity on the part of political leaders, and the importance of office­ seeking considerations. '4 Despite this, the notion of procedural rationality continues to induce opposition. One common complaint is that in an effort to explain every­ thing, rational choice theory becomes circular and nonfalsifiable (e.g., Green and Shapiro 1 994 ) . But there is no single rational choice "theory" that seeks to explain everything. Instead, the distinctive feature of the approach is the premise of optimizing behavior, so that the approach is best seen as a " family of theories or, better, a field of endeavor" ( Ferejohn and Satz 1 9 9 5 , 8 1 ) . Premises, of course, are judged by their success in generating falsifiable explanations. As will become clear as we proceed, the optimizing premise has been most productive in this respect. Another common complaint is that the emphasis on utility optimiza­ tion is inordinately restrictive because it precludes altruism and allows only for selfish behavior (e.g., Mansbridge 1 990; Monroe 1 9 9 6 ) . But as Tsebelis ( 1 990, 2 1 ) and others have pointed out, procedural rationality implies no such stipulations because ( unlike the substantive view of ratio­ nality) it does not address the question of goals. More generally, rational choice arguments are not restricted to utility defined solely as material economic interests, and in fact many non­ market applications carry no such restriction. During the forty years since Downs ( 1 9 5 7 ) was originally published, for example, students of politics have come increasingly to understand politicians to be fundamentally motivated by office-seeking concerns and, to a lesser extent, perhaps, by policy concerns. Even the most venal dictators often appear sensitive to officeholding considerations, although not exclusively so. ' 5 While utility may be defined in evolutionary biology in terms of gene reproduction (e.g., Dawkins 1 9 89 , Cronin 1 99 1 ), in the social sciences it simply refers to the preferences of the individual or set of individuals under discussion. r 4 . Our distinction between substantive and procedural rationality follows Riker ( 1 9 9 0 ) . Others make a slightly different distinction between " procedural " and " instrumen­ tal " rationality (e.g., Zagare 1 990), where the former refers to utility maximization with full information and complete certainty (following people like Simon) and the latter involves procedural rationality in Riker's terms. r 5. Consider the case of the late former president Mobutu Sese Seko of Zaire (now the Congo ) , who, had he been motivated solely by pecuniary greed, might reasonably have thought it prudent in the early 1990s to cash in and retire to Switzerland, where, free from the distractions of office, he would have been in a position to monitor and manage much more closely his considerable financial assets . Instead, he continued to reside in Kinshasa (punctu­ ated by lengthy trips abroad for health treatment), despite increasing challenges to his author­ ity and escalating threats to his personal security. There was an obvious revealed preference for officeholding here, independent of his other clear preference for personal wealth.

The Issues

19

Since there is obviously considerable variance in preferences across indi­ viduals, utility cannot be defined exclusively in simple income terms. Arguments couched in terms of procedural rationality can in fact sub­ sume a variety of goals, ranging from materialistic to idealistic or expres­ sive, and from egoistic to altruistic. '6 Thus, although it is typically taken to exemplify nonmaterial, idealistic values, religious behavior has effectively been analyzed in a rational choice framework (e.g., Stark and Finke 2000). Even martyrdom, apparently the ultimate irrational act for those who conceive of possible rewards as solely temporal, has fruitfully been exam­ ined in these terms (e.g., Iannaccone 1 9 9 s; R. Stark 1 9 9 6, esp. chaps. 8, 9 ) . Focusing o n more routine activities, Aldrich ( 1 9 9 7 ) and Schuessler ( 2ooo) have shown that consumer and voter behavior is often motivated by expres­ sive or symbolic, as opposed to instrumental, concerns. Voting in large electorates, especially, is more fruitfully seen as expressive than instru­ mental, since the odds of casting a decisive vote are exiguous. As Aldrich reminds us, "all votes are always wasted, if casting them is done for the purpose of affecting the outcome " ( 1 9 9 7, 3 7 8 ). Expressive participation is readily integrated into a rational choice framework, where the producers of participation (candidates, marketing experts, and the like) " strategically target the expressive motivation of prospective participants in order to elicit their participation " ( Schuessler woo, 6 3 ) . I n a similar vein, much behavior that might appear altruistic i s entirely consistent with an egoist's version of the Golden Rule: "Do as you would be done by. " Since social and political life obviously consists of more than one-shot encounters, apparent altruism of this form makes sense. Such behavior is predicated on expectations of reciprocity and is a form of insurance that helps govern future encounters by reinforcing reputations and trust (e.g., Wilson 1 9 7 8 , chap. 7; 0 . Stark 1 99 5 ) .I? The common r 6. The point is repeatedly misconstrued. For example, Sears and Funk ( 1 9 9 0 ) argue that self-interest has little impact on public opinion, and that the rational choice approach is therefore limited. Instead, "the general public thinks about most political issues, most of the time, in a disinterested frame of mind" ( r 7 o ) . However, this conclusion is predicated o n a definition of self-interest that includes only short- t o medium-term interests a n d material well-being, and refers only to the individual or that individual's immediate family. Excluded from the definition are long-term interests, nonmaterial well-being, and interests that affect the well-being of the individual's group but not necessarily the particular individual . This is inordinately narrow. r 7 . The label " reciprocal altruism" was proposed by Trivers ( r 9 7 r ) . As Wilson empha­ sizes, such altruism is ultimately self-serving in the sense that altruists expect reciprocation in the future either for themselves or for their closest relatives. Reciprocation among distantly related or unrelated individuals is the key to human society. The perfection of the social contract has broken the ancient vertebrate constraints imposed by rigid kin selection. Through the convention of reciprocation,

20

B E F O RE N O RM S

complaint that the rationality postulate applies only to the analysis of selfish, economic utility maximization is thus baseless. Others object that common views of procedural rationality are empiri­ cally deficient. Perhaps the best-known argument along these lines comes from Simon (e.g., 1 9 8 5 , 1 9 8 6), with his emphasis on bounded rationality ( see also Jones r 999 ). This proposal would carry more weight if it did not inaccurately attribute to procedural rationality the assumptions of com­ plete information and certainty. But even a cursory reading of Downs and the literature since reveals the centrality of arguments about imperfect information and uncertainty to rational choice models (e.g., Calvert 1 9 8 6) . Given this, " bounded rationality" is logically indistinguishable from procedural rationality, as commonly used. The point was shown very clearly thirty years ago by Riker and Ordeshook ( 1 9 7 3 , 20- 2 3 ) . Satisficing, to use Simon's term, i s thus merely a shorthand label for optimizing under conditions of limited information and uncertainty. ' 8 Finally, i t is often claimed that the idea o f procedural rationality i s culturally confined. Some seem t o believe that because different groups and political cultures emphasize different values, the possibilities for ra­ tional choice are restricted. Instead, different groups are said to process information in different ways, that is, they do not behave equally effi­ ciently. Thus, we are all familiar with stereotypes that claim, for example, that women are more emotional while men are more rational, or that people in some non-Western cultures are more " spiritual" while citizens of advanced industrial societies are more motivated by material concerns. But such claims conflate substantive with procedural rationality, and the latter is moot on the subj ect of values (i.e., goals), as we have already indicated. Other observers imply that the rules of decision making are dominated by unique configurations of enduring values or norms associ­ ated with different cultures. Were this the case, however, we would be combined with a flexible, endlessly productive language and a genius for verbal classification, human beings fashion long-remembered agreements upon which cul­ tures and civilizations can he built. ( 1 9 7 8 , 1 5 6) T 8. A related critique of the procedural position centers on the way decisions are " framed" (Tversky and Kahneman 1974, 1 9 8 6 ) . However, while the ways in which choices arc framed may affect the preference functions of actors, this critique is moot because the procedural position concerns behavior given a set of goals. For example, Downs is best read as an analysis of how citizens' preferences are framed by the behavior of parties and leaders . Aldrich's ( 1 993 ) emphasis on the impact of strategic politicians a l s o hears directly o n questions of framing. I n both instances, the framing i s embedded in a wider theory, unlike that offered by Tversky and Kahneman. There is strong evidence that an optimizing-actor premise outperforms arguments built around either Simon's bounded rationality or Tversky and Kahneman's framing argument (Wittman r 9 9 5 ; Lupia and McCubbins 1 9 9 7 ) .

The Issues

21

reduced t o idiographic descriptions a s opposed t o nomothetic attempts at explanation. In other words, we would be reduced to eschewing general­ izations, much in the mode of early studies of national character. An apparently more subtle claim comes from Eckstein ( 1 9 9 2 ) , who declares that in some cultures (say, among peasants or the poor) there is more concern with minimizing pain than with maximizing pleasure, and that the former is qualitatively distinct from the latter. However, the suggestion that there is a disjuncture between these two activities hinges entirely on the restrictive assumptions that utility is equated solely with " pleasure " and that all individuals perceive that they are confronted with the same choices. This, of course, ignores the essential fact that individu­ als in some circumstances face quite constrained opportunity sets (e.g., Popkin's [ 1 9 79] analysis of the choice options available to peasants ) . ' 9 For those living in a Hobbesian world and confronted with a compressed range of feasible alternatives ranging from horrendous to merely awful, minimizing pain is the same as maximizing utility. Both are surely optimiz­ ing forms of behavior.20 Institutional Constraints and Incentive Structures Actors do not optimize in an institutional vacuum. Instead, as we have already emphasized, choices are always constrained for all participants, although the severity of the constraints varies considerably across actors or groups of actors (Riker 1990, 1 7 2 ) . These constraints stem from insti­ tutional arrangements. Following Knight, we take a social institution most generally to be "a set of rules that structure social interactions in particular ways. " To be counted as an institution, " knowledge of these rules must be shared by the members of the relevant community or society " ( 1 992 , 2- 3 ) . This is T9. Analyses like Popkin ( T 9 7 9 ) , Tong ( T 9 8 8 ), and Lichbach ( T 9 9 5 , T 9 9 6 ) also belie common claims that optimizing accounts cannot be used to address the political behavior of the powerless. For a recent instance of such a claim, see the statement that " epistemological commitments to uncertainty, ambiguity, and messiness invite interpretivists to focus on social movements, political resistance, and modern povver in ways that are irrelevant to rational choice theorists " (Wedeen 2 0 0 2 , 7 2 6 ) . While it is the case that rational choice analyses have demonstrated no interest in messiness, analyses of uncertainty and ambiguity have long been central to the approach. 20. Consider the choices available to a woman confronting an unwanted pregnancy. Or consider a peasant in the Ming period choosing between castration (in the hope of gainful employment as a court eunuch ) and starvation (Tong I 9 8 8 ) . Eckstein's argument implies that rational decision making is impossible under conditions like these: utility cannot be optimized because none of the alternatives allows for the maximization of pleasure. The implication is odd.

22

B E F O RE N O RM S

obviously an inclusive definition that encompasses institutions ranging from those that govern marriage and family relations to those that regu­ late patterns of international trade. Further, it is a definition that em­ braces informal norms, social conventions, and the rules that guide politi­ cal and economic activity. Some of these institutions are embodied in law and thus subject to enforcement by the state; others (e.g., many social conventions, international agreements ) are largely self-administered be­ cause they are not subj ect to monitoring by external agents regarded by the relevant participants as authoritative; still others are a blend of the two (Knight 1 9 9 2 ; Calvert 199 5 ) . 2 1 It is, of course, difficult to sustain a watertight distinction between political and social institutions, since the former are a subset of the latter. With this in mind, we treat as " political " those institutions that contrib­ ute to decision making by regulating the aggregation of political prefer­ ences into coherent (not necessarily desirable ! ) outcomes. Examples of such political institutions abound. Considerable attention has been paid to the electoral laws that translate the preferences of individual voters into the selection of governments (e.g., Rae 1 9 7 1 a; Taagepera and Shu­ gart 1 9 89 ; Lijphart 1994; Cox 1 9 9 7 ) . Similarly, much of the study of legislatures centers on the rules and procedures (voting systems, commit­ tee systems, and the like) that decode legislation from the proclivities of particular legislators (e.g., Shepsle 1 9 79 ; Weingast and Marshall 1 9 8 8 ; Cox 2ooo). Parallel analyses center o n the ways i n which judicial institu­ tions affect public policies (e.g., Epstein and Knight 2ooo; Maltzman, Spriggs, and Wahlbeck 2000 ). In each of these cases, political institutions constrain the aggregation of preferences into outcomes. Different poli­ tical institutions can indeed generate markedly distinctive political out­ comes even in the face of similar distributions of preferences. It could perhaps be countered that any such differences across settings in institutional forms themselves simply reflect cultural variations. The problem then is that institutions are divorced from their political origins. We think institutions are more profitably cast as the result of conflict over the allocation of valued goods. Phrased differently, institutions reflect 2 I . Occasionally, cultures arc cast in terms that resemble this rules-of-the-game ap­ proach to institutions. For example, Greif treats cultures as the beliefs "that capture indi­ viduals' expectations with respect to actions that others will take in various contingencies" ( r 994, 9 r 5 ) . Chwe ( 2 oo r ) similarly analyzes cultures (manifested in rituals and ceremonies) as mechanisms that generate common knowledge that resolves coordination problems. Such usages of the term culture clearly diverge substantially from those common in the discus­ sions of political culture that we summarized earlier in this chapter and are fully consistent vvith a rational choice account.

The Issues

23

" the efforts of some to constrain the actions of others with whom they interact" (Knight 1 9 9 2, 1 9 ) . By this reckoning, institutions do not reveal cultural differences but are instead products of interdependent strategies of groups to achieve distributional gains ( Shepsle 1 9 8 9 ) . Moreover, as Knight points out, " to the extent that such rules can have substantive effects on social outcomes, the substantive content of those rules should reflect the self-interest that motivates these claims and actions " ( 1 9 9 2, 3 8 ) . Institutions thus acquire stability when groups with the resources to alter the rules of the game accept those rules. Clearly, institutional change alters the incentives that constrain politi­ cal actors, and it thereby modifies the behavior of those actors, even allowing for some stickiness that may slow the pace of the modification. We expect institutional change to occur when actors or groups with suf­ ficient power are able to challenge and transform those rules. Whereas the cultural argument essentially takes incentive structures as given, the institutional perspective directs our attention to the strategic choices faced by key political figures (e.g., Shepsle 1 9 89; Knight 1 9 9 2 ; Geddes 1994; Strom, Budge, and Laver 1994 ) and restores political consider­ ations to a central analytic role.22 Given uncertainty and imperfect infor­ mation, of course, institutional change does not guarantee the outcomes desired by those who initiate it ( Shepsle 1 9 89 , 1 3 8 - 4 3 ) . This fact serves as the basis for ongoing attempts to adj ust and reform rules and proce­ dures by those actors with the resources to do so. The basic point re­ mains: whether stable or evolving, institutions condition the distribution of both political and social resources. An Illustration: The Congress Party of India We digress briefly to consider the Congress Party of India. The case is of obvious intrinsic interest, given India's status as the most populous ongo­ ing democracy. Of equal importance, it illustrates with great clarity the issues at the heart of an institutional explanation. Founded in I 8 8 5, the Congress Party was the dominant force in Indian politics at the federal level from independence in 194 7 until 1 9 9 6. Indeed, 22. We prefer the institutional label for this account given the way in which it under­ scores the politics underlying the process. On occasion, others have adopted a different label for basically the same perspective. For example, Mishler and Pollack ( 2003 ) cast the distinc­ tion we have made between cultural and institutional approaches as one between "thick" and "thin" cultural approaches. While their thin culture approach also casts behavior and values as endogenous, in our view the label itself downplays the political manner in which institutions affect the distribution resources across groups.

24

B E F O RE N O RM S

with two exceptions ( 1 9 77-90 and 1 9 89 - 9 1 ) , i t held power continu­ ously from independence until 1 9 9 6. Further, the Indian case is notewor­ thy as the maj or democracy among the developing states. Except for the suspension of constitutional procedures during the Emergency of 1 9 7 5 7 7 , the outcomes o f general elections with full suffrage have been bind­ ing. In what is perhaps the key test, incumbents have actually transferred power to their opponents after electoral losses. Several factors have undoubtedly contributed to the ongoing strength of the Congress Party. These include its organizational age, its close identi­ fication with the struggle for decolonization, and the policy positions it has adopted since independence. Another factor, however, involves elec­ toral laws. National legislative elections in India are governed by a single­ member district (SMD ) , plurality, winner-take-all formula. As is well known, this is a format for translating votes into legislative seats that can manufacture majorities by amplifying the margins of large parties and penalizing minor parties (e.g., Duverger 1 9 6 3 ; Rae 1 9 7 1 a ; Riker 1 9 8 2; Taagepera and Shugart 1 9 89; Lijphart 1994; Cox 1 9 9 7 ) . It is thus an institutional mechanism that one might expect to serve well the interests of the Congress Party, as indeed it has.

TABLE 1 . 1 . Votes and Legislative Seats Received by the Congress Party of India in National Elections since I ndependence Year

Votes

Seats

Seats!Votes

1 952 1 95 7 1962 1 96 7 1 97 1 1 977" 1 980 1 9 84 1 989" 1991 1 996" 1 99 8 " 1 999"

45.0 47.8 44.7 40.8 43.7 34.5 42.7 48.1 39.5 36.4 28.8 25.8 28.3

74.4 75 . 1 73 . 1 54.4 6 7. 9 2 8 .4 66.7 76.6 3 7.4 44.5 25.8 26.0 2 1 .0

1 . 65 1.57 1 . 64 1.33 1 .5 5 .82 1 .5 6 1 .5 9 .95 1 .22 .89 1.01 . 74

Source: Data for the elections from 1 952 through 1 9 9 1 are from Brass 1 9 94 (tables 3 . 1 and 3 . 2 ) . Figures for 1 9 9 6 , 1 9 9 8 , and 1 999 are from the Statistical Reports on these elections issued by the Electoral Commission of India, and available from the Commission at http://www.eci .gov.in. "These five elections led to the formation of non-Congress con­ trolled govemments.

The Issues

25

Until the general election of 1 9 77, India was routinely described as having a virtual one-party or a dominant-party system, the party con­ cerned being the Congress. This perception of invincibility had many sources, not least of which is the fact that with the exception of the 1 9 67 election, when it received 54 percent of the legislative seats, the legislative majority of the Congress Party easily exceeded two-thirds until 1 9 77 (table r . r ) . Yet it is striking that the Congress Party has never received a majority of votes cast. Even in the first five elections listed in table r . r , the Congress Party vote averaged only 44· 4 percent, a plurality, t o be sure, but well short of 50 percent. Yet, as the figures show, the SMD electoral system translated these votes into seats most impressively for the Congress Party in these first five elections to yield an average legislative majority of 69 percent (and a mean seats/votes ratio of r . 5 4 ) . The effectiveness o f the SMD system i n manufacturing decisive legisla­ tive majorities from electoral pluralities is underscored when we examine more systematically all thirteen of the elections listed in table r . r. Specifi­ cally, an analysis of the votes-seats relationship following the procedure discussed by Simon Jackman ( r 9 9 4 ) points to considerable bias. The relation between the Congress Party's votes share and the proportion of seats won in national elections since 1 9 5 2 is graphed in figure r . r . Were seats allocated proportionally to votes, the data would follow the 4 5 degree line that w e have drawn. However, the graph indicates instead that the Congress Party requires only 3 8 percent of the vote to garner a majority of legislative seats. Further, the proportion of seats won steps up dramatically as the votes share increases beyond 3 8 percent. Indeed, were the Congress Party ever actually to receive 50 percent of the votes in a national election, the patterns reflected in table r . r imply that it would receive a magnificent 79 percent of the seats ! 23 The common portrait of 2 3 . The votes-seats relation is estimated as ln(s/[1 - s] ) = In f3 + p ln(u/[1 - u] ) +

E,

(1)

where s i s the proportion o f seats received b y a given party, u i s the proportion o f votes received by that party, and {3, p, and E are parameters to be estimated ( see S . Jackman 1 9 9 4 , equation [4] ) . Note that In f3 measures the b i a s of the system: a zero value for this term implies that 50 percent of the votes generate 50 percent of the seats, while a positive value reflects an advantage to the party whose votes share is analyzed. The least-squares estimates of ( r ) using the Indian data in table I . I are: In f3 = 1 . 3 27 (s.e. = . T 6) ; p = 2.722 (s.e. = . 2 9 ) ; and the R� is . 8 9 (n = T 3 ) . The positive estimate for In f3 reveals a considerahle pro- Congress Party electoral system hi as, while the estimate for p is statistically indistinguishable from 3 , the value implied by the so-called cube law for two­ party contests, according to which "the proportion of seats won by the victorious party varies as the cube of the proportion of votes cast for that party over the country as a whole " (Kendall and Stuart l9 s o, T 8 } ) . As shown by S. Jackman I T 9 9 4 . equations r 51 and r 6] ) , the

26

B E F O RE N O RM S

1 .0

.8 ••

(/)

,

Cii Q) (J) 0

.6

�0..

.4

•

c 0

•

0

a:

•

•

•

.2

0.0 0.0

.4

.2

.6

.8

1 .0

P roportion of Votes Fig. 1 . 1 .

I ndian elections, 1 952-99

India (especially from 1 9 4 7 through 1 9 7 7 ) as having a dominant-party system clearly stems more from the skills of the Congress Party at win­ ning legislative representation than from its vote-getting prowess. Clearly, with a different electoral formula (involving, say, some form of propor­ tional representation), India would never have been described as a one- or dominant-party system. One might surmise that the adoption of the SMD electoral formula after independence and the fact that it has benefited the Congress Party is simple coincidence. After all, this is the same system as that employed in the United Kingdom, and there are other instances of diffusion within empires ( see, especially, Strang 1 9 9 0 ) . Perhaps, then, the adoption in India of the British SMD formula should occasion no surprise, and some have indeed taken it as an instance of cultural diffusion (e.g., Blais and Massicotte 1997). Unfortunately, the evidence belies any such simple pattern o f diffusion. While many institutional arrangements were borrowed from the British, preceding expressions are readily rearranged to estimate ( r) the seats share Congress ob­ tains with 5 0 percent of the vote and (2) the votes share required to generate 5 0 percent of the seats.

The Issues

27

India in fact had considerable experience (dating back to I909 ) with various forms of proportional representation, despite long-standing resis­ tance from the Congress Party. This experience was a response to commu­ nal differences in preindependence (and prepartition) India, a setting in which the Congress Party came increasingly to compete with the Muslim League. Up until independence, the Congress Party argued strongly for a cen­ tralized, unitary state built around majoritarian principles along British lines (e.g., Coupland 1 944 ) . Many officials in Britain, however, thought such an electoral system would be catastrophic for the large Muslim minority. As Weiner notes ( 1 9 89 , 1 8 2 ) , their concern stemmed less from any sense of altruism than from an interest in maximizing Muslim sup­ port for the British Raj . To obviate these concerns, they proposed various forms of proportional representation that included separate communal electorates. British officials were endorsed in this endeavor by Muslim interests, who feared a " Congress Raj " and therefore argued among other things that Muslim representation ought to be " commensurate not merely with their numerical strength, but also with their political impor­

tance and the value of the contribution which they make to the defence of the Empire. " 24 Indeed, by the 1 940s, given communal differences in In­ dia, serious proposals were advanced that India should adopt a highly decentralized constitutional and electoral structure, along the lines of the Swiss model ( Coupland 1944, 3 : 67- 7 2 ) . I t was against this backdrop that the SMD system was formally adopted by " India " through the mechanism of a constituent assembly dominated by the Congress Party. The final adoption of SMD, we under­ score, came after independence and the partition. " In rejecting some form of proportional representation as the electoral process-even though In­ dia already had a multiparty system-the Congress-dominated leader­ ship once again acted to maximize its own political fortunes, as we would expect any rational set of political leaders to do" (Park and Bueno de Mesquita 1 9 79 , 1 3 8 ) .25 One might be forgiven for concluding that a Congress Raj ensued. 24- This is from a statement by a Muslim deputation to Lord Minto in October 1906, quoted by Coupland ( 1 944, 1 : 3 4 ; emphasis added ) . 2 5 . For additional discussions o f these developments, see Coupland ( 1 944 ) , Weiner ( T 9 5 7, esp. chap. T T ), Austin ( T 9 66, esp. chaps. T, 6), Gautam ( T 9 8 T ), Morris-Jones ( T 9 8 8 ) , Weiner ( 1 9 89, esp. chap. 7 ) , and Brass ( 1 994, esp. Part 1 ) . We have, o f course, n o direct evidence that members of the Congress Party were aware of something like the cube law, and Kendall and Stuart's essay was not published until 1 9 5 0 . On the other hand, there is no reason to believe that key actors in the Congress Party were naive on the general point. As Kendall and Stuart observe, a variant of the cube law had been discussed much earlier by

28

B E F O RE N O RM S

The adoption o f the SMD formula after independence and the benefits it has generated for the Congress Party are thus far from a simple coinci­ dence. Instead, they underscore the point that institutions are adapted and modified by actors with the necessary resources in an effort to pre­ serve and advance their interests.26 Such behavior does not guarantee political supremacy in perpetuity, of course, and the Congress Party has not won every election in the period (it lost five of the thirteen listed in table 1 . 1 ) . By the same token, there can be little doubt that the electoral formula in whose adoption the Congress Party was so instrumental has been a key ingredient in the party's preeminence until recently in Indian politics. Finally, the paucity of effective challenges to the Indian electoral for­ mula is readily explained. The Congress Party has lost five general elec­ tions since independence (those in 1 9 77, 1 9 89 , 1 9 9 6, 1 9 9 8 , and 1 9 9 9 ) . On these occasions, the other parties o r coalitions that have formed governments have themselves thereby profited immensely from the for­ mula. Thus, when the Janata Party received an electoral plurality to defeat the Congress Party government in 1 9 77, its 4 1 . 3 percent of the votes yielded 5 4 · 5 percent of the legislative seats for a seats/votes ratio of 1 . 3 2. In the elections of 1 9 9 6, 1 9 9 8 , and 1999, the Bharatiya Janata Party ( BJP) won pluralities of seats . While these were smaller than some of the majorities achieved by the Congress Party in its heyday, the BJP has still received about one-third of the seats in the Lok Sabha. Further, these pluralities have been generated by seats/votes ratios well in excess of 1 .0 ( 1 .4 6 in 1 99 6, 1 . 3 0 in 1 9 9 8 , and 1 .40 i n 1 999 ) . Thus, given its votes share, the SMD system has appreciably amplified the legislative presence of the BJP in recent years. One would hardly expect a party forming a new government to tinker with or dismantle the rules which enabled it to form that government in the first place. Indeed, it is difficult to conceive of an electoral formula that might have generated more seats for the BJP in recent years. Even with a weakened Congress Party, then, the key political actors in India have no incentive to reform the existing SMD electoral system. Edgeworth ( 1 8 9 8 ) . Edgeworth's analysis, i n turn, was addressed and applied i n a Royal Commission on Systems of Election published by the British House of Commons in 1 9 1 0 . 26. The Congress Party has evinced a n ongoing interest in electoral engineering. For example, under Article 3 2 7 of the Constitution, itself drawn hy the Congress Party, Parlia­ ment (by simple majority) is charged with drawing and modifying constituency boundaries. Further, there is evidence to suggest that, absent an active gerrymander after the 1 9 6 2 election, the Congress Party would n o t have achieved its relatively narrow victory (just 5 4 percent of the seats ) in the r 9 67 general election (Bueno de Mesquita r 9 7 8 ) .

The Issues

29

A Brief Com parison of the Cu ltu ral an d I n s titutional Approaches

Our analysis centers on the contrast between a cultural interpretation and one that casts political behavior as optimizing within institutional constraints. We recognize that cultural arguments have occasionally been pitted either against dependency theory (e.g., Portes 1 9 7 6; Harri­ son 1 9 9 2 ) or against some form of historical materialism (e.g., Harrison 1 9 9 2; Street 1 994; and, more intermittently, Inglehart 1 990; Putnam 1 99 3 ) . The most commonly adopted contrast, however, is the one we employ between cultural and rational choice accounts (e.g., Almond 1 990; Pye 1 9 9 1 a ; Green and Shapiro 1994 ) . The comparison is useful on several counts . First, it helps explain the recent revival and current popularity of the cultural account. Many apparently find that argument congenial because, among other things, it contains a large dose of Weberian idealism and so is seen as an affirmation of the importance of " ideas. " The classic ex­ ample of this claim is, of course, the Weberian thesis linking Protestant (especially Calvinist) theology to the capitalist spirit, a thesis we examine more closely in chapter 2. This idealism is also often associated with populist sentiments, so that cultural accounts are additionally said to assist in " bringing the people back in" ( see lnglehart 1997, chap. 6 ) . The presumed congeniality of the cultural account is only enhanced when the ideas with which it is identified include norms of trust, civic-mindedness, and cooperation, all containing a high dose of altruism. When contrasted with a view of rational choice that casts actors as concerned solely with the cold and selfish calculation of a narrowly defined economic interest, the cultural view becomes irresistible. But this is all predicated on an excessively narrow view of rational choice. Despite common beliefs to the contrary, the cultural argument does not have sole claim on the proposition that ideas matter. For ex­ ample, the analysis of policy preferences is central to many rational choice accounts, and these preferences are never treated as uniformly insincere. Similarly, the cultural argument has no monopoly on the analy­ sis of altruistic behavior. Altruism is not incompatible with rationality, as noted earlier, unless analysts inappropriately choose to attribute a utility function to others, one that is arbitrarily restricted to the selfish maximi­ zation of economic utility. When rational behavior is more reasonably cast as optimizing under the conditions we have already examined, it is immediately apparent that neither the evaluation of preferences nor the analysis of altruism is the exclusive prerogative of the cultural account.

30

B E F O RE N O RM S

Second, the contrast between the cultural and optimizing accounts em­ bodies the distinction between behavioral and institutional approaches to political analysis. We have already observed that The Civic Culture (Al­ mond and Verba 1 9 63 ) is generally taken as epitomizing the behavioral mode. More recent studies of political culture track the behavioral ap­ proach rather closely with their presumption that preferences reflexively generate outcomes in an almost context-free manner. As Calvert puts it, to the extent that they address institutional issues at all, " practitioners [in the behavioral model assumed to various extents that institutions are defined by the regular behavior patterns of individuals, and not vice versa . Behav­ ioralism seems to view institutions not as constraints at all, but rather as the aggregated result of individual psychological propensities" ( T 9 9 5 , 2 2 1 ) .The institutional alternative obviously takes a completely different tack. Third, comparing the cultural and institutional accounts is instructive because the former has much more difficulty accounting for change, as we have already emphasized. If political cultures reflect long-standing clusters of norms, then observed change is difficult to explain without invoking a deus ex machina. Thus, analyses of racial politics in the Ameri­ can South cast in terms of enduring Southern values up until the 1 9 60s are less helpful in advancing our understanding of the impact of voting rights legislation in ensuing years (e.g., Davidson and Grofman 1 994 ) . Similarly, since a t least d e Tocqueville, America has been routinely de­ scribed as a nation of joiners. The claim has recently been made, however, that in the past two decades social participation has declined precipi­ tously, the culprit allegedly being mass overexposure to commercial tele­ vision (Putnam 1 9 9 s b, 2ooo) . We think that cultural explanations of changes that occur in the face of allegedly durable values are necessarily contrived. The institutional approach is not similarly limited. Institutional arrange­ ments reflect the outcomes of conflict among key political actors and are thus more directly explained. Because changes in incentive structures lead to changes in behavior, the argument is not subject to the conservative implications that encumber the cultural argument. The change is not al­ ways instantaneous, of course, but it does become amenable to coherent analysis. Our point is simple. The cultural and institutional approaches comprise distinct explanations. Further, the distinctions at stake reflect real issues, not arcane and pedantic fine points. We believe that the institu­ tional account easily outperforms the cultural interpretation. This book is intended to make that case.

PART ONE

C u ltu ral

AND

Social

Capital App roac h es

TWO

The Protestant Ethic Thesis Virtually all the modern world has been read into Calvinism; liberal politics and voluntary association; capitalism and the social discipline upon which it rests; bureaucracy with its systematic procedures and its putatively diligent and devoted officials; and finally all the routine forms of repression, joylessness, and unrelaxed aspiration. By one or another writer, the faith of the brethren, and especially of the Puritan brethren, has been made the source or cause or first embodiment of the most crucial elements of modernity. Michael Walzer ( 1 9 6 5 , 3 00 )

ith their shared emphasis on the key causal role of social and politi­ cal values, analyses couched in terms of political culture have a common origin in Weber's book The Protestant Ethic and the Spirit of Capitalism ( [ 1 9 0 5 ] 1 9 5 8 a ) . Sometimes, the link to Weber is explicit and direct. Harrison ( 1 9 9 2 ) and Fukuyama ( 1 99 5 , chap. 5 ), for example, spe­ cifically ground their analyses in Weber. On other occasions, the linkage is indirect. Thus, while Putnam ( 1 99 3 ) does not reference Weber, he does draw heavily on Banfield's ( 1 9 5 8 ) analysis of amoral familism in southern Italy. Still other scholars link the argument to Weber directly and indirectly. Accordingly, Granato, lnglehart, and Leblang ( 1 9 9 6 ) emphasize both Weber and the subsequent exegesis associated with McClelland's study of need for achievement ( 1 9 6 1 ) . But Banfield and McClelland themselves drew heavily on Weber, and their arguments are direct extensions of his. Indeed, as Hirschman put it, " in the 1 9 5 0s, . . . Weber's Protestant Ethic [was] modernized into David McClelland's 'achievement motivation' as a precondition of progress and into Edward C. Banfield's 'amoral familism'

33

34

B E F O RE N O RM S

a s an obstacle " ( 1 9 84 , 9 9 ) . The longevity of the Weber thesis i s thus remarkable. While our use of the term Weber thesis is routine, it is important to understand that he did not initiate the idea that confessional differ­ ences are associated with progress. Far from being original, the core of Weber's argument had been common in the Protestant German intellec­ tual circles of which he was a part since the late eighteenth century, with their confessional stereotypes of Protestants as active and productive and of Catholics as passive and unproductive (Munch 199 3 ). By the time of The Protestant Ethic, the linkage between religion and economics was commonplace (Nipperdy 1 99 3 ) . And the argument was hardly confined to Germany. Consider the views of the eminent English Whig historian Thomas Macaulay. The loveliest and most fertile provinces of Europe have, under her [the Church of Rome's] rule, been sunk in poverty, in political servitude, and in intellectual torpor, while Protestant countries, once proverbial for sterility and barbarism, have been turned by skill and industry into gardens, and can boast of a long list of heroes and statesmen, philoso­ phers and poets . ( 1 849, I : 3 0- 3 I )

Nor were these simply national differences . Whoever passes in Germany from a Roman Catholic to a Protestant principality, in Switzerland from a Roman Catholic to a Protestant canton, in Ireland from a Roman Catholic to a Protestant county, finds that he has passed from a lower to a higher grade of civilisation. On the other side of the Atlantic, the same law prevails. The Protestants of the United States have left far behind them the Roman Catholics of Mex­ ico, Peru, and Brazil. The Roman Catholics of Lower Canada remain inert, while the whole continent round them is in a ferment with Protes­ tant activity and enterprise. (Macaulay I 849, I: 3 I )

Similarly, in the United States, John Draper ( I 8 74 ) had already advanced the influential thesis that the Reformation gave rise to modern rational science (and thus to technological advances and economic growth) in the face of strong resistance from Catholic Europe. Acknowledging that the general argument has a long pedigree, our purpose in this chapter is to evaluate Weber's thesis linking Protestantism with capitalism in light of ongoing historical scholarship. Such an evalua-

The Protestant Ethic Thesis

35

tion is crucial given the current revival of cultural explanations of politi­ cal life that stem from Weber's thesis and that thereby view his thesis as a fundamentally accurate account of the phenomenon he sought to explain. The evaluation is also critical because Weber's thesis has thoroughly permeated contemporary social science. The term Protestant work ethic has long since become part of common usage, a label used even for psychological measures whose meaning is thereby taken to be self-evident ( see, e.g., Furnham 1 9 9 0 ) . Some have reported a significant Protestant ethic in such unlikely quarters as contemporary Egyptian Sunni radical­ ism ( Goldberg 1 99 1 ) . Others have suggested that Protestant ideas were decisive in modern state formation ( Gorski 1 99 3 , 1999; Philpott 2000 ) . Still others have declared that patterns of humor (e.g., jokes about the canny Scots) constitute prima facie evidence of an affinity between the Protestant ethic and the spirit of capitalism (Davies 1 9 9 2 ) ! " This chapter begins with two preliminary issues that have often im­ peded discussions of the problem: the ambiguity that surrounds much of Weber's argument and the proclivity of many commentators to accept the argument as an article of faith, regardless of the evidence. We then lay out the key elements of Weber's thesis, a task made necessary by the fact that they are repeatedly misconstrued. In evaluating the argument we address two broad questions. First, and most fundamentally, we examine patterns of economic growth within Europe since the early Middle Ages in order to evaluate whether there was a significant shift in economic growth and activity after the Reformation, consistent with Weber's thesis. Second, we ask if Calvinist teaching was sufficiently distinctive to be linked with particular outcomes that might be linked to the rise of capitalism. We conclude that the historical record reveals no qualitative shift in eco­ nomic activity in post-Reformation Europe. Further, Protestantism gener­ ally ( and Calvinism in particular) did not constitute a distinctive outlook in the manner suggested by Weber. Recent claims that directly or indi­ rectly draw on Weber's thesis are thus predicated on an argument whose empirical foundations are precarious. r. The term routinely appears in press accounts. For example, the highest court of appeal in Italy has ruled that influence peddling is not in itself a crime, although overstating one's power to exert influence remains punishable. According to Franco Ferrarotti, an Italian sociologist, " This is our version of the Protestant ethic. When a favor works success­ fully, it ceases to be a crime and becomes a work of art" (New York Times, April 20, 200 1 ) . And reporting on a study on the link between siestas and productivity, the Economist concluded that, humans being crepuscular, "the Protestant work ethic that drives those now living in colder climes to work throughout the day may actually be counterproductive. At least, that is what you should tell your boss when asking for a couch to be installed in the office" ( " Siesta Time," June 1, 2 0 0 2 , 7 6 ) .

36

B E F O RE N O RM S

Two Pre li m in ary I s s ues

So much has been written about Weber's thesis that another sortie may seem presumptuous. The issue is compounded by two interrelated prob­ lems. First, we must acknowledge that Weber's treatise is often ambigu­ ous. Second, much of the commentary on Weber reveals an unfortunate tendency toward sectarianism. These two points need to be addressed before the thesis itself can be evaluated, and we consider each in turn. Weber's work is, in broad terms, a debate with Marxism generally, and with historical materialism in particular. The Protestant Ethic comprises one important element of that debate ( see, e.g., Braude! 1 9 8 2, 4 0 2; Zeit­ lin T990) that offers a nonmaterialist interpretation of the transition from feudalism to capitalism, a transition central to Marxism. Weber and Marx both accepted the proposition that a discrete transition to capital­ ism did occur, but they differed radically over its timing and origins. In an equally general sense, the contrast in styles between Weberian and Marx­ ist arguments is another ingredient of the debate between them. Where historical materialism is typified by elegance and parsimony, Weber stresses complexity, contingency, and qualification, arguing that Marx­ ism is misleading in part to the extent that it oversimplifies excessively. While this penchant for qualifying details characterizes all of Weber, his " notorious eccentricity on thematic development" (MacKinnon 1 9 8 8 , 1 4 5 ) is nowhere more conspicuous than i n The Protestant Ethic. Even some of Weber's defenders acknowledge the problem. Thus, writing of The Protestant Ethic, Poggi contends: Its central argument is, I believe, clear and forceful. But Weber sur­ rounded it with such a wealth of secondary arguments, qualifications, digression, he developed it with-let us say it outright-such a show of diverse, profound, and sometimes highly esoteric learning, that its essen­ tial components and their relations are not easy to discern. ( 1 9 8 3 , x; emphasis in original)

Leaving to one side the disj uncture between these two sentences, Poggi's concession is surely correct: the text contains many ambiguities. Indeed, a close reading of The Protestant Ethic might convince some that there is no distinctive Weberian thesis. Every key point is provisional, subject to considerable qualification and elaboration. Indeed, it is repeat­ edly suggested that Weber intended no causal argument at all (see, e.g., Kalberg 1994, chap. 2). All of this means that every statement has an escape clause, which naturally leaves much room for textual interpretation

The Protestant Ethic Thesis

37

and may even help account for the volume of the secondary literature. However, as we have already made clear, the Weber thesis is simply the best-known exemplar of a long-standing and fairly clear argument linking religious values to the transition to capitalism. Further, the continuing attention accorded The Protestant Ethic stems from the basic theme that most contemporary social scientists attribute to it, namely, that value sys­ tems are fundamentally consequential. Absent such a theme, the text would certainly not continue to be invoked so often as the rationale for cultural accounts of political and social change, and would instead have long since been forgotten. Second, a palpable sectarianism permeates many discussions of the argument. For example, in an insightful survey, Cohen ( T 9 8 o) showed that, Weber to the contrary, there is much evidence that rational capitalism conceived in Weberian terms developed extensively in pre-Reformation Italy. Cohen's paper was subsequently attacked by Holton for perpetuat­ ing " the myth of Weber's supposedly monocausal 'Protestant ethic' expla­ nation of the development of modern capitalism" ( r 9 8 3 , r 6 6 ) . Holton further claimed that Cohen ignored " an important footnote within the Protestant Ethic, [in which] Weber rebutted the proposition 'that capital­ ism as an economic system is a creation of the Reformation' with the rather cryptic counterproposition that 'certain forms of capitalist business organi­ zation are known to be considerably older than the Reformation"' ( r 9 8 3 , r 6 8 - 69 ) . 2 In other words, Holton's 1 9 8 3 defense o f Weber rests on one of the many qualifying clauses in the original text. Having thus dealt with the heretic, the same Holton was able to conclude j ust two years later (without reference to Cohen 1 9 80 ) that " it remains striking how few of the main planks of [Weber's l substantive historical analyses have stood up well to empirical testing" ( r 9 8 5 , 1 4 1 ) . In another commentary centering on MacKinnon's ( r 9 8 8 ) analysis of Calvinism, Zaret asserted that MacKinnon's " errors" should " be under­ stood in terms of a bad sociology of knowledge implicit in his analysis and not necessarily as a reflection of his evident animus toward Weber" (Zaret 1 9 9 2, 3 69 ; emphasis added) . Thus accused of apostasy, MacKinnon was contrite. On Zaret's charge that some strange animus moves MacKinnon's cri­ tique of Weber's thesis-this is false. Indeed, on other fronts my preoc­ cupation with Weber's grand themes has led to the opposite charge, 2. This qualification appears in the body of the text and is thus not literally a footnote. It is, nonetheless, an aside (see Weber 1 9 5 8a, 9 1 ) .

38

B E F O RE N O RM S

that I am his acolyte. If anything, the second is closer to the truth. ( 1 9 94, 5 9 4 )

I t is difficult t o s e e what analytic insights are advanced by the exchange of such charges and confessions. In the next section we outline the essential points of Weber's thesis. This outline necessarily discounts the many escape clauses and empha­ sizes instead the themes that have engaged and molded subsequent schol­ arship, the themes that underlie the continuing attention to Weber's argu­ ment. Further, we treat his thesis not as an article of faith but as an empirical claim subject to disconfirmation, the validity of which hinges on the available historical evidence. We thus concur with Cohen that " a historical thesis like Weber's needs t o be tested through historical evi­ dence if it is to be tested at all. Otherwise, it is not a scientific hypothesis but a fairy tale about a faraway ancient land that we can never know" ( 1 9 8 3 , 1 8 4- 8 5 ) . 3

The We be r Thesis

Weber's most general purpose was to show that ideas affect history, and this purpose accounts for the ongoing attention to his work. In other writings, of course, he also addressed the impact of material and institu­ tional factors on historical change. Nonetheless, it is his " idealism" that most clearly distinguishes his approach from that of Marx. The Protes­ tant Ethic exemplifies his particular interest in religious beliefs, in this case those associated with Calvin and Calvinism as they unfolded from the mid-sixteenth through the seventeenth centuries. Calvin's reformed Protestantism departed from Luther in many re­ spects, ranging from the meaning of the sacraments to Calvin's endorse­ ment of theocracy. Weber argued ( 1 9 5 8 a, 9 8 ), however, that the most distinctive element of Calvin's theology was the doctrine of predestina­ tion ( see also White 1 9 9 2 ) , according to which salvation cannot come from the worldly efforts of the faithful but only from God, whose j udg­ ment can never be known. The elect are predestined to salvation by God alone; others are condemned to everlasting death. 3· However, we arc less persuaded by Cohen's more recent ( 2002) suggestion that historical evidence restricted to the English Puritans, and more specifically to the diaries of two particular English Puritan merchants, provide sufficient leverage to evaluate the broad historical validity of the Weber thesis.

The Protestant Ethic Thesis

39

For Weber, the key analytical problem posed by the dogma of predesti­ nation centered on reconciling that dogma with personal expectations: How was this doctrine borne in an age to which the after-life was not only more important, but in many ways also more certain, than all the interests of life in this world ? The question, Am I one of the elect ? must sooner or later have arisen for every believer and have forced all other interests into the background. And how can I be sure of this state of grace ? ( I 9 5 8a, I 09-I O )

While Calvin may have been certain of his own salvation, such certainty was much less assured for those who followed. The logical response to predestination would be a fatalistic and hedonis­ tic live-for-the-present perspective, but this hardly serves as a compelling mechanism to link Calvinism to entrepreneurship. Recognizing this, We­ ber distinguished the logical from the psychological consequences of reli­ gious beliefs ( I 9 5 8a, 2 3 2 ) . He then argued that the distinction was evident in the difference between Calvin's dogma, which he saw as persisting well beyond the Westminster Synod of I 64 7, and Calvinism, as reflected in pastoral teachings. While Calvin's dogma (or decretum horribile) consti­ tuted a logical structure, the Calvinism of pastoral teachings was more sensitive to the psychological and spiritual needs of lay Calvinists who would otherwise have been confronted with the crisis of proof of personal salvation defined in the dogma. By differentiating official dogma from pastoral teachings in this way, Weber recognized the contrast between those who produce ideas and those who popularize or consume them ( Oakes 1 99 3 ) . In their pastoral teachings, Weber claimed that Calvinists stressed two related themes. First, an absolute duty to consider oneself chosen was emphasized, and doubts were cast as the work of the devil, " since lack of self-confidence is the result of insufficient faith, hence of imperfect grace " ( 1 9 5 S a, 1 1 1 ) . Second, an intense, sustained, and coherent pattern of worldly activity was stipulated as the optimal path to that self-confidence. [The Calvinist] could not hope to atone for hours of weakness or of thoughtlessness by increased good will at other times, as the Catholic or even the Lutheran could. The God of Calvinism demanded of his believers not single good works, but a life of good works combined into a unified system. There was no place for the very human Catholic cycle of sin, repentance, atonement, release, followed by renewed sin. Nor was there any balance of merit for a life as a whole which could be

40

B E F O RE N O RM S

adjusted by temporal punishments or the Churches' means o f grace. ( 1 9 5 8a, l l 7 )

Weber thus saw a n irrational disjuncture between dogma and pastoral teachings, that is, between Calvin and Calvinism.4 Calvinism served unin­ tentionally to subvert and thereby distinguish itself from Calvin's dogma by sanctifying works as the means of glorifying God. By reducing uncer­ tainty about personal salvation in this manner, Calvinism provided a unique spiritual sanction for labor in this world. Although he recognized that the usefulness of works is gauged "primarily in moral terms," Weber insisted: In practice the most important criterion [for a calling] is found in private profitableness. For if that God, whose hand the Puritan sees in all the occurrences of life, shows one of His elect a chance of profit, He must do it with a purpose. Hence the faithful Christian must follow the call by taking advantage of the opportunity. ( 1 9 5 8 a, 1 62 )

The asceticism and frugality epitomized b y such a value structure were precisely Weber's requisites for successful capitalist entrepreneurs. Indeed, this set of values provided the psychological mechanism that uniquely defined the " spirit" of capitalism. Weber's classic entrepreneur is thus one who "gets nothing out of his wealth for himself, except the irrational sense of having done his job well " ( 19 5 8 a, 71 ) . 5 Thus it was that he proposed a connection between a Protestant ethic inherent to Calvinism and Puritan­ ism (and antithetical to Lutheranism or Catholicism) and the spirit of capitalism. We have to this point outlined the key elements of the Weber thesis as articulated by Weber. Some of these elements may, however, be unfamil­ iar to those who think of the argument as linking Protestantism and economic activity in a different and more general way that does not depend so decisively on adjustments to the doctrine of predestination. Indeed, j ust as Weber saw a crucial distinction between Calvin's theology and the pastoral teachings of Calvinism, we can also distinguish " pasto4· Weber emphasized the irrational nature of this disj uncture, noting that " we can find a parallel today only in the at bottom equ ally superstitious belief of the modern proletariat in what can he accomplished and proved by science " ( I 9 5 8a, 2 3 3 ) . 5 . Observe the parallels between Weber's description o f entrepreneurial values and McClelland's subsequent discussion of need for achievement, defined as "a desire to do well, not so much for the sake of social recognition or prestige, but to attain an inner feeling of personal accomplishment" ( T 9 6 3 , 7 6 ) .

The Protestant Ethic Thesis

4r

ral " versions of Weber's thesis from the original. Moreover, the pastoral transcriptions are largely responsible both for the ongoing vitality of the thesis exemplified, for example, in the studies noted earlier in this chapter as well as in more everyday understandings of the " Protestant work ethic. " 6 One o f the better-known instances o f pastoral Weberianism appears in McClelland's analysis of need for achievement, which associates self­ reliance with Protestantism generally.? For McClelland, Protestantism meant continued self-improvement and more. Specifically, in both its Lutheran and Calvinist forms, it represented a revolt against the institu­ tional church, a revolt in which individuals were encouraged to find divine guidance directly from the Bible without exclusive reliance on the interpretations of learned experts. McClelland further saw the Protestant abolition of the celibate priesthood as a decisive social change. The Protestant pastor could now give concrete examples of child­ rearing practices that might be emulated by his parishioners in a way that was formerly impossible under the celibate priesthood. The social mechanism was provided by which the new religious world-view could specifically affect socialization and thereby the motivation of the new generation. ( 1 9 6 1 , 5 0 )

The modification represented by McClelland's argument is clear. Protes­ tantism remains the decisive stimulus for the emergence of capitalism, but the essential distinction is now between all Protestants and Catholics, not between Calvinists on the one hand and Lutherans and Catholics on the other. An earlier version of pastoral Weberianism is associated with Tawney ( 1 9 26), and the Weber thesis is indeed described on occasion as the Weber-Tawney thesis. However, Tawney's argument is distinctive in yet another way. Specifically, Tawney argued (contrary to Weber) that reli­ gious activity was inherently inimical to economic growth. Tawney's explanation of economic growth after the Reformation thus empha­ sized the erosion of religious authority generally, not the emergence of doctrinal modifications associated with Calvinism or any other particular theology. Like Weber, however, Tawney did regard the challenges to 6. Pastoral Weberianism closely parallels what another recent study has labeled the " Common Interpretation" of Weber's thesis (Dclacroix and Nielsen 2oo r ) , as will become clear later. 7· According to Weiner ( r 9 66, 5 ) , McClelland described Protestantism as the " mental virus" that made modernization possible.

42

B E F O RE N O RM S

established religion associated with the Reformation a s the turning point in the rise of capitalism . � More recently, Schneider ( 1 990) h a s advanced a parallel argument that emphasizes the importance of those religious reforms that reached a cre­ scendo on "the eve of capitalism in the sixteenth and seventeenth centu­ ries " with the Reformation itself ( 1 990, 2 6 ) . In her view, salvationist Christian reformers progressively undermined and demonized European peasant animism. Where animist worldviews emphasized economic and social equality and an ideal of material reciprocity, the Christian reform­ ers stressed brotherly love and a trust in providence placing less weight on personal responsibility in the process. The reformers thus generated a new ethical system within which individuals were free to act more self­ ishly and rationally, as beliefs in earth spirits and ghosts faded. In evaluating Weber's thesis in either its original or pastoral versions, there are two broad questions to consider. First, and most fundamentally, did patterns of economic activity change dramatically after the Reforma­ tion ? This is a broad issue on which hinges the validity of Weber's argu­ ments and all of its variants. Second, and more narrowly, did Protestantism generally (and Calvinism in particular) give rise to a new worldview that attached great weight to achievement norms ? This issue is obviously at the heart of Weber's argument and those variants ( such as McClelland' s ) that assign special significance to Calvinism or Protestantism. We take up these questions in turn.

Was There an Econom i c Bre akth roug h afte r the Reform ation ?

Like Marx, Weber sought to explain a transition from feudalism to capi­ talism. Both agreed that there had been a discrete and identifiable transi­ tion, before which capitalism did not exist and after which it did. While Marx defined the transition in terms of materially based crises, Weber cast it in nonmaterial terms as arising from the Protestant Reformation in Europe in the first half of the sixteenth century. Economic growth was limited or nonexistent before the relatively abrupt and historically identifi­ able transition, a view compatible with popular images of the Middle Ages as stagnant. The transition itself, however, was a decisive moment 8. In a further parallel with Weber, the clements of Tawney's argument had already been anticipated. Andrew White ( 1 8 9 6 ) , the first president of Cornell University, claimed that all dogmatic religions, including Protestantism, were an impediment to scientific prog­ ress, which came only when religious authority generally was weakened.

The Protestant Ethic Thesis

43

..c

�

e <.9 u

.E

0 c 0 u w

800

1 000

1 200

1 400

1 600

1 800

2000

Year Fig. 2. 1 .

G rowth pattern implied by the Weber thesis

that triggered a profound turnaround, and subsequent economic growth was substantial. Figure 2. 1 displays the general pattern this implies. In evaluating an argument along these lines, the key issue centers on whether there was an identifiable, discrete transition, in this case to capi­ talism. As Braude! put it, "The whole thing would be much easier if the rise of capitalism dated straightforwardly from Calvin's letter on usury (about T 54 5 l · This would give us a turning point" ( T 9 8 2, s 6 8 ) . The problem is that there is no such clear turning point, whether we start with the mid- sixteenth century or somewhat later (to allow for a lagged ef­ fect) , or whether we search for some other transition induced along Marx­ ist lines. Instead, while there were episodic interruptions and although the experience was not identical everywhere, the evidence accumulated by historians shows that there had been considerable economic growth in Europe starting at least in the eleventh century. Part of the problem here is definitional, and applies to the analysis of a number of different kinds of transitions, whether to capitalism or, more recently, to democracy. Consider Marx's criteria for capitalism (con­ trasted with feudalism ) , identified by Britnell ( T 9 9 3 , 3 6o- 6 T ) :

44

B E F O RE N O RM S

Towns dominate rural life and are the main force for social change; Production is chiefly for exchange, and the final consumer is unknown to the producer; Production is on a large scale; Production is based on wage labor; Economic decision makers have a high propensity to innovate and accumulate; Both tenant and labor relations are governed by contract; and Skilled and unskilled workers depend on entrepreneurs who coordinate manufacturing.

The striking feature of most of these criteria (or any comparable ones) is that they are inherently continuous. Thus, as Britnell points out, we might take performance on each of the criteria to be appropriately classi­ fied as " capitalist" when some arbitrary threshold has been reached ( say 50 percent) and perhaps count overall performance as capitalist when a majority of the criteria have reached that threshold. But such approaches are ungainly, and it is not clear what analytic leverage we purchase in the effort to identify a clear turning point. We believe it more profitable to cast the issue in continuous terms. Specifically, we prefer an initial scenario along the lines sketched in figure 2 . 2, which specifies an increasing (compound) rate of growth over time. This formulation has two key features. First, it does not depict the first part of the period as stagnant (as is the case in figure 2 . 1 ) . Instead, following the compound principle, it reflects the fact that, starting from a low base in the early part of the period, the effects of initial growth are likely to be less dramatic than subsequent patterns. Second, the formulation in figure 2 . 2 does not involve the identification of a key turning point of the kind proposed by Weber, Marxists, and others. Such a casting obviously does not imply that capitalist development was a smooth process that operated similarly everywhere. Clearly, the changes here routinely involved strug­ gles and resistance, along with interruptions generated by exogenous forces, and there were substantial geographic variations to the overall pattern. Bearing this in mind, figure 2 . 2 does broadly summarize the avail­ able evidence, and does so without recourse to an ad hoc transition point. The consensus view among historians has for some time been that the Europe of the Middle Ages was far from dormant and in fact experienced considerable economic growth beginning in about the tenth century. Thus, in his general survey Cipolla ( 1 994, chap. 8 ) labels the years from about 1 000 to 1 3 00 as " the great expansion, " one that was interrupted during the subsequent T 50 years and then continued thereafter. Duby

The Protestant Ethic Thesis

45

..c

�

e (.9 u

.E

0 c 0 u w

800

1 000

1 200

1 400

1 600

1 800

2000

Year Fig. 2.2.

G rowth as a simple compounding process

( 1 9 74 ) identified the twelfth and thirteenth centuries as the "take-off, " and Lopez reached a similar conclusion: With the exception of approximately two centuries ( from the mid­ fourteenth to the mid-sixteenth, but the interval was not exactly the same in every part of Europe), the economy of Europe has been expand­ ing ever since the tenth century. Naturally the figures of our day dwarf those of the Industrial Revolution; these in turn are much larger than those of the Commercial Revolution. ( 1 9 7 6, 8 5 )

Indeed, the argument was presaged much earlier (e.g., Pirenne 1 9 3 7 ) , and more recent research has helped elaborate the mechanisms at work. Thus, the evidence suggests that the Reformation was preceded by a long period of economic growth, not stagnation, and the curve shown in figure 2. 2 broadly reflects the characterization in the last sentence from Lopez. Given the limitations of the documentary record, it is of course diffi­ cult to gauge the size of the changes here precisely. In general orders of magnitude, however, the available evidence indicates that for England, total national income grew approximately threefold from 1 0 8 6 to T 3 00

46

B E F O RE N O RM S

(an increase slightly ahead of population growth) , while the money supply grew even more substantially. From 1 3 00 to 1 4 70, total national income dropped slightly, but given major population losses, per capita income rose significantly ( see Mayhew 199 sa, 1 9 9 5 b ) . Changes such as these had ear­ lier led Cams-Wilson ( 1 9 67, 2 3 5 ) to date the beginnings of capitalism in the English cloth industry not from the fourteenth and fifteenth centuries but two centuries earlier. Similarly, careful analyses point to considerable parallels in the economic expansion of Italy and England by the early fourteenth century (e.g., Britnell 1 9 89 ; Persson 1 99 3 ) . I t is useful t o con­ sider some of the changes associated with this expansion. One such major change involved technology. Indeed, the Middle Ages saw remarkable technological innovation and diffusion, some indigenous and some borrowed from Asia. Much of this naturally centered on agri­ cultural production. The heavy plow, the open fields, the new integration of agriculture and h erding, three field rotation, modern horse h arness, nailed horseshoe and the whipple tree had combined into a total system of agricultural exploitation by the year r r oo to provide a zone of peasant prosperity stretching across northern Europe from the Atlantic to the Dnieper. (White 1 9 7 2 , 1 5 3 )

Other innovations had more general applications to production and com­ mercial activity. These included water and windmills, the vertical loom, the spinning wheel, the compass, the clock, spectacles, printing, and new forms of ship design and navigation.9 Overall, such changes in mechani­ cal technology were linked intimately to economic expansion (Persson I9 8 8 ) and were of sufficient magnitude to lead Gies and Gies ( r 994 ) to conclude that the Europe of r 500 would have been unrecognizable to those who had lived at the beginning of the millennium. Although some of these changes and their diffusion followed an evolutionary pattern, they clearly involved risk-taking entrepreneurs. Innovation in the period was hardly limited to mechanical technology. The commercial revolution, which de Roover ( 1 9 4 2 ) dated to the thir­ teenth century, involved substantial changes in business practices and the organization of business enterprise. Originating in Italy, these include the evolution of permanent partnerships (as opposed to the formation of 9· The literature on technological innovation and diffusion during the Middle Ages is extensive . For a sampling, see Cipolla ( I 994, chap . 6); Gies and Gies ( I 994 ) ; Gimpel ( T 9 8 8 ) ; Persson ( T 9 8 8 ) ; and White ( T 972, T 9 7 8 ) .

The Protestant Ethic Thesis

47

partnerships for a single venture) , the adoption of bills of exchange, which made possible the geographic transfer of money without shipping actual coins, and the development of maritime insurance, which helped share risk with underwriters . All of these innovations hinged, in turn, on improved methods of record keeping. By the end of the thirteenth cen­ tury, debit and credit were recorded separately in the form of bilateral accounting, with full-scale double-entry bookkeeping following shortly thereafter ( de Roover 1 9 74 , chap. 3 ) . r o The adoption and diffusion of these changes during the later Middle Ages directly undermines portraits of the period as stagnant and associated claims that a capitalist spirit had yet to emerge. r r The label " medieval merchant venturers " itself implies innovation and significant risk-taking: "To venture was to take a chance, to hazard one's life or one's goods in an enterprise that might bring a worthwhile reward " ( Cams-Wilson 1 9 67, xvi ) . Cams-Wilson ( 1 9 67, xvi ) further observes that the link between risk and profits was already clear in Chaucer (ca. 1 3 40- 1 4 00 ) . U s moste putte our good i n aventure; A marchant, parde! may nat ay endure, Tmsteth me wei, in his prosperitee, Somtyme his good is drenched in the see, And somtym comth it sauf un-to the londe.

'2

Merchants o f the period routinely exhibited rational economic behavior, consisting "of conducting affairs in an informed and accountable manner in order to minimize the risks and maximize opportunities, of exercising continuous control over the progress of operations, and of never allowing oneself to be deterred in the pursuit of profit by moral, religious, or sentimental considerations " (Day 1 9 8 7, 1 8 0 ) . T O . Along with de Roover's foundational work, see Cipolla ( r 994, esp. chap . 7 ) ; Day ( r 9 8 7 ) ; Duby ( 1 974, esp. chap. 9 ) ; and Lopez ( 1 976, esp. chap. 4 ) . r r . Weber recognized the existence o f such entrepreneurs, but dismissed the point as inconsequential .

This kind of entrepreneur, the capitalistic adventurer, has existed everywhere. With the exception of trade and credit and banking transactions, their activities were predominantly of an irrational and speculative character, or directed to acquisition by force, above all the acquisition of booty, whether directly in war or in the form of continuous fiscal booty by exploitation of subjects. ( 1 9 5 8a, 20 ) Note that Weber's dismissal of the merchants as not capitalist hinges entirely on his judg­ ment of their activities as irrational. 12. Geoffrey Chaucer, " The Chanouns Yemannes Tale," The Canterbury Tales (New York: Modern Library Edition, r 9 2 9 ) , p r .

48

B E F O RE N O RM S

It bears emphasis that the innovations in business procedures just dis­ cussed, along with Day's description of merchant activity, speak directly to the Protestant ethic thesis. According to Weber, "The modern rational organization of the capitalistic enterprise would not have been possible without two other important factors in its development: the separation of business from the household, which completely dominates modern eco­ nomic life, and closely connected with it, rational bookkeeping" ( 1 9 5 Sa, 2 I - 22; emphasis added) . The adoption of permanent partnerships, bills of exchange, and insurance were, of course, key components in the insulation of business from households. Among other things, they helped distribute risk beyond households. All of the available evidence unambiguously indi­ cates that these innovations, along with developments in bookkeeping, predate the Reformation by a significant margin and thereby predate the Puritans by an even wider one. We have to this point emphasized that there is little evidence of eco­ nomic stagnation in Europe prior to the Reformation and no evidence of an abrupt economic transformation to capitalism thereafter. This is why our figure 2. 2 better represents the general trends in the period than does the pattern depicted in figure 2.1. Even so, figure 2. 2 reflects only the general pattern, and indeed, although it did not emanate in Weberian style from the Reformation and its aftermath, there was a major exoge­ nous shock. With some minor variations across regions, the general pat­ tern of economic growth was substantially interrupted for perhaps I 5 0 years, starting around I 3 oo. As a number of writers have observed, Europe was a crowded, indeed, overpopulated place by the beginning of the fourteenth century. Short­ ages of arable land, high food costs, and repeated famines were the order of the day, and population growth was arrested. While the economy did not contract, it did not continue to expand. Against this backdrop, the shock first appeared in T 3 4 8 in the form of the bubonic plague, com­ monly known as the Black Death, and recurred episodically for the re­ mainder of the century. The Black Death was a fundamentally exogenous shock that generated catastrophic human costs. It is estimated that Eu­ rope lost at least one-third of its population, and it was not until the sixteenth century that the size of Europe's population regained its I 3 4 0 level. Along with its intrinsic severity, this population loss had major repercussions for the subsequent fertility rate and age structure of the population, for land and grain prices, which plunged, and for the supply of tenants and labor more generally, which also plummeted ( see, e.g., Cipolla I 994, esp. chap. s; Day I 9 8 7, esp. chap. Io; Epstein I 99 I a;

The Protestant Ethic Thesis

49

Hatcher 1 9 8 6, 1994; Herlihy 1 9 9 7; Huppert 1 9 9 8 ; Livi-Bacci 1 9 9 2 , 445 0; McNeill 1 9 77, esp. chap . 4; Russell 1 9 8 5 ) . Beyond the human costs o f the plague, the ensuing labor shortage had major long-term consequences. Although the point often goes unacknowl­ edged, wage labor was by 1 3 00 already a common feature of urban life, a consequence of the growth of guilds in Europe. The guilds, especially the crafts guilds, were associations of employers that sought to monopolize a given trade and advance its interests. Among other things, guild masters established the criteria for training and promotion of labor through the apprenticeship and journeyman ranks to master. All of this involved wage labor ( see, e.g., Epstein 199 1 a; Lopez 1 9 76, 1 2 5 - 3 0 ) . To be sure, the guilds were a feature of towns and cities, while the European econ­ omy as a whole remained predominantly agricultural. At the same time, as Lopez notes, " between 1 000 and 1 3 00 it was the cities that blazed a trail toward recovery" ( 1 976, 1 8 8 ; see also Miller and Hatcher 1 99 5 ) . The widespread use of wage labor by 1 3 00 i s noteworthy in itself, since Weber defines the distinctive feature of modern Western capitalism as involving the organization of (formally) free labor. 13 The chronic labor shortages induced by the Black Death sharply accel­ erated the diffusion of wage labor beyond the towns. Acute reductions in the number of laborers and renters increased the bargaining power of those who survived with employers and landlords, and the period is often labeled the "golden age of labor" (Epstein 1 9 9 1 a, 2 3 2 ) . Agricultural rents dropped precipitously, while wages increased by a factor of two or three, despite strenuous ongoing government attempts to hold wages to their pre-plague levels. The combination of cheap land and expensive labor further intensified the pace of technological innovation and adop­ tion, and thereby improved productivity ( see, e.g., Britnell 1 9 9 6; Day 1 9 8 7, esp. chap. 10; Epstein 1 9 9 1 a, 2 3 2- 4 8 ; Herlihy 1997, chap. 2 ) . These changes are thus a component of the broader patterns of economic growth already described, and the expansion of wage labor was funda­ mental in this respect. Indeed, Britnell ( 1 99 6, 2 2 3 ) concludes, " The col­ lapse of serfdom was the most important thing that happened between 1 3 3 0 and 1 5 00. " To return, then, to the Protestant ethic thesis, the claim that a clear transition to capitalism occurred at some point after the Reformation is r 3 . " All these peculiarities of Western capitalism [the separation of business from households and bookkeeping] have derived their significance in the last analysis only from their association with the capitalistic organization of labour . . . . Exact calculation-the basis of everything else-is only possible on a basis of free labour" (Weber r 9 5 8a, 2 2 ) .

50

B E F O RE N O RM S

untenable, given the record. Further, the implicit portrayal of Europe as economically stagnant during the Middle Ages is equally indefensible. Instead, and bearing in mind the exogenous shock that came with the Black Death, economic expansion after about 1 000 followed the general more continuous pattern outlined in figure 2. . 2. . And as we trust we have made clear, this expansion and all that accompanied it were well under way before Calvin and his associates, not to mention the Puritans who followed. This leaves little for Weber's Protestantism to explain.

H ow D is tin ctive We re Protestan t Doctrine an d Te aching?

We turn now more briefly to evaluate whether Protestant doctrine repre­ sented a distinctive break with the past. In addressing this issue, there is no need to consider all the intricacies of Protestant (especially Calvinist) theology for two reasons. First, as we have emphasized, Weber made a clear distinction between Calvinist views reflected in official dogma and the Calvinism represented by pastoral teachings, and then linked pastoral Calvinism, not Calvinist theology, to capitalism. This obviates the need for a detailed discussion of the latter, which is not germane to the issues at hand. Second, debate over detailed theological niceties was restricted to a small number of the educated clergy and was in this sense an elite activity. Given the low levels of literacy of the time among the wider population (and even among a significant number of the clergy), it is unlikely that finer theological distinctions were appreciated beyond the confines of a narrow group. Indeed, this doubtless accounts for the rather loose fit between Calvinist theology and pastoral Calvinism and for Weber's em­ phasis on the latter. Bearing this in mind, it is often claimed that Protestantism offered a fresh doctrinal approach to two key issues relevant to capitalism and economic growth: usury and the determination of a "just price . " We address these issues in turn. Like many others, Weber ( 1 9 7 8 , 5 8 3 -9 0 ) was impressed both by the Catholic prohibition on usury and by Calvin's letter on the same topic, issued in about I 54 5. Calvin argued that charging interest on loans was not prohibited in all instances. At the same time, he did not believe that all restrictions on the charging of interest should be abandoned. Among other things, security on loans should be bearable by the borrower, the rate of interest should be modest (about 5 percent) and certainly not be higher than the rate allowed by civil laws, and the loan should be of potential benefit to the borrower as well as to the lender (Braude! T 9 8 2.,

The Protestant Ethic Thesis

5r

s 68 ; Gilchrist I 9 69 , 62- ?0; Samuelsson I 9 6 I , 9 I ) . Calvin's was thus a qualified endorsement of interest charges, and the key question is, Was this conditional endorsement a significant break with past practice s ? I t is true that the pre-Reformation Church opposed usury, and indeed the Scholastic writers devoted much attention to the topic. But it is also the case that, even at an early stage, Church leaders urged moderation in the enforcement of the prohibition. Consider the views of Thomas Aquinas ( I 22 5 - 74 ) . Human laws leave certain sins unpunished because of the imperfection of men; many useful things would disappear, in fact, if all improper operations were rigorously forbidden . This is why civil legislation has at times tolerated usury, not because usury is thought to be j ust but so as not to hinder the advantages that so many derive from it. ( Quoted in Little 1 9 7 8 , 2 1 2; emphasis added)

Observe how, in this view, moderation in enforcement is urged on the

practical grounds that commercial activities require credit. But the Scholastic writers went beyond this to identify conditions under which interest charges might be permissible and to j ustify alternative proce­ dures that achieved the same goals (Noonan T9 5 7 ) . For example, while interest that was intrinsic to the loan remained impermissible, rules for identifying exceptions were developed that, because they were not integral to the loan, were labeled " extrinsic titles. " As Little points out, "The time had not yet come for a coherent theory of credit operations, but the net effect of the various extrinsic titles to interest, that is to say of a collection of exceptional cases, was to j ustify the trade of moneylending at least within the confines of a competitive market" ( I 9 7 8 , I 8 I ) . Along with the lists of permissible exceptions, alternative procedures were employed to achieve similar goals. For example, since they could not charge interest, the bankers shifted the basis of their operations to dealing in the exchange of money ( de Roover 1974, chaps. 9, 1 0; Day 1 9 8 7, chap. 8 ) . As one would expect, the cumulative effect of the exceptions and alternative procedures was gradually to narrow and thereby change "the meaning of usury until for all practical purposes it meant an exorbitant, i.e., an unjust charge, for lending money, which was a serious sin, as it has continued to be down to the present day " ( Gilchrist T 9 69 , 6 5 ) . Indeed, while the term usury origi­ nally referred only to the practice of lending money at interest, that usage is now obsolete so that in contemporary English the term refers to the lending of money at excessive interest. Thus, before the Reformation a revised view of interest had already

52

B E F O RE N O RM S

developed that permitted interest under certain circumstances, that al­ lowed for alternatives, and that was accommodating on enforcement. Indeed, this revised view was similar in its key elements to Calvin's views on usury, which, it will be recalled, retained many similar restrictions (Noonan 1 9 5 7, 3 6 5 - 67; Samuelsson 1 9 6 1 , 8 7 - 9 5 ) . On the issue of usury, then, the Reformation was hardly a fundamental break from the practices of the past but is better cast as no more than an episode in an ongoing evolution of commercial doctrine. The other key area in which Protestant teaching has often been taken to be a break with the past centers on the question of the just price. Here, the argument is that pre-Reformation Scholastic theologians saw the j ust price as the reasonable charge that would enable the producer to live on a scale appropriate to his station in life. Part of the rationale for the medi­ eval guilds, for example, was that they would help members achieve a livelihood commensurate with established standards. In other words, the price system was not intended to reflect market principles but was rather designed to maintain the social hierarchy ( see, e.g., Tawney 1 9 2 6, 40- 4 1 ; Weber 1 9 5 0, 1 3 6- 4 3 ) . Any shift i n emphasis toward market-based pric­ ing systems must therefore have been a product of the Reformation. Again, however, this depiction of the pre-Reformation Church's teach­ ing is inaccurate. As a number of analyses have demonstrated, the Scho­ lastic writers endorsed a market-based pricing system, based solely on competition but subj ect to regulation under defined conditions ( see, e.g., de Roover 1 9 5 8 ; Baldwin 1 9 5 9 ; Gilchrist 1 9 69 , 5 8 - 62; Little 1 9 7 8 , part IV) . '4 Regulation was necessary in cases of collusion. This was indeed part of their case against monopolies, judged as offensive because they involved collusion or conspiracy. Further, the guilds were typically condemned by the Scholastics specifically for their monopoly characteris­ tics. This justification for regulation simply underscores the primacy placed on open competition. The Scholastics further argued that regula­ tion could be j ustified during emergencies when key commodities were in short supply. During the late Middle Ages, major crop failures for cereals, the staple food, were a recurring problem, and such circum­ stances j ustified price controls . But these j ustifications for regulation, based on challenging monopoly behavior and on addressing emergen­ cies, serve simply to highlight the centrality of competitive markets for the Scholastics. These patterns make sense if the medieval Church is cast as a monopor4- Similar, but not identical, considerations were used in the justification of the fair wage (Epstein T99 r b ) .

The Protestant Ethic Thesis

53

listie firm that owned approximately one-third o f all cultivated land in western Europe byA. D . 900 ( Ekelund et al. 1 9 9 6, 8 ) . As such an organiza­ tional form, it served as a quasi government that supplied j urisprudence and other public goods that weaker and more fragmented " temporal " governments could not. Among other things, this implies that the Church played a major role in maintaining law and order along with a system of relatively consistent property rights. It is useful to remember that the medieval Church survived and prospered for approximately a thousand years. Moreover, since it owned such a large share of the European economy, its long-term prosperity hinged on the well-being of the econ­ omy as a whole, so that the Church as an organization had every incen­ tive to promote economic growth ( Ekelund et al. 1 9 9 6, 1 7 5 - 77 ) . We would thus expect an evolution of pre-Reformation Church doctrine that reflected an ongoing adaptation to commercial activities. In light of the preceding, we conclude that Weberian and related claims about the changes in religious teaching concerning economic is­ sues alleged to have been induced by the Reformation reflect a fundamen­ tal misunderstanding of the teachings of the pre-Reformation Church. This misunderstanding parallels the problems with Weber's views about the nature of pre-Reformation economic activity as noncapitalist. What is striking in the key areas of usury and the just price is the continuity in religious views before and after the Reformation. Further, this may be part of a more general pattern with the Reformation. On the basis of a much broader analysis, Spitz concludes that " compared with the dog­ matic struggles of the first to the fifth century, the Protestant Reformation was an intramural scrimmage " ( r 9 8 5 , 3 5 0 ) . Indeed, there is evidence to suggest that doctrinal issues may have been less central to the Reforma­ tion than strategic and fiscal concerns (Manchester 1 9 9 3 ) .

Con clusions

In this chapter we have outlined and evaluated Weber's thesis about Protestantism and capitalism in both its original form and its pastoral forms developed by later scholars. Our evaluation has centered on two broad questions . First, and most fundamentally, we examined patterns of economic growth within Europe since the early Middle Ages, to gauge whether there was a significant shift in economic growth and activity after the Reformation anticipated by a Protestant ethic hypothesis. Sec­ ond, we asked whether Calvinist teaching represented a distinctive new perspective on two key issues often linked to capitalism, credit and prices.

54

B E F O RE N O RM S

A s we trust we have made clear, there i s no evidence of a qualitative shift in economic activity in post-Reformation Europe. The historical record is instead broadly consistent with the pattern of growth since about A . D . 1000 that we outlined in figure 2.2. Further, Protestantism generally (and Calvinism in particular) did not introduce the distinctive new outlook on either credit or prices that has often been suggested by proponents of the Protestant ethic hypothesis. Even if we relax the criterion from the " rise of capitalism " to the emergence of a capitalist spirit, risk-taking behavior (surely part of any such spirit) hardly originated with the Reformation. 1 5 That there is no relationship between percentage Protestant and eco­ nomic performance across nineteenth-century European states is to be expected, given these patterns (Delacroix and Nielsen 200 1 ). Recent stud­ ies that either directly or indirectly invoke Weber's thesis in either its original or pastoral forms to j ustify claims about alleged consequences of cultural values or social capital for political and economic outcomes are thus appealing to an argument with no empirical foundation. But the legacy of the Weber thesis goes beyond the substantive issues that we have addressed here. Specifically, there are two fundamental methodological problems with the way in which Weber developed his empirical argument. These involve selection effects that are built into the structure of his analysis and an idiosyncratic use of the available data. Beyond their intrinsic interest, these problems are noteworthy because they presage similar issues that plague more recent studies that draw on Weber. First, as Braude! pointed out, " the argument works backwards from the present to the past" ( r 9 8 2, 5 67 ) . To frame his analysis, Weber began with a statistical study of Baden as of r 89 5 carried out by his student, Martin Offenbacher. This study reported, first, that Protestants were more likely to attend high school. Second, among all students who did attend high school, Protestants were more likely to be educated in the modern languages and science, while Catholics tended more toward tradi­ tional classical forms of training. Third, Protestants were described as wealthier than Catholics. Given these data, Weber fully endorsed Offen­ bacher's broad inferential sweep. r 5 . Common aphorisms about such behavior have a venerable history, indeed. For example, " nothing ventured, nothing gained" goes back at least to Chaucer ( d. T 400 ) : " He which that nothing undertaketh, no thyng acheveth " ( Troilus and Criseyde, ed. Robert K. Root [Princeton University Press, 1 9 26], bk. 2, 11. 8 07- 8 , 9 5 ) . The same aphorism is included in the first English compilation of proverbs, published originally in 1 5 62: "Nought venture, nought have " (John Heywood, The Proverbs ofjohn Heywood, ed. John S. Farmer [London: Gibbings, 1906], part T, chap. T T , 3 8 ) .

The Protestant Ethic Thesis

55

The Catholic is quieter, having less of the acquisitive impulse; he pre­ fers a life of the greatest possible security, even with a smaller income, to a life of risk and excitement, even though it may bring the chance of gaining honour and riches. The proverb says j okingly, 'either eat well or sleep well . ' In the present case the Protestant prefers to eat well, the Catholic to sleep undisturbed. ( Quoted in Weber 1 9 5 8a, 40-4 1 )

Using Baden at the end of the nineteenth century a s his springboard, Weber leapt quickly backward to a discussion of Benjamin Franklin, then further on back to the preacher Richard Baxter of Cromwell's England, and so on, finding connections along the way. To borrow Brandel's term, such a mode of analysis is charitably described as " retrospective. " The problem here, o f course, i s that such an approach t o the evidence incorporates huge selection effects . First, an outcome is described and then the investigator works backward, searching for clues about the ori­ gins of the outcome. The preferable approach is to start at the beginning with an analysis of the cases that have the potential to manifest the outcome subsequently, and then to explain why some cases experienced the outcome while others did not ( see, e.g., Geddes 1 99 1 ) . Unfortunately, the backward approach to empirical analysis is another of Weber's lega­ cies to more recent studies of the effects of values and culture. In his discussion of cultural accounts of poverty (which constitute the flip side of value accounts of achievement) , Billig has identified the issue clearly. Note how the logic works : These words-authoritarian, individualis­ tic, present-time-focused, etc.-characterize this thing called Hispanic American culture and lead inexorably to maladaptive types of eco­ nomic activity. These values and attitudes exist in the shared conscious­ ness of Hispanic Americans and result in a business ethos that obstructs the type of enterprise we see in more successful countries, presumably ones that have prior cultural traits that foster a better kind of economic activity. Since we have only recently become aware of what constitutes good kinds of activity, it must be the case that those societies whose cultures encouraged that kind were pretty lucky; the others, well, let us hope their cultures are flexible enough to get with the program. ( 1 994, 66 1 ; emphasis in original)

In the next two chapters, we discuss several instances of approaches that draw on Weber's methodological legacy of selecting an outcome and then working backward to locate its origins. There is a second fundamental problem with Weber's analysis of the

56

B E F O RE N O RM S

Baden data. As Samuelsson ( 1 9 6 1 ) has shown, those data do not j ustify the inferences drawn by either Offenbacher or Weber. Consider first the data on high school enrollments (table 2 . 1 ) . Comparing the first and last rows of the table, we see that although they were only 3 7 percent of the popula­ tion, Protestants comprised 48 percent of the high school enrollments. In contrast, while fully 61 percent of the general population, Catholics con­ tributed only 4 3 percent of the high school population. Thus, Weber con­ cluded that Protestants were more likely to attend high school. Looking further at the type of school attended, the religious difference is smallest for the Gymnasien, the high schools that emphasized classics ( especially Greek and Latin), while it is larger for the remaining four school types, all of which in varying degrees placed less emphasis on the classics and more weight on modern languages and science. All of this appears consistent with Weber's general argument, especially if we ignore the awkward little fact that these particular Protestants were neither Calvinists nor Puritans, but Lutherans (the fact may, of course, be less troublesome for some pasto­ ral versions of Weber's thesis, especially Tawney' s ) . But caution is i n order. A s Samuelsson pointed out, the problem with this analysis lies in the denominator, the total population of Baden. Atten­ dance at high schools was voluntary, and high schools were not located in all districts. Thus, not all residents of Baden had an equal chance of attending high school, and district of residence drastically affected the odds of attendance. Samuelsson therefore compared the religious compo­ sition of high schools with the religious makeup of the districts within which they were located, and reached the following conclusion.

TABLE 2. 1 . High School Enrollments i n Baden, 1 885-95, by Type and Religious Affiliation

Total male high school enrollment

Protestants

Catholics

Jews

Totals

48

43

9

100%

43 59 52 49 51

46 31 41 40 37

9.5 9 7 11 12

638 37.1

1 ,057 6 1 .4

26 1.5

Type of School Gymnasien Realgymnasien Oberralschulen Rcalschulcn Hohere Biirgerschulen

98.5% 99% 100% 100% 100%

Population Total population ( 1 ,000) Population ( % )

1 ,72 1 1 00 %

Source: Samuelsson ( 1 96 1 , L i 8 - 3 9 ) ; see also Weber ( 1 9 S8a, 1 8 9 ) for the same data including a typographical error favorable to \'Feber's hypothesis.

The Protestant Ethic Thesis

57

School by school and district b y district i t appears that the proportions of school children classified by religious faith are almost exactly the same as the corresponding proportion of the appropriate district. That the Protestants in Baden as a whole display a " school frequency" higher than their share in the aggregate population is thus due entirely to the fact that more Protestants than Catholics lived in districts where Realgymnasien, Hohere Biirgerschulen, and Realschulen were avail­ able. If one reckons not in terms of total population but of inhabitants of districts containing the respective categories of schools, there remain no differences worth mentioning. ( 1 9 6 1 , 1 4 1 - 4 2 )

S o much for reported religious differences in the high school enrollment patterns of Baden. We turn now to Weber's second claim centering on religious differen­ tials in wealth favoring Protestants. This claim also draws from Offen­ bacher's study of Baden and is based on taxation records. The relevant data, as reported by Samuelsson ( 1 9 6 1 , 1 4 3 ) and displayed in table 2.2, show that Offenbacher and Weber were selective in their use of these data. Specifically, they focused their attention solely on tax liability stem­ ming from capital gains (Kapitalrentensteuer), ignoring tax liability from other sources. This restriction was indeed crucial if they were to draw the desired inference, since 4 5 . 5 percent of the capital gains tax liability involved obligations by Protestants comprising 3 7 percent of the popula­ tion, while 4 6 . 2 percent of the capital gains liability was incurred by

TABLE 2.2.

Taxable Wealth in Baden, 1 897, by Type and Religious Affi liation

Total taxable wealth (millions of marks ) Total taxable wealth ( % )

Protestants

Catholics

Jews

Totals

1 ,600 34.0

2,800 59.6

100 6.4

4,700 100%

28.1

67.5

4.4

100%

37.2

55.3

7.5

100%

45.5

46.2

8.3

100%

638 37. 1

1 ,057 6 1 .4

26 1 .5

1 ,72 1 100%

Type o f taxable wealth 1 . Grund-, Hauser- und Gewerhesteuer (Real estate and commercia! taxes) 2. Spezielle Einkommensteuer ( Income taxes) 3 . Kapitalrentensteuer ( Capital gains taxes)

Population Total population ( 1 ,000) Population ( % )

Source: Data on the distribution of capital are from Samuelsson ( 1 96 1 ) , 143, a n d data on total population are from p. 1 3 8 in the same source.

58

B E F O RE N O RM S

Catholics compnsmg a larger 6 1 percent of the population. But, of course, this inference conveniently ignores other information in the table not so favorable to their hypothesis. Consider the figures for total taxable wealth, which reveal that tax liability from all sources was almost exactly proportional to the popula­ tion shares of Protestants and Catholics (especially if the Jewish minority is not included in the calculations ) . The religious distribution for income tax liability (Spezielle Einkommensteuer) also parallels the total tax liabil­ ity distribution quite closely. In the case of real estate and commercial tax liability ( Grund-, Hauser- und Gewerbesteuer) , we see a reversal of the pattern indicating that Catholics in Baden had more capital invested in this area than Protestants . One could presumably claim that the figures discussed by Weber are the most appropriate ones for this comparison, involving as they do capital gain from sources other than real estate and commercial activity. It is striking, however, that Weber himself drew no such inference. Besides, any such claim would be awkward since real estate and commercial activities involve capital. But there is a simpler explanation of these differences in the type of tax liability displayed in table 2 . 2 . Samuelsson reported that Protestants were twice as likely as Catholics in Baden to reside in communities of at least 2o,ooo people ( 24 . 3 percent versus T 3 percent) . In other words, the Protestant population was disproportionately urban. He further reported that a comparison of urban with rural Catholics uncovered the same differences in the forms of capital investment as Offenbacher found be­ tween Protestants and Catholics. Accordingly, he concluded that size of community, not religion, is the primary factor ( 1 9 6 1 , 1 4 4 ) . I n light o f this, Weber's inference about wealth differences between Protestants and Catholics stems from a selective use of his data. A more dispassionate evaluation of those data reveals that there were no differ­ ences in the gross capital holdings of Protestants and Catholics, and that any gross differences in the form of capital holdings stemmed from the more urbanized nature of the Protestant population. As we shall see in the next two chapters, however, the selective use of data has continued in more recent studies that draw on Weber's legacy. In the interim, we conclude that Weber's empirical springboard to his historical analysis was poorly anchored in the available data.

T H REE

Civic Vi rtue in Italy and the U nited States The most democratic country on the face of the earth is that in which men have, in our time, carried to the highest perfection the art of pursu ing in common the object of their common desires and have applied this new science to the greatest number of purposes. Alexis de Tocqueville ( [ I 8 3 5 - 40] I 9 4 5 , 2: I I 5 ) It may h e impossible to bring about the changes that are needed. There is no evidence that the ethos of a people can be changed according to plan. It is one thing to engineer consent by the techniques of mass manipulation; to change a people's fundamental view of the world is quite a different thing, perhaps especially if the change is in the direction of a more compli­ cated and demanding morality. Edward C. Banfield ( I 9 5 8 , I 5 7 ) M y argument i s . . . with critics who seek t o find a social unit i n the community where none exists, and with romantic city planners, abetted by nostalgic social critics, "\lvho \i\Tant to " revive" a sense of community that never \Vas save in their imagination.

Jeremiah looks b ackward to a golden age and sometimes forward to the possibility of redemption, but the present is for him, at best, a trough. Richard A. Posner ( 200 I, 3 I 9 )

e turn now t o more recent claims advanced o n behalf o f political culture. This chapter centers on analyses of civic-mindedness in Italy and the United States carried out by Robert Putnam, while in chap­ ter 4 we broaden our focus to include a wider spectrum of countries. 59

60

B E F O RE N O RM S

Putnam's analyses are important on at least two counts. First, they can be read as the most current in an ongoing series of such studies in these two countries. The Italian study, Making Democracy Work (Putnam 1 99 3 ) , builds explicitly on Banfield's classic inquiry completed four decades ago, while Bowling Alone (Putnam 2000) evokes themes commonly associ­ ated with de Tocqueville's Democracy in America. Second, these analyses have struck a most responsive chord, receiving an extraordinary amount of attention from scholars, social and political commentators, and the policy community. The Italian and American studies address different elements of civic community. The former advances the argument that the effectiveness of regional governments in Italy hinges critically on patterns of civic engage­ ment that can be traced back to a quite distant past. The latter focuses on trends in civic engagement over the more recent past (i.e., the last five decades ) to suggest that Americans are becoming increasingly isolated from each other. In each setting, Putnam explicitly links patterns of civic virtue to political performance. In this chapter we show that the evidence for these arguments is weak. Most notably, the conclusions advanced about Italy hinge on strong and untenable assumptions about the measurement of institutional perfor­ mance, the principal dependent variable, along with an idiosyncratic read­ ing of the historical record. Claims about declining civic virtue in the United States stem from a selective use of the available evidence. We address these issues in turn.

Civic Virtue in I taly

The purpose of Making Democracy Work (Putnam 1 99 3 ) is to explain differences in the performance of political institutions across Italy's twenty regions. These regions vary appreciably in terms of their wealth and social structure, and, according to Putnam, in terms of their political culture. Italy makes an especially interesting case given the creation of new regional governments in the 1 9 70s. This reform afforded the oppor­ tunity to conduct what he calls a " unique experiment" to assess the effect of political culture on institutional performance. Putnam's study employs an extensive data set that includes informa­ tion on a wide variety of social, economic, and political attributes of each region. Much of these data reflect aggregate characteristics of the regions, and this information is supplemented with survey data from selected regions. While the bulk of his empirical material refers to the contempo-

Civic Virtue in Italy and the United States

6r

rary period, he also draws on quantitative information dating back to the last part of the nineteenth century and augments this with qualitative material reflecting the historical record over an even longer period. The essential conclusion of the Italian study is that regional variations in political culture outperform economic development in explaining the per­ formance of regional governments since the 1 9 70s. Apparent support for this inference comes from the following regression of institutional perfor­ mance on current political culture and level of economic development. Performance = 0.00 + 1 .20 ( CCV) - 0 . 3 2 ( ED ) ( 1 .5 ) ( 5 . 7) R 2 . 8 6 , N 20, =

(1)

=

where Performance is Putnam's measure of institutional performance, circa 1 9 70, CCV and ED are his measures of current civic community and economic development, respectively, and t-ratios are reported in pa­ rentheses. When data for earlier cultural traditions are substituted for current patterns, similar estimates are obtained. Performance = - 0 .04 + 1 . 1 6 (ECV) - 0 . 1 9 ( ED ) (2.9) (0.5) R 2 = . 74, N = 1 7,

(2)

where ECV is Putnam's measure of earlier civic values from the turn of the century, and other terms are defined as in ( T ) . These estimates appear to suggest that while the effects of earlier civic traditions are weaker than those of current civic culture, both outperform economic development in the explanation of institutional performance. Some have argued that such estimates are simply a circuitous way of showing that there are major differences between northern and southern Italy, and that Putnam has done little more than document regional differ­ ences in government performance (e.g., Goldberg 1 9 9 6; Tarrow 1 9 9 6 ) . Indeed, the north-south distinction is highly correlated both with Put­ nam's measure of institutional performance (r = . 8 4 ) and his index of civic virtue (r .90 ) . ' When this regional distinction is added to model ( 1 ), however, we obtain the following. =

r. These correlations involve a dummy variable that equals one for northern Italy, and zero otherwise. Following Putnam ( I 9 9 3 , 2 I 4 , note s 6 ) , the north includes Toscana, Umbria, 1\hrche, and all regions northward.

62

B E F O RE N O RM S

Performance = - 0 . 1 2 + 1 . 1 1 ( CCV) - 0 . 3 3 (ED) + 0.22 (North) ( 3 ) (0.5) ( 1 .5 ) (4. 1 ) R 2 = . 8 6, N = 20,

where North is the regional dummy variable and other terms are as previ­ ously defined. Were the " culture " effect in the first column simply a func­ tion of north-south differences, we would expect the regional dummy in model ( 3 ) to have a pronounced effect, and we would also expect the magnitude of the coefficient for culture to drop precipitously. However, the estimate for the regional dummy variable is much smaller than its standard error, while the remaining two coefficient estimates are very similar to those in model ( T ) . Thus, Putnam's results do not simply reflect gross regional differences. The more fundamental problem with Putnam's analysis centers on the meaning of the three key variables identified previously. Institutional performance is the dependent variable to which an entire chapter is de­ voted ( 1 9 9 3 , chap. 3 ), and the measurement of this variable is critically flawed. Civic community and (earlier) civic traditions are the core ex­ planatory factors. Each of these three variables is a composite measure created by combining an array of indicators on the basis of a series of principal components analyses. Principal components, of course, is a statistical procedure that linearly transforms an original larger set of correlated variables into a substantially smaller set of uncorrelated variables. The goal is to condense and summa­ rize information with a smaller set of variables ( components) that reflect most of the information from the original set of variables (Dunteman 1 9 89 , 7; Flury 1 9 8 8 ) , and the key question with any component analysis is whether the estimates effectively summarize the original information in an interpretable manner. A subsidiary issue is whether a unidimensional solu­ tion is warranted (which implies that the bulk of the original information can be condensed into a single dimension), or whether a multidimensional solution better characterizes the information in the original set of vari­ ables. While a single dimension is obviously preferable on grounds of simplicity, it may not accurately summarize the data. Such a solution is therefore typically compared against a multidimensional alternative on such criteria as goodness of fit. Putnam analyzed a large number of indicators, from which he ex­ tracted a single measure of each of the three variables identified earlier. Perhaps the most striking feature of his analysis is that he proceeded by specifying in advance that no more than one component be extracted at each phase. He thus simply assumed throughout that each of the three

Civic Virtue in Italy and the United States

63

variables is unidimensional, the strongest possible assumption that could be made. Is such a strong assumption warranted ? Given his assumptions, it is evident that Putnam's component analyses cannot themselves be used as a test of the dimensionality of these variables.2 Thus, we need to determine whether the composite variables accurately reflect most of the information contained in the composite indicators from which they are formed or whether they form an indecipherable amalgam. Measuring Institutional Performance To construct his dependent variable, Putnam selected indicators based on four elements: policy process and internal operations, policy decision content, policy implementation, and bureaucratic responsiveness ( 1 9 9 3 , 66- 7 3 ) . ln all, 3 3 indicators were used t o reflect these different elements. Putnam defended the large number of indicators on the grounds that "no single metric, taken in isolation, would suffice to rate the regions fairly. Collectively, however, these indicators can undergird a broad-based as­ sessment of institutional performance" ( 66- 6 7 ) . 3 Since there are 3 3 variables b u t only 20 cases, the analysis necessarily proceeded in steps.4 First, Putnam considered 24 of the indicators, which he broke down into three subsets. Three factors were generated from separate component analyses on each of these subsets, with the proviso that no more than one component be extracted in each of the three anal­ yses. Along with the remaining 9 indicators, these three factors were then subjected to an additional component analysis, in which a one-component solution was again specified in advance ( 1 9 9 3 , 7 5 , table 3 . 2 ) . Scores from this final component analysis were used to form the composite measure of performance in all of Putnam's subsequent analyses. At no step in these analyses was the possibility of a multidimensional solution entertained. Table 3 . 1 reports the estimates from two different principal compo­ nent analyses. The left margin lists the variables analyzed by the four 2 . Indeed, principal components can on occasion generate a unidimensional outcome even when the variables are known to form more than one dimension (Armstrong T 9 6 7 ) . 3 · The meaning of a " fair" rating of the regions is unclear. If the broad range o f indicators is chosen to minimize reliability problems, there is little reason t o anticipate such problems given the aggregated nature of the data. If, on the other hand, the collection of indicators is chosen to minimize concerns about measurement validity, the procedure is not very helpful, for the reasons detailed later. 4· The steps arc necessary because when the number of cases exceeds the number of observations (which it does in the present instance by a large margin, with 3 3 variables and 2 0 cases), there is insufficient information to evaluate the model. In other words, an initial component analysis of all 3 3 variables would involve a hopelessly underidentified model.

64

B E F O RE N O RM S

TABLE 3 . 1 . Princi pal Component Analyses of Indicators of the I nstitutional Performance of Italian Regional Govern ments

Indicator Policy process 1 . Cabinet stability 2 . Budget promptness 3 . Statistical services Policy decision content 4. Reform legislation 5. Legislative innovation Policy implementation 6. Day care centers 7. Family clinics 8. Industrial policy 9 . Agricultural spending 1 0 . Health spending 1 1 . Housing development Bureaucratic responsiveness 1 2 . Responsiveness Eigenvalue Proportion of variance Cumulative prop. of variance

Note: N

Four-!' actor Solution (rotated Varimax loadings)

SingleFactor Solution

�

II

III

IV

.67 .57 . 79

.32 - . 03 . 65

. 77 . 65 .21

- .05 .19 .19

.07 . 69 .45

.87 .87

. 74 . 72

.42 .28

.12 .16

.22 .44

.86 . 63 .59 .47 .50 . 83

. 92 . 82 .27 .41 .33 . 80

.26 - . 02 .07 .4 1 .37 .22

- .06 .05 -.10 - . 65 . 76 -.31

.04 .00 .91 - .0 1 .02 .28

.61

.15

. 8S

.06

.17

5.95 .496 .496

5.95 .496 .496

1.61 . 1 35 .631

1.12 .093 .724

1 .03 .086 .810

2 0 . Italicized loadings > .60.

domains of performance suggested by Putnam: policy process, policy decision content, policy implementation, and bureaucratic responsive­ ness. The estimates displayed in the first column of the table are gener­ ated by assuming a unidimensional solution, and they correspond closely to those reported by Putnam ( 1 99 3 , 7 5 ) . 5 Specifically, 8 of the 1 2 indica­ tors have loadings greater than . 6o, and the extracted component repro­ duces half of the variance in the r 2. Masked by these numbers, however, is the existence of 3 additional components with eigenvalues greater than r .o, the common value for inclusion (Kim and Mueller 1 9 7 8 , 4 3 ) . 5 · The components in table } . I are extracted from a correlation matrix like that re­ ported by Putnam ( I 9 9 3 , I 9 9 l · Note that this is a pairwise correlation matrix; all of the correlations are based on the full 20 cases except for those involving Legislative innovation (variable 5 ) , for which data are missing for 5 cases. We explored two other sets of estimates. First, we imputed values for the missing cases from the remaining I I variables. Second, we removed Legislative innovation from the analysis and performed principal components on the remaining I I indicators. Apart from the obvious fact that the second procedure gener­ ates no loadings for the omitted variable, these alternative ways of handling the missing data yield estimated factor loadings very similar to those displayed in table 3· T .

Civic Virtue in Italy and the United States

65

As the estimates in the remaining columns of the table show, when we abandon the unidimensionality assumption, it becomes clear that Put­ nam's single-factor solution is untenable. Indeed, when we adopt the customary eigenvalue cutoff value of r .o, the figures suggest that a four­ factor solution is optimal. 6 Jointly, these four components reproduce over 8 o percent of the variance in the twelve indicators, a clear improvement in fit over the figure of 50 percent for the first component. While the first factor remains the largest, only half of the indicators load clearly on it. Observe further that the four factors are not clearly interpretable. Most notably, they do not reflect Putnam's original fourfold classification of the twelve items (policy process, policy decision content, policy implemen­ tation, and bureaucratic responsiveness ) . There is thus no basis for the use of a unidimensional measure of institutional performance.? Forcing a single-component solution in the manner j ust summarized simply generates an uninterpretable composite. The problem is com­ pounded when that component is subsequently combined (in a process which assumes a priori that the underlying factor structure consists of a single component) with eleven other indicators to form one overall sum­ mary measure of " institutional performance. " Because the analysis con­ flates assumptions and conclusions, and given that the assumptions are unwarranted, the measure of Putnam's principal dependent variable does not in fact tap institutional performance in the manner claimed. Measuring Civic Culture Applying parallel analyses to the four indicators of current civic commu­ nity (preference voting, referendum turnout, newspaper readership, and the frequency of sports and cultural associations), we find a pattern more 6. The factor loadings shown in table 3 . 1 from the four-factor estimates are from a varimax rotation. Estimates from oblique rotations are very similar, suggesting that the four factors are reasonably treated as orthogonal. 7· A less obvious issue concerns the use of composites generated by an initial principal cumponents analysis as indicators in subsequent component analyses, as in table 3 . 1 . For example, the measure of bureaucratic responsiveness analyzed in the table was itself formed from a prior component analysis of six indicators. As we have shown elsewhere (Jackman and Miller 1996, 640), a principal components analysis of these six indicators shows that they form three quite distinct clusters or components. l'vloreover, each of these three compo­ nents reflects a distinctive policy area (centering on agriculture, health, and vocational policy, respectively), and each is therefore readily understood. We further showed in Jackman and Miller ( 1 99 6, 64 1 ) that a similar problem occurs with the construction of the component " housing and urban development" shown in table 3 . 1 . We actually found evidence for two distinct but correlated factors (r = .4 1 ) for this element of Putnam's overall measure of institutional performance.

66

B E F O RE N O RM S

consistent with Putnam's argument. In other words, his prior assumption of unidimensionality seems more warranted. Thus, our estimates of an unrestricted principal component model show that the first component has an eigenvalue much greater than the cutoff of I .o, and it reproduces 8 5 percent of the variance in the four indicators; the second factor, by way of contrast, has an eigenvalue of only . 2 8 . A one-factor solution also fits relatively well with the five indicators of early civic traditions: strength of mass parties, incidence of cooperatives, membership in mutual aid soci­ eties, electoral turnout, and the longevity of local associations. Our esti­ mates of an unrestricted model indicate that the first factor reproduces 74 percent of the variance in the five indicators, while the eigenvalue for the second factor is . 7 8 . I n themselves, o f course, these patterns d o not imply that the two com­ ponents extracted reflect either current civic community or earlier civic traditions. For example, both measures include voter turnout as compo­ nents, but the connection between turnout and civic culture is quite un­ clear. Other studies have shown that voter turnout in general is more a function of institutional than cultural differences (Powell 1 9 8 6a; Jackman 1 9 8 7; chap. 5 , this vol . ) , and there is no reason to believe that turnout in Italy (either in elections during the earlier period or in refer­ enda during the later period) should be different. Indeed, Putnam's own data suggest that turnout in Italian referenda during the 1 9 70s and 1 9 8os varied with the number of propositions to be decided. This pattern is consistent with an institutional voter fatigue hypothesis and is clearly at odds with the proposition that turnout can be taken as a measure of cultural values. 8 Such questions of interpretability aside, there is at least a statistical rationale for the single-component solution employed by Put­ nam in the construction of his explanatory variables. The empirical diffi­ culty with Putnam's analysis thus appears to hinge on his dependent variable. Reassessing the Link between Culture and Performance We now turn from the internal structure of Putnam's measures to con­ sider briefly the linkages between them. As we indicated earlier, Putnam 8. Specifically, a pooled analysis of the live referenda from T 9 7 4 through T 9 8 7 across all twenty regions yields a regression coefficient of - 2.7 for the number of items per referendum, suggesting that turnout dropped by just under 3 percent for each proposition added to a referendum (information on the number of propositions is from Butler and Ranney 1 9 9 4 , appendix A, and the maximum proposition/referendum ratio was 5 in the period ) . We revisit the voter fatigue argument in more detail in chapter 5 ·

Civic Virtue in Italy and the United States

67

TABLE 3.2. T-ratios from Regressions of the I nstitutional Performance of Italian Regional G overnments and Its Com ponents on (a) Current Civic Values (CCV) and C u rrent Economic Development (ED) and (b) Early Civic Values (ECV) and C u rrent Economic Development Current Culture Dependent Variable Overall institutional performance Policy process Cabinet stability Budget promptness Statistical services Policy decision content Reform legislation Legislative innovation Policy implementation Day care centers Family clinics Industrial policy Agricultural spending Local health spending Housing development Bureaucratic responsiveness Responsiveness

Early Culture

CCV

ED

ECV

ED

5.7

1 .5

2.9

0.5

3.2 1.2 1.2

1.9 0.0 1 .3

1 .5 0.7 1 .4

0.8 1.3 0. 7

2.5 0.9

0.0 0.6

2.8 1.6

0.7 0.3

2.6 1 .4 1.9 0.4 3.0 2.5

0.7 0.5 0.7 0.7 1.7 0.8

4.0 1 .4 0.8 0.1 2.3 2.0

1.5 0.5 1.7 1.0 1.3 0.2

2.7

1.7

3.8

3.1

reports strong relationships between h i s composite o f institutional perfor­ mance and his composites of political culture ( both current civic commu­ nity and early civic traditions) . If these estimates are robust, we should obtain similar relationships between civic culture and the constituent components of institutional performance. If, on the other hand, the com­ posite measures lack coherence, such a disaggregation will generate a much more mixed set of estimates.9 To address this issue, we regress each of the individual indicators of institutional performance on current civic values and early civic traditions, controlling for economic development. Table 3 . 2 summarizes the results by reporting the t-ratios (absolute val­ ues) estimated from the different regressions. '0 9· Our analyses follow the logic of construct validation (see, e.g., Zeller and Carmines I 9 8 0 ). Thus, we arc concerned with evaluating ( I ) the coherence of the composite measures of cultural values and institutional performance taken separately and ( 2 ) the linkages be­ tween the indicators of cultural values and the indicators of institutional performance . I O . We report the absolute value of the t-ratios simply to summarize the various regres­ sion analyses. With I7 degrees of freedom, a t-ratio of 2. I I or greater is significant at or beyond the .o 5 level with a two-tailed test, and a t-ratio of r. 7 4 is significant at or beyond the . I o level. Following convention, we use 2.0 as an approximate cutoff point for statistical significance. As is clear in table 3 . 2 , the majority of the estimated coefficients have t-ratios

68

B E F O RE N O RM S

The first row o f table 3 . 2 shows the t-ratios from the regressions of the composite measure of institutional performance on current civic commu­ nity and early civic traditions, respectively, controlling for economic devel­ opment (these t-ratios correspond to those reported in equations [ r ] and [2] ) . The t-ratios for civic community and civic traditions in this top row are quite large ( 5. 7 and 2.9 ), while those for economic development are much smaller (less than 2 . 0 ) . Calculations like these form the basis for Putnam's conclusions, as noted earlier. A much more mixed picture emerges from the t-ratios for the individ­ ual components of institutional performance displayed in successive rows of table 3 . 2 . Of the twelve regressions in the first part of the table for current culture (with economic development controlled), only six have t-ratios for civic community exceeding 2.0. The results for the correspond­ ing regressions in the second part of the table involving early cultural traditions are even less encouraging. Of these twelve regressions, only five have t-ratios greater than 2.0. In all, there is little evidence to indicate that institutional performance depends in any appreciable manner on cultural traditions . While there is a statistical j ustification for the measures of civic community, the substan­ tive rationale for those measures is less clear. But the most glaring prob­ lem is that the measure of institutional performance cannot be justified on either substantive or statistical grounds. Absent a viable measure of the key dependent variable, these analyses provide no warrant for linking cultural values to political performance. The Historical Origins of Italian Regional Differences Putnam supplements the quantitative analyses with historical materials to conclude that the sources of differences in civic engagement across Italian regions have long-standing roots. The regions characterized by civic involvement in the late twentieth century are almost precisely the same regions where cooperatives and cultural associations and mutual aid societies were most abundant in the nineteenth century, and where neighborhood associations and religious confraternities and guilds had contributed to the flourishing communal republics of the twelfth century . . . . The astonishing tensile strength of civic traditions testifies to the power of the past. ( 1 9 9 3 , r 62 ) considerably less than 2 . 0 , which means that the signs o f most o f the regression coefficients and related estimates not reported in the tables are unreliable.

Civic Virtue in Italy and the United States

69

In other words, Putnam believes that regional differences in current civic traditions have a long shadow that can be traced back to the eleventh century ( 1 9 9 3, r 8o ). As a result, the parallel between a map of variations in civic norms as of about 1 3 00 (his fig. 5 . 1 ) and the present distribution as of around 1970 (his fig. 4 . 4 ) is " remarkable " ( 1 3 4 ) . We therefore think it fair to infer that he casts these cultural norms as durable. We take issue with this interpretation of the Italian materials on at least two fundamental and overlapping counts. First, Putnam's view of medieval communal republics is romanticized. Second, his claim for long historical continuities in civic-mindedness is difficult to sustain. We ad­ dress these interrelated issues in turn. Putnam is hardly alone in his nostalgia for the past. Indeed, nostalgia is a common feature of much communitarian commentary that claims that the West has experienced a loss of civic values in recent years and that urges a need to reclaim the role that community formerly played. The empha­ sis on reclaiming or restoring a sense of community implies that such a sense existed in the past, and indeed communitarians routinely cite late­ eighteenth-century America, the European Middle Ages, and classical Greece as cases in point. De Tocqueville's Democracy in America is per­ haps the best-known nineteenth-century exemplar of such a line of analy­ sis, but the tradition remains alive and weJ J . r r Phillips shows in great detail that this is a romanticized view that glosses over a historical record replete with considerable conflict and with glaring inequalities in re­ sources and participation. Defining a community as "a group of people who live in a common territory, have a common history and shared values, participate together in various activities, and have a high degree of solidarity" ( 1 99 3 , q ) , Phillips concludes that communitarian thought harkens back to a past that never was. Putnam's reading of " the flourishing communal republics of the twelfth century" exhibits a similar nostalgia and conjures images of northern " Italians " resolving their Hobbesian dilemmas in a collaborative and egali­ tarian framework that eluded those in the south. Recall his reference to flourishing " neighborhood associations and religious confraternities and guilds " of the twelfth century. The reference to guilds immediately raises a red flag, since as we noted in chapter 2, the guilds, especially the crafts guilds, were monopolies controlled by the guild masters that sought to I I . Visible contemporary versions of the argument include Bellah et a!. ( r 9 8 5 ) , Lane ( 2ooo ) , Macintyre ( r 9 8 r , r 9 8 8 ), Sandel ( r 996, 1 9 9 8 ) , and Taylor ( r 9 8 5 , r 9 8 9 ) . With his concerns about the effects of weak community norms in southern Italy and his bowling alone thesis about the decline of such norms in the United States, Putnam ( 1 9 9 3 , 2000 ) falls squarely in this Jeremiah tradition ( Posner 200 T ) .

70

B E F O RE N O RM S

advance the interests of a given trade by limiting competition. Among other things, guild masters established restrictive criteria for training and promotion of labor through the apprenticeship and journeyman ranks to master. While one person's collusion can be another's cooperation, the guilds are miscast when seen as egalitarian organizations promoting the interests of the community as a whole, unless of course one takes the mas­ ters of a particular trade to comprise the entire community. But the guilds, of course, are only part of the medieval communal era. In more wide-ranging analyses of the period, Brucker ( 1 999 ) and Muir ( I9 9 9 ) repeatedly characterize social relations as agonistic and faction­ alized, political life as coercive, authoritarian, and arbitrary, an order in which violence and brutality were commonplace. From city to city, the scenario varied only in the details: beleaguered regimes plagued by internal divisions, weak and indecisive leadership, the erosion of civic institutions and values, and the transfer of power from the community to an individual or family . . . . [Arbitrary and ruthless] tactics, employed by every regime from Sicily to Piedmont to enforce obedience and raise revenue, created a pernicious legacy for the future-a pervasive and deeply rooted distrust of, and hostility to, the state, its institutions, its operations, and its personnel. Subj ects did not perceive the state as a protector and defender, but as an exploiter and predator. (Brucker 1 9 9 9 , 3 69 - 70, 3 74 )

Were w e ( unwisely) t o adopt Putnam's view that current political culture can be traced back to the communal era, we would conclude that this era contained the seeds of amoral familism. As Muir wryly observes, " Ban­ field's 'amoral familism,' which Putnam establishes as the antithesis of civil society, represents a social system that medieval Italians would have understood almost instinctively" ( T 999, 3 8 T ; see also Spruyt T994, T 3 63 9 ) . Putnam's nostalgia for the medieval period resembles that of other communitarians and is equally unwarranted. It is, of course, possible that Putnam simply delved too far back into the past to locate the sources of regional differences in civic culture. Perhaps by moving forward several centuries he might have located a better starting point reflecting regional differences more favorable to the claim that cultural values have remarkable " tensile strength. " Unfortu­ nately, even the more recent historical record is inconsistent with his claims for durable continuities in civic-mindedness. For example, the fascist regime initiated by Mussolini in 1 9 1 9 - 2 1 and lasting through T 9 4 3 was consciously totalitarian. While it received a

Civic Virtue in Italy and the United States

71

measure o f popular support (the magnitude of which is difficult t o calcu­ late) , the regime relied heavily on police, who had wide latitude. Further, the fascists devoted considerable effort to mobilizing youth and other segments of the population into newly minted party organizations ( see, e.g., Morgan 199 5; Whittam 1 99 5 ). Even if the regime did not achieve all of its stated goals, we would expect these activities to have had a pro­ nounced negative impact on social capital, as evidenced by interpersonal trust and patterns of voluntary association. This implies that social capi­ tal is endogenous. It further implies a substantial discontinuity in patterns of social capital even during the twentieth century, one straddled by Putnam's statistical analyses of the effects of civic virtue in the late nine­ teenth century on political performance in the T 9 8os. Indeed, given Put­ nam's claims about the " tensile strength" of cultural values, it is interest­ ing to contemplate how he might have related regional differences in civic virtue as of the I 9 3 0s to subsequent variations in performance. Similar discontinuities were also evident in the nineteenth century. As Tarrow reminds us, every regime that governed southern Italy from the Norman period until the I 8 6os was foreign and " governed with a logic of colonial exploitation" ( I 99 6, 3 94 ) . Further, major changes were intro­ duced in southern Italy after I 8o 5, including a restructuring of property rights that concentrated resources and power in the hands of landowners: "The phenomenon of southern latifundism [large estates] , far from being a relic of medieval times, is of more recent origin-the intended and unintended consequences of the political changes in the nineteenth cen­ tury " (Sabetti 1 9 9 6, 3 0 ) . The nineteenth century thus brought an in­ creased concentration of wealth to many parts of southern Italy, which, in turn, actively exacerbated the marginalization of peasants and day labor­ ers. Again, these were fundamental changes that we would expect to lower social capital in the form of interpersonal trust and patterns of voluntary association. But a conclusion like this, of course, also returns us to social capital as an outcome of changes in circumstances, not as an immutable source of such problems as poverty and poor government. Given the above, the historical record does not in fact warrant the inferences drawn by Putnam. The evidence we have briefly summarized indicates that his reading of medieval communal republics is nostalgic: these were hardly political orders imbued by civic-mindedness. Beyond this, the record belies the notion that regional differences in social capital can be traced back to the distant past. Indeed, there is no evidence for the long historical continuities in civic-mindedness claimed by Putnam, and there is every reason to believe that levels of social capital fluctuated across time and regions in response to broader circumstances.

72

B E F O RE N O RM S

Civic Virtue in the U n ited States

More recently, and by way of elaboration on Making Democracy Work, Putnam has turned his attention to questions of civic association and social capital in the United States, following closely in the footsteps of de Tocqueville. His crucial claim is that social capital in the United States has dramatically declined over the last two decades. As more civic genera­ tions are replaced by their less civic progeny, Americans have increasingly taken to " bowling alone, " a pattern heralded by declines in both trust and group membership (Putnam 1 9 9 5a, 2ooo ) Y We are not the first t o observe the striking disjuncture between the Italian study and this claim. Where cultural values in Italy are cast as incredibly durable, " Civic America" has apparently evaporated over the course of j ust one generation. The talk about " Bowling alone, " and to a lesser extent the article itself, directly contradict the logic of Making Democracy Work. In Putnam's Italian model, the kind of overnight deterioration of civic virtue that he proposes regarding America would be inconceivable-once civic virtue is in place it is incredibly durable over the centuries . ( Lemann 1 9 9 6, 2 5 ; see also Sobel 200 2 )

Italians have apparently been fortunate t o avoid exposure t o American network television. Leaving this to one side, the bowling alone argument outlined in Put­ nam's original paper immediately struck a responsive chord among public officials at the highest levels. 13 For example, President Clinton reportedly used it to prepare for his 1 9 9 6 State of the Union address, I4 while Senator Bill Bradley, in a speech to the National Press Club on February 9, r 99 5 , T 2 . Surrounding information on Putnam's thesis is available at www.bowlingalone.com, which offered (as of November 2003 ) the following two " surprising facts " : ( r ) "Joining one group cuts in half your odds of dying next year, " and ( 2 ) "Ten minutes of commuting reduces social capital by ro percent." For those considering the possibilities of redemption, this site also provides a direct link to www.bettertogether.org. Among other things, this second site announces a book, Better Together (Putnam and Feldstein 2003 ), that " describes a dozen innovative organizations from cast to west and north to south that arc re-weaving the social fabric of our country, and brings the hopeful news that our civic institutions are taking new forms to adapt to new times and new needs. " I 3 . Political commentators did not share the initial enthusiasm for the howling alone argument. Sec "The Solitary Bowler, " J'.conomist, february r 8 , r99 5 , 2 r - 22; Lemann ( r 9 9 6 ) ; Samuelson ( r 9 9 6 ) ; Stengel ( r 9 9 6 ) . r 4 - Statement by White House Press Secretary l\1ike lVIcCurry in a press briefing o n January 1 9 , 1 9 9 6 , 1 : 4 5 P.M.

Civic Virtue in Italy and the United States

73

had already expressed concern that " like fish floating on the surface of a polluted river, the network of voluntary associations in America seems to be dying. " More recently, the second Bush administration launched its " Faith-based initiative, " an effort designed to build social capital by elimi­ nating " the federal government's discrimination against faith-based orga­ nizations . " ' 5 And, having j ust launched a USA Freedom Corps, President Bush called on all Americans to perform at least four thousand hours of volunteer service over their lifetimes in his 200 2 State of the Union ad­ dress. r6 These responses are noteworthy public expressions of concern about community values because they require no tangible commitment of resources. As Lemann puts it, "if Putnam is right that as local associations go, so goes the nation, his work suggests the possibility of solving our problems through relatively low-cost association-strengthening initiatives that don't require taxes. This makes a wonderful message " ( 1 9 9 6, 2 5 ) .I? Putnam begins his analysis of trends with a discussion of voter turnout in presidential elections, which he claims has declined by about 2 5 per­ cent since the r 9 6os. He regards this trend as a leading indicator of more profound problems. Like the canary in the mining pit, voting is an instructive proxy mea­ sure of broader social change . . . . Declining electoral participation is merely the most visible symptom of a broader disengagement from community life. Like a fever, electoral abstention is even more impor­ tant as a sign of deeper trouble in the body politic than as a malady itself. ( 20oo, 3 5 )

While the claim of declining electoral participation is common, the evi­ dence for it is less clear-cut. Specifically, the measure of turnout is the ratio of total votes cast in presidential elections to the size of population T 5. Rallying the Armies of Compassion, available from the White House Web page at www.whitehouse.gov/news/reportslfaithbased.html. " Our goal is to energize civil society and rebuild social capital, particularly by uplifting small non-profit organizations, congrega­ tions, and other faith-based institutions that are lonely outposts of energy, service, and vision in poor and declining neighborhoods and rural enclave s . " r 6. Washington Post, lVlarch I J , 2 0 0 2 , A27. In January 200 3 , the President's Council o n Service a n d Civic Participation w a s created under the rubric of the Freedom Corps with the purpose of recognizing and encouraging outstanding volunteer service and civic participa­ tion, and Putnam himself was among those named to the council (see www.u safreedomcorps .govIfor_ volunteers/awards_scholarships/presidents_council. asp ) . r 7 . Judging b y his subsequent support for President Bush's USA Freedom Corps initia­ tive, Putnam seems to agree: "We can get away with a much smaller government because we have this tradition [of charity and volunteerism] " ( quoted in the Washington Post, March T?, 2002, A6 ) .

74

B E F O RE N O RM S

o f voting age, which i s taken a s the eligible population. As McDonald and Popkin ( 200 1 ) have shown, however, the size of the population of voting age includes many people (most notably, noncitizens and felons) who are ineligible to vote. This fact is of special significance because the proportion of ineligible voters in the population of voting age has been growing over time. With these ineligibles removed from the denominator, there has been no decline in voter turnout since 1 9 7 2 . We agree that electoral participation is a crucial leading indicator, but discussions of its decay in the past three decades are greatly exaggerated. When cast in broader terms, the bowling alone thesis has an equally shaky empirical footing. Specifically, Putnam is selective in the evidence about civic-mindedness that he presents . Looking at a broader range of the available information, sustained trends in civic norms are hard to find. Further, nothing in the record indicates a global decline in group membership over the last three or more decades. To be sure, membership in some groups has fluctuated over time, and some kinds of groups have experienced dwindling popularity while others have been in expansion mode. But broad claims about social capital must rest on overall patterns of group membership, not on the fate of particular groups. We address these issues in turn. Social Norms and Trust in the United States Putnam identifies two related aspects of the concept of social capital: expressions of community norms and patterns of civic engagement. To measure the former, he relies heavily on responses to a question asked by the General Social Survey (GSS) and several other polling agencies: " Gen­ erally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people ? " (Davis and Smith 1 99 3 , 203 ) . Putnam reports that levels of social trust in the United States, as reflected in this item, have declined by about a third ( 2ooo, 1 40, fig. 3 8 ) . Indeed, when we regress the average level of trust for each survey on year, we obtain the following estimates. Trust = 7.713 - .004 (Year) (3.8) R 2 = .48, N = 1 8 ,

where Trust is the average number of individuals per survey indicating that most people can be trusted, Year is the year of the survey, and the t-ratio is reported in parentheses. The coefficient for Year is negatively signed and

Civic Virtue in Italy and the United States

75

statistically significant ( t = 3 . 8 ) , implying that levels of trust did decline ( by about four-tenths of a percent per year) between 1 9 7 2 and 1 99 8 . ' � Observe, however, that Putnam relies primarily on this single indicator to gauge trends in social trust. One ordinarily would be reluctant to draw broad inferences about trends in social values on the basis of a single survey item in this way. The preferred strategy that helps maximize reli­ ability is to gauge responses to a series of items that reflect the underlying value. Indeed, elsewhere in Bowling Alone, he attaches great weight to the " core principle " that "No single source of data is flawless, but the

more numerous and diverse the sources, the less likely that they could all be influenced by the same flaw" ( 2ooo, 4 1 5 , emphasis in original ) . I n this connection, i t i s noteworthy that along with the Trust item, the GSS each year asks a wide variety of other questions. While many of these are not appropriate measures of the views of individuals about those around them, two questions that immediately precede the Trust item in the GSS survey questionnaire seem especially germane to the question at hand (Davis and Smith, 1 99 3 , 202- 3 ) . These two questions are worded as follows: ( r ) "Would you say that most of the time people try to be helpful, or that they are mostly just looking out for themselves ? " (hereafter Helpful­ nes s ) ; and (2) "Do you think most people would try to take advantage of you if they got a chance, or would they try to be fair ? " (hereafter Advan­ tage) . While these two questions do not explicitly employ the term, they clearly bear directly on issues of trust, and indeed they are often used in conjunction with the trust item to form a general measure of interpersonal trust (e.g., Brehm and Rahn 1 9 9 7; Paxton 1 999 ) . I f Putnam's broad inference about dwindling social capital is correct, given the decline in the single trust item, we would anticipate a similar de­ cline in the proportion of people who believe that others are helpful along with an increase in the number who believe others will take advantage of them. Accordingly, table 3 . 3 presents the results from three separate regres­ sions of Trust, Helpfulness, and Advantage on Year, for the period from 1 9 7 2 through 1 99 8 . The first row shows the estimates reported earlier for purposes of comparison, and the sign of the coefficient for Year is negative and statistically significant. In the next row, however, we see that responses to the Helpfulness item have not on the average changed over time (the Year coefficient is - . oo r , with a t-ratio of o . 69 ). The estimates in the third row appear consistent with Putnam's hypothesis: the coefficient suggests that the proportion of people who believe that others will take advantage of them has been increasing, and this trend is statistically significant T

8. The GSS data are available in electronic form from Davis and Smith ( T999 ) .

76

B E F O RE N O RM S

TABLE 3.3. Average Levels of Trust, Helpfulness, and Advantage Regressed on Year, 1 974-94 Dependent Variable Trust Helpfulness Advantage

Year Coefficient -

. 004 " - .001 .002"'

T-Ratio

R2

3 .81 0.69 3.0 6

.48 .0 3

.37

Note: N = 1 8 .

�·coefficients a r e more than twice their standard errors.

( t-ratio = 3 .0 6 ) . However, inspection of the scatter plot suggests that the relationship between Advantage and year is nonlinear. '9 From 1 9 7 2 through 1 9 8 6, the percentage o f people believing that others will take advantage of them declined; only in the ensuing years ( from 1 9 8 7 to 1 99 8 ) is the pattern consistent with Putnam's hypothesis. Thus, when we broaden the scope of the analysis to three items, we obtain a considerably more mixed picture about trends in social norms relating to trust. This mixed pattern is reflected in the relationships among these three indicators. As we have emphasized throughout, cultural explanations place great weight on the idea that values form coherent clusters. Thus, in the case of trust in others, we expect respondents who believe that " most people can be trusted" also to believe that " people try to be helpful " and that " they try to be fair. " In other words, we expect a degree of consis­ tency within this set of norms . Bearing this in mind, consider the two separate correlation matrices shown in table 3 · 4 · The first presents the correlations among Trust, Help­ fulness, and Advantage using the pooled individual-level data, while the second reports the correlations among the three at the aggregate level. In the first matrix using individual-level data, the relationships among the three measures are in the expected direction. Those individuals who be­ lieve others are helpful are also likely to trust them, and not to believe that they would take advantage of them. The moderate absolute magni­ tude of these figures (about · 3 7) is reasonable given that the correlations are based on individual-level data . Thus, the three indicators appear to cluster together quite nicely. However, the estimates from the aggregate data imply a quite different picture. Although the relationship between Helpfulness and Advantage is in the correct direction ( - . 7 5) and quite strong, the correlation between 1 9 . When we include a quadratic term in the regression to capture the nonlinear rela­ tionship between Advantage and Year, the R2 increases by almost one-third. Neither of the other two regressions (with Trust and Helpfulness) exhibits this degree of nonlinearity over time.

Civic Virtue in Italy and the United States

77

TABLE 3 . 4 . Correlations among Trust, Helpfulness, a n d Advantage a t the Individual and Aggregate Levels, 1 9 74-94 Pooled individual-level data (n = 25, 1 7 1 ) Advantage Helpful Trust Trust Helpfulness Advantage

1.0 .36 - .3 7

1.0 - .38

1.0

Aggregate Times-Series Data (n = 1 8 ) Advantage Helpful Trust 1 .0 - .20 - .27

1.0 - . 75

1.0

Trust and Helpfulness is incorrectly signed ( - . 20 ), given the corresponding individual-level figure. While it is correctly signed, the aggregate correla­ tion between Trust and Advantage is particularly weak (only - . 2 7 ) , a fig­ ure that is lower than even the average individual-level correlation of - . 3 7· This suggests that the Helpfulness and Advantage items provide a more consistent picture of how people relate to each other than does the Trust item, and that the former items therefore better reflect social trust, broadly conceived, the quantity Putnam believes to be in decline. The correlations between the first two items are in the expected direction and reasonably strong at both individual and aggregate levels. Although the measure of Trust clusters well with Helpfulness and Advantage at the individual level, it is negatively related to Helpfulness and only weakly correlated with Advantage at the aggregate level. It is quite a leap from these patterns to the broad conclusion that levels of interpersonal trust are hemorrhaging in the United States. Group Membership in the United States Putnam's case for declining levels of interpersonal trust is accompanied by his additional argument that, over the last two decades, levels of group membership have also fallen, so that Americans have become increasingly isolated and disconnected from each other. He emphasizes that this trend toward disengagement is substantial, claiming that overall group member­ ship in the United States dropped by about one-fourth between 1 9 74 and 1994 ( 1 9 9 5 b, 6 6 6 ) . 20 These conclusions about group membership are flawed on three inter­ related counts. First, our analysis of the available group membership data suggests that even with education controlled, there is little, if any, overall change in membership rates in the United States in recent years. Second, 20. The claims about declining group membership seem to be widely accepted. Even in an otherwise critical evaluation of Bowling Alone, Durlauf ( 2002, 260) allows that " No one can go away from the book unpcrsuadcd that for the United States as a whole, participation in voluntary organizations has experienced a profound decline in the last 30 years . "

78

B E F O RE N O RM S

that participation rates have dropped for some groups does not in itself reflect declining social capital, since we would expect fluctuations in group membership with variations in political, economic, and demo­ graphic conditions. For example, all else equal, we would predict a mem­ bership decline for such groups as the Boy Scouts as the population ages. Finally, Putnam is again selective in the evidence he presents, attaching great social weight to the decline in league bowling ( 1 99 5a, 70) , but discounting the Sierra Club as a mere " mailing list" organization and, indeed, generally dismissing the significance of the growth of a variety of newer organizations.

Has Group Membership Declined? Putnam's original evaluation of changes in group membership rates relies heavily on what has come to be known as the GSS Standard Question on group membership : " Now we would like to know something about the groups or organizations to which individuals belong. Here is a list of various organizations.2I Could you tell me whether or not you are a member of each type" (Davis and Smith 1 99 3 , 3 69 ) . After pooling the surveys for the period 1 9 74 through 1994, Putnam ( 1 99 5b, 66 5 ) esti­ mated the effect of Year on the responses given for each type of group (controlling for education ) and compared the size of the logistic regres­ sion coefficient for the variable Year with a baseline of 1 .00 as a means of comparison. As figure 3 . 1 indicates, we obtain similar estimates:22 Frater­ nal groups seem to have experienced the most marked decline, while nationality and " other" groups show modest increases. Using individual­ level data, our results thus closely parallel Putnam's.23 2 1 . Sixteen groups are covered in the GSS General Question on group membership, including fraternal groups, service clubs, veterans' groups, political clubs, labor unions, sports groups, youth groups, school service groups, hobby or garden clubs, school fraterni­ ties or sororities, nationality groups, farm organizations, literary, art, discussion or study groups, professional or academic societies, church-affiliated groups, and a residual category for any other groups. The group membership question has not been used since I994· 2 2 . We follow Putnam's ( r 9 9 5 b, fn. II, 6 8 I ) coding of the indicator for education, which is a trichotomous measure where the lowest value represents less than twelve years of education, the middle value indicates the completion of twelve years, and the highest value represents more than twelve years of education. \'Ve arc unable to reproduce his results for the " all groups" category presented in his figure I ( 6 6 5 ) . All other estimates we obtain are similar, but not identical, to those he reports. 2 3 . It is important to reiterate that in absolute terms membership levels have not been declining. As Samuelson ( r 9 9 6) correctly points out, the presence of the control for educa­ tion used to produce figure 3 . r means that the " decline " in social capital that Putnam claims to have occurred actually results from the fact that membership rates have not risen as quickly as education levels. Hence the " decline. "

Civic Virtue in Italy and the United States

79

1 . 02 c

0 ·u; "' <J)

1 .01

2'

1 . 00

0, 0

"tj

·o, .2 .s

.99

-0;

.98

.E @:

.97

<J) r

"X w

.96

"

�

·a �

-"'

u

� -"' u

�

•.=J "" � �

0 �

-5 V>

"

-�

"

� ·o;

� � 3

5 �

t � "" C/)

§

u..

2

� Vl

Fig. 3 . 1 .

Membership trends ( 1 974-94) by type of group (education control led)

Since his argument centers on declines in overall patterns of group membership, we should see a similar decline in membership rates over time at the aggregate level. Table 3 . 5 thus displays the average member­ ship by year of survey for fraternal groups, literary groups, service clubs, and membership across all groups. 24 We have selected the first three categories to represent the variation in membership trends at the individ­ ual level depicted in figure 3 . 1 , with fraternal groups having the largest decline, followed by literary and service groups. We also include a mea­ sure of the average number of groups to capture general trends in group membership. Column 2 of table 3 · 5 shows that in 1 9 74 approximately 1 3 . 7 percent of those sampled were members of a fraternal group, and this figure declined by about 6 percentage points to a low of 7·9 percent in 1 9 9 3 , providing preliminary support for Putnam's thesis. However, T 9 74 ap­ pears to have been an unrepresentatively high point from which to start, because membership rates for the next four surveys hovered between 24. Nlembership rates across all groups were constructed by first summing the sixteen group categories. Given the skewed nature of the data, values of 6 or more were recoded to 5, resulting in a more balanced distribution. The average number of group memberships was then calculated for each year by dividing by the number of respondents .

80

B E F O RE N O RM S

1 0 . 7 percent and 9 · 9 percent. Comparing these figures with the average membership rate for the last four GSS surveys ( excluding the 1 9 9 3 low) of 9 · 5 percent, we find only a slight drop (about one-half of a percent) in aggregate membership in fraternal organizations over the twenty years from 1 9 74 to 1 9 9 4 . Participation i n literary organizations shows even less, if any, decline. The membership rate for the first four surveys is about 9 percent, while the corresponding figure for the four most recent surveys is 9 . 8 percent, an increase of about 8 percent. However, excluding the I 99 3 survey with its unusually high estimate of 1 1 .4 percent, we see that about 9 · 3 percent of those surveyed were members of literary groups in recent years, indicat­ ing that any increase in membership in such organizations between 1 9 74 and 1994 was slight. There is also a barely perceptible increase in rates of membership in service organizations, with about 9 . 1 percent of the popu­ lation participating between 1 9 74 and 1 9 7 8 , compared with a 9 · 5 per­ cent rate based on the 1 990, 1 99 1 , and 1994 surveys (again, the figure from the 199 3 survey is disproportionately high ) . With the exception of fraternal groups, then, this analysis identifies no significant decline in the average level of participation over time. The last column of table 3 . 5 presents the membership rates for all groups from 1 9 74 to I 9 9 4 · In 1 9 74 , respondents belonged to an average of 1 ·94 groups, a figure that fell to 1 . 5 9 by 1 9 80, the lowest membership rate for all years . After 1 9 80, .

TABLE 3 . 5 . Average Levels of Membership in F raternal, Literary, Service, and "All" G roups, by Survey Year Type of Group Year 1 9 74 1 975 1977 1978 1980 1 983 1 9 84 1986 1987 1 988 1989 1990 1991 1993 1 9 94

Fraternal .1 3 7 .1 0 7 .1 02 .099 .1 04 .094 .09 1 .09 0 .09 3 .084 .09 1 .09 7 .09 0 .079 .1 00

Literary

Service

.092 .0 8 9 .090 .0 8 8 .084 .1 00 .087 .089 .073 .0 8 5 .096 .089 .092 .1 14 .0 98

.08 9 .08 3 . 1 07 .084 .08 9 .1 03 . 1 04 .113 .09 1

.1 10 .097 .097 .08 9 .124 .100

All Groups 1 .94 1 .77 1 .8 3 1 .73 1 .59 1 .8 6 1 .76 1 .8 7 1 .6 6 1 .75 1 .79 1 .76 1 .69 1 .8 9 1 .8 7

Civic Virtue in Italy and the United States

8r

TABLE 3.6. Regressions of Rates of Membership in Selected G roups on Survey Year, with and without a Control for Average Level of Educati on, 1 974-94 Dependent Variable Fraternal Fraternal Literary Literary Service Service All groups All groups

Year - .003 ( 1 .42 ) - .00 1 ,. ( 3 .9 7) - .002 ( 1 .1 7) .0003 ( 0.9 6 ) .004 ( 1 .55 ) .0008 ( 1 .22 ) - .0 3 1 ( 1 .22 ) - .0001 ( 0.0 1 )

Education .03 ( 1 .04 )

.03 ( 1 .3 5 )

- .05 ( 1 .2 6 )

.42 ( 1 .22 )

Note: n = 1 5 . Coefficients with t-ratios in parentheses. •:· coefficient has a t-ratio > 2.0.

however, group membership averaged 1 . 79 , with the latest two surveys ( 1 99 3 and 1 994) showing that on average Americans belong to 1 . 8 8 groups. While there is certainly variation from year to year in group membership rates, one would be hard-pressed to take these figures to indicate a systematic decline. Perhaps our failure to find a trend in these data reflects the fact that the figures in table 3 . 5 reflect gross rates of group membership, without controls for respondents' education levels. Recall that the estimates in figure 3 . 1 show the effect of Year on individuals ' responses to questions about their group membership, controlling for level of education. Table 3 . 6 displays analogous figures for the aggregate level, obtained by regress­ ing membership rates in different groups discussed earlier on the year of the survey, with a control for education.25 For purposes of comparison, we also report the simple regressions of group membership on Year. The figures in the first row of table 3 . 6 show that membership rates in 2 5. Education is measured here in terms of the mean level of educational attainment for each year. Average levels of educational attainment have of course been increasing, and Education is highly collinear with Year. Given the presence of apparently influential data points in table 3 . 5, we report robust regression estimates.

82

B E F O RE N O RM S

fraternal groups have not declined over time, even with education con­ trolled: while the sign of the coefficient for Year is negative ( .003 ), the coefficient itself is not statistically significant ( t-ratio 1 .4 ) . The same pattern holds for literary group membership, for which the coefficient is negatively signed ( .002) but statistically indistinguishable from zero ( t-ratio r . 2 ) . Membership rates in service groups are quite similar. Although there is a slight positive trend over time (the coefficient for Year is .004 ) , indicating a net increase from 1 9 74 to 1994 in service group membership at about an annual rate of ·4 percent, the estimate is not significant (t r . 5 ). Finally, the estimated Year coefficient for overall group membership levels in table 3 . 6 is negative ( .03 ), but statistically indistinct from zero (t T . 2 ), which indicates no significant time trend one way or the other. Even without education levels controlled, there is only one significant coefficient for Year in the estimates in table 3 . 6, and that is for member­ ship in fraternal groups. The remaining three bivariate coefficients, includ­ ing that for overall group membership, fall far short of providing any systematic evidence for the claim of waning participation in groups. Indeed, while it is not apparent in the figures just discussed, there is reason to believe that overall rates of group membership may have actu­ ally increased over the period. Baumgartner and Walker ( T 9 8 8 ) pointed out some time ago that the Standard Question used to assess membership rates has not been modified since it was first introduced in 1 9 7 4· We noted earlier that the question asks respondents about sixteen possible kinds of groups (including a residual " other" category) to which they might belong. Further, the question allows only a single affiliation within each category ( for example, a respondent is allowed to name only one service or sports group to which she belongs ) , even within the residual category. Thus, as Baumgartner and Walker observe, the question is blind to the many new types of groups that have mushroomed since T 9 7 4 , such as civil rights, environmental, and consumer movements. This, coupled with the fact that respondents are restricted to naming no more than a single group membership within a given category, implies that the Stan­ dard Question has, over time, become an increasingly biased measure of group membership, one that understates " the explosive growth of the number of groups within certain types " (Baumgartner and Walker 1 9 8 8 , 9 T 4 ) . Their comparisons of the Standard Question with an alternative that does not suffer these restrictions suggests that the bias is quite sub­ stantial and growing. 26 -

=

-

=

=

-

=

2 6 . These issues arc further discussed in Smith ( I 990) and Baumgartner and Walker ( I990).

Civic Virtue in Italy and the United States

83

Thus, whether or not we adjust for trends in levels of educational attainment, there is simply no consistent support for the claim that group membership rates in the United States, measured by the Standard Ques­ tion, have headed south over the last two decades. The time trends for different types of group membership rates are of mixed sign and fail to meet conventional criteria for statistical significance, and there is simply no trend to report for the measure of overall group membership ( see also Ladd 1 9 9 9 ) . 2? Indeed, if Baumgartner and Walker are correct, group membership may have actually increased over the past three decades .

Are Particular Groups Special? Although we thus find no general trend in group membership rates, perhaps declining participation rates in key activities and organizations herald dwindling social capital. Putnam himself makes j ust such a case with reference to patterns of church membership and to the National Congress of Parents and Teachers (PTA), an organization that he high­ lights. His argument implies that such groups are so central to American life that trends in their membership can be taken as leading indicators of trends in social capital. We address trends in church and PTA member­ ship in turn. Putnam attaches great significance to patterns of religious participa­ tion, devoting an entire chapter to the issue on the grounds that " churches and other religious organizations have a unique importance in American society" ( 20oo, 6 5 ) . The chapter includes two charts (figs. 12 and 1 3 ) covering the years since 1940. Figure 1 2 suggests that church membership has declined by between 5 and 10 percentage points since about 1 9 60 ( depending on whether data from church records or sample surveys are used ), while figure 1 3 indicates that the proportion of adults attending church at least weekly has dropped by perhaps 10 percentage points in the same period. Putnam attributes these shifts primarily to generational suc­ cession: " For the most part younger generations . . . are less involved both in religious and in secular social activities than were their predecessors at the same age " ( 2ooo, 79 ) . 2 7 . While the GSS Standard Question forms the centerpiece o f Putnam' s ( 1 99 5 b ) origi­ nal analysis of declines in group membership, that question is not used to analyze trends in group membership in Bowling Alone. In that more recent study, only one graph employs the Standard Question, and this is simply to show that group membership rises and then declines with increasing age (Putnam 2000, 249, fig. 7 0 ) . At one other point, Putnam concludes on the basis of the GSS and related data that "the net decline in formal organiza­ tional membership is modest at best" ( 2ooo, 5 9 ) . But by way o f summary only four pages later he concludes that "the broad picture is one of declining membership in community organizations" ( 6 3 ) .

84

1 776

1 860 1 850

Fig. 3.2.

B E F O RE N O RM S

1 890 1 870

1916 1 906

1 952 1 926

1 9 80 1 970

1 990

The churching of America, 1 776- 1 990

While Putnam cites a good deal of relevant research, it is striking how little this research informs his analysis. 28 Take trends in church member­ ship. The comprehensive study of religious participation in America since 1 77 6 by Finke and Stark ( 1 9 9 2 ) paints a significantly different picture. Figure 3 . 2 displays the trend in religious adherence rates in the United States since 1 77 6 constructed by Finke and Stark.29 The most impressive feature of this graph is the consistency with which religious participation has increased over the last two centuries. Beginning with a 17 percent adherence rate at the end of the American Revolution, by 1 8 6o it had more than doubled to 3 6 percent. After the Civil War, the adherence rate contin­ ued to grow to its current high of 64 percent in 1 9 70. Finke and Stark ( 1 99 2 ) label this growth in adherence rates the " churching of America. " A full account o f this surge in religious participation would take u s far afield. Briefly, Finke and Stark trace the growth summarized in figure 3 . 2 to the emergence o f an increasingly competitive religious market. In colo28. We are not the first to make this observation. Posner ( 2 00 T , J T T ) is particularly struck by Putnam's failure to engage the issues raised and data reported by Ladd in a variety of reports, culminating in Ladd ( I 9 9 9 ) , while Durlauf ( 2002, 2 6 5 ) characterizes the exten­ sive use of secondary sources as "typically polemical rather than scholarly. " 29. Church adherence rates from I 7 7 6 to I 9 8 o are from Finke and Stark ( I 992, I 6 ) . The T 9 7 0 and r 9 9 0 figures are from Ladd ( T 9 99, 44 ) .

Civic Virtue in Italy and the United States

85

nial America, New England's Congregational churches enjoyed virtual monopoly status and experienced a correspondingly low adherence rate. With the expansion of religious freedom following the American Revolu­ tion, other religions were free to compete for adherents. The result was a substantial decline in Congregationalist membership, and the hitherto " dissident upstarts " -the Baptists and Methodists - " rocketed from 1 2 percent o f all adherents in 1 776 to 4 1 percent in 1 8 5 0 " ( Finke and Iannaccone 1 99 3 , 3 0 ) . As figure 3 . 2 shows, over the same years the overall adherence rate doubled to 34 percent. The religious market has continued to become increasingly competitive in the century and a half since.3° Of course, as Putnam notes, membership may have become increas­ ingly nominal: hence we need also to consider trends in attendance rates. Here again, however, his analysis is idiosyncratic. On the basis of a much more detailed analysis of the available data, Greeley ( 1 9 8 9 ) concludes that, with one partial exception, there has been no systematic change in overall patterns of church attendance since 1940. The exception involved attendance rates among American Catholics, which declined from 1 9 69 to 1 9 7 5 , reflecting disappointment over the papal encyclical letter on birth control of 1 9 6 8 , that is, reflecting a doctrinal issue. Greeley con­ cludes that "with the exception of the dramatic Catholic change between 1 9 6 8 and 1 9 7 5 (an episodic event) , patterns of American church atten­ dance are remarkably stable-straight lines with only one deviation, and that ended by 1 9 7 5 " ( 1 9 89 , s 6) . This lack o f a systematic trend i n overall church participation implies, among other things, that Putnam's claims about basic generational differ­ ences are misplaced. Instead, to the extent that age is a factor, its impact comes through pronounced life-cycle effects on religious involvement, such that religiosity increases with age. Given this life-cycle pattern, mi­ nor temporal fluctuations in participation with shifts in the age structure of the population are to be expected. 3 • Such fluctuations should not, however, be confused with a generational process in which overall pat­ terns of church membership or attendance are in decline. 30. In addition to Finke and Stark ( 1 9 9 2 ) , see Finke and Iannaccone ( 1 993 ) , Iannaccone ( 1 9 9 1 , 199 5 ), Iannaccone, Finke, and Stark ( 1 9 9 7 ) , Stark a n d Finke ( 2000 ), and Warner ( 1 9 9 3 ) for discussion of the ways in which the structure of religious markets affects religious behavior. The general argument about the effects of religious monopolies originates with Adam Smith ( 1 976, hie 5 , chap. I, part 3, art. 3 ) . 3 r . Life-cycle (as opposed to generational ) effects on church attendance arc well docu­ mented. In addition to Greeley ( 1 9 8 9 ) , see Hout and Greeley ( 1 9 8 7 ) , Firebaugh and Harley ( 1 9 9 1 ) , and Stolzenberg, Blair-Loy, and Waite ( 1 9 9 5 ) . Putnam cites these latter studies, but they appear not to inform his analysis.

86

B E F O RE N O RM S

Perhaps recognizing that overall rates o f religious participation have not experienced the dramatic shift he describes, Putnam offers an alternative line of argument about religion and social capital. It is widely acknowl­ edged that there has been a major shift within American Protestantism where the established, mainline denominations have suffered substantial membership declines while the appeal of more evangelical and fundamen­ talist groups has grown dramatically (see, e.g., Greeley 1 9 8 9 ; Finke and Stark 1 9 9 2 ) Y Phrased differently, the former have lost market share to the latter. Putnam believes that this denominational shift itself may have con­ tributed to the erosion of social capital. In both evangelical and mainline congregations, the religiously in­ volved learn transferable civic skills, such as management and public speaking, but mainline Protestants are more likely to transfer them to the wider community . . . . Thus the fact that evangelical Christianity is rising and mainline Christianity is falling means that religion is less effective now as a foundation for civic engagement and " bridging " social capital. ( 20oo, 7 8 )

Observe the shift in emphasis: even i f overall rates o f religious participa­ tion have not declined, social capital is dropping because Americans are increasingly attracted to the wrong kinds of churches . The argument is, at best, ungainly. We believe it more defensible to conclude that religious participation is incidental to social capital. This judgment avoids ad hoc claims like the one j ust discussed and obviates the need to explain away other awkward patterns. Perhaps the best way to see this is to cast the problem in a more comparative framework. Using data from the 1990- 9 3 World Values Surveys, table 3 . 7 displays rates of church attendance for six countries, consisting of the five cases examined in the original Civic Culture study ( Almond and Verba T 9 63 ) along with Sweden. In considering this table, recall further that Almond and Verba rated the United States and the United Kingdom as the most civic of their cases, and Italy and Mexico the least civic. (We also hazard the guess that, had it been one of their cases, they would have rated Sweden as highly civic. ) The first panel o f table 3 . 7 shows that the relationship between age and reported church attendance is not a peculiarly American phenome­ non. In all six countries, attendance increases with age. Of more interest, perhaps, the highest rates of attendance are reported for Mexico, fol­ lowed closely by the United States and Italy, while the lowest rates are 3 2 · Mainline Protestant denominations include Congregationalists, Episcopalians, Lu­ therans, lVlethodists, and Presbyterians.

Civic Virtue in Italy and the United States

87

TABLE 3.7. Rates of Church Attendance (% attending at least once a month) for Select Countries, by Age, from 1 990-93 World Values Survey 1 8- 34

3 5 - 54

55+

49 ( 54 1 ) 44 (813) 14 (465 ) 17 (717) 59 (92 1 ) 9 (352)

60 (639) 48 (713) 25 (48 6 ) 29 (713) 67 (462) 9 (392)

63 ( 627) 65 (492) 31 ( 524 ) 55 ( 671 ) 70 (148) 17 ( 242 )

Men

35 (212)

37 (213)

52 ( 1 33 )

Women

48 (200)

51 ( 2 1 7)

72 ( 1 30)

Men

36 (185)

39 ( 1 27)

52 (93)

Women

56 (215)

67 ( 155)

81 (136)

Country

A: Country Rates b y Age United States Italy United Kingdom West Germany Mexico Sweden

B : Italian Rates b y Region, Age, and Gender Northern Italy

Southern Italy

Note: Cell entries are the percentage of respondents reporting that they attend church as least once a month, and numbers in parentheses identify the numbers of repondents on which the percentages are based.

reported for Sweden and the United Kingdom.33 If, following Putnam ( 2ooo ), we take religious participation to reflect social capital and civic­ mindedness, we arrive at a baffling ranking of cases that is completely at odds with the ranking offered by Almond and Verba ( 1 9 6 3 ) .34 3 3. The low rates of church attendance in these two cases are best explained in terms of the structure of their religious markets. Both the United Kingdom and Sweden have estab­ lished churches, and Swedish Lutheranism in particular enjoys a virtual monopoly status. Among other things, the Lutheran clergy in Sweden are well-paid civil servants who enjoy the right to strike . Such monopoly status depresses participation (on these issues, see Iannaccone, Finke, and Stark 1 9 9 7 ) . 3 4 · We recognize that Almond a n d Verba's surveys were completed more than thirty years earlier than the surveys used in table 3 . 7. This, however, cannot help explain the ranking of cases in table 3 ·7 without auxiliary arguments that would account for ( r ) a hemorrhaging of social capital in the United Kingdom (and presumably Sweden ) and ( 2 ) an apparent concurrent and rapid maturation of social capital in Mexico and Italy. Even if such

88

B E F O RE N O RM S

The second panel o f table 3 . 7 pursues this line of thought a step further. Here, we report church attendance rates in Italy by region, gen­ der, and age, which allows a direct comparison of Making Democracy Work with Bowling Alone. Net of the age and gender differences in this panel, observe that church attendance is higher in southern Italy than in the north. Indeed, it is distinctively high among southern Italian women, especially those fifty-five or older. Thus, if we follow Putnam ( 2ooo ) and employ church attendance as an important measure of social capital, then we are led inexorably to the conclusion that there is more social capital in southern Italy than there is in the north, which of course contradicts the central empirical claim in Putnam ( 1 99 3 ) . 3 5 We are, however, reluctant t o draw such a n inference. Taking participa­ tion in organized religion as a key index of social capital involves a series of assumptions that are altogether heroic and generates indecipherable empirical patterns. We do not think there is more social capital in south­ ern Italy than in the north. We are not persuaded that Sweden suffers a distinctive social capital deficit. The proposition that involvement in mainline religious organizations generates a better form of social capital than does engagement in more fundamentalist denominations strikes us as a stretch. The only defensible conclusion is that rates of religious participation reflect the structures of religious markets and have no bear­ ing on social capital, one way or the other. Along with religious participation, PTA membership is also taken as a leading indicator of social capital. Indeed, Putnam argues that the PTA " has been an especially important form of civic engagement in twentieth­ century America because parental involvement in the educational process represents a particularly productive form of social capital" ( 199 s a, 69 ) . 36 Further, " the explosive growth of the PTA [from 1 9 1 0 to 1 9 60] was one arguments did not elude us, the comparatively high social capital ranking of the United States implied in table 3 · 7 means that Putnam's ( 20oo) concerns with dwindling social capital are misplaced . Aud the proposition that levels of civic virtue in Italy improved in j ust over thirty years is inconsistent with Putnam's ( 1 9 9 3 ) claim that regional differences in civic community have persisted over the very long haul. 3 5. To the best of our knowledge, most southern Italian women attendees are Catholic, and few belong to evangelical Protestant denominations. Thus, in Putnam's terms their high rates of church attendance cannot reflect a suboptimal form of social capital. Further, Putnam seems to regard church attendance among Catholics as desirable from a social capital perspective. For example, he estimates that about a quarter of American Catholics are Hispanic: "Their involvement means that the Catholic Church is once again playing an important role in connecting immigrants to the broader American society " ( 2ooo, 7 6 ) . 3 6. Among other things, w e presume that PTA membership could be seen as a particu­ larly productive form of social capital to the extent that it helps generate human capital, although Putnam does not address this issue.

Civic Virtue in Italy and the United States

89

of the most impressive organizational success stories in American his­ tory " ( 2000, s 6 ) . From I 9 60 to about I 9 80, however, the proportion of parents of school-age children who were PTA members plummeted by about 50 percent, and membership rates have not shifted appreciably since ( 2ooo, 5 7, fig. 9 ) . This much-discussed claim, o f course, takes the decline i n PTA member­ ship as symptomatic of a more general decay in social capital-social net­ works, norms, and trust. Indeed, if it is the case that social trust and civic engagement are strongly correlated, and if PTA membership is a leading indicator of civic engagement, we should find a significant link over time between levels of trust and PTA membership. However, examination of the plot between PTA membership rates and the average levels of trust re­ ported in the GSS ( 1 9 72-9 8 ) suggests no such relationship. And the simple correlation between the two series is only . 1 3 , a figure that is statistically insignificant at any meaningful test level. That fluctuations in PTA member­ ship rates track levels of trust so poorly in the United States suggests that the former is a poor proxy for social capital, more generally conceived. Instead of reflecting shifts in social capital, we think that recent varia­ tions in PTA membership stem fundamentally from changes in the labor force. Consider the following. Between 1 9 59 and 1999 the percentage of women 1 6 and older who were members of the civilian labor force in­ creased steadily from 3 7 to 6o percent, whereas the percentage of male labor force members aged I 6 or more declined slightly, from 84 to about 75 percent. Since the overwhelming majority of PTA members (especially active members) are women, and given that PTA membership is sensitive to women's marital status and labor force participation ( Crawford and Levitt 1 999 ), this means that over the last forty years the supply of potential members of the PTA has declined. Further, women' s labor force participation has been increasing steadily with time ( see, e.g., Costa 2ooo ), so that the correlation between the two over the period is an almost perfect · 9 9 · This obviously means that we cannot statistically distinguish the meaning of " time " from that of women's labor force participation. By the same token, however, it is clear that in this context time itself has no intrinsic meaning but serves simply as a proxy for a variety of possible factors, the most obvious of which is women's labor force participation. Indeed, when we regress PTA membership (in mil­ lions) on the percentage of women in the labor force for the period 19 5 9 through 1 9 9 8 , w e obtain the following estimates:37 3 7 · PTA membership data were provided b y the National PTA headquarters, Chicago; and labor force participation rates for women and men were extracted from the Bureau of

90

PTA R2

=

B E F O RE N O RM S

20. 4 5 - .25 (Women in labor force) ( 1 0. 0 ) . 72, N 4 0.

=

=

These figures are similar to those obtained from regressing PTA member­ ship on time, although women's labor force participation provides a mildly better fit.3 8 One might conclude that because time is statistically indistinguishable from women's labor force participation in these data, it is reasonable to draw inferences about declining social capital in recent decades. But then we have to confront the fact that PTA membership does not map meaning­ fully onto levels of interpersonal trust, which muddies the relevance of PTA membership for social capital. We need also to acknowledge that the decline of PTA membership does not in itself imply a general decline in levels of parental involvement with the education of their children. Indeed, focusing exclusively on National PTA figures may be quite misleading. There is considerable evidence to indicate that the National PTA has become a minority player in this niche, as large numbers of school parent-teacher groups have disaffiliated themselves from the Na­ tional PTA and formed local unaffiliated groups (Ladd 1999, 3 1 - 4 3 ) . In this connection, evidence from the Roper Center shows that parental par­ ticipation in school activities has actually increased significantly in the past two and a half decades. In 1 9 69, only 1 6 percent of those surveyed in a Gallup poll reported that they had attended a school board meeting, while by 199 5 this figure had more than doubled to 39 percent (Ladd 1999, 3 9 ) . Finally, a s already noted, w e need t o remember that while women's labor force participation is a variable with clear substantive meaning, the vari­ able time in the present context has no such intrinsic meaning but serves simply as a proxy for variables like labor force participation. We are therefore inclined to the view that PTA membership has de­ clined as women's labor force participation has increased. No more, no less. The PTA, after all, is an organization with a " traditional constitu­ ency of white middle-class women who do not work outside their homes" ( Crawford and Levitt 1999, 2 8 2 ) . As the supply of such individuals has Labor Statistics Web site (http://www.bls.gov/data) (using Series ID LNS I I 3 0000 2 ) . The slight decrease in male labor force participation reflects the aging of the U.S. population over the period. 3 8. The estimated R' from a regression of PTA membership on time for the same period is .67. The decline over time in PTA membership is not linear but was steepest for the first half of the period, and a quadratic regression of membership rates on women's labor force participation (or time ) describes the data almost perfectly.

Civic Virtue in Italy and the United States

91

contracted, so too has the PTA, while other more locally based groups have flourished.39 Thus, drawing broad alarmist conclusions from the PTA data about dwindling social capital seems entirely gratuitous.

Beyond League Bowling A third problem that plagues Putnam's analysis of group membership is selection bias. Most notably, he places considerable weight on the fact that between 1 9 8 0 and 1 9 9 3 league bowling declined by 40 percent while the total number of bowlers increased by TO percent. He further claims that " the broader social significance" of these trends " lies in the social interaction and even occasionally civic conversations over beer and pizza that solo bowlers forgo " ( 2000, I I 3 ) . 4° Missing from this congenial portrait is any allusion to the drive home after the beer. League bowling may have started to decline around I 9 8o, but that also is the year in which Mothers Against Drunk Driving was founded. Two decades later, the membership of MADD has skyrocketed, so that it is now a large organization with more than 6oo chapters nationwide.4' A key difference between league bowling and MADD, of course, is that the latter repre­ sents a newer type of organized activity. The selectivity here is clear: the alleged social capital implications of membership declines in older organi­ zations are lamented, while the significance of more current membership groups is downplayed or ignored. The former are thus cast as the sole mechanisms that can generate robust and meaningful social networks. Completely brushed to one side is the possibility of a shift over time in the kinds of group ties that might generate social capital ( see, e.g., Wuthnow 1998). Environmental groups are an interesting case in point. While some of these organizations are older, all have experienced rapid rates of growth m recent years. Yet their significance seems to elude Putnam. For 3 9 · Evidence also indicates that the rise in women's labor force participation has more generally resulted in a decline in social capital produced within the home in the form of entertaining friends, neighbors, and relatives ( Costa and Kahn 200 I ) . 4 0 . From Putnam's analysis, it appears that beer and pizza are even more significant for the owners of bowling lanes, since league bowlers consume three times as much of these commodities as do nonlcague bowlers, " and the money in bowling is in the beer and pizza, not the balls and shoes " ( 2000, I I 3 ). On the bright side, there is evidence that the quality of league bowling has increased dramatically in recent years. According to a report on Na­ tional Public Radio, the annual number of perfect ( 3 oo ) games increased from about 900 thirty years ago to over 3 4,000 in I999 (NPR 2000 ) . 4 r . NIADD is listed a s one o f the large U.S. membership associations by Skocpol ( I 999, 72-7 5 ) , where " large " organizations are those with membership lists greater than I percent of the adult population. For more information on the organization itself, see www.madd.org.

92

B E F O RE N O RM S

example, the Sierra Club i s discounted a s a " mailing list" organization with few consequences for social capital because it is "not really an organization in which members meet one another" (Putnam 1 9 9 5 b, 666) .42 But even at the national level, the Sierra Club actively engages members. Thus, its Outing Department offered 3 4 0 trips in 1 9 9 6, in­ cluding eleven days of backpacking in the Arctic National Wildlife Ref­ uge and a seven-day Arizona Trail Family Service trip. Such activities would appear to provide at least as much opportunity for civic engage­ ment among Sierra Club members as do the occasional civic conversa­ tions over beer and pizza that Putnam associates with league bowling. Local chapters of the Sierra Club provide even more opportunities for civic participation. To take j ust one instance, Pettinico lists the following activities organized by the Sierra Club's Los Angeles Chapter in one weekend (May 1 7- 1 9 , 1 9 9 6 ) : " 2 1 day hikes ( including one for singles only ) , two evening hikes, three bicycle trips ( one for singles only), one bird watching walk, four trail repair outings, a nature camera excursion, a wilderness first aid workshop, a nature knowledge workshop, a back­ packing class, a camp fundraiser, two camping trips, and one weekend trip to Catalina Island" ( 1 99 6, 2 7 ) . These are not simply the activities of a mere " mailing list" organization. Equally troubling is Putnam's treatment of rates of youth participa­ tion. He concludes from his analysis of GSS and National Election Study data that "the very decades that have seen a national deterioration in social capital are the same decades during which the numerical domi­ nance of a trusting and civic generation rborn before 1 9 4 01 has been replaced by the dominion of 'post-civic' cohorts " ( r 99 5 b, 677 ) . Putnam ascribes this generational shift in civic virtue to television, noting that the proportion of American homes with television jumped from about r o percent i n 1 9 5 0 t o 90 percent i n 1 9 60. Thus civic participation dropped as the children of the " trusting and civic generation " were replaced by those of a couch-potato "post-civic" generation. Participation in Little League Baseball provides an interesting way to assess any shift from the civic to the post-civic generation, since data are available from 1 9 3 9 to the present.43 Figure 3 · 3 presents a graph of Little League Baseball participants by year. The striking feature of the graph is the consistency with which participation in Little League ( indicated by the 4 2 . According to the Sierra Club's office in San Francisco, membership increased from 1 8 2 in 1 8 9 2 to a high of 629 , 5 3 2 in 1990. By 1 9 9 5 , its membership was 5 8 7,499 . 4 3 . Little League data were provided by the Little League Baseball International Head­ quarters, Williamsport, PA; soccer participation information ( discussed later ) was obtained from the United States Youth Soccer Association, Richardson, TX.

Civic Virtue in Italy and the United States

93

3000

2500 "'

., " �

2000

" 0

E

2

1 500

"'

D � E 0) ::8

r

1 000

/

/

/

/

/ -

__., � I

,.. \ I

I

I

/

(

500 /

0 1939

1 947

1 943

Fig. 3.3. 1 9 39-99

-­

/

I J

Baseball Soccer

1 95 5

1 95 1

1 963

1 959

1 97 1

1 967

1 979

1 975

1 987

1983

1 995

1991

1 999

Membership in Little League and United States Youth Soccer,

dashed line) has increased over the years. Beginning with j ust one league with 4 5 players in 1 9 3 9 , participation grew to about 2 million in 1 9 8 5 , from which i t declined slightly to r . 8 million i n 1 9 8 8 , subsequently in­ creased to 2 . 6 million by 1 9 9 5 , and then tapered off at about 2 . 5 million. This trend is especially noteworthy given the other organizations that have formed and compete for members. Little League Softball was formed in 1 9 74 with 29 ,69 6 members and a s o f 1 9 9 9 had 3 9 0,000. Even more remarkably, U.S . Youth Soccer participation has mushroomed to 2.4 mil­ lion, a twentyfold increase over j ust two decades ( Stengel 1 9 9 6, 3 5 ) , a trend indicated by the solid line in figure 3 - 3 - A striking feature of figure 3 . 3 is that the growth in U.S . Youth Soccer has had very little effect on participation in Little League Baseball. Overall, participation has in­ creased consistently since the League's formation in 1 9 3 9 . One might object that figures like those i n figure 3 . 3 simply reflect demographic shifts. As the age structure of the population becomes youn­ ger, the supply of children available for such organizations as Little League and Youth Soccer increases. However, the age structure of the population is actually becoming older. Indeed, when we regress baseball participation (in tens of thousands) on year, controlling for the number of children enrolled in kindergarten through eighth grade (in thousands) for the period T 9 64 through T 9 9 8 , we obtain the following estimates:

94

B E F O RE N O RM S

Baseball = - 8 575.6 + 4 . 3 7 (Year) + .0034 ( Enrollment) ( 4 . 47 ) (28.6) R2 = . 9 6, N = 3 5 .

Thus, participation i n Little League i s partly a function o f the available supply of children. By far the more important factor, however, is time. Holding constant the number of students enrolled in kindergarten through the eighth grade, participation in Little League Baseball has been increas­ ing by roughly 44 ,000 members per year.44 This constitutes a remarkable pattern of growth. Such an expansion would appear to have immediate implications for social capital. We emphasize that as participation in Little League and Youth Soccer has grown, so too has parental involvement in the form of coaching, refereeing, and related roles. Lemann emphasizes the activities this entails: " As a long-standing coach . . . I can attest that it involves in­ cessant meetings, phone calls, and activities of a kind that create links between people which ramify, in the manner described by Putnam, into other areas" ( 1 99 6, 2 5 ) Data from the American Youth Soccer Organiza­ tion on players ( in tens of thousands), coaches ( in thousands ) , and referees ( in thousands) from 1 9 64 to 1 9 9 9 underscore the magnitude of this phe­ nomenon.4i Figure 3 · 4 shows that, starting in 1 9 64 with j ust 1 3 5 players, I 8 coaches, and 9 referees, the American Youth Soccer Organization has steadily increased its membership, and as of 1 9 9 8 - 99 boasted over 62o,ooo players, 9 5 ,000 coaches, and almost 4 8 ,ooo referees. Thus, not only are the children of the post-civic generation participating at increasing rates, but their parents are joining in as coaches and referees. As this direct participation expands, so too does the number of other family members who attend soccer games, further enhancing civic engagement. We could obviously elaborate this discussion with numerous other examples, but we think the general point is clear. Voluntary organizations have lives of their own. While some may lose ground over time, new organizational forms arise and fill an overlapping or even slightly differ­ ent niche.46 In other words, there are substitution patterns stemming from competition. .

4 4 - The estimates for an identical analysis with U.S . Youth Soccer participation indicate that, controlling for kindergarten through eighth grade enrollment, membership has in­ creased by about 1 0 ) ,000 per year since the league's formation. 4 5. Between 1 9 64 and 1 9 7 8 membership data were collected every five years; after this date, they were gathered on an annual basis. Data provided by the American Youth Soccer Organization, Hawthorne, CA 9 0 2 5 0 . 4 6 . F o r a n accessible listing of patterns of growth a n d decline among major member­ ship groups in the United States, see Skocpol ( 1 999, appendix 2A).

Civic Virtue in Italy and the United States

95

1 20

1 00

80 �

:;; .n E " :;;;:

'

60

'

40

20

I ,' / �v

0 1 9 64

1 97 3

Fig. 3.4.

--

I

/

'

I

./

/

Players (in ! 0000s) Coaches (in 1 000s) Referees (in 1 000s)

1 9 76

1 970

1 9 67

/

/'

'

1 979

1 994

1 98 8

1982 1 985

1 99 1

1 997

Membership in American youth soccer, 1 964-99

In case after case where a group that's been important in the past now finds itself losing ground, or at least struggling to maintain its place, inves­ tigation shows that the main cause is simply strong competition. The PTA has been getting beat by local entrepreneurs who are more concerned with "hometown " than with Chicago headquarters. The old mainline churches are getting beat by all sorts of religious newcomers . And the Elks and Masons are losing out to the Sierra Club. ( Ladd 1999, 5 2 )

O f course, we can always selectively identify particular organizations that are experiencing declines in support. But it is critical not to focus primar­ ily on such cases at the expense of other organizations that do not fit some predetermined trend. The fact that particular organizations may lose membership in a given period does not in itself mean that member­ ship organizations in general are on the wane. And indeed when we do cast the net a little more broadly, it is hard to come by good evidence that Americans are decreasingly a nation of j oiners.

Con clusions

In this chapter we have examined Making Democracy Work and Bowling Alone, two studies designed to contribute to the study of political cultures by addressing the political consequences of civic community. The former

96

B E F O RE N O RM S

claims that the effectiveness of regional governments in Italy hinges on patterns of civic engagement that can be traced back to a quite distant past. The latter focuses on trends in civic engagement over the more recent past (i.e., the last five decades ) to conclude that Americans are becoming increasingly isolated from each other and hence are now less virtuous citizens. While there is a disj uncture between the two studies over the durability of social capital, Putnam insists in both studies that civic virtue impinges directly on political performance. We have shown in this chapter that the evidence for these arguments is weak. In the case of Making Democracy Work ( 1 9 9 3 ) , the statistical results hinge on restrictive and unreasonable assumptions. Indeed, the " tests " are not really tests because the assumptions preclude all but the reported outcomes. When we adopt a more standard approach to the analysis, there is simply no basis for the claim that regional differences in the performance of Italian regional governments stem from long-standing regional differ­ ences in civic virtue. The further claim that regional differences in commu­ nity involvement reach well back to early modern Italian history is simi­ larly belied by the record. Even painting with a broad brush, there is no sustained evidence either that northern Italians have been distinctively virtuous citizens or that they have consistently outpaced the residents of southern Italy in this respect. We have also shown that the evidence offered in Bowling Alone is less complete than it might first appear. The basic issue here is that informa­ tion is marshaled selectively. For example, the claim that levels of interper­ sonal trust have been declining in recent years stems from an analysis of responses to a single survey item. When we broaden the scope to incor­ porate other survey questions addressing trust, convincing support for the reported decline is more difficult to find. Similarly, using the best available data, we find no sustained evidence of a decay in overall levels of group membership. This is also the case when we examine particular forms of participation highlighted by Putnam. Despite his claims to the contrary, religious participation has not been waning in recent years: indeed, church membership has grown remarkably since the time of de Tocqueville. Further, the connection between religious participation and social capital is far from self-evident. To be sure, PTA membership has dropped in recent years as more women have entered the labor force, and league bowling has become less popular than it once was. But such de­ clines have been accompanied by growth in other areas, including envi­ ronmental groups and parent-organized youth groups. Bowling Alone constitutes a striking metaphor, but the underlying empirical claims are wide of the mark.

Civic Virtue in Italy and the United States

97

In view of this, we conclude that the two most visible recent country studies fail to make the case for political culture. Along the way, we have shown that there is little evidence either for enduring regional patterns of civic virtue in Italy or for the proposition that recent years have seen a pronounced erosion of community values in the United States. Italy and the United States are key cases within which to address these issues, of course, and focusing on these settings in more detail has enabled us to address elements of the problem at a more disaggregated level than would otherwise be possible. Nonetheless, these two countries hardly exhaust the possible varieties of political cultures. Accordingly, we turn now in chapter 4 to an evaluation of claims about the impact of culture and values in a broader comparative framework.

FOU R

Civic Virtue, G rowth, and Democratization What, then happened to great civilizations like the Renaissance in Flor­ ence ? The Florentines lost interest in achievement. Their dreams changed. They became more concerned with love and friendship, with art, with power struggles. David C. McClelland ( 1 9 6 1 , 4 3 7 ) Neither rapid economic growth nor effective government can develop, or, if introduced, will be long sustained, without the widespread diffusion in the rank and file of the population of those qualities we have identified as those of the modern man. In the conditions of the contemporary world, the qualities of individual modernity are not a luxury, they are a necessity. They arc not a marginal gain, derived from the process of institutional modernization, but are rather a precondition for the long-term success of those institutions. Alex Inkeles and David Horton Smith ( 1 974, 3 1 5 - I 6)

e turn now to a different setting for further empirical evaluation of the cultural argument. Where chapter 3 addressed patterns within each of two particular national contexts, our attention now shifts to national-level variations across a broader range of countries. Such varia­ tions, of course, were at the heart of earlier studies that sought to explain the impact of cultural values on the economic and political performance of societies as a whole. McClelland ( 1 9 6 1 ) thus concluded that national differences in rates of economic growth are in good part a function of country differences in the diffusion of achievement-oriented values. In a minor variant on the theme, Hagen claimed that the Industrial Revolu-

Civic Virtue, Growth, and Democratization

99

tion occurred first in Britain " because British people were inwardly differ­ ent from those on the Continent" ( 1 9 67, 3 7) . Similarly, Almond and Verba ( 1 9 6 3 ) argued that the prevalence of participatory orientations within societies is a key factor in helping to understand national differ­ ences in democratic performance. Perhaps the best-known current empirical work along these lines would be the studies carried out by Ronald Inglehart and his associates. These have involved the collection of extensive survey data on mass values in several countries, a major undertaking. The surveys themselves initially focused on European societies (the Eurobarometer studies), but in more recent years the net has been cast more widely in the World Values Surveys. It was on the basis of these surveys that Inglehart ( 1 9 8 8 ) announced the " renaissance " of political culture. This research program has advanced a series of claims. For example, Inglehart has argued in a variety of settings that mass values have moved in a postmaterialist direction in the last few decades, a proposition that has generated considerable discussion. More central to our purposes, however, is the broader contention that values directly affect economic and political performance. It is this contention that places the current analyses squarely in the tradition of the earlier studies by McClelland ( 1 9 6 1 ) and Almond and Verba ( 1 9 6 3 ) . This chapter evaluates recent attempts t o link cultural values t o eco­ nomic growth and democratic performance. We show that these efforts are plagued by a variety of problems. Among other things, different analyses employ idiosyncratic measures of culture, those measures can seldom j ustify the interpretation assigned to them, and research designs are plagued by problems of temporal ordering where " explanatory " vari­ ables too often postdate the phenomena they are invoked to explain.

Values an d Econom i c G rowth

Following very much in the footsteps of Weber and McClelland, Ingle­ hart has argued there is considerable evidence from the recent past that values profoundly affect economic growth. Specifically, he concludes that " economic development itself is influenced by cultural variables . . . . The available evidence supports Weber's insight that culture is not just a consequence of economics; it can shape the basic nature of economic and political life " ( 1 990, 6 5 ) . A s substantiating evidence, Inglehart compared the modest correlation

B E F O RE NOKMS

IOO

o f - . 2 2 between the level o f economic development and rates o f eco­ nomic growth with a " far stronger " relationship between his measure of culture (as postmaterialism, r 9 8 o ) and economic growth during the pe­ riod 1 9 6 5 through 1 9 8 4 (r - . 5 4 ) . He further compared an insignifi­ cant partial correlation of - . 1 3 between wealth and growth, with post­ materialism held constant, to a significant partial correlation of - . 5 2 between postmaterialism and growth, with wealth held constant. Ingle­ hart inferred from this that while a country's wealth may affect the rate at which its economy grows, any such effect is indirect, and that wealth affects growth " only insofar as it brings cultural change" ( 1 990, 64 ) . 1 Cultural factors are thus assigned a crucial explanatory role. An odd feature of his analysis is that this reported effect hinges on an association between ( r ) a measure of culture circa 1 9 80 and ( 2 ) a mea­ sure of economic growth that covers the twenty years from 1 9 6 5 to 1 9 84 . This means that fully three-quarters o f the growth reflected i n this mea­ sure occurred before measurement of the cultural values was taken. In other words, the measure Inglehart uses to predict growth largely post­ diets growth, which is empirically awkward. A more prudent evaluation of his argument comes from asking whether the configuration of mass value priorities prevailing in the early 1 9 8os is systematically related to economic growth rates in the following period. To gain a preliminary sense of the evidence, we therefore regress r 9 8o- 2ooo GDP per capita growth rates on the percentage endorsing Materialist values, circa 1 9 80, controlling for the natural logarithm of 1 9 80 real GDP per capita. 2 The estimates we obtain are =

Growth 1 9 8 0-2000 = - 0 . 1 9 + 0.00 ( % Materialist8 0 ) + 0 . 02 (GDP 8 0 ) ( 1 .6) (4.3 ) N = 1 7,

where t-ratios are reported in parentheses. Clearly, these figures offer scant support for Inglehart's conclusion. Employing his basic model, but with a more defensible temporal ordering of the variables, we see that the estimate for proportion Materialist is statistically insignificant by conven­ tional criteria, while the estimate for the level of national wealth is highly significant. Of course, this is hardly a conclusive test, since it reflects the most rudimentary of models. r. A more common interpretation, given endogenous-growth models of economic growth, is that a control for initial wealth is necessary to allow for the effects of conditional convergence on growth (see, e.g., Barro 1 9 9 7 ) . 2 . Data o n the proportion materialist are from lnglehart ( 1 9 9 7 ) , a n d G O P data are from Heston, Summers, and Aten ( 2 0 0 2 ) . We employ a robust regression estimator.

Civic Virtue, Growth, and Democratization

ror

Against this backdrop, the analyses subsequently reported b y Granato, Inglehart, and Leblang ( 1 9 9 6 ) appear to represent a major improvement. Especially notable is their explicit evaluation of the cultural explanation against a maj or rival, as represented by endogenous growth models of scholars like Barro ( 1 99 1 , 1 9 9 7 ) and Levine and Renelt ( 1 99 2 ) . These models regress economic growth rates over a given period on a set of economic, human capital, and other variables measured at the beginning of that period. It is in the context of such models that Granato, Inglehart, and Leblang report a significant, independent effect of culture on growth. Their analy­ ses expand and modify Inglehart ( 1 990) in the following way. Where the earlier discussion of links between culture and economic growth concen­ trated largely on the role of postmaterialist values, Granato, Inglehart, and Leblang focus on the role of both postmaterialist values and achievement orientations. In considering the latter, their analyses can be seen as an attempt to blend Inglehart's ( 1 990) discussion with the issues that moti­ vated McClelland ( r 9 6 r ) . Their empirical estimates lead them to conclude that while postmaterialism has no systematic effect on growth (consistent with the estimates j ust reported), achievement orientations do play a cru­ cial role. Thus, while Granato, Inglehart, and Leblang's empirical focus shifts notably from Tnglehart ( 1 990), they converge on the same basic conclusion: culture matters. Is this a defensible judgment ? Granato, Inglehart, and Leblang's dependent variable covers the years from 1 9 60 through 1 9 89 , following Levine and Renelt ( 1 9 9 2 ) . Their data on initial GDP and on primary and secondary school enrollment ratios are for 1 9 60 ( " the beginning of the time period," as Granato, Inglehart, and Leblang properly emphasize) , while their data on investment rates are averaged over the years from 1 9 60 to 1 9 8 9 . In stark contrast, data for the two measured components of cultural values are drawn from the World Values Survey for r99 0 . That these variables are measured for 1 990 may not be obvious since it is reported only in the notes to Granato, Inglehart, and Leblang's appendix table 1 : hence, the point may escape those readers who make the usually plausible assumption that all the explanatory vari­ ables temporally predate the dependent variable. 3 Thus, a major problem of temporal ordering appears to carry over from Inglehart ( 1 9 9 0 ) . Perhaps labeling Culture, 99 0 as a " s ignificant predictor" of Growth, 9 60_8 9 (as do Granato, Tnglehart, and Leblang) is less problematic than we 3 · A series of World Values Surveys was also carried out between r 9 8 r and r 9 8 3 , but this earlier series did not include 7 of the 2 5 countries analyzed by Granato, lnglehart, and Leblang (see Abramson and lnglehart r99 5 , 9 7 - 9 9 ) . The 7 countries not surveyed in r 9 8 r through T 9 8 3 are Austria, Brazil, China, India, Nigeria, Switzerland, and Turkey.

102

B E F O RE NOKMS

believe. Indeed, were there evidence that the values at hand are largely time-invariant, the use of Culture, 99 0 as an approximate proxy for Cul­ ture, 9 60 might be warranted. However, there is considerable evidence to indicate that these values can vary substantially over time, as a function of both long- and short-term patterns ( see Clarke and Dutt 1 9 9 1 ; Clarke, Dutt, and Kornberg 1 9 9 3 ; Duch and Taylor 1 99 3 ; Abramson and Ingle­ hart 1 99 5 ; Clarke, Dutt, and Rapkin 1 9 9 7 ) . Granato, Inglehart, and Leblang themselves emphasize that " central elements of culture can and do change . " If so, Culture, 99 0 is an inadequate proxy for Culture, 9 60. A trait measured at time, cannot be used to explain an " outcome" measured at time, _ , . Fortunately, there is another way to evaluate Granato, Tnglehart, and Leblang's argument. McClelland ( 1 9 6 1 , 9 0 ) provided data on national levels of need for achievement collected for 19 50. These data are based on codings of themes emphasized in textbooks read by children from second­ through fourth-grade levels and are fully discussed by McClelland ( 1 9 6 1 , chap. 3 ) . The 1 9 5 0 date for these figures is almost ideal for present pur­ poses, as the children exposed to these readers circa 19 50 would be adults and at least potential labor force members by 1 9 60, and they would play an increasingly important role in economic patterns over the ensuing years ( 1 9 60 to 1 9 8 9 ) . McClelland's 1 9 5 0 data are available for 19 of the 2 5 countries examined by Granato, lnglehart, and Leblang.4 Accordingly, table 4 . 1 displays the estimates obtained when economic growth rates ( 1 9 60- 8 9 ) are regressed on GDP per capita and school enrollment ratios ( r 9 6o), domestic investment ( 1 9 60 - 8 9 ) , and McClelland's measure of need for achievement ( 1 9 5 0 ) . The first column shows the OLS estimates, while the second contains the robust estimates . 5 These estimates employ

4· McClelland also reported need for achievement estimates for r 9 2 5 . In fact, he was especially attentive to issues of temporal ordering. Consider the following: The estimates of n Achievement [ r 9 2 5 ] are positively correlated with subsequent economic growth and very significantly so . . . . On the other hand, n Achievement level as estimated from the r 9 5 0 readers is not related to previous economic growth. The difference in the two sets of correlations is particularly important theoretically because it hears on the issue of economic determinism . . . . 1v1arx appears to have been somewhat premature in dismissing psychology as a major determinant in his­ tory. (McClelland r 9 6 r , 9 3 ; emphasis in original ) Even s o , his conclusions a r c difficult t o sustain in light of analyses like Barro ( r 997 ) . Among other things, McClelland relied heavily on zero-order correlations. 5· The robust regression procedures we use are discussed by Berk ( r 990), and the estimators are generated by the Huber ( r 9 8 r ) procedure. The estimated standard errors these robust procedures generate are of course based on asymptotic theory. In view of this, and given the small number of cases in the analyses, our inferences rely less on the standard

Civic Virtue, Growth, and Democratization

103

TAB LE 4. 1 . Regressions o f G rowth on Initial G D P, School Enrollment Ratios, I nvestment, and Need for Achievement Levels, 1 9 50 Economic Growth, 1 960-89

Dependent Variable GDP per capita, 1 960 Primary education, 1 960 Secondary education, 1 9 6 0 Investment, 1 960-89 Need for achievement, 1 950 Constant

R2

(a)

(b)

- .3 5 " ( 3 .1 ) 1 .6 6 ( 1 .6 ) 2.58" ( 3 .5 ) 9.44'" ( 2.7) - .0 6 ( 0.2 ) - . 65 ( 0.5 ) .72

- .38"" (4.6 ) 1 .2 1 ( 1 .6 ) 2.2 9 "" (4.3 ) 1 0.83 ''" (4.2 ) - .05 ( 0.2 ) - .25 ( 0.3 )

n/a

Source: Data arc from Granato, Inglchan, and Lcblang ( 1 996, appen­ dix table 1 ) , Leblang's website (sobek.colorado. edu/-- leblang), and Mc­ Clelland ( 1 9 6 1 , 9 0 ) . Note: Column ( a ) = O L S estimates, colwnn (b) = Robust regression estimates. (Coefficients [t-ratios]; N � 1 9 ) >:·Coefficients have t-ratios > 2.0.

Granato, Inglehart, and Leblang's data, and the regression model matches closely their table T, model 3 , except that the N is T 9 rather than 2 5 . Comparing across columns, i t i s evident that the estimates are quite robust and that the core endogenous growth model performs well, even with the reduced number of cases. Only the coefficient estimates for primary school enrollments are of borderline statistical significance (with t-ratios j ust above I . 5 ), although they remain substantively significant in the sense that completion of primary school is a prerequisite for secon­ dary school enrollment ( Barra 1 99 7, 1 9 - 20 ) . Second, and most impor­ tant, there is no evidence here of a need for achievement effect: the relevant coefficient estimates are incorrectly signed, and the t-ratios hover below o . 5. When achievement motivation is thus analyzed on the basis of a defensible temporal ordering of the measured variables, there is no evidence that it influences economic growth one way or the other. Given this noneffect, how can we account for the connection between growth from 1 9 60 through 1 9 8 9 and values for 1 9 9 0 , as described by errors and are informed primarily instead on comparisons of shifts in the estimated coeffi­ cients across estimating procedures. All calculations were made with the STATA statistical package.

I 04

B E F O RE NOKMS

Granato, Inglehart, and Leblang ? This question i s best addressed i n terms of the alternative perspective we have developed that draws on analyses like Hirschman ( r 9 8 4 ) and Gambetta ( r 99 3 ) and that endogenizes val­ ues. Recall that in Granato, Inglehart, and Leblang's analysis, achieve­ ment motivation is gauged from a question asking respondents to rank " qualities which children can be encouraged to learn at home. " It is said to increase directly with the proportion stressing " thrift, saving money and things, " and " determination, " while it also increases inversely with the proportion emphasizing the traditional social values of " obedience " and " religious faith . " Our hypotheses are straightforward. First, economic growth generates higher levels of achievement motivation, thus defined, because it creates a climate where norms favoring saving and " determination" make sense. In contrast, saving is often suboptimal (and indeed there are fewer incen­ tives for saving) in periods of poor economic performance with their typically higher rates of inflation and unemployment. 6 Second, invest­ ment in human capital generates higher levels of such achievement moti­ vation insofar as it fosters the diffusion of achievement-oriented norms and undermines more " traditional" social values . We anticipate this ef­ fect to be most pronounced with investment in secondary (as opposed to primary) education, which has a clientele closer to adulthood and thus closer to entering the labor force. These two hypotheses are readily evaluated using the data reported by Granato, Inglehart, and Leblang in their appendix table r . Table 4 . 2 displays estimates obtained when achievement motivations from 1 990 are regressed on economic growth rates for r 9 6o through r 9 89, and school enrollment ratios for 1 9 60. The OLS estimates in the first column of the table are consistent with both hypotheses. Economic growth has the most pronounced positive effect and has a t-ratio considerably greater than 2 . 0 . Both of the estimates for school enrollment ratios are positive, but that for primary school enrollment is smaller than its standard error. The t-ratio for the secondary enrollment coefficient is greater than 2 . 0 , and the estimated coefficient itself is four times the size of that for pri­ mary enrollment (these two right-hand variables share a common met­ ric ) . Overall, this simple model fits the data quite well, and the R2 is . 6 r . The second column o f the table displays the robust estimates for the model and shows that the parameter estimates are relatively stable. 6 . Granato, lnglehart, and Leblang recognize the " possibility that economic growth might be conducive to thrift," but assert that the rationale for such an argument is " less obviou s . " The assertion is puzzling.

Civic Virtue, Growth, and Democratization

I05

Before we embrace the proposition that " achievement norms" are the phenomenon generated by the explanatory variables in table 4 . 2, however, one remaining empirical issue warrants clarification. The measure em­ ployed by Granato, Inglehart, and Leblang is generated from aggregate responses to only four of the eleven available items. Specifically, Granato, Inglehart, and Leblang ( I ) sum the percentage of respondents in each coun­ try emphasizing the virtues of " thrift, saving money and things " and " deter­ mination, " ( 2 ) sum the percentage of respondents stressing " obedience " and " religious faith, " and then subtract ( 2 ) from ( I ) . The implicit assump­ tion here, of course, is that these four items constitute a single coherent cluster of attitudes across and within countries, so that a high country score indicates that there are many individuals who attach a high value to thrift and determination and who also place much less weight on obedi­ ence, religious faith, and so forth. How reasonable is this assumption ? The first column of table 4 · 3 displays the loadings from a principal components analysis of the four measures of achievement motivation for the 25 cases examined by Granato, Inglehart, and Leblang. These esti­ mates provide some apparent evidence for a relatively coherent cluster of attitudes, as the first factor has an eigenvalue of 2.4 8 , and no additional factors with eigenvalues greater than 1 . 0 are extracted. The loadings for the first two measures (thrift and determination ) have the same sign (nega­ tive) and are relatively strong ( - . 8 2 and - . 6 I , respectively) . The loadings for the third and fourth measures (religion and obedience) are also in the

TABLE 4.2. Regressions of Need for Achievement, 1 990, on Prior G rowth and Initial School Enrollment Ratios Need for Achievement 1 990 Dependent Variable Economic growth, 1 9 60- 8 9 Primary education, 1 960 Secondary education, 1 9 60 Constant

R'

(a) .22" (4.8 ) .20 ( 0.7) .81" (2.9) - 1 .1 6 '' (4.2 ) .69

(b) . 1 9 '' ( 3 .6 ) .2 1 ( 0.6 ) . 8 8 '' ( 2 .6 ) - 1 .1 4 '' ( 3.4 )

n!a

Source: Data are from Granato, lnglehart and Leblang ( 1996, appendix

table 1 ) and Leblang's Web site (sobek.colorado.edu/ �leblang/). Note: Column (a) = OLS estimates, column ( b ) = Robust regression estimates. (Coefficients [ t-ratios] ; N = 2 5 ) >:· coefficients have t-ratios > 2.0.

ro6

B E F O RE NOKMS

expected direction (positive) and quite high ( . 84 and . 8 5 , i n turn) . More­ over, the substantive interpretation of this factor appears consistent with Granato, Inglehart, and Leblang's argument: a country with a high factor score is one where citizens attach less importance to thrift and determina­ tion and place more weight on religion and obedience. Unfortunately, we find very little evidence of comparable patterns within these same 25 countries. In fact, in only 3 cases (India, Nigeria, and South Africa ) did we find the same pattern. To illustrate, the esti­ mates for Nigeria are shown in the second column and suggest a single­ factor solution. While the loadings are weaker than those found in the aggregate analysis (not surprisingly, given that these involve individual­ level data) , the four items do cluster similarly in the Nigerian and aggre­ gate analyses. For 2 2 of the 2 5 countries, however, there is no such structure across the four items, and in fully r 7 of these 2 5 cases, a two-factor solution is warranted (i.e., there are two components with eigenvalues greater than r .o). The remaining three columns in table 4 · 3 illustrate the diversity of factor solutions. In both France and Mexico, we obtain a two-factor solu­ tion. In the former, thrift and determination cluster into one factor and religion and obedience cluster into another, and the loadings for thrift and determination are of opposite signs! For Mexico, the first factor comprises determination, religion, and obedience (correctly signed), but there is a second that solely consists of thrift. Norway, in contrast, returns a single­ factor solution, but one loading is incorrectly signed (given Granato, Inglehart, and Leblang's argument) , and only one exceeds . 6o. At the individual level, then, these four items fail to cluster into the single coherTABLE 4.3.

Principal Components Analyses of Four Values I ndicators, 1 990

Variable

Aggregate Data

Component Thrift Determination Religion Obedience Eigenvalue

N

Individual Patterns in Selected Countries Nigeria

II

- . 82 - . 61 . 84 . 85 2.48 25

n/a "

France

II - .40 - .70

n/a "·

Mexico II

.16 .09

. 63 . 63

. 66 - . 80

1 .44 1,001

1 .2 6 1,002

. 73 - . 80 .1 6

.07 1 .04

Norway

II .00

.92

II .52

- . 66 .70 . 6S

- .3 5 - .03 - .25

- . 73 .47

1 .3 5 1,531

1 .04

1 .3 3 1,239

Source: Data are from the World Values Survey, 1 9 9 0 (ICPSR study no. 6 1 60 ) . Note: Italicized loadings > . 60 .

n/a "·

.55

�C Only components with eigenvalues > 1.0 displayed. Loadings for France and 11exico from varimax rotation procedure.

Civic Virtue, Growth, and Democratization

I 07

ent dimension anticipated by Granato, Inglehart, and Leblang for almost all of the cases that they consideo Finally, recall that the 4 items examined in table 4 · 3 are a subset of I I items that could have been employed. The principles guiding Granato, Inglehart, and Leblang's selection of this subset are, however, unclear. Why, for example, is " determination and perseverance" included as part of the achievement syndrome when " hard work " and " feeling of responsi­ bility " are not, when Granato, Inglehart, and Leblang themselves empha­ size that " hard work " is " conducive to economic growth " ( 1 9 9 6, 6 1 3 ) ? Although we do not display the estimates here, a component analysis of all I I (aggregated) items across all 25 cases generates three factors. Fur­ ther, the loadings have " hard work" on the second component (along with " imagination" and "tolerance " ) and thus counterintuitively distin­ guish " hard work" from " determination. " The corresponding compo­ nent analyses of all I I items within each of the 2 5 countries generate a miscellany of solutions and are thus substantively uninterpretable. As we have emphasized, the core of the political culture account ad­ dresses the prevalence of value clusters within countries, clusters that assume political and social meaning insofar as they are widely shared across individuals. There is simply no evidence of any such coherent clustering in Granato, Tnglehart, and Leblang' s data to warrant either their narrower claims about achievement motivation or their broader claims about political culture. Given this, Granato, Inglehart, and Leblang's argument that culture drives economic growth is unconvincing, and their empirical analyses are fatally flawed. Specifically, ( I ) the measure of their principal explanatory variable constitutes an ex post " predictor" ; ( 2 ) an ex ante measure of achievement motivation has no systematic effect on subsequent economic growth; ( 3 ) evidence suggests that their measure of culture is more profit­ ably cast as an outcome of economic growth; and ( 4 ) the meaning of that measure of culture is obscure ( Granato, Inglehart, and Leblang to the contrary, it certainly cannot be interpreted as a unidimensional gauge of achievement motivations) . Consider now another attempt t o link culture t o economic growth of­ fered by Swank ( I 99 6 ) . Swank agrees with Inglehart and his collaborators that culture affects growth. However, he proposes that the relevant cul­ tural values involve " communitarian " norms, as opposed to either postma­ terialist or achievement norms. This clearly represents an abrupt shift: 7 . Parallel inconsistencies i n the measurement of cultural values between aggregate and individual levels for a set of Central American countries are reported by Seligson (200 2 ) .

ro8

B E F O RE NOKMS

where achievement norms center on individualistic orientations, com­ munitarian orientations would seem to involve more civic values that incor­ porate norms of altruism, cooperation, and the like. 8 Insofar as these are norms that help individuals avoid collective action problems ( Swank r 9 9 6, 6 7 1 ), they involve the norms of communities rather than of individuals. Swank's analysis departs from those of his predecessors in another way. Where those previous analyses employed survey data from a variety of countries to generate direct national-level scores for different norms, Swank adopts a much less direct approach. Specifically, he proposes that communitarianism has two elements: " corporatism" and " Confucian­ ism . " While opining that culture is "multifaceted," he actually relies on two simple dummy variables to reflect the presence of communitarian values in either a corporatist or a Confucian guise. However, Swank's classification of countries has a decidedly ex post flavor, one that is typical of the tendency toward ex post explanation in the cultural account that we have already noted. Consider his treatment of " Confucianism, " which as it turns out is simply represented by a dummy variable that distinguishes China, Japan, and Korea from the other countries in his analysis. Leave to one side the fact that these three cases are not equally Confucian. It is noteworthy that Confucian values have often been seen as inimical to growth. Weber, of course, went to great lengths to distinguish Puritanism from Confucianism: " Confucian rationalism meant rational adjustment to the world; Puritan rationalism meant rational mastery of the world" ( r 9 5 r , 24 8 ) . Indeed, to provide further substantiation of his argument in The Protestant Ethic, he de­ voted the entire last chapter of The Religion of China to the contrast between the two worldviews . The theme is echoed in more recent treat­ ments. Thus, according to Pye: The Confucian ideal was eminently appropriate for an agrarian society but was detrimental to the development of commerce and industry. Eventually the Confucian tradition of distrust for all that might unsettle the agricultural order worked against the Chinese in their confronta­ tion with the modern industrial and technologically oriented West. ( r 9 9 r b, 3 4 )

A major difficulty with this description i s that i t squares poorly with the strong economic performance of many East Asian economies, the so-called 8. The shift from postmaterialism to communitarianism is perhaps less abrupt, given Tnglehart's claim that postmaterialists display a heightened concern with community.

Civic Virtue, Growth, and Democratization

1 09

tigers, from 1 9 60 to about 1990. There are, of course, different ways of accommodating this new information, One might, for example, conclude that there is no causal link between values and economic growth. Another tack, perhaps more consistent with Festinger ( 1 9 5 7 ) , involves an about­ face on the content of Confucianism to render the latter consistent with the pattern of economic growth already observed. Such is the strategy pro­ posed by pundits like Kahn and implicitly endorsed by Swank: Most readers . . . are familiar with the argument of Max Weber that the Protestant ethic was extremely useful in promoting the rise and spread of modernization. Most readers, however, will be much less familiar with the notion that has gradually emerged in the last two decades that societies based upon the Confucian ethic may in many ways be superior to the West in the pursuit of industrialization, affluence, and moderniza­ tion. (Kahn 1979, 1 2 1; emphasis added)

The " thing" labeled Confucianism is now refurbished as a pro-growth value system, and the East Asian economic "miracle" is thereby ex­ plained. The ex post reasoning here is remarkable.9 What can we learn from Swank's empirical analyses? Table 4 ·4 displays four sets of OLS estimates. The figures in column a contain the estimates for the baseline model of economic growth with which Granato, lnglehart, and Leblang began their analysis. Column b shows the estimates obtained when the two dummy variables for Corporatism and Confucianism pro­ posed by Swank are added to the model; they are similar to those he reports in the third column of his table I, except that they include insignificant parameter estimates for secondary enrollment ratios and investment. Com­ paring the two columns, we see that the coefficients for the two dummy variables are statistically significant and that their addition to the model 9· More recently, and to accommodate the Fast Asian economic crisis of the late T990s, Pye has suggested that the effects of values on economic performance hinge on the circum­ stances of the day. At the same time, he concludes that several of the affected economies have recovered rapidly, a fact that " reflects in part the same cultural factors that contributed to the rapid growth of recent decade s " ( 2ooo, 2 5 5 ) . This conclusion is, of course, at odds with his claim (quoted earlier) of a decade ago (Pye 1 9 9 r b ) , and again the reversal appears to be an ex post adjustment to recent patterns of growth . While there is now less reference to the Asian tigers, the label itself has migrated as Ireland has assumed the mantle of the Celtic tiger (e.g., O 'Hearn T 99 8 ) . Another discussion with much the same tenor centered o n alleged growth-enhancing qualities of Hindu asceticism over forty years ago (sec Singer 1 9 5 6; Goheen et a!. 1 9 5 8 ), a view that stands in stark contrast to Weber's ( 1 9 5 8 b ) own treatment of the topic. That we have heard less of this particular claim more recently doubtless reflects India's weaker economic performance in the ensuing years.

IIO

B E F O RE NOKMS

TABLE 4.4. Regressions of G rowth on Initial G D P, School Enrol l ment Ratios, I nvestment, and Other Sundry C haracteristics Economic Growth, 1 9 60- 8 9 Dependent Variable G D P p e r capita, 1 960 Primary education, 1 9 6 0 Secondary education, 1 9 60 Investment, 1 960- 8 9

(a)

(b)

(c)

(d)

- . 63 » (4 .4 ) 2.69 ''" (2.2) 3.27" (3.2) 8.69 (1.8)

- .30» (2.7) 3 . 3 0 "' (3.9) .33 (0.4) - 1 .48 (0.4) .69 (1.8) 2 . 9 8 '' (5.7)

-.61» (3.8) 2 .42 (1.8) 3 .40* (3.2) 1 0 . 04 (1.8) - .26 (0.5)

- . 70 ( 0.71 .63

.49 (0.7) .87

-.83 (0.7) . 64

- .2 9 "" (3.3) 1 .64 "' (2.0) .45 (0.7) 2 .46 ( 0.7) .34 (1.1) 2 .40 " ( 5 .4 ) - 1 .3 4 '' (3.5) 1 .5 0 '' (2 . 3 ) .92

"" Corporatism" " Confucianism" " Failed Protestant colonies " Constant

R2

Source: Data are from Granato, lnglehart, and Leblang ( 1 996, appendix table 1), Leblang's website

(sobek.colorado.edu/ � leblang/. Accessed May 2003), and from Swank ( 1 99 6 ) . Note: Coefficients [t-ratios]; N = 25. �·coefficients have t-ratios > 2.0.

improves its fit substantially (the R" increases from .63 to . 8 7 ) . This would appear strongly consistent with Swank' s conclusions. Recall that he claims that both Corporatism and Confucianism have separate, additive effects on growth. As shown in column c of table 4 . 4 , however, w e find n o evidence of a Corporatism effect o n growth when the coefficient for Confucianism is constrained to zero: indeed, the appli­ cable estimate is incorrectly signed and smaller than its standard error. A similar set of estimates is obtained when the set of Confucian economies is excluded from the analysis and the model in column c reestimated (figures not shown ) . Swank's estimates thus appear to be less robust than he believes, and a comparison of columns b and c in table 4·4 reveals that " Confucianism" is the deus ex machina of his analysis . Further perspective on the issue comes from an additional variable. As Swank observes, the 25 economies under consideration encompass some non-Western economies. Among these are several former British colonies, within which there were considerable Protestant missionary efforts dur­ ing the nineteenth and early twentieth centuries. ' 0 While some have arro. The literature on these activities is extensive. See, e.g., Ajayi ( r 9 6 5 ) , Du Plessis ( r 9 r r ), Duvall ( r 9 2 8 ) , Messmore ( r 9 0 3 ) , Mitchell ( r 8 9 9 ) , Ross ( r 9 8 6 ) , and Stewart ( r 89 9 ) .

Civic Virtue, Growth, and Democratization

III

gued that missionary activities were on balance growth-enhancing (e.g., Willems 19 5 5 ; Cole I 9 6 I ), these particular efforts were ultimately unsuc­ cessful, as most people in the colonies involved did not convert, thereby rej ecting the entrepreneurial values highlighted by Weber. We would on the average thus expect lower growth rates in these economies. Column d of table 4 · 4 reports the OLS estimates for a model contain­ ing a dummy variable that identifies these " failed Protestant colonies. " Comparing these figures with those in column b shows that " Confucian­ ism" still exerts a marked effect while the coefficient for " failed Protes­ tant colonies" is also correctly signed (negative) with a t-ratio of 3 · 5 · The fit for the model as a whole increases from . 8 7 to . 9 2 , a statistically significant improvement. Following Swank's logic, we might therefore conclude that both factors are important to a complete explanation of economic growth. Caution is in order, however. The three countries identified by Swank ( China, Japan, and South Korea) are indeed often labeled " Confucian, " even though Japan and Korea are not predominantly Confucian (e.g., Pye I 9 8 5 ) . It is also true that the Protestant missions generally failed in the three countries we have identified (India, Nigeria, and South Africa), although Christianity in a more general sense has been successful in South Africa. r r However, these two binary variables are linked in a more funda­ mental manner. The so-called Confucianism dummy actually identifies the three cases with the three highest scores on the dependent variable, namely, economic growth for I 9 6o through I 9 89 . Conversely, the dummy variable we have labeled " failed Protestant colonies" actually designates the three cases with the lowest rates of economic growth. These two " explanatory" variables are thus simply recoded versions of

the dependent variable.

12

Swank's analysis brings new meaning to the problem of selecting on extreme values of the dependent variable. The difficulty with selecting observations for analysis on the basis of such values is well-known (e.g., Geddes 1 99 1 ) . Here, we are asked to go the next step. Identify a subset of such cases and give this subset a name (any name will do) . Create a binary variable that equals one for cases in this subset, and zero otherwise, and r I . The three countries we have identified are unusual in another sense: they are the only three cases for which the components of need for achievement used by Granato, Inglehart, and Leblang cluster together in the manner they anticipated, as we indicated earlier. r 2 . This fact is evident in Granato, lnglehart, and Leblang's figure r ( r 9 9 6, 6 r 2 ) , which plots economic growth against their achievement motivation index. In this figure, China, Japan, and Korea stand out as a distinct cluster in the northeast corner, as do India, Nigeria, and South Africa in the southwest corner.

II2

B E F O RE NOKMS

include i t on the right a s a n explanatory variable. Results are guaranteed, and indeed it is not surprising that " Confucianism" drives Swank's em­ pirical analysis. Whether or not these " Confucian " economies comprise the Elect in a Calvinist sense, then, they certainly are the Select. Our addition of a parameter for " failed Protestant colonies " (for want of a better term ) serves simply to round out Swank's analysis and further improve the fit by taking care of those cases with low scores on the dependent variable. Under no circumstances, of course, do such efforts constitute an explanation: economic growth is not usefully explained by itself. Instead, Swank's analysis simply typifies the ex post tendencies typical of so many cultural explanations of economic development, and it carries them to their inescapable conclusion. Finally, Knack and Keefer ( 1 9 9 7 ) provide perhaps the most sophisti­ cated test of the cultural approach to growth. They evaluate the effect of two measures of social capital-trust and civic association-on the eco­ nomic performance of twenty-nine market economies during the period 1 9 80 through 1 9 9 2, concluding that both factors make sizable indepen­ dent contributions to economic growth and that these effects are rela­ tively robust to outliers and alternative model specifications. Their contribution is noteworthy on two counts . While they employ the World Values Survey (WVS) for their measures of social capital, unlike Granato, lnglehart, and Leblang, Knack and Keefer are keenly aware of the potential endogeneity problems that arise from using the 1 990 WVS surveys to construct variables that " explain " economic growth over the previous decade (Knack and Keefer 1 99 7 , 1 2 5 9 - 60 ) . Where possible, their measures of trust and civic association are based on responses to the 1 9 8 r WVS surveys, and only as a last resort do they employ the 1990 WVS surveys. Second, Knack and Keefer are sensitive to issues of internal valid­ ity. In contrast to previous studies, they explicitly ground their work on the endogenous growth models of Barro ( T 99 T ) and incorporate controls for initial level of wealth, the price level of investment, and primary and secon­ dary schooling ratios into their analyses. The first two columns of table 4· 5 present the estimates we obtain using data provided by Knack to replicate their table I for the effects of trust and civic cooperation, respectively, on economic growth for 1 9 80 through 1 99 2 (Knack and Keefer 1 9 9 7, 1 26 1 ) . The estimates w e obtain are almost identical to those reported in their table I: social capital, in the apparent form of levels of trust and civic association, has a significant positive effect on growth rates even with controls for initial levels of wealth and human capital. While Knack and Keefer's analysis marks a clear improvement over

Civic Virtue, Growth, and Democratization

rr3

previous studies, we do not believe that it appreciably advances the cul­ tural argument, because it, too, is plagued by measurement problems. Although they minimize the ex post problems of Granato, Inglehart, and Leblang's analysis, their use of multiple indicators in the construction of their measure of civic association is problematic. Specifically, it comprises an index formed from responses to these five questions that ask respon­ dents to rank the degree to which a behavior can " always be j ustified" or " never be j ustified" on ten-point scale (Knack and Keefer 1 9 9 7, 1 2 5 6 ) . a ) claiming government benefits which you are not entitled to b ) avoiding a fare o n public transport c ) cheating on taxes if you have the chance TAB LE 4.5.

Trust, Civic Virtue, and G rowth, 1 980-92 (a)

(b)

(c)

(d)

Real GDP p e r capita 1 9 8 0 (OOOs) log GDP per capita 1 9 8 0

- 0 . 3 6 ''" ( 2 .4 )

- 0.27 ( 1 .9)

- 0.32 ''" ( 2. 1 )

- 0 . 3 3 ':· (2.2)

Price level of investment 19 8 0

- 0 .0 4 ''" (3.8) 6 . 1 9 '' (5.3) 2.19 ( 1 .2)

Primary schooling 1 9 60 Secondary schooling 1 9 6 0

- 0.03 ''" (2.5) 5 . 9 '' (4. 5 ) 3 . 4 6 ''"

Secondary schooling 1 9 8 0 0 . 0 8 '' ( 2 .4)

Civic Constant

R2

- 0 .93 (0.7) .63

- 0 . 0 4 ''" ( 3 .4 )

3 . 74 ( 0 .4 ) 7 . 9 1 ''" (3.5) 0.01 ( 0 .4 )

6.57 (0.8) 8 . 24 ''" (3.2)

(f)

- 1 .9 2 ( 1 .6) - 0 . 04 '" (3.2)

- 1 .9 5 ( 1 .6 ) - 0.0 4 ''" (3.5)

13.32 ( 1 .0 ) 8 . 1 7 '" (3.0) 0.00 (0.2)

16.1 1 (1.3) 8 . 14 '" (2.9)

(2.0)

Primary schooling 1 9 8 0

Trust

- 0.0 4 ''" (3.2)

(e )

0.27 ''" (2.5) - 1 0.47" (2.0) .5 4

0.14

(1.1) - 0.95 ( 0. 1 ) .59

- 8.95 (0.8) .60

- 9 .09 (0.8) .57

0.10 (0.9 ) - 1 5 .44 (1.1) .57

Source: Data for columus (a) a n d (b) were generotJSiy provided by Stephen Knack. In columns (c) through ( f ) we u s e t h e Penn World Tables version 5 . 6 (Summers a n d Heston 1 9 9 1 ) a n d the Barro-Lee data s e t as sources for economic and educational data, respectively (these data are available from the National Bureau for Economic Research). Note: The 1980 secondary schooling figure for Switzerland is missing from the Barro-Lee data, and we use an estimated value of .9. Similar estimates to those reported in columns (c) through (f) are obtained when Switzerland is dropped from the analysis. Barro ( 1 997) models substitute a logged version of Real GDP 1 9 8 0 a n d primary a n d secondary schooling 1 9 8 0 for G D P 1 9 8 0 a n d the 1 960 primary a n d secondary school measures used in the Knack and Keefer models. (Coefficients [t-ratios] ; N � 2 9 ) >:·Coefficients have t-ratios > 2.0, calculated using White-corrected standard errors, following Knack and Keefer ( 1 997).

I I4

B E F O RE NOKMS

d ) keeping money that you have found e) failing to report damage you've done accidentally to parked vehicle

It is striking that Knack and Keefer ( r 997, I 2 5 6) used only responses to these five situations, when the WVS question actually includes responses to twenty-four different situations. Why were " claiming government bene­ fits " and " avoiding a fare" included, while " accepting a bribe " and " throw­ ing away litter in a public place " excluded? Nowhere are the criteria for the selection of the five items discussed in the text, and the reader is simply left to assume that the five items employed to construct the civic measure were the only ones available. Setting aside this selectivity question, consider whether the Knack and Keefer measure truly represents a coherent cluster of values. While we find that the five measures cluster at the aggregate level, the measure is much less coherent at the individual level. Individual-level associations are, of course, typically weaker than those observed at the aggregate level. How­ ever, in two countries (Argentina and Brazil) we found that two of the items were negatively related. In each country, respondents who believed it is justified to " claim government benefits which you are not entitled to " tended also to think that one was never j ustified in "keeping money that you have found. " r3 Coupled with issues in the selection of the five items employed, such sign reversals cast serious doubt on the validity of Knack and Keefer's measure of civic cooperation and hence on the meaning of the estimates in the first two columns of table 4 · 5 · In addition to these measurement problems, Knack and Keefer's analy­ sis employs a peculiar specification of the endogenous growth model. Knack and Keefer ( 1 99 7 ) report that they draw directly on Barro ( 1 99 r ) , and indeed they do use some of the same variables, such as initial level of wealth (following Barra's conditional convergence hypothesis) and pri­ mary and secondary schooling (based on the importance of human capi­ tal in fostering economic growth) . However, the estimates in the first two columns of table 4· 5 diverge from Barra's in a key respect. Instead of using a measure of human capital at the beginning of the relevant I J . With a little ingenuity, one can develop explanations for this sign reversal: perhaps confidence in government institutions is so low that people do not feel the same sense of civic responsibility toward the government as they do toward each other. But then one has to account for the fact that the negative relationship between the two items also implies that people who feel that one is never j ustified in claiming government benefits to which they are not entitled arc actually more likely to believe that one is always justified in keeping money that they have found than are those who feel that they are always justified in claiming government benefits ! It is clearly difficult to avoid ad hoc arguments in explaining patterns like these.

Civic Virtue, Growth, and Democratization

II5

period, as does Barro ( 1 99 1 ) , Knack and Keefer elect without explana­ tion to introduce a twenty-year lag. Specifically, they use primary and secondary school enrollment measured in I 9 6o to explain economic growth from 1 9 8 0 to I 9 9 2 . Unfortunately, their conclusions about so­ cial capital and growth hinge entirely on this idiosyncratic and unex­ plained treatment. This is shown in columns c and d of table 4 . 5 , where we substitute I 9 8o measures of human capital for the I 9 6o measures used in the first two columns, Comparing the two sets of models, initial level of wealth and the price level of investment remain negative and statistically signifi­ cant. However, there is an interesting change in the estimates for human capital. While primary school ratios for 1 9 60 are a significant predictor of growth in the models with trust ( column a) and civic ( column b ) , primary school ratios measured in 1 9 8 0 are not (column c o r d ) . The opposite pattern holds when secondary school enrollment ratios 1 9 8 0 are substituted for the 1 9 60 figures. The specification used in columns c and d is not only closer to Barro's model in form, but it also provides esti­ mates that more closely resemble his. Specifically, as noted earlier in connection with table 4 . 1 , we would expect the effects of primary school ratios on economic growth to be indirect (and hence not significant in a single-equation format) because primary schooling is a prerequisite for secondary schooling, which has a significant, positive effect on economic growth (Barro 1997, 1 9 - 2o ) . r4 The most striking difference between the two sets of models across the first four columns of table 4 · 5 involves the coefficients for social capital. The estimated coefficient for trust plummets to .01 (and is much smaller than its standard error), a reduction of about 8 5 percent from its original value of .o8 ( compare columns a and c ) . This is of special note given the primary emphasis Knack and Keefer assign to trust ( 1 9 9 7, 1 2 5 2- 54 ) . While the decline is less precipitous, the coefficient for civic cooperation also drops substantially and is no longer statistically significant when the measures of human capital are for 1 9 8 0 (compare columns b and d of table 4 . 5 ) . Thus, notwithstanding Knack and Keefer's ( 1 9 9 7, 1 2 6 5 - 66 ) claim about the robustness o f their model, their reported estimates for r 4 - Since primary schooling is a prerequisite for secondary schooling, and given that this is a relatively wealthy sample of countries, there is in fact little meaningful variance in primary school enrollment ratios in either 1 9 60 or 1 9 80, while there is much more variance in the secondary enrollment ratios in both periods. In light of this restricted variance for the primary school data, there is little statistical reason to expect a pronounced effect of this variable on growth or any other outcome, and there is every reason to believe that the Knack-Keefer estimate for this coefficient is far from robust.

II6

B E F O RE NOKMS

social capital are driven entirely by their singular decision to employ a twenty-year lag for their measures of human capital. As a final check on the robustness of these figures, the last two columns of the table display the estimates from a model that, following Barra ( 1 99 7 ) , includes the initial level of wealth in logarithmic, rather than raw score, form. With this slight modification, as one would expect, the magni­ tude of the coefficient for the effect of initial wealth on growth increases substantially, although the t-ratios dip a little below 2.0. Coefficients for the other control variables remain quite similar to those reported in col­ umns c and d. Most important, the coefficients for the social capital vari­ ables are reduced even further below the values reported in the two middle columns of the table. This simply reinforces our basic conclusion that social capital is incidental to the explanation of growth, and that Knack and Keefer's suggestion to the contrary stems solely from their unusual and unexplained use of a twenty-year lag for human capital.

V alues an d Democrati c Perform an ce

We turn now to assess recent studies claiming that civic culture enhances democracy. These parallel the analyses of economic growth that we have j ust considered in many respects, except that they consider a broader array of cultural values. Recall from the last section that the analyses of growth centered on postmaterialism, achievement norms, communitari­ anism, and, to a lesser extent, Protestantism. In contrast, Inglehart identi­ fies the following seven indicators of mass political culture as potentially relevant to democratic performance. r . Levels of overall life satis faction 2. Levels of interpersonal trust 3 · Support for revolutionary change 4· Support for the current social order 5. Levels of political discussion 6. Levels of postmaterialist values 7· Proportion of the population Protestant

We have already stressed the consistent claim in cultural accounts that indicators like these form a coherent cluster. Thus, Inglehart describes the first five of these items as constituting "a syndrome of positive attitudes toward the world in which one lives " ( r 990, 4 r ) . Similarly, postmaterial-

Civic Virtue, Growth, and Democratization

I I7

TAB LE 4.6. Product-M oment Correlations among Seven Components of Political C u lture, ci rca 1 980

1. 2. 3. 4.

Satisfaction Trust % Revolutionary <Jio Conservative 5 . % Materialist 6 . % Women discussing 7 . % Protestant

.55 - .57

.36 - .66 .62 .76

2

3

4

5

6

7

22

18 18

18 18 18

21 19

20 18

24 22 18 18 21 20

- .52 .43 - .43 .45

.61

- .40 .41 - . 45 - .24

17

15

17

15 18

- .3 6 .55

- .6 1

.48

- .68

. 76

Source: Data on the first six indicators are from lnglehart ( 1 990, chap. 1 ) ; information on Protestant­ ism is the natural logarithm of the precentage Protestant scores from Taylor and Hudson ( 1 9 72 ) . Note: Entries below t h e main diagonal a r e pairwise correlation coefficients, a n d entries above the main diagonal indicate the number of cases for which those correlations are calculated .

ists are said to place special weight on quality of life issues and on the sense of community, both of which involve the altruism, cooperative spirit, and general civic-mindedness typically associated with civic virtue. Finally, Protestantism is also described as part of this syndrome ( s o ) . ' 5 I s there evidence that these measures of political culture form an underly­ ing coherent cluster of attitudes, as claimed by Inglehart ? Table 4 . 6 dis­ plays the correlations among the seven indicators of political culture. Be­ cause some of these indicators are available for more countries than others, we report pairwise correlation coefficients below the main diagonal and indicate the number of cases over which the correlations are calculated above the main diagonal . 1 6 Given the coding of the variables, the signs of the coefficients are all as expected. Yet the most striking feature of these correlations is their comparatively modest size, given that they are calcu­ lated for national aggregates. Ignoring signs, the 2r correlations in the table range from . 24 to . 7 6, with a mean value of . 5 2. Only 2 (approxi­ mately ro percent) exceed . 70, which translates into an R2 of . 5 0. This means that only in these 2 pairings of indicators is more than half the variance shared. Across the remaining r 9 pairings, the shared variances are r 5. Measures of the first six indicators of political culture are from variou s tables in Inglehart ( r 990, chap . r) and refer to the early r 9 8os. Following Inglehart ( r 990, s r ), figures on political discussion rates arc for women only. Data on the last indicator, percent­ age Protestant circa r 9 6 5 , are from Taylor and Hudson ( r 9 7 2 ) . Given its distribution, the variable u sed in table 4 . 6 is the natural logarithm of the percentage Protestant (plus r ) . r 6. Despite his claim that these variables form a coherent cluster, Ingle hart does not himself report a correlation matrix (or factor analysis) that includes all the variables shown in table 4 . 6 . A factor analysis of these variables seems unwise in view of ( r ) the lack of structure in the correlations in table 4.6 and ( 2 ) the variation across the items in the incidence of missing data that is evident a hove the main diagonal.

II8

B E F O RE NOKMS

considerably smaller. The modest size o f most o f the correlations in table 4 . 6 challenges the assertion that these indicators are all components of a single enduring and distinctive cluster of cultural traits. An alternative way of gauging whether these indicators form a well­ defined configuration is to see if they are generated by the same underlying economic conditions. Civic virtue should be more widespread within the wealthier democracies, all else equal, for the reasons advanced by Inglehart (among other things, national wealth takes time to accumulate and is therefore a long-term condition that changes comparatively slowly) . By the same token, civic virtue should be less sensitive to shorter-term economic conditions (e.g., economic growth in the recent past) or economic condi­ tions that are more prone to temporal fluctuation (e.g., levels of unemploy­ ment ) . Moreover, the claim that the indicators form a coherent cultural cluster necessarily implies that they should each respond in a similar man­ ner to the same economic conditions. Table 4 · 7 reports the estimates from regressions of the indicators of political culture on economic conditions. These estimates involve sepa­ rate regressions of the first six components from table 4 . 6 (all of which entail values measured in r 9 8o or very shortly thereafter), but exclude the measure of Protestantism, which is for circa r 9 6 5 . The six elements are regressed on national wealth, as indexed by the logarithm of real GDP per capita for r 9 8o; economic growth over the previous decade, as mea­ sured by growth in real GDP per capita ( r 9 7o- 8 o ) ; and the level of unemployment for r 9 8 o . Il The cultural argument, of course, implies that the explanatory variables in table 4· 7 should have similar effects on each of the six component indicators. From the first row of the table, we see that national differences in overall levels of life satisfaction are sensitive to the level of economic development and economic growth over the prior decade but do not vary with levels of unemployment. Specifically, levels of satisfaction are higher in wealthier countries and in those countries that experienced lower growth rates over the previous ten years. This pattern is repeated in the estimates for the percentage materialist. 1 8 In contrast, when we turn to levels of interpersonal trust in the second row of the table, only the coefficient for unemployment is statistically significant. Levels of distrust apparently increase with unemployment but are insensitive to either the I J . Data for levels and growth of real GDP are from Summers and Heston ( r 99 r ) ; those for levels o f unemployment arc from OECD ( r 99 r ) . r 8 . The sign reversals here merely reflect the coding o f the dependent variable (percent­ age materialist minus percentage postmaterialist) , so the coefficients in the fifth row show that postmaterialism increases with level of development and decreases with prior growth.

Civic Virtue, Growth, and Democratization

I I9

TABLE 4.7. Regressions of Com ponents of Political Culture, circa 1 980, on Basic Economic Conditions Dependent Variable

Per Capita GDP 1 9 8 0

GDP GrO\vth

1 970- 80

Unemployment

Constant

R2

N

Satisfaction

1 . 1 9 "" (2.6) 0.73 (0 . 1 ) - 3 .93"" (2.0) 10.80 (1.8) - 1 9. 1 6 "" (2.6) 1 1 .94 ( 1 .2)

-2.96"" (2.2 ) -28.91 (1.1 ) 1 0.20 (1.5) 25.53 ( 1 .2 ) 57.9 8 ''" (2.6 ) - 4 1 .4 1 ( 1 .2 )

- .02 (0.5 ) - 2.46'; (2.6) 0.52'; (2.2) - 1 .84'; (2.5 ) 0.29 (0.4) - 1 .88 ( 1.9)

- 2 . 76 (0.6) 56.65 (0.7) 35.94 (1.8) - 7 1 .53 ( 1 . 2) 1 8 1 .3 3 ''" (2.5 ) - 3 0 .42 (0.3 )

.60

20

.37

18

.58

16

.55

16

.64

19

.46

18

Trust

o/o Revolutionary IJ"o Conservative % Materialist % Women discussing

Source: for the dependent variables, see table 1 . GDP data are from Summers and Ileston ( 1 9 9 1 ) and unemployment data are from OECD ( 1 9 9 1 ) . '"Coefficients have 1-ratios > 2 . 0 . (Coefficients [1-ratios] )

level of development or growth in the recent past. A similar pattern is obtained with the percentage of women discussing politics in the bottom row of the table (its coefficient has a t-ratio of 1 .9 ), suggesting that political interest declines with increasing unemployment. Finally, the two items on support for the existing social order behave in a parallel manner (reflecting in part the fact that they are based on different responses to the same question ) . Support for the current order is higher in wealthier coun­ tries and in those with lower unemployment, but it does not vary with recent growth. ' 9 Overall, the patterns apparent i n table 4 · 7 reinforce the inferences drawn from table 4 . 6. The six " components" of political culture do not form a coherent general structure : instead, they are sensitive to different factors. Even the first three components in the table do not cluster to­ gether, despite Inglehart's ( 1 990, 4 4 ) use of these three as a summary measure of political culture. Moreover, the similarities that are evident in table 4 · 7 are not clearly interpretable: why, for example, should life satisfaction and materialism respond differently to economic conditions than do levels of trust and the percentage of women who discuss politics ? Finally, insofar as growth in the recent past and levels of unemployment reflect shorter-term economic patterns, the fact that these two variables r 9 · The same patterns as those reported in table 4 - 7 are obtained if the measure of economic growth is taken to cover an even shorter period ( r 9 7 5 - 8 o ) that focuses on economic performance after the oil shock of 1974-

120

B E F O RE NOKMS

influence some but not all of the components i s incompatible with the interpretation that the latter represent enduring cultural values.20 What then of Inglehart's claim that culture influences the success of democracy? His argument depends on two pieces of evidence. First, he reports a correlation of . 8 5 between life satisfaction and democracy ( 1 990, 4 2 ) . Second, he asserts that the effects of economic development on democracy are mediated by the proportion of the labor force in the tertiary sector and by " civic culture " (as reflected in levels of life satisfac­ tion and interpersonal trust, and support for revolutionary change) . How­ ever, the meaning one attaches to these associations hinges decisively on how one identifies " s uccessful democracies. " The particular measure used by Tnglehart ( 1990) is a count of the years of continuous democracy from 1 9 00 through 1 9 8 6. Unfortunately, the measure and the research design within which it is employed contain several anomalies. First, as one would correctly expect given the title of his book, the countries analyzed by lnglehart consist almost exclusively of the advanced industrial societies (the notable exceptions in his figure r . 5 are Argentina, Hungary, and South Africa llnglehart 1 990, 4 2] ) . Their levels o f democratic performance a s o f the early t o middle 1 9 8os were, as a result, all roughly equivalent ( see Bollen 1 99 3 ) . However, under this procedure, Belgium, which is coded as continuously demo­ cratic since 1 9 00, is measured as being three times more democratic than France, whose democratic history is coded as beginning in r 9 5 �L The j udgment that Belgium really was approximately three times more demo­ cratic than France by 1 9 8 0 is untenable, given any reasonable set of criteria for democratic performance. Second, lnglehart's use of 1 900 as the starting date ignores the fact that the meaning of democracy has evolved considerably over the past century, especially in terms of what Dahl ( 1 9 7 1 ) has labeled inclusiveness. When 20. Clarke and Dutt ( 1 99 1 ) found a similar problem. The resu lts from their pooled cross-sectional time-series analysis of eight countries over the period 1 9 7 6 through 1 9 8 6 indicate "that sharp increases i n unemployment i n the early 1 9 8os make i t appear, based on the four-item measure, that there was a substantial shift toward postmaterialism in several countrie s . " They conclude that the upward trend stemmed more from "the failure to include an unemployment statement in the measure " than from changes in underlying values ( 1 9 9 1 , 9 1 8 ) . In another pooled analysis of some of these items from eight countries, Clarke, Dutt, and Kornberg ( 199 3 ) provide further evidence of the sensitivity of these items to short-run economic and political circumstances including inflation, unemployment, fluc­ tuations in support for political parties, and rally-around-the-flag events. O uch and Taylor ( 1 9 9 3 ) reach the parallel conclusion that apparent shifts in postmaterialism stem dispropor­ tionately from the inclusion of an inflation item in the index (see also Clarke, Dutt, and Rapkin 1 9 9 7 ) . All of these analyses underscore the sensitivity of lnglehart's measures to short-term rather than long-term economic conditions.

Civic Virtue, Growth, and Democratization

121

one recalls the typical restrictions o n suffrage based o n gender, income, and race a century ago, the implications of this evolution become self­ evident: since 1 9 00, none of the countries so coded by Inglehart has been continuously democratic in a contemporary sense. This problem under­ mines all attempts to identify a binary starting date after which a country can be considered democratic and before which it cannot. 2 ' Third, lnglehart's analysis does not help account for levels of demo­ cratic performance because his research design selects cases according to their (high) values on that dependent variable (see Geddes 1 99 1 ) , Varia­ tions in political culture (or any other potential explanatory variable) cannot logically be used to account for fundamentally uniform degrees of political democracy. That Tnglehart's measure includes more variance does not obviate the problem, It simply means that his measure inadver­ tently conflates democracy with at least one other variable-political stability (Bollen and Jackman 1 9 8 9 ) .22 Given these limitations, the associations reported by lnglehart should most plausibly be taken to reflect covariation between certain cultural values and stability, not between culture and democratic performance. They therefore do not address the issues that have motivated scholars, at least since Almond and Verba, concerning the impact of democratic (i.e., civic) values on democratic performance. Further, these associations can­ not logically sustain the causal argument at stake, because that argument implies that the distribution of values identified in surveys administered in the early 1 9 80s gave rise to the stability of constitutional orders over the preceding eighty-five years. Apparently dissatisfied with the potpourri of cultural attributes ad­ dressed by lnglehart, Muller and Seligson ( 1 9 9 4 ) seek t o narrow the field. Specifically, they take issue with Inglehart's argument that civic cultures 2L That the meaning of democracy has evolved considerably over the last century does not mean that the concept is inherently two-dimensional (lnglehart T997, T 6 6 ) . And even if it were, the subsequent use of a single (unidimensional) measure of democratic stability is unwarranted. 2 2 . Duch and Taylor point to a related problem: " Recent survey evidence from the Soviet Union and Eastern and Central Europe suggests the mass public embraces postmateri­ alist values in spite of the serious economic penury they have endured" ( 1 9 9 3 , 7 5 3 ) . lnglchart explains this apparent anomaly b y noting that "the crucial factor i s security during one's formative years, and it is dear that the communist regimes of Eastern Europe provided a relatively secure existence " (lnglehart T 99 2, T 4 , i n D u c h a n d Taylor T 9 9 3 , 7 5 3 ) . I n making this argument lnglehart confounds economic security and economic develop­ ment. The result is an unfalsifiablc argument: " Economic security drives postmatcrialism, but if we see high levels of postmaterialism and low economic development at the same time, then factors other than economic wealth have promoted economic security " (Duch and Taylor T 9 9 3 , 7 5 3 ) .

I22

B E F O RE NOKMS

i n general influence the process o f democratization. They conclude in­ stead that only one cultural element seems to play a causal role in democ­ ratization, namely, the proportion of the population that prefers gradual social reform. Other cultural variables-such as the level of overall life satisfaction, the level of interpersonal trust, the proportion supporting revolutionary change, and the proportion wishing to defend their " pres­ ent society . . . against all subversive forces " -have no such effects, ac­ cording to Muller and Seligson. They further argue that income inequal­ ity is a key explanatory variable, such that the process of democratization is inhibited where income inequality is high, but we focus primarily on their claims about culture. Muller and Seligson's inferences are based on the estimates from a series of regression analyses that regress degree of democracy in the 1 9 8os on degree of democracy for the 1 9 70s, along with a series of cultural and other characteristics for a sample of 2 5 countries. After deleting a series of statistically insignificant variables on the basis of a stepwise regression analysis, they focus on the " trimmed" set of estimates displayed in the first column of table 4 . 8 . However, a n alternative set o f estimates generated b y median regres­ sion shown in the second column of table 4 . 8 indicates a somewhat different pattern. The estimated coefficients on culture and ethnic frac­ tionalization in column b are about a third smaller than the correspondTABLE 4.8. Estimates of M u l ler and Seligson's Basic M odel Democracy 1 9 8 0s Dependent Variable Democracy 1 970s Income share o f top 2 0 % Belief i n gradualism Ethnic fractionalization Constant

R'

(b)

(a)

.29 I 1 .2 ) - 1 .66 '' (2.1) .39 (0.9) - .12 (0.8) 1 1 1 . 80 I 1.9) .64

.32 '' (2.5) - 1 .60 '' (4 . 1 ) .62 '' (2.8) - . 1 6 '' (2.2) 94.34 '' (3. 1 ) .89

Source: Data from Muller and Seligson ( 1 994) and Taylor and Hud­

son ( 1 972, table 4. 1 5 ) . Note: Column ( a ) � O L S estimates, column ( b ) sion estimates. (Coefficients [t-ratios]; N = 2 5 ) >:· coefficients have !-ratios > 2.0.

�

Median regres­

Civic Virtue, Growth, and Democratization 1 00

(f) 0 co OJ

Ar

•

. nt

60

• •

ro

()) 0

Portug •

• •

Hondur •

>. u 0 0 E

•

Spain •

80

123

Panama •

40

• •

20

0 0

20

40

60

80

1 00

Democracy 1 970s Fig, 4, 1 ,

Democracy 1 980s against Democracy 1 9 70s

ing estimates in column a. These shifts in the coefficients with the median regression procedure are of special interest given that this estimator is less sensitive to peculiarities in the distribution of the dependent variable than is OLS.2l Additionally, two of the four coefficients ( including that on culture) are now smaller than their standard errors, while a third is barely larger than its standard error. Only the coefficient on income inequality retains a plausible t-ratio (although that ratio is about half the size of the comparable t-ratios for the first two columns ) . These shifts in the median regression coefficients suggest that the figures in the first column of table 4 . 8 are less robust than they appeared to Muller and Seligson. To probe the sources of the differences across estimators in table 4 . 8 , w e plot measured democracy i n the r 9 8os against measured democracy in the 1 9 70s. The pattern is displayed in figure 4 . 1 and reveals the prob­ lem starkly. Muller and Seligson proceed as if there are 25 cases in the 2 3 . Median regression (also known as least absolute value regression) fits hy minimiz­ ing the sum of absolute (as opposed to squared) residuals. The resulting estimates arc thus less sensitive than OLS to skewed dependent variables, paralleling the differences between medians and means as measures of central tendency (see Berk 1 9 9 0 ) . All calculations were made with the STATA 7.0 statistical package.

1 24

B E F O RE NOKMS

analysis, and indeed they claim that their sample o f countries " contains reasonably large variation on change in democracy that corresponds to the global trend of the 1 9 70s and r 9 8os" ( 1 994, 64 6 ) . However, fully 2 0 of those 2 5 cases are o n o r around the main diagonal. I n other words, any discernible variance in democratization over the period occurs in only 5 of the cases, a small minority of the total. Further, of the 20 cases on the main diagonal, r 5 are bunched together in the extreme northeast corner of the plot. With its five parameters, the model in table 4 . 8 thus comes very close to fitting an individual parameter for each of the 5 cases that exhibit any meaningful variance in change in democratization and is neither an effective nor an elegant summary of general patterns in the data.24 The consequences of the restricted variance in democratization be­ come clear when the basic model in table 4 . 8 is revised to include country dummy variables for the two cases with the most extreme residuals in figure 4 . 1 , namely, Spain and Argentina (these two cases could equiva­ lently be removed from the analysi s ) . 25 Table 4 · 9 displays the results. Even with the OLS figures in column a, the coefficients on income inequal­ ity and belief in gradualism drop substantially. The median regression estimates in column b are even less ambiguous: the coefficient on lagged democracy is effectively T .o, and all of the remaining three substantive coefficients plunge dramatically in magnitude. The coefficient on income inequality drops from - . 9 1 to - . 2 5 ; that for belief in gradualism goes from .4 2 to . 1 4 ; and the estimated coefficient for ethnic fractionalization is reduced from - . 1 7 to - .07. In other words, with just two cases re­ moved, we see that in this data set democracy in the 1 9 8os is basically a perfect function of democracy in the 1 9 70s and nothing else, consistent with the pattern in figure 4· r . Thus, there is simply no basis for the conclusion that the cultural variables examined by Muller and Seligson, including " belief in gradualism," have any discernible effects on rates of democratization. Indeed, in broader terms, any such effects would be difficult to inter24- One way in which Muller and Seligson evaluate the robu stness of their estimates is by examining the residuals (see their fig. 3 ) . However, because their model is overfitted in the manner j ust described, the residuals arc not especially informative. Given the ovcrfitting and the lack of variance in democratization, we would expect the residuals to appear to be relatively well-behaved, and this is indeed the case. The problem we have just described doubtless accounts for the fact that when we attempted to estimate this model with a robust estimator, the model did not achieve convergence. 2 5 . This, of course, is a standard diagnostic procedure, and the t-ratios for these coun­ try dummy variables are equal to the studentized residuals for the same observations (see, e.g., Roll en and Jackman r990, 2 64 ) .

Civic Virtue, Growth, and Democratization

125

pret, since cultural values are supposed to reflect deep-seated social norms that evolve very slowly over time, if they change at alL We would indeed be j ustly skeptical of the validity of any measure of support for gradual reform that was itself subj ect to pronounced temporal fluctua­ tion. Given this, it is difficult to develop a convincing substantive ratio­ nale for the proposition that entrenched cultural norms could in principle explain both democratic collapses and subsequent democratic restora­ tions in the short term (that is, from the 1 9 70s to the 1 9 8 o s ) . For ex­ ample, the coup that preceded the 1 9 8 2 restoration in Argentina that figures so prominently in Muller and Seligson's analysis occurred only six years earlier. It is implausible to argue that there was a massive shift in cultural values involving beliefs about gradualism, a shift logically im­ plied by their argument, that could account for both events. Similar considerations apply to cases like Chile (not included in Muller and Seligson's analysis ) , where the military j unta lasted a little longer ( sixteen years ) . Since beliefs about gradualism are deep-seated, it is especially awkward in this instance to link nonchanges in a potential explanatory variable with changes in a given outcome ( democratization) over a rela­ tively short period. TABLE 4.9. Estimates for a Modified Specification Using Mu ller and Seligson's Data Democracy 1 9 8 0s Dependent Variable Democracy 1 9 70s Income share of top 2 0 % Belief in gradualism Ethnic fractionalization Spain dummy Argentina dummy Constant

R'

(b)

(a)

. 9 5 ''" (4.0) - .25 (0.4) .14 (0.6) - .07 (0.9) 3 5 .4 8 '' (2.9) 3 5 .46 '' (4.3) 7.42 (0.2) .72

. 5 6 '' ( 3 .0 ) - .9 1 ( 1 .6 ) .42 I 1 .7) - . 1 7 '' (2.3) 1 7 . 74 I 1 .4 ) 1 8 .93 ( 1 .8 ) 55 .02 I 1 .5 ) .91

Source: Data from Muller a n d Seligson ( 1 994) and Taylor and

Iludson ( 1 972, table 4. 1 5 ) . Note: Column ( a ) � O L S estimates, column ( b ) sion estimates. (Coefficients [t-ratios] ; N = 2 .5 ) *Coefficients have t-ratios > 2.0.

�

Median regres­

I26

B E F O RE NOKMS

The most recent example of a n attempt to connect values with demo­ cratic performance is reported by Inglehart ( r 9 9 7 ) . This analysis is note­ worthy on at least four counts. First, it includes data through I 9 9 5 . Second, i t draws o n newer survey data from a considerably expanded set of cases covering more than forty countries. Third, Inglehart examines the impact of cultural factors on both democratic stability and changes in democracy. Fourth, in contrast to the variety of cultural factors that we discussed earlier in this section, the newer analysis centers primarily on two cultural syndromes: levels of overall life satisfaction and levels of interpersonal trust. Of these, Inglehart attaches special significance to the former. Satisfaction with one's life as a whole is far more conducive to political legitimacy than is a favorable opinion of the political system itself . . . . Politics is a peripheral aspect of most people's lives; and satisfaction with this specific domain can rise or fall overnight. But if one feels that one's life as a whole has been going well under democratic institutions, it gives rise to a relatively deep, diffuse, and enduring basis of support for those institutions. Such a regime has built up a capital of mass support that can help the regime weather bad times. Precisely because overall life satisfaction is deep rooted and diffuse, it provides a more stable basis of support for a given regime than does political satisfac­ tion. ( r 997, r 7 6- 7 7 )

Not surprisingly, overall life satisfaction features centrally i n the analysis. In particular, Inglehart reports that life satisfaction is associated with democratic stability, from which he infers that higher levels of such satisfac­ tion enhance the long-term viability of democratic institutions. He further reports that satisfaction is negatively associated with short-run transitions toward democracy. This leads him to a second conclusion that authoritar­ ian regimes become vulnerable to replacement with decreasing levels of mass satisfaction, but that the viability of the resulting ( democratic) re­ placement regime hinges on whether levels of satisfaction increase substan­ tially after the transition. The second conclusion is somewhat unwieldy, given the associated claim that levels of satisfaction exhibit considerable temporal stability. Although it is required by his first inference, this associ­ ated contention logically implies that sufficient increases in mass levels of satisfaction after democratic transitions required to enhance the viability of the new regime are unlikely to occur. Let us therefore briefly consider the two maj or claims. First, the suggestion that life satisfaction impinges on the long-term

Civic Virtue, Growth, and Democratization

1 27

viability of democratic regimes echoes Inglehart's ( 1990) earlier linkage of values with democratic stability and suffers from the same problems. The claim most notably invokes a value measured in 1 990 to " explain" the stability of democratic institutions, almost all of which was experienced over the previous seventy years. While Inglehart offers a j ustification of this usage, it is not a compelling one, because it leaps from limited evi­ dence on temporal stability over an approximately twenty-year period to make the much more extensive claim of temporal stability over the seventy-five years from 1 9 2 0 through 1 99 5 .26 The j ustification is also selective: for example, evidence from panel data using similar items sug­ gests considerable response instability over time at the individual level (see, e.g., Clarke and Dutt 1 9 9 1 ) . 27 Just as achievement norms as of 1 990 cannot be invoked t o explain economic growth over the previous three decades, life satisfaction gauged in 19 9 0 cannot plausibly be summoned as an explanation of democratic stability over the preceding seven decades. Observe further that the newer measure is simply an updated version of Inglehart's index of democratic stability that now covers the years from 1 9 2 0 to 1 9 9 5 and is subject to all of the same problems that we have already addressed. For example, there remains no clear rationale for specifying a precise date on which democracy was initiated. Belgium is thus still cast as more than twice as democratic as France, an indefen­ sible j udgment given any standard criteria. Taken in conjunction with the temporal ordering problem, this means that there is no credible evidence to suggest that the long-term viability of democratic institutions hinges in any way on national levels of overall life satisfaction, or on any other cultural value. The second new broad empirical claim is that low levels of satisfaction undermine authoritarian regimes and thereby enhance the prospects for democratic transitions, at least in the short run. The basis for this conclu­ sion is an analysis of changes in democracy from 1 990 to 199 5, as reflected in changes in national ratings of political rights and political liberties 26. In making his case for temporal stability, lnglehart further observes that the over­ time correlations of about ·9 for national levels of trust and satisfaction are similar to the over-time correlation for levels of GNP per capita. He then analogizes that, since levels of wealth arc "a relatively stable attribute of most societies " ( r 997, r 8 6 ), cultural values must be stable as well. But that even he does not take this claim seriously is evident from the fact that he devotes his subsequent chapter to an analysis of cross-country differences in eco­ nomic growth rates. 27. In addition, our table 4 · 7 shows evidence suggesting that levels of satisfaction arc sensitive to short-term patterns of economic growth in the preceding period. Again, this evidence is inconsistent with the argument that levels of satisfaction are stable over the very long haul.

I28

B E F O RE NOKMS

assigned by Freedom House over these two years. Inglehart ( r 997, 203 ) regresses changes in democracy on levels of satisfaction and trust, the proportion of the work force in the service sector, the proportion of the age-eligible population in higher education, and per capita GNP. The esti­ mates obtained are similar to those shown in column a of table 4 . 1 0 and suggest that of all the explanatory variables considered, only levels of satisfaction have a significant, negative effect on changes in democracy. In contrast to Inglehart's estimates involving levels of democratic sta­ bility, the figures in column a of this table are not plagued by problems of temporal ordering. Even so, they are incomplete on two counts. First, the dependent variable involves change scores that are subject to floor and ceiling effects. For example, the T9 9 0 Freedom House ratings for most West European democracies were at or close to the maximum ( demo­ cratic) value, so that they could not even in principle increase over the next five years. It is customary when analyzing such change scores to include a control for the first value in the regression analysis ( see, e.g., Jackman r9 8 o ) , which in this case is the r990 democracy rating from Freedom House. This control enables us to gauge how much democratizaTABLE 4. 1 0. Regressions of Changes in Democracy, 1 990-95, on C ultural and Economic Characteristics Changes in Democracy, 1 9 9 0 - 9 5 Dependent Variable

(a)

Democracy 1 9 9 0 Satisfaction 1 990 Trust 1 9 9 0

o/o i n Service Sector % i n Higher Education GNP per capita 1 990 (OOOs )

- .0 8 '' ( 2.2 ) - .02 (0.4) - .08 (0.8) .06 ( 1 .5 ) .02 (0.4)

(h) - 1 .07" (8.4) - .0 1 (0.4) - .03 ( 1 .0 ) .12 ( 1 .7) .06 (1.9) . 1 4 '' (2.6)

Log per capita GNP, 1 9 9 0 Constant

R2

8 . 5 8 *' (2.1) .40

Source: Data are from Inglehart ( 1 997, appendix 3 ) .

5 . 5 7 '' (2. 1 ) . 74

(c)

(d)

- 1 .08* (8.7) - . 006 (0.2) - . 03 (0.9) .06 ( 1 .2 ) .02 (1.1)

- 1 .03 '' (12.1)

1 . 72 * (4.9) -4.90 ( 1 .5 ) .82

1.91" (5.5) -4.65 ( 1 . 7) .80

Note: O L S estimates; t-ratios calculated using heteroskedasticity-consistent standard errors. (Coeffi­ cients [t-ratios]; N = 42) "'Coefficients have !-ratios > 2.0.

Civic Virtue, Growth, and Democratization

1 29

tion took place, given the initial democracy level of each country, Such a control is especially important when the change score is calculated from variables like the Freedom House ratings that have upper and lower limits . We therefore control for initial level of democracy in column b of table 4 . r o . Comparing columns a and b, we see that this control results in a precipitous drop in the coefficient estimate for levels of satisfaction (from - .o 8 to o r ) , accompanied by a substantial increase in the coefficient estimate for per capita GNP. These shifts are accompanied by an equally dramatic increase in goodness of fit (the R2 shifts from .40 to . 74 ) . Diagnostic checks on the estimates i n this column reveal that the linear term for per capita GNP conceals a nonlinear relationship. Such a pattern has been well-documented in cross-country studies of democracy, which suggest that economic development has a declining marginal effect on democratic performance ( see, e.g., Jackman 1 9 7 3 ; Bollen and Jackman I 9 8 s ; Burkhart and Lewis-Beck 1 9 9 4 ) . Accordingly, we transform per capita GNP, and the modified estimates are shown in column c of table 4 . r o . These revised figures reveal an even more pronounced GNP effect, and there is a further appreciable improvement in the fit of the model (the R2 increases from ·74 to . 8 2 ) . Finally, there is no evidence of the culture effect on change in democracy proposed by Inglehart. Neither of the esti­ mates for the two cultural variables in column c approaches statistical significance, and neither the fit of the model nor the estimate for initial level of democracy or for per capita GNP is affected appreciably by the inclusion of the other independent variables (compare columns c and d ) . Given the foregoing, w e believe there is n o systematic evidence that links cultural values either to the longer-term viability of democratic institutions or even to shorter-term transitions to democracy. The evi­ dence on economic development is interesting, however, indicating as it does that the often-noted connection between national wealth and demo­ cratic performance has implications for shorter-term patterns of democra­ tization in more recent years . Specifically, the estimates in table 4 . 1 0 show that national wealth increases the prospects for democratization, even within the last fifteen years. -

.

Con clusions

In this chapter we have evaluated recent cross-country studies contending that values affect economic growth and democratic performance. We conclude that none of these analyses makes an effective case for a cultural

I30

B E F O RE NOKMS

interpretation. Instead, the claims o f each o f the studies are driven by one or more enigmatic empirical decisions without which the argument does not work. These decisions, in turn, are idiosyncratic to the study being considered. First, while all the studies suggest that " culture " makes a difference, what they actually count as culture is another matter. In the case of economic growth, Inglehart ( r 990) initially attached great significance to materialist values. The subsequent analysis by Granato, Inglehart, and Leblang ( 1 9 9 6 ) , however, shifts attention to achievement orientations, but in so doing asserts an ongoing consistency with Inglehart's original study. For Swank ( r 9 9 6 ) , culture is embodied in communitarianism, which, with its emphasis on group values, would seem to represent quite a break with Inglehart's initial materialism, which has a more individualis­ tic focus. Knack and Keefer ( 1 9 9 7 ) place most weight on levels of trust, with a subsidiary emphasis on civic-mindedness. The studies of democrati­ zation are similarly variegated. Inglehart originally invoked a number of values ( including satisfaction, support for the current order, and material­ ism), arguing that they are all part of a coherent underlying worldview. Muller and Seligson ( r 994 ), however, shift attention to support for grad­ ual change as the key cultural value enhancing democracy. Most recently, Tnglehart ( 1 9 9 7 ) has changed tack yet again, assigning most weight to levels of satisfaction and (to a lesser extent) trust. While each of these studies thus claims to establish the importance of culture, the operational procedures they actually employ reveal that each of them means something different by the term. Our appraisal has shown that the results obtained are a function of these operational idiosyncra­ sies . We have further shown that the different approaches employed are far from interchangeable manifestations of a coherent underlying set of cultural norms. Second, following the anthropological analyses of culture from the r9 50s and r 9 6os on which they draw, the analyses we have examined betray a fundamental tendency toward ex post explanation.28 Thus, like many other commentators, Swank invokes Confucianism as responsible for the growth of certain East Asian economies, an assertion that makes sense only in the light of the growth already observed in those economies. Given their earlier growth rates, such a claim for the countries involved would have seemed preposterous in, say, 1 9 50 or 1 9 60, even though their cultures then were presumably no less " Confucian" than they are now. 2 8 . For discussions of the ways this problem is manifested in the older anthropological studies, see among others, Spiro ( T 9 87, esp. chaps. T, 2 ) , Brown ( T 9 9 T ) , a n d Billig ( r 994 ) .

Civic Virtue, Growth, and Democratization

13 r

And if Confucianism was responsible for growth in the 1 9 70s and r 9 8 os, what leverage does it provide on the Asian economic crisis of the late 1990s? A similar ex post flavor permeates the work by Inglehart and his associates and (to a lesser extent) by Knack and Keefer, which employs cultural values measured at the end of a given period to explain growth or democratization in the preceding decades . It is difficult to take such conclusions seriously. Finally, the studies we have examined are plagued by the presence of influential cases, that is, by the presence of a small minority of countries that are driving the reported results. As a result, the inferences drawn do not reflect general patterns in the data. The problem is conspicuous, for example, in the Muller and Seligson study, in which only a small minority of the countries examined actually experienced democratization in the period. As we have shown, the claimed cultural effect in this study dis­ appears with j ust two cases removed. The problem is equally glaring in Swank's analysis, in which the measure of Confucianism is simply a recoded version of the dependent variable that singles out the three cases with the highest rate of economic growth in the period. Economic growth is thus inadvertently invoked to explain itself. These problems taken as a whole generate a set of noncumulative results and thereby signify an empirical research program grounded on a set of ad hoc assumptions. This may seem a strong judgment, but one comes away from studies like these with the distinct impression that proponents of the cultural approach are willing to do whatever it takes to spawn the desired results. If each spawning requires a different stimulus, that fact, to the extent that it is recognized at all, seems implicitly to be regarded as intrinsic to the research endeavor. Perhaps this analytic strat­ egy is to be expected, since the theoretical foundations on which the cultural perspective rests are as tenuous as the empirical foundations we have evaluated. We have suggested that a more plausible approach to the kinds of values we have discussed is to treat them as endogenous, that is, to cast values as a response to the conditions within which people find themselves and the constraints and opportunities that thereby confront them. We provided some evidence along these lines earlier in this chapter by showing that achievement norms, measured in 1 990, are plausibly viewed as a function of economic growth in the preceding decades along with levels of human capital (see table 4 . 2 and the surrounding discussion) . Yet this evidence is weak and incomplete, given our reservations about the particular measure of achievement norms on which it is based ( see table 4 . 3 ) . We can pursue the question a little further. Recall that in his analysis of

I32

B E F O RE NOKMS

democratization, Inglehart ( r 9 9 7 ) places considerable emphasis o n levels of overall life satisfaction (or sense of well-being ) . Indeed, he takes satis­ faction as the principal measure of political legitimacy, one that signifies diffuse levels of support for the existing order. If this is the case, the endogenous-values perspective would lead us to cast levels of satisfaction as a positive function of democratic and economic performance. Since this variable is measured for 1 990, it can usefully be employed to illus­ trate the endogenous-values position. Accordingly, we regress national levels of life satisfaction on national levels of democratic performance for r9 90 and per capita GNP for r 990. This yields the following estimates: Satisfaction 1 990 = 8 . 00 + 3 . 6 5 (Democracy 1 990) + 0.78 ( GNP 1 990) (2 . 3 ) (4.7) ( 1 .2 ) R 2 = . 7 1 , N = 42,

where t-ratios are reported in parentheses .29 These figures are strongly consistent with an endogenous-values argu­ ment. They show that levels of overall satisfaction (or subjective sense of well-being) increase monotonically both with degree of democracy and with national wealth. Observe that the R2 is . 7 1 , which means that the fit of the model is good. These figures therefore provide a reasonable basis for inferring that levels of self-reported well-being are a function of demo­ cratic and economic performance. We recognize that the estimates are generated from a fairly simple and highly aggregated model, and we can imagine different ways of elaborating that model. On the other hand, the figures do have the singular virtue of reflecting data where the explana­ tory variables do not postdate the outcome under discussion. A recent analysis using disaggregated data reinforces this conclusion. Mishler and Rose (200 1 ) analyze levels of political trust reported in sample surveys from ten postcommunist states in eastern and central Europe and from the former Soviet Union. Their research design is well-suited to the question: the postcommunist states all obviously experienced fundamental institutional discontinuities in the recent past, discontinuities that enable Mishler and Rose to distinguish the endogenous-values perspective from the cultural alternative. They find little evidence that trust in political institutions stems from long-standing cultural patterns. Instead, trust is " substantially endogenous, " resulting largely from the political and eco29. These are OLS estimates using data from lnglehart ( I 997, app. 3 ), the data we also employed in table 4 . I o .

Civic Virtue, Growth, and Democratization

13 3

nomic performance of institutions and governments , The policy implica­ tions are straightforward. Trust can be built more surely and swiftly than the decades or genera­ tions suggested by cultural theorists. Trust can be nurtured by improv­ ing the conduct and performance of political institutions. Governments can generate public trust the old-fashioned way: They can earn it by responding promptly and effectively to public priorities, rooting out corrupt practices and protecting new freedoms. ( 5 6)

Given the analyses of this chapter, we conclude that further attempts to refine and test the case that cultural variations generate the kinds of economic and political outcomes addressed here are likely to be unpro­ ductive. Cultural values are more profitably seen as a function of the political and economic circumstances that different people face. In this reckoning, mass political behavior is generated less by different cultures and much more by different institutional environments. That is the issue to which we turn in part 2.

PART TWO

THE

I n stitutional Alternative

F I VE

Institutions and Voter Turnout

O olitical participation manifests itself in many ways. Voting, signing I petitions, working in political campaigns, and attending political rallies are some of the more widely discussed forms ( see, e.g., Verba, Nie, and Kim 1 9 7 8 ; Dalton 200 2 ) . Even so, voting occupies a special place in the analysis of political participation, and for good reason. As Putnam reminds us, " voting is by a substantial margin the most common form of political activity, and it embodies the most fundamental democratic principle of equality" ( 2000, 3 5 ) . Further, rates of voter turnout are often taken as a leading indicator of broader patterns of participation. Thus, as we ob­ served in chapter 3, Putnam begins his analysis of civic disengagement in the United States with a discussion of how, " like the canary in the mine pit, " rates of electoral abstention signal " deeper trouble within the broader body politic " ( 2000, 3 5 ) . Against this backdrop, rates of electoral participation vary dramati­ cally across the established democracies. As shown in table 5 . I , the aver­ age turnout in national elections from 19 50 through 2000 was 9 3 percent in Italy, 6 5 percent in France, and less than 50 percent in the United States and Switzerland. What drives these differences across countries ? Our argument starts from the premise that political institutions as embodied in electoral laws play a powerful role in shaping the distribution of incen­ tives for people to participate in the political system. They do so in two complementary ways . First, institutions directly affect the incentives of individual citizens contemplating whether to vote. The presence of a compulsory voting law, for example, typically enhances voter turnout (even when the penalties for noncompliance are minor) by its direct 137

I38

B E F O RE NOKMS

marginal impact o n the incentive structures of citizens who would other­ wise be nonvoters. Second, at times the effect of institutions on turnout is indirect, in the sense that they alter the incentives of parties and candi­ dates to mobilize voters, which in turn influences turnout. Thus, in single­ member district systems, some legislative districts may be considered so safe for incumbents that nonincumbent parties and candidates move re­ sources (e.g., time and money) away from those districts to others where they believe their resources have more potential to affect the outcome by mobilizing voters. Here, the institutional impact is indirect: electoral laws induce lower levels of mobilization effort in the first set of districts, which in turn inhibits voter turnout. Perhaps the earliest systematic institutional account of electoral partici­ pation was offered by Gosnell ( r9 30 ), whose impressive analysis of Euro­ pean political systems showed that levels of political participation varied consistently with variations in type of electoral system. Following this line of thought, Powell ( r 9 8 6a) and Jackman ( r 9 8 7 ) concluded that variations in turnout rates across the industrial democracies in the r 9 6os and r 9 70s are largely a function of institutional arrangements embodied in electoral laws, including the number of political parties, the presence of compulsory voting laws, and the degree of electoral disproportionality. Among stu­ dents of American politics, much attention has centered on the impact on turnout of legal factors such as residency requirements and registration procedures ( see, e.g., Wolfinger and Rosenstone r 9 8o; Squire, Wolfinger, and Glass r 9 8 7; Nagler r 99 r ; Teixeira r 99 2; Highton and Wolfinger 1 99 8 ) , both of which have powerful implications for how institutions affect political participation. The principal alternative against which we evaluate this institutional view is the cultural approach that stresses linkages between enduring values and participation. Indeed, these linkages have long been central to key analyses of mass political behavior. Almond and Verba ( 1 9 63 ) repre­ sents the classic statement of the relationship between culture and politi­ cal participation, while Inglehart ( 1 9 9 7 ) and Putnam ( 1 99 3 , 2000 ) exem­ plify the most recent variants. In participatory cultures, according to this view, citizens are more politically satisfied with their institutions (what­ ever the particular form that those institutions take) and are therefore more politically efficacious. Cultures that foster such values thereby en­ hance participation in general and voter turnout in particular. In contrast, political cultures in which more passive values are the norm inhibit par­ ticipation, which is why turnout rates are often taken as symptomatic of broader patterns of political and social engagement.

Institutions and Voter Turnout

I39

We proceed as follows. First, we examine the connections between institutional factors and voter turnout rates in 22 democracies with popu­ lations greater than one million from I 9 5 0 to 2000. Second, we consider whether cultural factors add to our understanding of the mechanisms that generate electoral participation.

I n s titutions an d Voter Turn out,

1 950-2000

We begin by extending our earlier analyses of the impact of institutional factors on turnout to cover the period I 9 50 through 2000. The original analysis centered on a cross-section of democracies from 1 9 70 through I 9 8o and included a replication for the preceding decade (Jackman I 9 8 7 ) . We subsequently elaborated the analysis i n three ways (Jackman and Miller I 9 9 5 ). First, we expanded the temporal coverage to include the I 9 8os, retaining the focus on cross-sectional decade-by-decade averages. Second, we expanded the geographical coverage to include three countries that had experienced democratic restorations in the I 9 70s ( Greece, Portu­ gal, and Spain ) . Third, we recast the analysis into a pooled time-series cross-sectional framework, thereby avoiding any issues in the original approach associated with averaging by decade. We expand these analyses further in this part of the chapter. Most notably, we focus on a substantially longer period, covering the fifty years from I9 50 through 2000. Our analyses center on five institutional fac­ tors: nationally competitive elections, electoral proportionality, number of parties, unicameralism, and compulsory voting laws.

Nationally Competitive Elections. Voter turnout should reflect the structure of political competition. Powell ( 1 9 8 6a) and Jackman ( 1 9 8 7) addressed this question by evaluating the extent to which electoral dis­ tricts are nationally competitive. ' We take the level of competition pri­ marily as a reflection of the laws governing the translation of votes into seats. Broadly speaking, one can distinguish between systems of propor­ tional representation (PR) and single-member plurality ( SMP ) . A key feature of PR systems is that electoral districts are constructed so that each district elects more than one representative. Citizens often cast their r. Powell also examined the closeness of elections, to check whether close elections give citizens more reason to vote and give parties more reason to mobilize voters. However, he found no evidence to support this expectation.

I40

B E F O RE NOKMS

votes for parties rather than for individuals, and the number o f votes a party receives determines (according to the particular formula employed) the number of seats in the legislature the party acquires to represent voters at the national level.2 In SMP, on the other hand, each district elects a single representative, and individuals typically cast their votes for specific candidates. These differences in electoral laws have important consequences for the behavior of voters. According to Cox: PR facilitates the emergence of multiple parties and gives those that enter incentives to disperse across the ideological spectrum, while SMP tends to limit the number of viable parties in any given district at two and gives them incentives to converge on the median voter. . . ; hence, one expects ideologically more focused parties under PR. This greater ideological focus may in turn foster stronger links with social groups. ( r 999, 3 9 9 )

Under PR, voters are therefore more likely t o be able t o identify strongly with a political positions outlined by a particular party and/or to be strongly opposed to a platform subscribed to by a party, and both of these increase the incentives for individuals to vote. In contrast, SMP systems encourage parties to shift their platforms to the political center, thereby minimizing the likelihood that a citizen can vote for a political party that closely resembles his or her positions on the issues (or vote against one that clearly opposes those positions), thus decreasing the incentives to vote. Differences in electoral systems at the district level also have important consequences for the electoral behavior of political parties. According to Powell: Where the chief executive is chosen by simple maj ority or plurality vote, all regions should be equally important ( e.g., France ) . In countries where the chief executive is chosen by the legislature, as in the various parliamentary systems, the question becomes the nature of the constitu­ encies electing the legislators. With proportional representation from the nation as a whole or from large districts, parties have an incentive to mobilize everywhere. With single-member districts, some areas may be written off as hopeless . ( r 9 8 6a, 2 r ) 2 . For a good discussion o f the different electoral formulas employed to translate votes into legislative seats, see Lijphart ( I 994, chap. 2 ) .

Institutions and Voter Turnout

14 1

This is consistent with Gosnell's much earlier ( 1 9 3 0) argument that PR provides incentives for parties and candidates to mobilize voters in all districts, which increases turnout. 3 While it is unlikely that any one vote will be decisive in shaping the overall outcome of an election,4 some electoral systems raise the potential decisiveness of the marginal vote. The degree of proportionality in the translation of votes into seats in the lower legislative house is central in this respect. Most electoral systems produce a degree of disproportionality in favor of the largest parties, but some systems generate a good deal more than others (Rae 1 9 7 1 a ) . s Highly disproportional systems require minor parties to accumulate many more votes to achieve a given degree of legislative representation, thereby di­ minishing the benefits of voting for the supporters of those parties. The greater the disproportionality, then, the more likely are the votes of minor party supporters to be wasted. Note that this calculus extends to the behavior of political parties (active and latent) as well. Most notably, one would expect the campaign appeals and mobilizing activities of minor parties to reflect their chances of achieving a measure of legislative repre­ sentation. Disproportionality in the translation of votes into legislative seats should therefore lower turnout.

Electoral Proportionality.

Number of Parties. Institutional arrangements that allow elections a more decisive role in government formation should increase voter turnout. The principal factor here is the number of political parties. As Downs ( 1 9 5 7, 1 5 5 ) argued, voters in multiparty systems that produce coalitions face a fundamental problem: they do not directly select the government that will govern them. Instead, they vote for parties that then select a government in the legislature, so that " ambiguity and compromise are introduced on a secondary level whenever coalitions are formed. " Elec­ tions accordingly play a less decisive role in government formation within multiparty systems. This also implies a paradox for multiparty systems. 3 · C o x ( 1 999, 399) suggests that " close districts under SMP . . . should have higher turnout levels than typical districts under PR . . . which in turn should have higher turnout levels than in safe districts under SMP. "' This argument is appealing on theoretical grounds. However, Powell's ( r 9 8 6a) empirical results indicate that, when considered cross-nationally, the closeness of elections has no bearing on turnout. In contrast, as noted earlier, Powell ( r 9 8 6a) does find a significant effect of nationally competitive districts on turnout. 4· See Aldrich ( 1 9 9 3 , 1 9 9 7 ) for interesting discussions of the implications of this point. 5. On this issue, recall our discussion in chapter r of the biases favoring the largest party in India.

14 2

B E F O RE NOKMS

When we contrast PR and SMP, " the type of political system which seems to offer the voter a more definite choice among policies in fact offers him a less definite one. This system may even make it impossible for him to choose a government at all. Instead, it may force him to shift this responsi­ bility onto a legislature over which he has very little control between elections " (Downs I 9 5 7, I 5 6) . Where elections are less decisive in the process of government formation, citizens have fewer incentives to vote. Multipartyism should therefore depress turnout.

Unicameralism. Beyond the issue of the decisiveness of elections in gov­ ernment formation is the question of the decisiveness of the governments that ensue.6 Jackman ( r 9 8 7) suggested that the degree to which the first legislative body is constrained or checked by other institutions is the relevant quantity. That is, unicameralism is crucial to producing decisive governments. Where there is no second house (as in New Zealand or Sweden ), governments based on the first house do not have to compete and compromise with another legislative chamber. In contrast, where there is strong bicameralism (as in Germany and Switzerland), legislation can only be produced by compromise between members of the two houses. This means that elections for the lower house play a less decisive role in the production of legislation where bicameralism is strong. Uni­ cameralism should therefore foster turnout. The preceding four factors are somewhat interrelated, for the reasons advanced by Duverger ( r 9 6 3 ) , Rae ( r 9 7 r a ) , and Lijphart ( I 994 ) , among others. However, these interrelations are not overwhelming, and most democracies include mixtures of these four features (Lijphart 1999 ), which means that their separate effects on voter turnout can fruitfully be ex­ plored. On a more substantive level, these correlations serve as a reminder of the trade-offs between proportional representation and plurality sys­ tems, trade-offs involving the representativeness of election outcomes and the effectiveness of governments. These trade-offs make it difficult to make a strong case for one system over the other. We are suggesting that the two forms of electoral systems also have trade-offs for voter turnout: some features typically associated with PR enhance turnout (nationally competi­ tive elections and electoral proportionality), as does at least one feature of SMP systems (the smaller number of parties) . 6 . We are hardly the first t o address the role o f electoral decisiveness. Most attention, however, has centered on the role of decisiveness in the process of government formation rather than on the decisiveness of the governments that ensue (e.g., Dahl 1 9 66; Rae 1 9 7 r b ; Strom 1 9 9 0 ) . And electoral decisiveness is seldom seen as part of the institutionally induced incentive structure that shapes the volume of electoral participation.

Institutions and Voter Turnout

I43

Compulsory Voting Laws. Some democracies (Australia, Belgium, Italy, and, since the constitution of I 9 7 5 , Greece ) have laws that mandate voting, as did the Netherlands until I 9 70. One expects that the presence of such laws should increase turnout ( Gosnell I 9 3 0, I 84 - 8 5 ; Tingsten 1 9 3 7, chap. 4; Jackman 1 9 8 7 ) . However, while the presence of manda­ tory voting laws does provide a disincentive for nonvoting that should increase turnout, the penalties for disobedience are neither uniform nor always severe, and there is thus no reason to expect universal compliance with these laws. During the period 19 50 through 2000, for example, the average turnout in Australia was 84 percent. While this is a high figure, it also indicates that a substantial number of Australian citizens did not vote, and to our knowledge none of them was prosecuted for this trans­ gression. In Italy, " the only penalty incurred by an offender is to have his name posted outside the town hall in his commune of residence, and to have his 'certificate of good conduct,' now fallen largely into disuse, stamped 'Did not vote' for five years" (Seton-Watson I 9 8 3 , I I I ) . Analysis Our empirical evaluation centers on voter turnout in the established democ­ racies with populations of more than one million that are identified in table 5 . I . As in our earlier analyses, 2 of these 22 democracies have atypical patterns of voter turnout: Switzerland and the United States. We directly address the idiosyncratic features of Swiss and American politics later.

Measures The dependent variable is voter turnout, 19 50- 2000, expressed as a per­ centage of the eligible population for each election. Observe that the eli­ gible population is not restricted to those registered to vote, but includes all of those population members of voting age. 7 Note further that the turnout figures refer to elections for the first (or lower) legislative house. We adopt a minor modification over our earlier procedure: where we originally used average turnout in presidential elections from France and the United States, here (as noted previously) we employ turnout in lower house elec­ tions for all of the countries included in the sample to maximize consis­ tency across units of analysis. Thus, the cases we analyze include non­ presidential-election-year elections. Any distortions produced by the U.S. 7 . This procedure follows the treatments by Powell and Jackman. Others have em­ ployed the size of the registered electorate as the denominator (e.g., Blais 20oo, chap . r ), but discrepancies between the size of the electorate and that of the age-eligible population are themselves a function of legal and institutional arrangements.

I 44

B E F O RE NOKMS

TAB LE 5. 1 . Turnout Rates, Electoral Disproportionality, and Number of Parties for 22 Established Democracies, 1 9 50-2000 Country Australia Austria Belgium Canada Denmark Finland France Germany Greece Ireland Israel Italy Japan Netherlands New Zealand Norway Portugal Spain Sweden Switzerland United Kingdom United States

Turnout ( # elections )"

Electoral Dis proportionality

Number of Parties

84(20) 85(15) 87( 1 6 1 67( 1 6 1 84(20) 78 ( 1 4 ) 65 ( 1 1 1 81(13) 85(10) 75 ( 1 5 ) 8 1 ( 14 ) 93(12) 70( 1 8 ) 84(14) 84(17) 80(12) 8 1 ( 101 76 ( 8 ) 83(16) 47( 1 3 1 75 ( 1 4 1 48(26)

9.0 2.5 3.1 1 1 .6 1 .7 3.0 10.6 2.5 8.3 3.5 1.8 3.7 6.2 1 .4 1 1 .2 4.2 4.6 7.9 1 .9 2.5 11.1 5.0

2.5 2.6 5.7 2.4 4.6 5.1 3.8 3.1 2.2 2.8 4.9

4 .4 3.0 4.7 2.0 3.5 2.9 2.7 3 .4 5.3 2.1 1.9

<�figures in the first column a r c average turnout levels for 1 950-2000, a n d entries in parentheses indicate the number of elections on which the averages are based. Turnout levels reflect total turnout divided by the size of the population of voting age.

case are effectively addressed by the dummy variable discussed later. As one would expect, the model specified underpredicts turnout in France ( by j ust over 7 percent) ; however, inclusion of a dummy variable for France to control for its peculiar institutional arrangement provides only a marginal increase in model fit (the R2 increases from . 79 to . 8 r ) at a substantial cost in terms of parsimony. Total turnout figures are taken from the Interna­ tional Institute for Democracy and Electoral Assistance Web site. 8 Mean 8. There arc minor errors and omissions in the IDEA Web site (http ://www.idca.int/ index.htm) . For example, Swiss turnout figures fail to take into account the correct date for the enfranchisement of women, and they incorrectly estimate turnout in the T 9 7 T election as 97 percent. Also, the r 9 9 0 Greek election and the second r 9 8 2 Irish election are missing from the data base, and the I99 5 Japanese upper house (Sangiin) election is inappropriately listed as a lower house (Shugiin) election. To supplement and correct these figures we use Mackie and Rose ( I '}'} I, I 9 9 7 ) , Wilfried Derksen's Elections around the World Web site (www.electionworld.org.), Wolfram Nordsieck's Web site, Parties and Elections in Europe

Institutions and Voter Turnout

I4 5

turnout values by country for I 9 5 0 through 2000 are listed i n the first column of table 5 . I , along with a count of the number of elections included for each country and country values for two key independent variables. To gauge the degree to which there are nationally competitive districts, we employ Powell's ( 1 9 8 6a) figures, with four exceptions ( discussed la­ ter ) . Countries with national elections by PR, a national pool for some legislative districts, or a simple national presidential vote are assigned a score of 4; those with PR in large districts receive a score of 3 ; countries with PR and three to five members per district are scored 2; and countries with SMP districts receive the lowest score of I . The greater the score, the higher the expected level of turnout. Values for this variable are listed by Powell ( T 9 8 6a, 3 8 ) . 9 O n e of the difficulties with the measure developed b y Powell ( 1 9 8 6a) is that it is not readily applied to mixed systems. Germany before the 19 57 elections, Italy after 1 99 3 , Japan after 1 9 9 4 , and New Zealand after 1 9 9 3 each adopted electoral systems that elect some representatives using PR and some using SMP. In Italy, for example, the elections for the lower house ( Chamber) are carried out using a system where 4 7 5 deputies are elected by plurality and 1 5 5 by PR (Mackie and Rose 1 9 9 7, 6 8 ) . In Japan, 3 00 members of the s ao-member lower house are elected by SMP and 200 are elected by PR (Mackie and Rose T 997, 8 o ) . To reflect these conditions, we assign these cases a score of 1 . 5 for the nationally competi­ tive districts measure during the years when the mixed systems are in operation. Electoral disproportionality is measured using a formula developed by Gallagher ( 1 99 1 , 1 9 9 2 ) . The higher the score, the greater the level of disproportionality. r o According to Lijphart ( 1 994, 6o- 6 1 ) , although there are a number of different measures of disproportionality, the least­ squares index is superior to the alternatives. Summary measures of (www.parties-and-elections.de/indexe .html ) , various issues of Electoral Studies, and the

European .Journal of Political Research. 9· Our source on Greece for assigning this variable a score of 2 is Featherstone ( r 9 8 7 ) . Using the information i n l\1ackie and Rose ( r 9 9 r , 5 03 - r r ) , w e have given Portugal a score of 3, which is the same as the score assigned to Spain by Powell. ro. Gallagher's least-squares index (LSq) is calculated as follows. LSq

=

\lfL: ( V;

-

S;)l

For each party involved in an election, the percentages of seats won is su btracted from the percentage of votes obtained. These deviations are squared and then summed across all of the parties that contested the election. This amount is then divided by two, and the square root of this number constitutes the least-squares index. We are grateful to Arend Lijphart for providing the data on disproportionality and on the effective number of parties for 1 9 5 0 through r990. The figures for the r990s were calculated by the authors.

I4 6

B E F O RE NOKMS

disproportionality for the period r 9 5 0 through 2000 are listed i n the second column of table 5 . I Multipartyism is gauged with a measure of the effective number of political parties in the legislature developed by Laakso and Taagepera ( 1 9 79 ) . The higher the score, the greater the distance from a pure two­ party system. ' ' Figures for the period 1 9 9 I through I 9 9 7 are calculated using the same formula and based on data from Mackie and Rose ( 1 997) ; party data for more recent years ( I 9 9 8 - 2ooo ) are from the sources listed in footnote 8. Summary measures for the I 9 5 0- 2ooo period are listed in the third column of table 5 . I . Unicameralism i s measured using the criteria proposed by Lijphart ( 1 9 84 , 2 1 2- 1 3 ), who identifies two features as particularly important. First, he determines the extent to which bicameralism is asymmetrical in a manner that favors the lower house or symmetrical as in Belgium and Italy. The more symmetrical, the stronger the bicameralism. Second, he determines whether bicameralism is " congruent," in the sense that both houses are very similar in composition, or if the two houses reflect very different sets of interests. The more congruent, the weaker the bicam­ eralism. Employing these criteria, Lijphart assigns the highest score of 4 for unicameral arrangements; 3 for congruent and very asymmetrical bicameralism; 2 for incongruent and asymmetrical bicameralism; 1 for bicameralism; and o for strong bicameralism. Country scores are made on the basis of Lijphart's empirical classifications in his tables 6 . 3 and 6.4, except that Sweden is scored 4 after 1 9 70 because that country adopted a unicameral legislature; Denmark is scored as r until April 1 9 5 3 , after which it is assigned a score of 4 ( see Lijphart 1 9 84, 2 1 3 , for more details on the scoring system ) . '2 Compulsory voting laws are represented by a dummy variable that equals one where such laws are present and zero otherwise. Four coun­ tries receive a score of one on this variable throughout the 19 50- 2000 period: Australia, Belgium, Greece, and Italy (Mackie and Rose 1 9 9 1 , 5 09 ) . The Netherlands i s coded as one until 1 9 70, when the compulsory voting laws were revoked, and zero thereafter. I I . The Laakso and Taagepera measure (see also Taagepera and Shugart I 9 8 9 , chap. 8 ) i s calculated a s follows .

N'

1 =

'LS,"

where S, is the proportion of seats obtained by the i-th party. For each election, the propor­ tions are squared, summed, and then divided into 1 . I 2 . Both Greece and Portugal have unicameral legislatures ( Crewe I 9 8 I , table I 0 . 2 ) a n d hence receive scores of 4 · Spain h a s a bicameral legislature, a n d w e have assigned i t a score of I in light of the discussion in Donaghy and Newton ( I 9 87, chap. 4 ) .

147

Institutions and Voter Turnout TABLE 5.2.

Institutions and Voter Tu rnout, 1 9 50-2000 Estimatora

Independent Variables Nationally competitive districts Electoral disproportionality Multipartyism Unicameralism Compulsory voting

OLS

OLS

Robust

4. 1 1 ''" ( 6.2) -.14 (0.9 ) - 1 . 82''" (3 . 7) 4.38 ''" (10.1 ) 20.64 ''" ( 13.3)

1 .20''" (2.6) - .60'" (5 . 5 ) - 1 . 03'' (3 . 3 ) 1 . 5 8 ,. (5 . 8 ) 12.42 ''" ( 14.1 ) - 27. 1 6 '" (10.1 ) - 25.50''" ( 12.7) 77. 3 8 " (44. 1 ) .79 1 1 8 .4

1 .34 ''" (3 . 1 ) - .5 8 '" (6.2 ) - .9 7'" (3.0 ) 1 .49 ''" (5.2 ) 12.24 ''" ( 1 1 .7) - 28 .6 6 '" ( 14.0 ) - 25 . 72 ''" (15.3 ) 77.00 " (42.1 )

Switzerland United States Constant

R2 F

59.76"" (22.7) .56 56.3

1 66 . 1

Note: n = 324. aMetric coefficients ( t-ratios ) : starred coefficients are more than twice their standard errors. Esti­ mates in the first two columns are calculated using White-corrected standard errors, and those in the third column are robust regression estimates.

Coefficient Estimates Table 5 . 2 presents three different sets of estimates for our basic institu­ tional model in which turnout rates are regressed on the five explanatory variables j ust described. 1 3 The first two columns of the table evaluate the sensitivity of the coefficients for these five variables to the treatment of Switzerland and the United States in the analysis: these two countries are not singled out as potentially distinctive in the estimates shown in the first column, while they are treated as potentially distinctive in the second column with the inclusion of the two country dummy variables.14 1 3 . We also estimated the model with controls for changes i n age eligibility. The esti­ mates for these variables neither were significant nor affected the parameter estimates for the institutional variables. They are therefore excluded from this and subsequent analyses. 1 4 - The estimates in the first two columns of the table are estimated using White­ corrected standard errors (White r 9 8 o ) . This procedure is designed to correct for hete­ roskedasticity across countries and is suggested as a correction procedure to pooled cross-sectional time-series data by Beck and Katz ( 1 9 9 5 ) . Autocorrelation should not be a significant confounding factor. First, we do not have a lagged dependent variable in the equations, which decreases the likelihood of autocorrelation ( sec, e.g., Chatterjee and Price 1 9 9 1 ) . Further, the time intervals are irregularly spaced. Nevertheless, as a check we

I4 8

B E F O RE NOKMS

Comparing the first two columns o f the table, it i s clear that the esti­ mates are highly sensitive to the treatment of Switzerland and the United States, j ust as they were in our earlier analyses. For example, the coeffi­ cient for nationally competitive districts in the first column is three times the size of the corresponding coefficient in the second column, and the estimated effect for compulsory voting is also much larger in the first column than it is in the second. In contrast, the coefficient for electoral disproportionality is much smaller in the first column than in the second. These patterns, combined with the two highly significant country dummy estimates in the second column, indicate that the model does not fully account for the remarkably low rates of turnout in Switzerland and the United States. When the estimates are adjusted for these two countries, however, the institutional model appears to work well. In addition, the dummy variable adj ustments for these two countries are similar: each has a turnout rate about 2 5 to 2 7 points below that implied by their other institutional characteristics. Further, a comparison of the second and third columns of the table shows that the institutional model is quite robust. The OLS estimates are remarkably similar to those generated with a robust regression estimator, in terms of both their size and their statistical significance. These results provide strong support for the proposition that political institutions play a maj or role in structuring rates of political participation. All of the parameter estimates are correctly signed and have acceptable t-ratios. Turning to the individual parameter estimates, it is clear that the struc­ ture of electoral districts matters: electoral systems that foster competi­ tion nationally thereby also encourage higher rates of voter turnout. In addition, the estimates for disproportionality are quite strong: the greater the disproportionality in the translation of votes into seats, the lower the level of turnout. With these two factors controlled, multipartyism inhibits turnout, a pattern we interpret to mean that, at the margins, increasing multipartyism decreases the salience of elections in the process of govern­ ment formation, which results in lower turnout. The coefficient for uni­ cameralism is highly significant. Again, we take this as evidence that the salience of elections bears directly on turnout. Since governments in uni­ cameral settings are less constrained than others, there is a clearer link performed Cochrane-Orcutt regression and obtained substantively identical results. As a second check on the robustness of the results, we reestimated the model using feasible generalized least squares, specifying corrections for autocorrelation (first-order) and hctcro­ skedasticity. The results were very similar to those reported in table 5 .2. As a result, we report the White-corrected estimates, because they also enable us to generate a useful measure of overall goodness of fit in the form of the R'.

Institutions and Voter Turnout

149

between legislative elections and policy outcomes and thus a clearer incen­ tive for citizens to participate. Along with these four factors, the presence of mandatory voting laws has a pronounced, positive effect on turnout. Finally, as we have already indicated, the dummy variables for Switzer­ land and the United States display large, similarly sized negative coeffi­ cients ( and t-ratios ) . Overall, these results are encouraging. Most notably, they indicate that the institutional model is quite robust in the following sense. The model was first developed and evaluated employing data from just one decade, the 1 9 70s (Jackman 1 9 8 7 ) . We subsequently expanded the analysis to cover three decades and three additional cases (Jackman and Miller 199 5 ) . Here, w e have shown that the estimates apply well to a period of five decades, so that we have expanded the number of observations ( election years) by almost 40 percent over the number employed in our previous analysis. Finally, the model continues to perform well in terms of its fit. Using five institutional variables and two country dummy variables, the model accounts for 78 percent of the variation in rates of voter turnout across 3 24 elections held during a fifty-year period.

The American an d Swiss Cases

The preceding analyses treat the United States and Switzerland as distinc­ tive from the general pattern. In other words, the particular institutional determinants of participation that we have addressed to this point are insufficient as accounts in these two settings. At the same time, the fact that the country dummy estimates for these two countries are of almost the same size is noteworthy ( see the second and third columns of table 5 . 2 ) . Does this mean that the United States and Switzerland are sui generis ? It is obviously difficult to provide a systematic evaluation of this question, given that only two cases are involved. Moreover, Switzerland and the United States differ from each other on a number of key institu­ tional dimensions, each of which can plausibly be linked to turnout. Voter registration in the United States is voluntary, for example, while in Switzerland it is not. ' 5 Similarly, while both are federal systems, power is especially decentralized in Switzerland, which reduces the salience of national elections in the formation of public policy. Finally, the bearing of elections on government formation at the federal level is singularly r s . Teixeira ( I 992, chap. 4) estimates that reform of voter registration procedures in the United States would increase turnout in federal elections by approximately 8 percent.

I 50

B E F O RE NOKMS

limited i n Switzerland, where the current governing coalition has been in place continuously since I 9 5 9 · Keeping these differences i n mind, what the two countries d o share is a distinctively high frequency of elections, compared with the other coun­ tries we have examined. Along with elections for national, cantonal, and communal representatives, Switzerland makes extensive use of the refer­ endum. Since I 8 4 8 , there have been more than 4 6 3 nationwide referen­ dum elections in Switzerland, nearly ten times more than were held in second-place Australia. Swiss use of the national referendum is thus ex­ ceptional. Moreover, Swiss referenda typically include multiple ballots ( so-called omnibus referenda ) on a variety of issues. Thus, on November 2 8 , 1 99 3 , federal referenda were held on six different issues ranging from banning advertising on alcohol and tobacco products to adopting a fed­ eral value-added tax on all goods and services (Kobach 1 9 9 4 , 1 29 ) . Referenda are not, o f course, held a t the federal level i n the United States. Even so, ballots for federal elections typically reflect elections for multiple offices. Combined with other statewide races, primary elections, local elections, and the increased use of the initiative process in many states ( especially in the West), elections are thus much more demanding and frequent in the United States than they are in most other democracies. The problem is one of long standing: "In Europe [with the exception of Switzerland] the voter is not given an impossible task to perform on election day. He is not presented with a huge blanket ballot as are the voters in many of the American states " ( Gosnell 1 9 3 0, r 8 6 ) . Fifty years later, Crewe emphasized the same issue: "No country can approach the United States in the frequency and variety of elections, and thus the amount of electoral participation to which its citizens have a right" ( r 9 8 r , 2 3 2 ) . Along with increasing information costs for voters (Watten­ berg 2002), this high frequency of elections reduces the incentive for parties to engage in get-out-the-vote mobilizational efforts. One of the most effective means of lessening mobilizational incentives in a system is to lower the expected benefit of mobilization by lessening the decisiveness of the election. Decisiveness can be lowered in two main ways: first, by creating longer ballots with separate votes for each office; second, and more potently, by holding elections for different offices at different times. The U. S. appears to be in the most advanced state of electoral disaggregation along both dimensions. ( Cox I 9 9 9 , 4 T T ; emphasis added)

We believe that this high frequency of elections in Switzerland and the United States depresses voter turnout. That we are dealing with only two

Institutions and Voter Turnout

151

cases means, of course, that it i s difficult to obtain a precise quantitative estimate of the magnitude of this turnout-inhibiting effect. The number of elections in the two countries is manifested in somewhat different ways (national referenda versus elections for a variety of offices) . Further, the number of elections at a given time is not uniformly distributed in either case across federal units. In the United States, for example, many states do not employ the initiative process, and there is also considerable vari­ ance in the use of the initiative across those states that do employ it. Beyond this, the number of choices presented to voters in each country has been increasing over time, especially in the recent past. Bearing all these issues in mind, we take the distinctively high vol­ ume of elections itself to be the critical factor. Consider the United States. The available evidence indicates that election calendars, off-year state­ wide races, and statewide primaries systematically influence turnout lev­ els in j ust the manner anticipated by the high-frequency-of-elections argu­ ment. For example, Boyd finds that even party primaries appreciably reduce turnout. A single statewide race contested in both parties lowers the probability of voting by four percentage points in a typical spring and fall primary. Overall, spring and fall primaries reduced turnout by about five percent­ age points in each of the presidential elections of 1 9 7 6, 1 9 80, and 1 9 8 4 . ( 1 9 8 9 , 7 3 8 ; see also Boyd 1 9 8 6)

Boyd interprets this pattern to reflect opportunity costs imposed by fre­ quent elections on the campaign organizations that mobilize voters. If the resources (in the form of money and volunteer labor) available to elec­ toral organizations are limited, then an increased election frequency im­ plies that those resources will be spread more thinly over a number of campaigns, thereby inhibiting the opportunities for mobilizing voters in each of them . 16 Further evidence for a voter fatigue effect comes from an analysis of levels of political information among potential voters in a generally high-frequency election environment. Focusing on California elections over time, Nicholson ( 2003 ) shows that levels of issue aware­ ness fluctuate systematically with the number of initiative items on the ballot. All else equal, increasing numbers of items on the ballot generate r 6. The election frequency argument is distinct from the related issue of voter roll off in American elections. Rolloff refers to voters who go to the polls but choose not to vote on some of the items on the ballot and who thereby fail to complete their ballots ( for a recent discussion of rolloff patterns, see Wattenberg 2002, chap. 6 ) . In contrast, our analysis centers on rates of voter turnout, that is, the proportion of age-eligible adults who make it to the polls in the first place.

B E F O RE NOKMS

152 TABLE 5.3.

Basic I nformation about the Swiss Experience with Referenda, 1 879-2000

Number of referendum elections' Annual average number of referendum electionsb Average number of propositions per election'

1 8 791 900

190120

192140

194160

196180

19812000

29

21

36

48

53

54

1 .38

1 .05

1 . 80

2.40

2.65

2 . 70

1 .34

1 .29

1 .42

1.31

2.19

2.96

Source: S e e note 1 7 . 8Number of dates on which referendum elections were held (ignoring t h e number of referendum items on the ballot) bAnnual average number of dates on which referendum elections were held cAverage number of distinct referendum items on the ballot per referendum election (range: 1 - 7)

lower levels of political information, as a voter fatigue account would lead us to expect. The Swiss experience with national referenda is also consistent with the election-frequency argument. As observed earlier, Switzerland makes considerable use of the referendum: table 5 ·3 displays the basic informa­ tion for the years from r 8 79 through 2000. '7 The first row reports, in twenty-year intervals, the number of times a referendum election was held. The second column shows the annual average number of times a referendum election was held. In the bottom row we report the average number of distinct propositions included on the ballot in each referen­ dum election. Four points are clear from the table. First, the Swiss have always made extensive use of the referendum device. Even in the last two decades of the nineteenth century, referendum elections were held at an annual aver­ age rate of almost r · 4 · In other words, there were at that time almost three such elections every two years . Second, the use of the referendum has been increasing, so that by the last two decades of the twentieth century, the annual average rate was 2. 7, almost double the rate of a century before. Third, there has also been a long-standing use of omnibus elections, that is, referendum elections that called for a vote on more than one issue. Even in the last part of the nineteenth century, the average number of items per referendum election was L 3 4 · Fourth, the use of such omnibus elections has also been growing, especially since about 1 9 60. In fact, the average frequency of such elections has more than 17. The patterns in table 5 · 3 are taken from two sources. Figures from r 8 79 through 1 9 9 3 are from Kobach ( 1 994, table 4-r ) ; figures after 1 9 9 3 are from www.admin.ch/ch/d/ pore/val. These sources also report the turnout data examined later.

Institutions and Voter Turnout

I53

TABLE 5.4. Regressions o f Turnout i n Swiss Referenda o n the Number of Referendum Items in the E lection and the Frequency of Elections in the Recent Past, 1 879-2000 Number of reierendum items in election (log) l'vlore than two reierenda in last 12 months"

- 7. 8 0 ''" (6.5) - 3 .40 ''" (2.3)

Turnout in previous referendum election Constant

R2 F N

55.29 ''" (45.5) .15 27.75 241

- 5 . 76 '" (4.8) - 2 . 9 1 ''" (2. 1 ) 0.36" (6.3) 3 6 . 0 1 ''" ( 1 1 .0 ) .27 35.23 240

>:·Starred coefficients are more than twice their standard errors, and t-ratios (calculated from '\XThite­ corrected standard errors) are shown in parentheses immediately below them. aA dummy variable that equals 1 if there were more than 2 referendum elections in the previous 1 2 months, and 0 otherwise. The number of referendum elections in the previous 12 months ranges from 0 to 6, with a median value of 2 .

doubled over a century (compare the figure of 2.96 for 1 9 8 1 through 2000 with the figure of 1 . 3 4 for 1 8 79 through 1 900 ) . Thus, Switzerland was already a high election-frequency environment at the end of the nineteenth century, and that environment has simply intensified over the course of the twentieth. Given this environment, we can use the data summarized in table 5 · 3 t o develop a more systematic sense of the effects of election frequency on voter turnout. For each of the 24 1 referendum elections held from 1 8 79 through 2000, data are available on turnout, expressed as the percentage of the eligible voters participating in that election . 1 8 The question then is whether the frequency of elections inhibits turnout. Table 5 ·4 reports two sets of estimates. In the first column, we regress turnout on two variables: the number of referendum items in the elec­ tion (logged) and a dummy variable set equal to one if there had been more than two referendum elections in the previous twelve months and zero otherwise.19 The estimates are instructive: both of the explanatory r 8 . These data are from the same sources used for table 5 · 3 · Observe that this is a different ratio from that used in the analyses of turnout reported earlier in the chapter (which centered on the percentage of the age-eligible population voting) . For most of the period, many adults were ineligible to vote. 11ost notably, women were not fully emanci­ pated until r 9 7 r , following the referendum of February 7, r 9 7 r , in which 66 percent of the (male) voters approved women's suffrage. In an earlier referendum held on February r , 1 9 5 9 , women's suffrage had been defeated by about the same margin. 1 9 . As shown in table 5 . 3 , the number of distinct referendum items on the ballot per referendum election ranges from r to 7, with an overall mean of 1 .9 and a median of r. The

I 54

B E F O RE NOKMS

variables have sharp negative effects a s anticipated by the election­ frequency argument. Further, as shown in the second column of the table, these negative effects persist and are only mildly attenuated when we include the turnout rate from the previous referendum election as a control for any time trends in turnout. These figures thus indicate that, taking into account the turnout rate in the previous referendum election, voter participation in a Swiss referendum is systematically dampened when voters are asked to consider a number of measures and when there has been an above-average number of referendum elections in the previ­ ous twelve months. And all of this, we reiterate, occurs in a setting where, even at their least frequent, referenda occur much more often than anywhere else. 20 While we recognize that there may be other ways of evaluating the argument, these patterns for Swiss referenda over a period of 1 20 years are striking, especially considered in conj unction with the available evi­ dence for the United States. We are therefore inclined to the view that the distinctively low rates of voter turnout in Switzerland and the United States have an institutional source that takes the form of a remarkable number of elections. Given the evidence, such a conclusion is more plau­ sible than any of the common alternative explanations cast in terms of an " exceptional ism" that emphasizes distinctive cultural configurations.

Does Culture Offe r a S upe rior Exp lan ation of Turnout?

Our analyses to this point indicate that the institutional model provides an effective explanation of variations in turnout across democracies, and that even the Swiss and American exceptions are plausibly linked to a particular institutional feature that they share. Nonetheless, we recognize logarithmic transformation helps reduce the skewness in this variable, given that the mean is almost twice the size of the median. The number of distinct referendum elections held in the previous twelve months ranges from o to 6, with an overall mean of 1.9 and a median of 2. Thus, the dummy variable distingu ishes those elections held at a time when there had been an above-average number of referendum elections in the previous twelve months from other times. While autocorrelation docs not appear to be an issue, there is evidence of hctcro­ skedasticity in the original estimates: accordingly, all t-ratios in the table are calculated u sing White-corrected standard errors. 20. Recall that in chapter 3 (footnote 8 ) , we observed a similar pattern for referenda in Italy. Specifically, a pooled analysis of the five referenda from 1974 through 1 9 8 7 across all twenty regions yielded a regression coefficient of - 2. 7 for the number of items per referen­ dum, suggesting that turnout dropped by j ust under 3 percent for each proposition added to a referendum.

Institutions and Voter Turnout

I55

that the political culture studies discussed in earlier chapters are not without apparent empirical support. We now need to assess the ability of the institutional approach to account for political participation with cul­ tural factors controlled. Before doing so, we briefly consider a proposal offered by Radcliff that culture can be captured in a series of country-specific terms. The rationale is apparently straightforward: In the debate over institutions versus culture . . . it is perfectly reason­ able to conceive of fcountryl dummies as encapsulating political cul­ ture . In other words, a dummy for, say, "Norwayness " means pre­ cisely that: the enduring norms that shape behavior in that country. ( 1 9 9 6, 7 2 2 )

Radcliff accordingly evaluated the relative importance of political culture and institutions by estimating a model including both institutional vari­ ables and a series of dummy variables representing each of the countries in the analysis. Swank ( 1 9 9 6 ) similarly employed dummy variables to isolate the effects of " Confucian" and " social corporatist" political cul­ tures on levels of economic growth (see chap . 4, this vol. ) . Evidence that this approach has n o bearing o n our analysis o f voter turnout can be generated from the robust regression estimates shown in the third column of table 5 . 2. That estimator proceeds by down-weighting observations that might be driving the estimates (that is, by down­ weighting influential cases ) . Were the figures shown in the table masking country differences of the form proposed by Radcliff, there would be a large number of cases with small weights, and the size of those weights would cluster cross-sectionally by country. However, the mean weight used in the estimates shown in the third column of table 5 . 2 is .92, only three country years have weights less than ·4 (the smallest is . 3 4 ) , and there is no pattern to these weights.2r We therefore conclude that our institu­ tional model outperforms a country dummy approach. Thinking about the problem in broader terms, it remains unclear how 2 I . To gain some perspective on these numbers, the weights generated by robust regres­ sion estimates of the model in the first column of table 5 . 2 that excludes dummy variables for Switzerland and the United States have a mean of .8 5 and a minimum of zero (in fact, fully I I country years have weights less than o . o i , which means that they have been dropped from the analysis completely: all I I of these cases involve elections in Switzerland and the United States). There are 22 election years with weights less than .4, of which 8 involve Switzerland and IO the United States ( underscoring the distinctiveness of these two countries that we have already discussed ) .

156

B E F O RE NOKMS

well the country dummies could i n principle serve a s an unambiguous proxy for culture. One might cast them as measures of culture (however one chooses to define that term), but they are j ust as readily cast as measures of a host of other factors, including political institutions. Rad­ cliff to the contrary, they certainly cannot be taken unambiguously as measures of " the enduring norms that shape behavior " in each of the countries identified. Instead, country dummies in a pooled framework are generally taken to reflect ignorance and are " inserted merely for the purpose of measuring shifts in the regression line arising from unknown variables" (Kennedy 1 9 9 8 , 2 2 7 ) . This is why we spent some time seeking to understand and explain the coefficients displayed in table 5 . 2 for Switzerland and the United States: estimates like these have no intrinsic meaning. Taken by themselves, significant country dummies simply indi­ cate that country A differs from country B, which, as an explanation, is pre-embryonic. After all, one of the fundamental goals of comparative political analysis is to replace (proper) country names with variables (see, e.g., Przeworski and Teune 1 9 70, chap. r ) . Approaches like Radcliff's and Swank's thus give us no purchase on the issue at hand. We accordingly turn to two more direct analyses of the possible effects of cultural values. First, we evaluate whether culture, defined in terms of McClelland's ( T 9 6 T ) classic study of need for achievement, impinges on turnout over the same period. Second, drawing on data from the World Values project (lnglehart 1 9 9 7 ) , we evaluate the bearing of values on rates of political participation in the 1 990s.

Need for Achievement McClelland's ( r 9 6 r ) analysis of need for achievement is one of the earli­ est efforts to quantify cross-national differences in cultural values and is very much in the spirit of the arguments laid out by Weber in his Protes­ tant ethic thesis. Indeed, McClelland argued that need for achievement is the primary component of the Protestant ethic.22 McClelland developed country scores for need for achievement as of 1 9 5 0 from content analyses of themes emphasized in textbooks read by children from second- through fourth-grade levels ( r 9 6 r , chap. 3 ) . On the basis of these scores, he linked need for achievement to a variety of different outcomes. Swanson ( r 9 7 1 ) extended this analysis to report that 22. The same argument was advanced by Swanson ( 1 9 7 1 , 6 2 2 ) . Recall from our discus­ sion in chapter 4 surrounding table 4 . 1 that we were unable to find any support for McClelland's claim that need for achievement fosters economic growth.

Institutions and Voter Turnout TABLE 5.5.

Need for Achievement and Voter Turnout

Variable Nationally competitive districts Electoral disproportionality Multipartyism Unicameralism Compulsory voting Swiss dummy U . S . dummy Need for achievement, 1 950"

Rl F N

I57

1 9 60s

1 9 70s

1 980s

1 990s

2 . 0 '' (2.6) - 1 .0" (3.4) - 2 .4 '' (2.3) 2 . 8 '' (4 . 1 ) 1 5 . 3 '" (7.3) - 1 8.2" (5.7) - 1 5 . 1 ,. (3.8) 1.6 ( 1 .0 ) .83 24 1 .9 55

2 . 2 '' (2.0) - . 7'' (3.0) - 1 . 9 '' (2.5) 1 . 6 '' (2.8) 1 1 .4 ''" (5.7) - 3 3 . 1 '' (11.1) - 2 7 . 8 ''" (6.7) -1.7 (0.9) .88 96.6 67

0.9 (0.7) - . 5 '' (2.0) - 2 . 6 '' (2.6) 1 . 6 '' (3. 1 ) 1 1 . 3 '' (5.9) - 34 . 5 '' (15.3) - 27 . 0 '' (5.9) - 7. 7" (3.0) .85 615.2 67

0.8 (0.7) - . 5 '' (2.7) - 1 .5 ( 1 .6) 1.2 I 1.6) 1 5 . 3 '' ( 6 .4) - 34 . 5 '' ( 1 1 .2 ) - 2 8 . 2 '' (5.4) -3.9 ( 1.3) .87 72 . 8 57

Note: Metric coefficients ( t-rarios) * Coefficients are more than twice their standard errors. Estimates are calculated using \XIhite­ corrcctcd standard errors. "Need for achievement data are from McClelland ( 1 9 6 1 , 90, table 3 . 4 )

need for achievement is significantly correlated with political outcomes, including type of constitutional system. Since McClelland's data were collected in 19 50, the children exposed to the values he analyzed would be adults and potential voters by the early 1 9 60s. Moreover, need for achievement scores are available for 2 1 o f the 2 2 countries included in our sample (all nations except Israel) and thus provide an interesting vehicle for evaluating the effect of values on turnout. Because our analyses of turnout cover the years after 1 9 50, we need information about cultural values that apply to the early I 9 sos in order to avoid endogeneity problems. To the best of our knowledge, McClelland's are the only data available that deal so directly with cul­ tural values for such a large set of cases in this period. 23 To isolate potential generational effects that are part of the cultural argument, we employ a decade-by-decade evaluation of the effect of cultural values on participation. Table 5 ·5 presents estimates of the effects 2 3 . The other major early study of values is, of course, Almond and Verba ( r 9 6 3 ) . However, their study i s restricted to five countries. Values have more recently been a central focus of the World Values Surveys, which we examine in the next section. Country scores for need for achievement, circa T 9 5 0, are provided in McClelland ( r 9 6 r , 90- 9 T , table 3 .4 ) .

158

B E F O RE NOKMS

o f need for achievement on turnout in the r 9 6os, 1 9 70s, r 9 8os, and 1990s, controlling for institutional factors. These roughly correspond to the need for achievement levels of individuals in their twenties, thirties, forties, and fifties in each of the countries included in the sample. Although the estimates for the institutional factors vary by decade, overall they are quite similar to those reported in table 5 . 2 . Bearing this in mind, the figures in table 5 . 5 show that for elections held in the 1 9 70s and 1990s, the estimates for need for achievement are negative (consis­ tent with the cultural argument) but statistically indistinguishable from zero. For the 1 9 8os, however (when the cohort from McClelland's study was in its forties ) , there is a statistically significant negative coefficient for need for achievement, hinting at a possible generational effect. However, for the 1990s the need for achievement coefficient is again indistinguish­ able from zero. Since by the 19 9 0s the cohort from McClelland's sample was in its fifties and approaching the height of its participation levels, this pattern obviously cannot sustain a generational argument. More gener­ ally, and when taken as a whole, the estimates from table 5 . 5 provide no support for the view that values in the form of levels of need for achieve­ ment have any systematic bearing on rates of voter participation. Survey Data on Civic Values As a final check on the relative merits of cultural and institutional ac­ counts of voter turnout we turn to data from the World Values Surveys conducted by Inglehart and his colleagues (Inglehart 1 9 9 7; Inglehart, Basafiez, and Moreno 1 99 8 ) . While McClelland's data are informative because they were collected for 19 50 (so that they predate the bulk of our data on voter turnout), they have two limitations. First, they center on only those values associated with achievement, and second, they originate in content analyses of textbooks rather than in more direct measures of values that can be gathered in public opinion surveys. Against this background, the 1990-9 1 World Values Surveys include data for most of the countries that we analyzed in table 5 . 2 . These data have two advantages: ( r) they come from direct public opinion questions about values, and ( 2 ) they provide information about a broad array of potentially significant values. The primary drawback of the data is that they were collected around 1990, so that they cannot be employed to account for turnout since 1 9 5 0 . We therefore confine our attention here to patterns of turnout since 1990. While this entails a loss of information about rates of political participation over the preceding four decades, it

Institutions and Voter Turnout

1 59

does ensure that our analyses are not subject to the issues associated with ex post " explanations. " A s i n discussions o f the components o f civic culture that g o back at least to Almond and Verba ( 1 9 6 3 ), we focus on three particular values that would seem to tap into different aspects of participatory norms and that therefore have a potential bearing on voter turnout. These are levels of trust in others, levels of satisfaction with one's life, and rates of politi­ cal discussion. Data on each of these three variables are available from the 1990- 9 1 version of the World Values Survey and are described in Inglehart ( 1 9 9 7 ) . 24 Table 5 . 6 displays four sets of regression estimates. Since we are re­ stricting our attention to turnout from 1990 to 2000 in order to avoid endogeneity problems, the first column of the table shows the estimates for the basic institutional model for the 1 990s. Successive columns ( b - d ) display the estimates obtained when each of the three cultural variables i s considered i n turn. From the first column, it is evident that even when estimated with the restricted number of elections held during the 1 990s the baseline institu­ tional model performs well. All of the coefficients are correctly signed, and all but one are more than twice the size of their standard errors (note that the exception, the coefficient for nationally competitive districts, comes close to meeting this test with a t-ratio of 1 .9 ). Indeed, the param­ eter estimates are similar to those we have reported for earlier decades in the past, and the R2 of . 8 9 indicates that the fit of the baseline institu­ tional model remains good for the 1 990s. The remaining three columns in this table report the estimates obtained when each of the three cultural variables from the World Values Surveys, measured circa 1 990, is evaluated in turn against the baseline model. Here, the evidence is clear-cut. First, comparing all the columns it is clear that adding the cultural variables has no discernible effect on the model fit: the R2 is . 8 9 with or without cultural variables included. Second, none of the t-ratios for the estimates for culture exceeds the conventional 24- The data are described in Inglehart ( 1 997, appendixes 3 ltable A.2j and 5 ) . Percent Trusting comes from variable 94, which asks " Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people ? " Our measure is the percentage of respondents indicating that "most people can be truste d . " Satisfaction is the average level of satisfaction ( based on a ten-point scale) self-reported by respondents to the following question (variable 9 6) : "All things considered, how satisfied arc you with your life as a whole these days ? " Political Discussion Rate is based on variable IO ( " When you get together with your friends, would you say you discuss political matters frequently, occasionally, or never ? " ) .

r 6o TABLE 5.6.

B E F O RE NOKMS

Culture, I nstitutions, and Voter Turnout in the 1 990s

Variable National competitive districts Electoral disproportionality Multipartyism Unicameralism Compulsory voting Switzerland United States

(a) 1.9 ( 1 .9 ) - . 7 '' (4.6) -2.1" (3.3) 1 . 1 ,, (2.0 ) 24.7 '' (6.6) - 30.9 '' ( 1 1 .0 ) - 2 8 . 6 '' (5.3)

% Trusting

(b) 2 . 3 '' (2.3) - . 7 '' (4.6) - 2 .7" (3.6) .9 ( 1 .6 ) 27.7" (6.5) - 2 8 . 9 '' (9.3) - 2 9 .7" ( 5 .4) .1 ( 1 .5 )

(c) 2 . 0 '' (2.0) - . 7'' (5.0) - 2.4 '' (3.7) .9 ( 1 .6) 2 5 . 6 '' (6.8) - 32 . 6 '' (9.5) - 3 0. 1 '' (5.3)

1.8 ( 1 .2 )

Satisfaction Political Discussion

Rl F N

(d) l.5 ( 1 .4 ) - . 7 '' (4.3) -2.1" (3.3) l . l" (2.0) 25.4'' (7.3) - 3 1 .4'' ( 10.8) - 2 8 . 8 '' (5.3)

.89 66.1 50

.89 48.5 50

.89 52.2 50

.2 ( 1 .5 ) .89 61.1 50

Note: Metric coefficients ( t-ratios) ::·coefficients arc more than twice their standard errors. Estimates arc calculated using White­ corrected standard errors.

yardstick of 2.0, while the F-ratios for the second through fourth columns of the table are appreciably lower than the F-ratio for the baseline model in the first column. Third, comparing the estimates for each of the institu­ tional variables across the table shows that each remains quite similar to those generated for the baseline model and displayed in the first column. Taken together, these patterns provide no evidence that national differ­ ences in cultural values as of 19 9 0 had any systematic effect on turnout over the ensuing decade.

Con clusions

The analyses in this chapter provide scant support for the view that rates of political participation hinge meaningfully on cultural values. To be sure, we do find a mild suggestion that values may make a difference in our analysis of McClelland's data on need for achievement, but our other tests

Institutions and Voter Turnout

r 6r

provide no corroborating evidence. That is, when we adopt a country­ specific approach to political culture and when we employ data from the World Values Surveys, there is no systematic evidence for a cultural inter­ pretation of turnout. This is consistent with Powell's ( r 9 8 6a) cross­ national survey analyses, which indicated that while Americans have the edge in participatory attitude structures, such attitude structures are not, in turn, systematically linked to turnout rates. Crewe ( 1 9 8 1 , 260) had previ­ ously noted that, if anything, cultural norms appear to be inversely related to voter turnout, although his conclusion was based on only four cases ( Germany, Italy, the United Kingdom, and the United State s ) . A similar conclusion is implied by Sidjanski's observation that general political activ­ ism during the early T 9 70s was actually higher in Switzerland and the United States than it was in six other European democracies studied ( 1 9 79 , 1 0 7 ) . I n this chapter, w e have provided more extensive evidence that re­ inforces this conclusion. At the same time, we have provided additional evidence that levels of voter turnout reflect institutional arrangements as embodied in electoral procedures, and our model appears to offer a feasible account of the observed patterns in the established democracies over the last half cen­ tury. Specifically, we have identified five institutional arrangements that systematically impinge on turnout. We have further argued that the lower-than-predicted rates of turnout in two cases (Switzerland and the United States) are best explained by the distinctively high volume of electoral choices that confront citizens in those two countries. This, of course, is simply another institutional feature that is manifested in those two particular settings. As we indicated at the beginning of this chapter, it is our view that the effects of institutional arrangements operate in two distinct, but overlap­ ping ways. Institutions sometimes affect turnout directly, by defining an incentive structure that emphasizes particular benefits or costs of voting. The presence of mandatory voting laws would seem to be a clear case in point. At other times, institutions provide incentives for parties and candi­ dates to mobilize votes, and thereby influence turnout in a more indirect manner. Electoral systems that provide parties incentives to concentrate their campaign efforts in particular geographic areas at the expense of others typify such indirect incentive systems. This leads us to a final point. As we have emphasized throughout this book, cultural explanations focus on the durability of values and thereby lead us to anticipate limited changes in political behavior over time in particular settings. The institutional perspective, by contrast, implies that changes in behavior are induced by changes in the rules. As is well

r 62

B E F O RE NOKMS

known, institutional arrangements o f the sort we have focused on are themselves typically durable. Nonetheless, they are modified on occasion, and when they are, political behavior appears also to follow suit. Thus, voter turnout declined in the Netherlands after the abolition of manda­ tory voting in 1 9 70, while it increased in Sweden after the abolition of the upper legislative house in the same year (Jackman 1 9 8 7, 4 1 6 ) . Similarly, after the democratic restorations in Greece, Portugal, and Spain in the 1 9 70s, patterns of voter turnout in those countries appear to have been governed by the same mechanisms accounting for cross-country turnout differences in the established democracies. 25 In this chapter, we have provided rather detailed evidence on the same point for Switzerland. Turnout in Swiss national referenda is quite sensitive to the number of referendum items on the ballot and the number of referendum elections in the previous year (the more items and recent elections, the lower the turnout) . This implies that political behavior responds quickly to even short-term changes in the incentive structure. Thus, contrary to the cultural account, political participation is most profitably cast as a function of the institutional environment. Typically, that environment is relatively stable over time. When it is modified, how­ ever, the political behavior of citizens and other relevant political actors (parties and candidates for office) adapts to the modifications, and the adaptation occurs in remarkably short order.

2 5 . Indeed, we showed in an earlier analysis of the counterfactual (Jaekman and Miller 5) that had different institutional procedures been adopted in these three countries after they underwent their democratic restorations, our model would have been unable to ac­ count for the observed patterns of voter turnout. I9 9

SIX

Institutions and Support for Parties of the Extreme Right with Kari n Vol pert

T he revival of extreme right-wing parties in a variety of West European I democracies during the 1 9 8os caught many politicians and opinion leaders off guard. It also contradicted well-known scholarly claims about the end of ideology, the advent of postindustrial society, and the growth of postmaterialist values (e.g., Bell 1 9 60, 1 9 7 3 ; Inglehart 1 9 9 7 ) . Given grow­ ing abundance and accompanying worker satisfaction after 19 so, tradi­ tional forms of polarization stemming from material scarcity were sup­ posed to have been superseded by such postmaterial concerns as the need for freedom, participation, and self-actualization ( Straussman I 9 7 S ) . These and associated changes were thought to have generated new politi­ cal alignments involving innovative parties on the left of the political land­ scape (see, e.g., Kitschelt 1 9 8 8 ) , accompanied by a decline in support for right-wing parties. In retrospect, these expectations seem to have been part of a golden age that began around 19 so, but that ended abruptly after the oil shock of the early 1 9 70s, an age in which economic performance in the ad­ vanced capitalist societies " surpassed all historical records" (Maddison 1 99 1 , 1 67 ) . By the 1 9 8os, " old" economic concerns like unemployment had returned to center stage in general. These materialist issues are, for reasons detailed later, likely involved with the concurrent return of the extreme right. Beyond this, the resurgence of extremist politics in this form is note­ worthy given debates over the factors that facilitated precursor political

1 64

B E F O RE NOKMS

movements i n Europe, especially i n Germany, following the First World War. As we observed in chapter r , the claim that these movements stemmed from deep-seated cultural characteristics has a long pedigree, represented most recently in Goldhagen's ( 1 9 9 6 ) widely discussed thesis that Nazism appealed to and would not have been possible without fundamental traits of German national character. Yet institutional expla­ nations of the electoral success of earlier extremist movements have an equally long pedigree. Most notably, the early electoral successes of the Nazis have repeatedly been cast as a function of the highly proportional electoral arrangements of Weimar Germany ( see, e.g., Hermens 1 9 4 1 ; Rustow r 9 5 o ) . Given these conflicting interpretations o f their precursors, contemporary extremist political movements constitute an important set­ ting within which to evaluate the merits of cultural and institutional accounts of political behavior.

Con tem porary Parties of the Extre me Right

Before proceeding, it is important to identify the parties with which we are concerned. By the term extreme right, we refer to parties located at the right tail of the left-right political scale (Castles and Mair T 9 84; Mair and Castles 1 9 9 7 ) , whose electoral policy programs often include a neofascist or antisystem component (Ignazi 1 9 9 2 ) . While the populist and national­ ist appeals of these parties vary slightly from one setting to another, they typically involve xenophobic stances that focus on immigrants and for­ eign workers, especially those not of European descent. As Husbands has put it, "What unites all of these parties is their particular commitment to some sort of ethnic exclusionism-a hostility to foreigners, immigrants, Third-World asylum-seekers, and similar outgroups-as well as aggres­ sive nationalism or localism " ( T 9 9 2, 2 6 8 ) . Eatwell refers to this as a " holistic" form of nationalism, one typically based on an ethnic concep­ tion of nationhood. This form of nationalism " stresses conversion, expul­ sion, or worse of the ' Other' and the defence of a traditional conception of community" ( 2ooo, 4 1 3 ) . The exclusionism inherent in this worldview makes immigrants a particularly likely target. Thus, it is not surprising that immigration is taken to be " the extreme right's issue par excellence" (Hainsworth T 9 9 2b, 7 ) , a view echoed in most subsequent studies of extreme right party ideology (e.g., Mudde 2ooo; Swyngedouw and Ivaldi 200 T ). Our analysis of these political movements centers on all national parliamentary elections in the sixteen major West European democracies from T 9 70 through 2000.

Institutions and Support for Parties of the Extreme Right

r 65

While the percentage of the total vote cast for parties of the extreme right in national European elections since 1 9 8 0 has typically been small, many of these parties have been far more successful at garnering electoral support than observers had anticipated. Their resurgence has indeed at­ tracted considerable attention, because the combination of xenophobia and populist antisystem sentiments embodies a blunt challenge to norms of tolerance in liberal democratic societies. A particularly dramatic recent success was achieved by the French National Front in the first round of the 2002 presidential election, when Jean-Marie Le Pen came in second ahead of Lionel Jospin, the sitting prime minister, and not far behind Jacques Chirac, the sitting president . 1 The extreme right resurgence has also included participation in government by the parties involved or their successors, as in the cases of Italy in 1994, Austria in early 2000, and most recently the Netherlands in July 2002.2 With these exceptions, how­ ever, levels of support for the extreme right have been insufficient to result in direct cabinet involvement. Nonetheless, the significance of recent right-wing extremism should not be minimized by the fact, say, that French voters overwhelmingly endorsed Chirac against Le Pen in the second round of the 2002 election, or that occasional direct participation in government may have moder­ ated the policy positions of some of these parties. Indeed, there are at least two reasons for believing that this revival has appreciable indirect political effects by which extreme right parties influence the behavior of larger, more established parties. First, fragmentation (the number of parties in a given party system) and polarization of party systems (the proportion of legislative seats held by extremist parties) have been identified as factors that undermine political stability, defined as government duration ( see, e.g., Taylor and Herman 1. For a description of this event, see "After the Cataclysm, " Economist, April 27, 2002, 2 5 - 27 . Le Pen was overwhelmingly defeated in the second round of the election on May 5 by Chirac. 2. The government formed in Italy after the 1994 general election included the National Alliance Party, successor to the neofascist Italian Social l'vlovement. The Austrian Freedom Party associated with Jiirg Haider was part of the governing coalition from Febru ary 2000 through most of 2002. The short-lived coalition government formed in the Netherlands in July 2002 after the 2002 general election (and which failed just three months later) included four cabinet members from the Pim Fortuyn List, a new political movement with a strongly anti-immigrant plank whose leader (Fortuyn) was assassinated j u st prior to the election. For discussion of the Italian and Austrian cases, respectively, see Newell ( 2000 ) and Luther ( 2ooo ) . In both instances, the inclusion of these parties in the ruling coalitions elicited considerable criticism from other European governments (see " Mussolini is Dead-for the l'vloment," Economist, June r r , 1994, 46, and "A Conundrum for Austria-and for Eu­ rope, " Econornist, February 5, 2000, 4 5 ) .

r66

B E F O RE NOKMS

1 9 7 r ; Powell r 9 8 6b; Laver and Schofield 1 990; Warwick 1 9 9 2a ) . These patterns are hardly surprising. As Sartori and Warwick have pointed out, the presence of antisystem parties in legislatures reflects ideological divi­ sions within governments themselves, which in turn directly influences their duration. Governments terminate because they contain member parties that can­ not agree with one another on government policy. Polarization plays an indirect, although powerful role in this account because it indicates the root cause: the size of the extremist or antisystem presence in the legislature . The greater the antisystem presence, the more necessary it becomes to include most of the remaining party spectrum if a pro­ system maj ority government is to be formed. Because the entire party system is ideologically polarized, however, coalitions formed on this basis tend to be ideologically diverse. (Warwick 1 9 9 2a, 3 4 7 ; see also Warwick 1 9 9 2b; Sartori 1 9 7 6 )

Second, Kitschelt ( 199 5 ) has argued that when they engage i n strategic convergence around the center to maximize their rates of participation in governments, mainstream conservative or Christian democratic parties can provide a niche for the extreme right. To minimize the size of this niche and thereby forestall extreme right political movements, the estab­ lished parties have an incentive to move closer to the policy position of those movements. When they do move in this manner, extreme right positions obtain a place in the mainstream political agenda and thus accrue a degree of legitimacy. In this vein, Schain describes how the conventional parties of the right in France, the RPR and the UDF, had by 1 9 8 6 attempted to preempt the National Front by calling for greater efforts to encourage immigrants to return " home," to reduce welfare benefits to resident immigrants, to make naturalization more difficult, and to rein in " delinquency" among immi­ grants.3 The Communist and Socialist response was somewhat softer but hardly ignored the issue, with calls for tighter controls on illegal immi­ grants ( Schain 1 9 8 7, 24 2 ) . By 1 99 1 , in response to conservative claims that she was " soft" on immigration, the Socialist prime minister Edith Cresson went on the offensive against illegal immigrants and proposed that de­ tained illegals be returned to their countries of origin on chartered planes 3· In 1976, the Gaullist bloc became the Rassemblement pour Ia republique ( RPR), while the Union pour Ia democratic fran�ais (UDF) emerged in 1978 from a federation of centrist and conservative groups.

Institutions and Support for Parties of the Extreme Right

r 67

(Brechon and Mitra 1 99 3 , So). Similarly, both the Free Democrats and the Social Democrats of Germany substantially modified their positions on immigration and on political asylum during the 1 990s by endorsing more restrictive immigration policies (Backer 2000, r r o - r r ). And in Switzer­ land during the 1 990s, the established parties on the right also began campaigning on issues of immigration and to a lesser extent law and order ( Gentile and Kriesi 199 8 , 1 3 6- 3 7) . Even without direct participation i n national governments and al­ though they might technically be marginalized, then, parties of the ex­ treme right can exert a substantial impact on the mainstream political agenda as more established parties adopt part of their policy positions in an effort to forestall them. This boosts their own legitimacy and, in the process, reduces the unpalatability of xenophobic and related political appeals as those appeals become part of the core political dialogue.

Expl ain i ng Con temporary Parties of the Extre me Right

How, then, is the recent revival of right-wing extremism in Western Eu­ rope explained? Most accounts have focused on particular parties in a given country and have typically emphasized explanatory factors idiosyn­ cratic to the case under discussion.4 These studies have generally not concerned themselves with institutional sources of support, which, to­ gether with their case-oriented approach to the subject, means that they are cast implicitly in more cultural terms. On occasion, the cultural argument is more explicit in the case studies. For example, the relative weakness of the extreme right in the Nether­ lands during the 1990s has been attributed to the absence of a Dutch nationalist subculture. One reason for this may be the long tradition of the Netherlands as a trading nation, with an international orientation that does not allow for narrow-minded nationali sm. Also, though a reasonably young coun­ try within its current borders, the Netherlands has never been threat­ ened in its national identity or integrity, except for the five years of German Nazi occupation. This persistent feeling of " national security " may explain the absence of the national question from the political agenda. (Mudde and Van Holsteyn 2000, r 64 - 6 5 ) 4 · Relevant country studies include Betz ( 1 990 ) , Mayer and Perrineau ( 1 9 9 2 ) , Minken­ bcrg ( 1 9 9 2 ) , Schain ( 1 9 8 8 ) , Vocrman and Lucardic ( 1 9 9 2 ) , a n d Wcstlc a n d Nicdcrmaycr ( 1992).

r68

B E F O RE NOKMS

This claim, of course, invokes a familiar set of images o f the Netherlands as distinctively tolerant. By the same token, the images offer little leverage in helping understand the obvious appeal of Pim Fortuyn, a right-wing populist with a starkly anti- immigrant program. In the parliamentary election j ust following his assassination in May 2002, Fortuyn's party (organized j ust three months earlier) finished second with I 7 percent of the votes, ahead of each of the parties that had for the preceding eight years constituted the coalition government. Juxtapose Mudde and Van Holsteyn's description of the Netherlands with a contemporaneous analysis of Belgium, where the extreme right has been a political force for a longer period. There, the relative success of the Vlaams Bloc has been explained in terms of cultural changes that origi­ nated in the r 9 6os. Secularization, the breakdown of traditional religious and sociopoliti­ cal barriers (i.e., " depillarization " ) , the economic crisis, computeriza­ tion, technocratization, and media-ization, the rise of a dual society and neoliberalism, and the legitimization of a politics of exclusion com­ bined to lead to the formation of value reorientation on the part of lower and lower-middle classes. ( Swyngedouw r 9 9 8 , 7 3 )

Most o f the items o n this list o f cultural changes adumbrated by Swyngedouw have a familiar ring because they feature in so many discus­ sions of the advanced democracies, discussions going back at least to David Riesman's image of the " lonely crowd " ( r9 s o ) . Observe the appar­ ent implication of Swyngedouw' s analysis either that these changes were consequential in Belgium but not in the adjacent Netherlands, or that secularization, the economic crisis, computerization, neoliberalism, and the like occurred in the former but not in the latter. Other analysts have gone beyond case studies to cast the net in broader terms . 5 While their discussions emphasize a variety of themes, a recurring motif has been a " mass society" interpretation that casts right-wing ex­ tremism as a grass roots political response to insecurity. Basically, this argument extends Swyngedouw's interpretation of events in Belgium be­ yond that particular case. The specific claim is that while many benefit from rapid economic growth and the introduction of new technologies into the workplace and elsewhere, these same processes marginalize siz­ able minorities both economically and socially. Thus, according to Betz: 5 · More general studies include Betz ( I 9 9 3 ) , lgnazi ( I 9 9 2 ) , and the essays in von Beyme (T988).

Institutions and Support for Parties of the Extreme Right

r 69

Contemporary societies' demands for flexibility and mobility have led to the fragmentation and decomposition of traditional milieus and so­ cial institutions while furthering tendencies towards the individualiza­ tion of life's chances and life's misfortunes . . . . [The marginalized in­ clude l young people, unskilled or semiskilled low income workers, elderly people drawing small pensions, farmers fearing economic and social downward mobility, and lower level employees . ( r 990, 47-4 8 )

Marginalization and the ensuing social isolation make such groups more susceptible to new collective identities and simple solutions such as those offered by extreme right-wing ideologies or organizations that claim to reaffirm traditional values (Westle and Niedermayer 1 9 9 2 , 9 5 ) . The argument is cultural on at least three counts. First, the posited link between insecurity or marginalization and support for extremist political movements presupposes that the former (like the latter) varies across national contexts, since a constant cannot explain a variable and given the observed variance in support for extremist political movements. Yet the mass society interpretation offers no explanation for why significant numbers of the French or German populations should be marginalized while their counterparts in Britain or the Netherlands are not, beyond the implicit suggestion of cultural differences. Second, the argument has a social capital flavor to it: sizable minorities are marginalized and thereby detached from their communities. The ensuing diminished levels of inter­ personal and political trust make these majorities distinctively susceptible to extremist political appeals. Third, and perhaps most important, the mass society interpretation casts voting as overwhelmingly expressive. Support for extremist move­ ments is taken as a response to broader cultural changes from voters seeking scapegoats for current ills along with simple solutions to policy issues that are inherently complex. Thus, such support is more the product of an emotional form of protest voting, a form with little policy content. The mass society argument has a familiar ring because it was central to earlier efforts to understand extremist politics generally and the rise of European fascism in particular ( see esp. Adorno et a!. I9 s o; Kornhauser I9 5 9 ; Lipset r 9 6o ) . We nonetheless find the interpretation wanting be­ cause it casts right-wing extremism as a more or less spontaneous mass­ based emotional phenomenon typically manifested in some countries but not in others for unspecified historical or cultural reasons. At the same time, it overlooks the systemic political and economic conditions that create opportunities for political entrepreneurs to mobilize extremist po­ litical movements in the first place.

1 70

B E F O RE NOKMS

Indeed, the mass society interpretation o f the Nazis' electoral success in the I 9 20s and early I 9 3 os has been challenged on largely the same grounds ( see esp. Hamilton 1 9 8 2; Brustein 1 9 9 6; King et al. 200 2 ) . The available data on the more recent period are at best mixed. For example, Republi­ kaner support among low-income Germans is not significantly higher than among more prosperous voters (Westle and Niedermayer 1 9 9 2 ) . Studies of National Front voters in France during the 1 9 8 0s similarly show that differences in levels of support for the National Front across occupational, educational, and age groups are minor to nonexistent (Mitra I 9 8 8; Lewis­ Beck and Mitchell 199 3 ). A cross-country study of support for the extreme right in the middle 1990s reports more consistent effects of unemploy­ ment, age, and gender (with unemployed, younger, or male voters being more supportive) , while the effects of education on support varied by country in an idiosyncratic manner (Lubbers, Gij sberts, and Scheepers 200 2 ) . Thus, the available evidence does not consistently suggest that the extreme right is dependent on and springs from a marginalized and inse­ cure social base as suggested by the mass society interpretation. We believe it is more profitable to examine the effects of structural, system-level conditions on extreme right party support. Specifically, it is important to explore how party systems and particular economic and social conditions influence the electoral success of these parties. Such factors do not provide a complete explanation, as we note later, but they do contribute critical information on the circumstances that provide op­ portunities for entrepreneurs to mobilize visible political movements. Fur­ ther, this focus directs our attention to elements that are, in varying degrees, amenable to direct policy intervention. This is a crucial feature for those interested in minimizing the electoral success of the extreme right, one that distinguishes explanations centering on party systems and economic conditions from those emphasizing societal and cultural traits. We address the political and economic factors in turn. Political Conditions As is the case with any newer or smaller political parties, the success of those on the extreme right should hinge critically on the barriers to entry in the legislature that they face. Higher barriers deter electoral support for parties outside the mainstream among potential supporters concerned with wasting their votes. Two political conditions seem especially ger­ mane in this respect. First, to what extent do electoral systems encourage a proportional translation of popular votes into legislative seats ? Second, to what degree does multipartyism prevail in the party system as a whole ?

Institutions and Support for Parties of the Extreme Right

I7I

These two political conditions have, perhaps, received more scholarly attention than others. They have been causally linked in Duverger's ( I 9 6 3 ) well-known proposition that single-member district, plurality methods foster two-party systems, while more proportional electoral procedures promote multipartyism. This proposition has a long history and has also generated a good deal of discussion since Duverger's original inquiry. 6 While smaller parties are underrepresented everywhere, the extent of this bias increases with electoral disproportionality. In highly dispropor­ tional systems, voting for minor parties becomes a rather futile act: "The electors soon realize that their votes are wasted if they continue to give them to the third party, whence their natural tendency to transfer their vote to the less evil of its two adversaries" (Duverger T 9 6 3 , 2 2 6 ) . The wasted vote phenomenon, b y the same token, deters political entre­ preneurs from mobilizing electoral support through the channel of a minor party (Blais and Carty 1 99 1 ) . If disproportionality in electoral systems thus discourages multipartyism in general, we would expect it to dampen the prospects for smaller parties of the extreme right in particular. Were there a perfect relationship between electoral disproportionality and multipartyism, we would not also have to consider the effects of the latter on support for the extreme right. However, as Duverger, Sartori, and many others since have emphasized, there is a variety of " propor­ tional representation" schemes that vary considerably in the forms of multipartyism they generate. The relationship between these two factors is consequently far from perfect. For example, the simple correlation between Lijphart's preferred measure of disproportionality and the effec­ tive number of parliamentary parties is - ·4 5 for the seventy electoral systems in his analysis ( 1 994, 7 6 ) . While this correlation has the expected sign, it is hardly strong. We therefore need to explore any additional effects of the number of political parties on support for parties of the extreme right. This enables a direct evaluation of the view that multi­ partyism provides a larger opportunity than is typical of other party systems for political entrepreneurs from smaller parties ( including those of the extreme right) to mobilize electoral support. There are three different ways in which these political variables might influence support for the extreme right. The first possibility is that each has a simple direct and independent effect. Second, if Duverger's basic argument that disproportionality inhibits multipartyism is correct, then 6. See, among others, Rae ( r 9 7 r a ) , Riker ( r 9 8 2 ) , Duverger ( r 9 8 6 ), Sartori ( r 9 8 6 ) , Lijphart ( T 994), and Cox ( T 9 9 7 ) .

I72

B E F O RE NOKMS

disproportionality may exhibit no direct effect o n extreme right support with multipartyism controlled, suggesting that the impact of dispropor­ tionality is wholly indirect and mediated by multipartyism. The third possibility is drawn from Lijphart's ( r994, 76-77) conclusion that dispro­ portionality and multipartyism are loosely interdependent. In contrast to the first two patterns, this would lead us to expect a statistical interaction between the political variables, such that the magnitude of the effect of each depends on the value of the other. The specific interaction is one where the dampening effect of disproportionality on extreme right sup­ port increases with multipartyism, while the positive effect of multiparty­ ism diminishes with rising disproportionality. Economic Conditions The basic component of economic performance that should most closely influence the success of extreme right-wing parties is unemployment. These parties are distinguished by their neofascist and anti system stances, which in recent years have typically centered on immigrants and foreign workers. All else equal, we would anticipate political scapegoating of this sort to find much more fertile ground when jobs are scarce than when they are plentiful. The targets involved are, after all, visible and readily linked to job losses. Thus, with its key political slogan "Two million immigrants are the cause of two million French people out of work, " the French National Front explicitly linked the two and criticized both the left and the right for their ineffective responses to unemployment prob­ lems ( Mitra 1 9 8 8 , 5 1 ) . The German Republicans have been equally blunt and explicit on the linkage: " Eliminate unemployment: Stop immigra­ tion ! " (Betz 1 99 3 , 4 1 6) . Higher levels of unemployment would seem to provide the optimal conditions for mobilizing protest campaigns on the extreme right with appeals like these. As with the recent success of right-wing parties, of course, substantial unemployment came as a surprise to many. Much attention centered in­ stead on new issues said to surround postindustrial and postmaterial soci­ eties, issues that transcend such mundane questions as jobs. Since the early 1 9 70s, however, levels of unemployment have increased substantially in Western Europe. Where average levels of unemployment in the European Community hovered between 2 and 3 percent in the early 1 9 70s, they had risen dramatically to around ro percent by the middle r 9 8os; only the smaller countries of the European Free Trade Area avoided increases of this magnitude. The problem has been especially severe for youth, manual

Institutions and Support for Parties of the Extreme Right

1 73

workers, and women. 7 Given these patterns, unemployment has become a central political issue. � None of this assumes that parties of the extreme right depend dispropor­ tionally on unemployed, blue-collar, lower-income, or younger voters, although there is some evidence that unemployment, youth, and being male do increase the propensity of individuals to vote for these parties (Lubbers, Gij sberts, and Scheepers 200 2 ) . Nor, since we are dealing with a form of protest voting, are we proposing that unemployment and extreme right support are linked by a sociotropic mechanism, according to which the " voter is influenced most of all by the nation's economic condition " (Kinder and Kiewiet 1 9 8 1 , 1 3 2) . Instead, we are simply proposing that high aggregate rates of unemployment reveal mediocre economic perfor­ mance that provides an especially propitious context for political crusades of the form favored by the extreme right, whose electoral support we therefore expect to increase directly with unemployment.9 Immigration Immigration has been at the core of the contemporary right's appeal. Even allowing for minor cross-country variations, this has served as the extreme right's defining issue, so that it is plausible to anticipate that the success of the extreme right would hinge on the size of the immigrant population. All else equal, larger immigrant populations provide a more fertile context within which to mount xenophobic political appeals. Fur­ ther, since the campaigns of extremist parties often link levels of immigra­ tion and unemployment, we explore whether these campaigns are more successful when higher levels of unemployment are coupled with larger immigrant populations. At the same time, we recognize that there is a potential and distinctive 7· The most thorough treatment of these issues is offered by Layard, Nickell, and Jackman ( 1 9 9 1 , 2, 7, 2 8 8 - 90 ) , who compare unemployment trends across the EC, EFTA, the United States, and Japan (EFTA includes Austria, Finland, Norway, Sweden, and Switzer­ land) ; age and gender differences in unemployment; and differences by occupational stand­ ing (in most cases, the ratio of manual to non manu al unemployment rates is approximately 2 : 1 ) . See also Nickell ( 1 9 97, 1 9 9 8 ) . 8 . Indeed, by the early 1990s i t was described a s " Europe's most pressing domestic problem" ( " Getting Europe Back to Work, " Economist, August 2 8 , 1 9 9 3 , 4 3 -44; see also "Jobless Europe," Economist, June 26, T 9 9 3 , T 7 ) . 9 · A parallel argument i s made by Lewis-Beck and Mitchell ( 1 9 9 3 ) in their analysis of support for the bench National Front. Comparing aggregate data across departments, they find that support for the FN is a function of unemployment and immigration levels, along \Vith crime rates.

I 74

B E F O RE NOKMS

endogeneity problem when immigration levels are cast i n this manner. These levels are, after all, a function of immigration control policy among other things, an area that is especially politicized (Money I999 ). More­ over, we have already observed that mainstream parties participating in governments often pursue more restrictive immigration control policies or adopt more restrictive rhetoric on the issue to maintain their own electoral advantage and to forestall or reduce electoral gains by extremist parties. To the extent that they engage in these tactics, the question of what is giving rise to what becomes somewhat muddied. Bearing this in mind, we delay our evaluation of the potential role of immigration until we have provisionally estimated a basic model.

D ata

We examine all general election years from r 9 70 through 2000 for r 6 West European countries with populations larger than one million. This yields r 4 2 country elections for analysis, and the data are organized in an unbalanced or a pseudopooled design, so labeled because the intervals between elections vary both within and across countries. ' 0 Measuring the dependent variable requires the classification of political parties' ideological posi­ tions. Of the alternative ways of addressing this question, a commonly employed method of constructing party location scales relies on the views of country experts. The most systematic and widely used ratings by ex­ pert j udgments of parties' positions on a left-right scale are reported by Castles and Mair ( r 9 84 ) , who asked a sample of expert raters to place all of their country's parties on the following ten-point left-right scale: ultra­ left (o- 1 . 2 5 ) ; moderate left ( 1 . 2 5 - 3 . 7 5 ) ; center ( 3 . 7 5 - 6. 2 5 ) ; moderate right ( 6. 2 5 - 8 . 7 5 ); ultra-right ( 8 . 7 5 - r o) . " Since a number o f small parties are not included in Castles and Mair's

Extreme Right- Wing Party Support (ERPS).

r o . The countries are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom. r r . Castles and Mair ( r 9 8 4 ) report that respondents were able to assign numerical values to their country's parties according to instructions even though this procedure does not take into account the possible multidimensional meaning of the left-right language ( see also l'v1air and Castles I 9 9 7 ) . For a summary and evaluation of various methods of con­ structing policy scales, see Laver and Schofield ( r 990, appendix B) and Gabel and Huber ( 2ooo ) , and for evidence of the salience of left-right orientations to voters, see Huber ( r989).

Institutions and Support for Parties of the Extreme Right

I75

scales, the sample of ERPs identified through expert judgments is ex­ panded using Ignazi's ( r 9 9 2) criteria. Besides spatial location, Ignazi relies on the analysis of electoral policy programs to identify the family of ERPs in Western Europe. To be categorized as an ERP, the most right-wing parties in a given country had either to " fulfill the historic-ideological fascist criterion or to exhibit a delegitimizing impact, through a series of issues, values, and attitudes (rather than a structured and coherent ideol­ ogy), which undermines system legitimacy" (Ignazi r 99 2, 1 2 ) . The ERPs thus include two variants: the older version consists of parties with an explicit fascist imprint, while the newer version comprises more recently formed parties with no direct fascist associations but with a right-wing populist anti system attitude. Further, the evidence suggests that the resur­ gence of the extreme right in recent years has disproportionately entailed parties of the second form.n However, it is important not to overstate the significance of this distinc­ tion, since both versions of the extreme right emphasize a xenophobic nationalism, as we have already observed. Indeed, efforts to sustain the distinction empirically too easily become strained. Consider the most recent such effort by Golder ( 200 3 a ) , who distinguishes neofascist from populist parties. Populism is distinctive in terms of its " appeals to the people" and its antisystem stance. However, Golder concedes that main­ stream parties also " appeal to the people from time to time " and simulta­ neously claims that the antisystem stance of the extreme right does not mean " that it is antidemocratic in any sense " ( 2003 a, 4 4 7 ) . The distinc­ tions are at best ungainly and are certainly difficult to substantiate empiri­ cally. 1 3 In the following analyses, we focus on parties of the extreme right, taken as a group. In identifying extreme right parties, we have relied primarily on Ignazi's classification, which is based on an extensive list of sources. We r 2 . In addition to lgnazi ( r 9 9 2 ) , see Taggart ( I 99 5 ) and especially Kitschelt ( 1 99 5 ) . It is thus not surprising that, following its participation in the Berlusconi government in 1994, the Italian Social Movement was officially replaced by the National Alliance in 1995 in a formal effort to sever the link between the extreme right and the Mussolini legacy ( " Fini Snuffs Out Fascist Flame to Try to Boost Party," Financial Times, January 26, I 99 5 , 2 ) . r 3 . Golder's own figure I indicates that h i s neofascist category comprises a small component of extreme right support. Looking at his data another way, the correlation between populist support and total extreme right support is ·97 (N = I 5 I ) , while that between neofascist and total extreme right support is j ust . I 7 . When populist support is regressed on total extreme support, the estimated coefficient is ·9 5 ( s . e . : . 0 2 ) , with a con­ stant of - . 3 6 ( s . e . : . 0 7 ) ; these figures are close to r .o and o, respectively. Alternative estimators of these associations that accommodate the left-censored nature of the data imply the same conclusion (we address the censoring issue later) .

I76

B E F O RE NOKMS

TABLE 6. 1 . Extreme Right Parties Included in the Analysis Country

Party

Austria Belgium Denmark France Germany

Freedom Party Flemish Block Progress Party National Front National Democratic Party Republicans National Political Union Party of the Progressives Italian Social Movement Northern League Centre Party Progress Party Christian Democratic Party Popular Alliance [now Popular Party] New Democratic National Action/Vigilantes National Front

Greece Italy Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom

have also consulted other studies that appeared after lgnazi's study was completed.'4 The full list of ERPs included in our analysis is displayed in table 6. r . ' 5 Electoral disproportionality can be evaluated in a variety of ways. Most of these center on comparisons of the differences between the share of votes received by a party and its share of legislative seats. We employ a slightly different approach that entails the electoral threshold, that is, the minimum level of support that a party needs to gain legislative representation. Our choice is governed by two consider­ ations. First, Lij phart ( r 994, r o 7- r 3 ) provides considerable evidence to show that electoral thresholds and disproportionality are closely linked. ' 6

Effective Threshold (THRESH) .

I + Along with Taggart ( r 99 5 ) , these supplementary sources include country studies that appeared in two special issues of Parliamentary Affairs (vol. 4 5 , July r992, and vol. 5 3 , July 2000 ) , Hainsworth ( T 99 2a, 2000), Merkl and Weinberg ( T 9 9 2 ) , a n d Betz a n d lm­ merfall ( r 9 9 8 ) . r 5 . Data o n the percentage of votes these parties received i n each general election year from r 9 70 through 2000 are taken from lvlackie and Rose ( r 9 9 r , r 9 9 7 ) . Figures on the most recent elections are from www.electionworld.org/ (Wilfried Derksen, Elections around the World ) , and www.parties-and-elections.de/indexe.html (Wolfram Nordsieck, Parties and Elections in Europe ) . r 6 . Further evidence o n the link between effective thresholds and disproportionality comes from our own results : when a direct measure of disproportionality is substituted for the electoral threshold measure, the estimates are very similar to those we report later. On

Institutions and Support for Parties of the Extreme Right

1 77

Second, thresholds are a feature of electoral systems that are subject to policy adjustment and are readily interpretable. As Taagepera puts it, " Mention a 5 percent effective threshold to the leader of a party that usually obtains 3 percent of the votes and he or she knows very directly what it would mean for the party" ( 1 9 9 8 , 3 9 8 ; see also Lijphart 1 9 9 7 ) . No measure of disproportionality has an equally intuitive meaning. Data on effective thresholds, by country and electoral system, are from Lijphart's ( 1 994, r o - 4 7) tables 2 . 1 - 2. 8 , using the procedure proposed by Taagepera ( 1 99 8 ) and Lijphart ( 1 999, 1 5 3 - 54 ) , and updated by the authors for the r 990- 2000 elections. To gauge the de­ gree of multipartyism, we employ a measure of the effective number of political parties in the legislature. Developed by Laakso and Taagepera ( r 979 ), this measure takes account of the relative size of the parties in a given party system. The "effective number of parties " is calculated as follows.

Effective Number of Parliamentary Parties (ENPP) .

ENPP =

1

­

Ls?

where s; is the proportion of seats (lower house) of the i-th party. Data on elections from 1 9 70 through 1990 were generously provided by Arend Lijphart and updated by the authors for the elections between T 99 T and 2000. Since these figures, like those for effective thresholds, are by elec­ toral system rather than by country, they often vary within countries across the years from 1 9 70 through 2000. ' 7

Unemployment (UNEM) . Figures o n unemployment are taken from the Economic Commission for Europe. ' � These unemployment rates are based the measurement of disproportionality, see Gallagher ( 1 9 9 1 , 3 8 - 4 0 ) . Note that the number of electoral systems is greater than the number of cases because the key features of electoral systems are not time-invariant. T 7 . See footnote T 5, this chapter, for the Web addresses for the sources of data that we used to calculate the most recent country values for effective threshold and effective number of parties. 1 8 . Economic Commission for Europe, Economic Survey of Europe in I 9 9 0 - I 9 9 I (New York: United Nations, T99 T ) , appendix table A. T 2 . The only case for which unem­ ployment data are missing is Switzerland, 1 9 7 1 . More recent data are from the UN Eco­ nomic Commission for Europe, 1-:conomic Survey of t:urope 2 0 0 3 No. I, appendix table A.1o, available at http://www.unece. org/ead/ead-h.htm. We substituted the Swiss unem­ ployment figure for 197 5 for the missing 1 9 7 1 value, but none of the later results hinge on this interpolation.

178 TAB LE 6.2.

B E F O RE NOKMS

Sum mary Statistics for All Variables (N

Variable Extreme right support (ERPS) Log extreme right support (lnERPS)" Effective threshold (THRESH) Effective number of parties (ENPP ) Unemployment rate (UNEM) Immigration level (IMMIG)h

=

1 42)

Minimum

l\1aximum

l\1ean

Standard Deviation

0.0 0.0 0.7 1.7 0.3 0.2

44.5 3.8 37.5 1 1 .1 2 1 .2 1 9 .4

4.79 1 .08 8.84 3.93 6.50 4.47

7.98 1.11 10.32 1 . 62 4.22 4.25

<�To preserve a true zero score, this is the natural logarithm of (ERPS + 1 ) . 1w = 130

on estimates by the OECD and are expressed as a percentage of the total labor force. Figures on immigration are taken from SOPEMI reports (various years), supplemented by country studies, and reported in Golder ( 2003 a ) . • 9 The size o f the immigrant population is measured a s the percentage of foreign citizens by election year.

Immigration (IMMIG) .

Summary statistics for all of the variables for the T 42 elections between 1 9 70 and 2000 are reported in table 6 . 2 . Not unexpectedly, the distribu­ tion of the measure of extreme right party vote in the first row of the table is heavily skewed right: for example, the mean (4. 79 ) is substantially smaller than the standard deviation ( 7 .9 8 ) . In the analyses that follow, we have therefore transformed this variable logarithmically, having first in­ cremented the raw ERPS scores by r to preserve the zero point. Basic statistics for this transformed form of the dependent variable are also displayed in table 6.2 (row 2 ) .

Coeffi ci e n t Es tim ates

Our discussion to this point implies that aggregate levels of support for parties of the extreme right are a function of two interrelated political factors. We expect ERPS to be higher the lower the electoral threshold (i.e., the more proportional the electoral system) and the greater the number of 1 9 . SOPEMI is the acronym o f the Systeme d'observation permanente d e s migrations internationales, sponsored by the Organisation for Economic Cooperation and Develop­ ment ( OECD ) .

Institutions and Support for Parties of the Extreme Right

1 79

parliamentary parties. Whether these effects are additive or jointly condi­ tioned is an empirical issue to be investigated. We have also hypothesized that ERPS is a positive function of the level of unemployment. Since the data vary spatially and temporally, we adopt a pooled cross­ sectional time-series format. The principal potential problem that we need to address is heteroskedasticity, so we include dummy variables for each country in the model (as recommended, for example, by Stimson 1 9 8 5 ) . Along with helping to address this statistical problem, these dum­ mies reflect other country differences not captured by the specified inde­ pendent variables. Such differences include minor cross-country varia­ tions in classifying employment status and other unmeasured differences across countries. 20 We return to this issue later. Our basic model specifies electoral support for parties of the extreme right as a function of THRESH, ENPP, and UNEMP. We anticipate the coefficient for THRESH to be negative, while the coefficients for ENPP and UNEMP should be positive. As indicated earlier, we also include a series of 1 5 country dummy variables (the country dummy variable for the other country is absorbed into the constant) for purposes of estima­ tion, although we do not report these coefficient estimates in the table. Observe that the effects of the two political variables are initially assumed to be additive, that is, these effects are presumed independent of each other. Since we have considered the possibility that the effects of the two political variables are jointly conditioned, we also estimate and report a model that includes a parameter for an interaction term between the two political variables. Given the inclusion of the country dummy variables, the typical estima­ tion procedure for models like these involves a least-squares dummy­ variable (LSDV) approach. We are reluctant to rely solely on a least­ squares estimator here, however, given the distribution of the dependent 20. In addition to the raw unemployment rates we use, OECD reports a series of " standardized" unemployment rates that are intended to control for small cross-country differences in the reporting of unemployment. Unfortunately, the spatial and temporal coverage of the standardized figures is limited. Since the standardization is intended to minimize cross-country differences in reporting, however, any such differences are also controlled with the country dummy variables that we employ. A more complete series based on the standardized figures that covers all of the countries we examine except for Greece and Portugal is provided by La yard, Nickell, and Jackman ( r 9 9 r , 5 2 6- 29 ) . A regression of our measure on the standardized version (N = 89) yields an estimated constant of . o r (standard error, . r r ) , a slope of r . o r ( standard error, . o r ) , and a n R� of .9 8 . Thus, i t is not surprising that substituting the standardized figures for those we usc later (and thereby excluding Greece and Portugal from the analysis) provides estimates very similar to those reported in tables 6.2 and 6.3 . We therefore conclude that the standardization procedure involves quite minor adjustments .

I80

B E F O RE NOKMS

variable. While the logarithmic transformation reduces the skewness of this measure considerably, for a third of the election years (4 8 of the I 4 2 ) parties o f the extreme right garnered n o electoral support a t all. Because the dependent variable is left-censored in this way, ordinary least squares can generate biased and inconsistent estimates, the magnitude of the bias hinging on the ratio of censored to uncensored observations (Amemiya 1 9 84 ; Kmenta 1 9 8 6) . Following other studies, w e could modify the design by excluding instances where the dependent variable is left-censored and focus instead on variations in ERPS across those election years where extreme right parties received at least s o m e electoral backing ( see, e.g., most of the country studies, or Carter 2002 ) . However, this strategy involves eliminat­ ing cases on the basis of (zero) values of the dependent variable, and it thereby introduces a selection bias into the analysis ( see, e.g., Berk 1 9 8 3 ; Geddes I 9 9 I ) . We need instead t o examine all cases that have been at risk of experiencing successful extreme right parties in order to understand more fully why these parties have been more successful in some countries or periods than others. Accordingly, we retain all election years in the analysis and employ an estimator that accommodates the limited nature of the dependent vari­ able but that also retains the variation in its uncensored component ( as, say, a logit model would not ) . Since the censoring always occurs at the value of zero, the most obvious candidate is the Tobit procedure, a maximum-likelihood estimator for (left- or right-) censored dependent variables. The Tobit estimates for the additive model are displayed i n the first column of table 6- 3 - All of the three substantive coefficients have the ex­ pected sign, and the model as a whole has a log likelihood of - 9 8 . p. The coefficients on both unemployment and the effective number of parties are statistically significant, with t-ratios of 4 . 8 and 4 . 3 , respectively. In con­ trast, the t-ratio for the coefficient on effective threshold is less than one. Additional perspective on these summary statistics comes from a baseline Tobit regression of the dependent variable on the country dummy variables only, which generates an overall log likelihood of - I I 8. 74 · The estimates for the additive model thus represent a considerable improvement over this baseline. At the same time, it is also evident that little of this stems from the 21

2 I . The Tobit model was first used by Tobin ( r 9 5 8 ), after whom it is named, in an analysis of the purchase of consumer durables. Useful discussions are available in Amemiya ( r 9 8 4 ) and Kmenta ( r 9 8 6 ) .

Institutions and Support for Parties of the Extreme Right TABLE 6.3.

Sources of Support for Parties o f the Extreme Right, 1 9 70-2000

Regressor Effective threshold Effective number of parties Threshold

X

r8r

Additive Model

Nonadditive Model

- .005 (0.3) .283 ''" (4.3)

.073 '' (2.3) .420 '" (5.4) - .027'' (3.0) . 0 8 8 '"

number of parties

Unemployment Country dummies" Constant Log likelihood

.095 ''" (4.8) 1 .408 ''" (5.4) - 9 8 . 52

(4.6) 1 .049 '" (3.8) - 94.24

Note: The dependent variable is the natural logarithm of (ERPS + 1 ) . Tobit coefficients i t-ratios ) : >:·Coefficients are more than twice their standard errors of estimate. (N = 142)

3Estimates for the 15 country dummy variables included in columns (1) and ( 2 ) are not displayed to simplify the presentation.

separate effects of the political variables; most of the improvement reflects the positive coefficient on unemployment. The second column of table 6. 3 shows the Tobit estimates for a non­ additive specification according to which the effects of the two political variables (THRESH and ENPP) are jointly conditioned. Comparing the two columns indicates that the nonadditive model has an improved fit over the first one, since the overall log likelihood is - 9 4 . 24 . Further, the coefficient for the interaction term is more than twice the size of its standard error, with a t-ratio of 3 .o. 22 This suggests that the nonadditive model of political effects offers a reasonable interpretation of the data . Because these estimates are for an interactive model, however, it is difficult to gauge whether the coefficients are correctly signed from the information in the second column of table 6. 3 alone. With the additive specification in the first column, of course, the effects of the two politi­ cal variables are represented by their respective separate coefficients. In 22. As a probe for robustness, we reestimated the model with a median regression estimator (on which sec Bcrk 1 9 9 0 ) . The resulting estimates arc quite similar to those shown in table 6. 3 . The median regression estimator ( also known as least-absolute value regres­ sion) is a maximum-likelihood procedure that fits by minimizing the sum of absolute (as opposed to squared) residuals and is thus less sensitive to extreme values on the dependent variable than OLS. As a further robustness check, we reestimated the second model in table 6. 3 removing, in turn, each of the three countries (Austria, Denmark, and Spain) with average logERPS scores of 2.0 or more. The Tobit estimates thus obtained closely resemble those estimated for all countries in table 6. 3 .

r82

B E F O RE NOKMS

contrast, with the nonadditive variation, there i s no single coefficient for either of the two variables. Instead, the influence of each is conditional on the values of the other ( see, e.g., Friedrich r 9 8 2 ) .23 What meaning then should we attach to the coefficients for the politi­ cal variables in the second and third columns of the table ? Essentially, the more complex formulation implies that the size of the dampening effect of electoral thresholds on support for the extreme right increases directly with the number of parties. At the same time, the coefficients in the second and third columns of the table establish that multipartyism encour­ ages voting for the extreme right most within proportional electoral sys­ tems; among systems with moderate to high effective thresholds, varia­ tions in the number of parties have much less bearing on support for the extreme right. To what extent do levels of immigration affect these patterns ? Bearing in mind the potential endogeneity problems to which we alluded earlier, we now add immigration to the analysis. Table 6.4 displays the relevant Tobit estimates. The first column reports the figures for a model where the effects of immigration and unemployment are conditional upon each other. 24 Cast in this manner, levels of immigration do not appear to be systemati­ cally related to support for the extreme right, since the estimates for both TMMTG and UNEMP '' TMMTG are smaller than their standard errors. In other words, there is little evidence in these data that the combination of high levels of both immigration and unemployment boosts the success of extremist parties. However, seen in simpler terms, there is some evidence that levels of immigration matter. The figures in the second column of table 6.4 show that the coefficient for IMMIG is correctly signed, and its t-ratio increases to r . 6 when an additive effect of this variable is specified. To be sure, the estimate falls short of conventional levels of statistical significance, even 2 3. In a commentary on our original analysis of support for the extreme right, Golder ( 2oo3 b ) has suggested that our estimates imply peculiar outcomes for particular values of each of the two explanatory political variables. There are two problems with his critique. First, the lower bound for the effective number of parties is not o, as he graphs it, but basically 2.0 (one election year has a smaller value of r .7 2 ) , so that his graphs are mislead­ ing in that the peculiar outcomes they identify arc outside the range observed in the data. Second, this problem with Golder's depiction of the data is compounded by the fact that the effects of the two political variables are specified as conditional on each other, so that we need to be concerned with their j oint distribution. As a simple plot of the two variables indicates, there arc no election years with j ointly high values on both the threshold and number of parties variables, so the empirical conditions he considers do not exist. 24- Since immigration data are unavailable for IJ of these election years ( Golder 2003 a, 4 6 3 ) , the number of election years analyzed in table 6.4 is I 3 0.

Institutions and Support for Parties of the Extreme Right

r83

employing a one-tailed test. Nonetheless, since the coefficient on immigra­ tion levels is positive, there is at least a hint here that immigration may have a mild effect at the margin on support for parties of the extreme right. There is one other issue. Although they are not displayed in the first two columns of tables 6 . 3 and 6.4, the estimates there are from models that include a full set of country dummy variables. Does this constitute evidence for some residual effect of cultural values on support for the extreme right ? We think not. First, the effects of the systematic variables in the second column are pronounced even with the country dummies included in the analysis, and the unlisted country dummies from the estimates in the second column follow no pattern that could readily be explained in terms of cultural differences. Second, as we emphasized in chapter 5, country dummies have no intrinsic substantive meaning, and simply indicate that country A differs from country B, which does not constitute even the beginning of an explanation. Any interpretation of such coefficients is thus necessarily conjectural. But conjecture does not lead inexorably to culture. Consider j ust one alternative scenario. The systemic political and economic influences we TABLE 6.4. 1 970-2000

I m migration and Support for Parties of the Extreme Right,

Regressor Effective threshold Effective number of parties Threshold

X

number of parties

Unemployment Immigrants Unemployment

X

immigrants

Country dummies" Constant Log likelihood

Nonadditive Model

Additive Model

.072"" (2.3) .41 8 ''" (5.2) - .02T (3.0) . 1 00''" (3.3) .077 ( 1 .3 ) .001 ( 0. 1 )

.072" ( 2 .4 ) .4 1 9 ''" (5.3) - .027" (3.1) . 1 02 '' (4.9) .080 (1.6)

.629 ( 1 .7) - 8 1 .23

.612 ( 1 .9 ) - 8 1 .24

Note: the dependent variable is the natural logarithm of (ERPS + 1). Tobit coefficients I t-ratios): >:·Coefficients are more than twice their standard errors of estimate. (N = 1 3 0 )

aEstimates for the 1 5 country dummy variables included i n columns ( 1 ) a n d (2) are not displayed to simplify the presentation.

r 84

B E F O RE NOKMS

have considered represent opportunities for political entrepreneurs on the extreme right to mobilize electoral support. However, the presence of such opportunities does not in itself guarantee the presence of entrepre­ neurs who can take advantage of them. This suggests that the dummy variables may, among other things, reflect unmeasured country differ­ ences in the supply of such entrepreneurs.25 In all, the estimates in tables 6. 3 and 6.4 thus suggest that support for the extreme right is a function of the electoral threshold, the effective number of parties, and the rate of unemployment. The figures in table 6.4 also hint that immigration levels may have a slight effect at the margin. The estimates in both tables further establish that the multiplicative specifica­ tion of the two political effects better represents the data than the additive specification. While the effects of the two political variables are thus more subtle than implied in the simple additive formulation, they are consistent with our basic expectations.

I m plications

Our analyses have centered on the systemic conditions that have favored parties of the extreme right in West European politics from r970 through 2000. We have shown that electoral and party-system factors interact with each other to account for the response that these parties have generated from voters. Specifically, electoral disproportionality (through the mecha­ nism of thresholds) increasingly dampens support for the extreme right as the number of parliamentary parties expands. At the same time, multi­ partyism increasingly fosters parties of the extreme right with rising elec­ toral proportionality. Our analyses also indicate that higher rates of unem­ ployment provide a favorable environment for these political movements. As we have already emphasized, the unemployment effect does not imply that parties of the extreme right have drawn the bulk of their support from economically marginalized individuals. While there is some 2 5 . Others have also addressed the role of entrepreneurs in the current context. For example, Lubbers, Gij sberts, and Scheepers (2002) stress the role of effective organization (which includes information on the charisma of the party leader) for extreme right party success. However, their measure of party organization comes from expert j udgments solic­ ited in 2000, which are then used to explain the success of parties in the middle 1990s. Since the success of these parties has already been observed, the expert judgments have a decid­ edly ex post flavor. In our view, while conditions that might be propitious for entrepreneurs can be identified in advance, determining ex ante which particular entrepreneurs will be able to capitalize on those conditions is much more difficult. Instead, the supply of effective entrepreneurs is best treated as stochastic.

Institutions and Support for Parties of the Extreme Right

r85

evidence that unemployed individuals were more likely to vote for ex­ treme right parties in the mid- r 9 90s (Lubbers, Gij sberts, and Scheepers 2002), our own aggregated data obviously do not speak directly to this issue. The unemployment effect we have reported is instead most profit­ ably seen as reflecting the influence of broad economic conditions. In our view, higher rates of unemployment epitomize uneven economic perfor­ mance that fosters support for the extreme right by providing the pretext for political entrepreneurs to mount the xenophobic political appeals that characterize these political movements. Increasing unemployment is sig­ nificant because it provides a fertile environment for such appeals. The unemployment effect may also help explain why the resurgence of extreme right parties in the late T 9 70s and during the T 9 8os was not widely anticipated. Many observers had focused instead on " new" nonmaterial issues, the emergence of which was said to be generated by fresh forms of political organization associated with the affluence that comes with postin­ dustrial and postmaterial society (Bell 1 9 60, 1 9 7 3 ; lnglehart 1 9 9 7 ) . These fresh political forms were expected to find expression on the left of the po­ litical spectrum (Kitschelt 1 9 8 8 ) . All these expectations, of course, were implicitly predicated on the assumption that the prosperity of the golden years starting around 1 9 5 0 would persist indefinitely. The decline in full employment that began in the late T 9 70s serves as a reminder that the older problems of industrial society continue to await resolution and that con­ flict due to material scarcity is far from obsolete. The unevenness of post­ war economic performance also cautions us against drawing sweeping conclusions about cultural changes involving new postindustrial or post­ materialist orders said to come with affluence. Our estimates suggest that one result of the job decline is a climate conducive to xenophobic and antisystem political appeals from the extreme right, a pattern with a de­ pressingly long pedigree. like unemployment, electoral proportionality and multipartyism are systemic factors that together facilitate parties of the extreme right in the manner we have described. This pattern is encouraging for those of us who view these parties as a blight on the politics of liberal democracies, because, unlike more deeply rooted cultural values or the social patholo­ gies discussed in mass society accounts of extremist politics, all three of these factors can be adjusted through policy intervention. These patterns are notable, because electoral proportionality is some­ times offered as the guiding principle of electoral systems, a value to be maximized above all others. We trust that our analyses serve to underscore the trade-offs between efficiency considerations and the " function of mir­ roring " in any representational system, to use Sartori 's term ( T 9 6 8 b,

r86

B E F O RE NOKMS

4 69 ) . 26 The recent electoral resurgence o f parties on the extreme right i s a vivid reminder that proportionality is not without its costs. Casting the net more broadly, Sartori noted over three decades ago that the electoral system in general is " the most specific manipulative instrument of politics" ( 1 9 6 8 a, 273 ) .27 Given this observation and Lijp­ hart's analysis, electoral disproportionality appears to be the most readily modified of the three systemic factors we have identified. Specifically, Lijphart shows that changes in the effective threshold for parliamentary representation influence electoral proportionality most immediately, but that the latter is also sensitive to the electoral formula and assembly size. While the relationships are somewhat weaker, Lijphart also establishes that the effective number of parliamentary parties responds to the same three elements. Unemployment rates may be the least tractable from a policy standpoint. Even so, there is evidence that levels of unemployment reflect political choices as embodied in such broad factors as the partisan composition of governments, at least in the short run (Alesina 1 9 8 9 ; Alt 1 9 8 5 ; Hibbs 1 9 8 7; Korpi 1 99 1 ) . There is further indication that unem­ ployment rates are sensitive to active (and cost-effective) intervention in such concrete labor market programs as adult training and recruitment subsidies (see, e.g., Calmfors and Nymoen 1990; jackman, Pissarides, and Savouri T 990; Layard, Nickell, and jackman T 9 9 T ) . We do not, of course, mean to suggest that the possibilities for (and effectiveness of) electoral engineering or other policy innovations are unlimited (Shugart 1 9 9 2; Lijphart 1 9 9 4 , 1 3 9 - 5 2) . Electoral laws are often somewhat sticky. And when reform is undertaken, the process is typically surrounded by considerable uncertainty so that the key political actors involved often miscalculate.28 But we do wish to emphasize the point that since the electoral successes of extreme right-wing parties are a 26. See also Sartori ( 1 994, esp. chap . 4 ) . There is a further distinction between propor­ tionality from the perspective of ( 1 ) political parties (the conventional approach ) and ( 2 ) individual voters. For a n analysis o f the possible trade-offs between these two forms of proportionality, see Riedwyl and Steiner ( 1 99 5 ) . 2 7 . This claim i s o f more than academic interest. The view that electoral laws can be manipulated to secure political goals is, of course, widespread (see, e.g., " Electoral Reform: Good Government ? Fairness? Or Vice Versa. Or Both . " Economist, l'vlay 1, 1 99 3 , 1 9 - 2 1 ) . 2 8 . One o f the most celebrated such miscalculations was made b y the Liberal Party of the United Kingdom in the early part of the twentieth century in its initial failure to endorse proportional representation prior to the introduction of full adult suffrage. By 1 9 24, when the Liberals finally collectively endorsed proportional representation, their moment had passed. As Butler concludes, " From the point of view of the proportional representa­ tionalists, the tragedy was that by the time the Liberals became enthusiastic converts it was j ust too late; the bulk of the Labour supporters of PR had, of course, become recusants " ('963, 47).

Institutions and Support for Parties of the Extreme Right

r87

function i n part o f conditions that are themselves manipulable, then those successes could be circumscribed by changes in these conditions.29 And so we come full circle to the issues with which we began this chapter and that have motivated our analyses throughout this book. Our conclusion that the success of extreme right parties stems from factors that are at least to some degree manipulable implies that this success is not due to exogenous cultural forces that, if they change at all, do so at a glacial pace as successive generations replace their predecessors. Instead, the revival of these parties reflects conditions that are themselves endoge­ nous, and thereby amenable to intervention and change.

29. The results from the 1994 general election in Italy might seem to contradict our argument, given the success of the National Alliance (successor to the MSI) following electoral reforms designed to restrict smaller parties. However, it is important to note that, because of the reforms, the National Alliance formed a pre-election coalition with the Northern League and Forza Italia that allowed them, among other things, to field j oint candidates throughout Italy (on these developments, see "The Rise of Italy's Right, " f:cono­ mist, February 19, 1994, 5 5 - 5 6) . The reforms themselves are discussed in " Italy: Rules of the New Game," Economist, June 26, 1 9 9 3 , 5 6- 5 8 , and by Calise ( 1 99 3 ) .

SEVEN

Culture, Institutions, and Political Behavior

T hroughout this book we have focused on the fundamental contrast

I between a cultural reading of political behavior and one that casts political behavior as optimizing within institutional constraints. Many are attracted to a cultural reading on the grounds that it emphasizes the role of ideas in political life. This idealism is often linked to populist sentiments, so that cultural accounts are additionally said to assist in " bringing the people back in" (see lnglehart 1 9 9 7 ) . For some, the cultural argument becomes even more attractive when linked to civic virtue in the form of such other-regarding norms as trust and cooperation, under the guise of which political actors altruistically pursue some common good. Perhaps it is not surprising that, when compared with a view of rational choice in which actors are concerned solely with the cold and selfish calculation of a narrowly defined economic interest, many find the aura of the cultural view irresistible. As we trust we have made clear, however, such a conclusion begs a number of vital questions, and it hinges on an excessively narrow view of rational choice. However tantalizing at first blush, the specter of political actors altruistically pursuing a common good (often labeled in terms of a national interest) squares poorly with long-standing conceptions of poli­ tics as centering on conflict over the allocation of scarce resources (e.g., Lasswell 193 6; Easton 1 9 57). By all accounts, this conflict revolves around interests and worldviews, which implies among other things that the cul­ tural argument has no unique claim to the proposition that ideas matter. Second, the cultural argument has no monopoly on the analysis of altruistic behavior. Instead, altruism is quite compatible with a rational r88

Culture, Institutions, and Political Behavior

r89

choice perspective, unless w e take the inordinately narrow and arbitrary view that this perspective can only accommodate the selfish maximiza­ tion of economic utility, and thereby we high-handedly choose a utility function for others. When rational behavior is more reasonably cast as optimizing under the conditions we have considered, it is immediately apparent that neither the evaluation of preferences nor the analysis of altruism is an exclusive prerogative of the cultural account. Third, while a term like cooperation may add to the allure of the cultural account because it implies a group rather than an individualistic basis to behavior, the picture is considerably muddied when the noun collusion is substituted for cooperation. The line separating cooperation from collusion is murky at best, and the two terms typically refer to patterns that are behaviorally indistinct. Indeed, the key distinction gener­ ally involves a judgment about the goals underlying the patterns of behav­ ior: cooperation has congenial overtones while collusion is always cast in pejorative terms. The explanation of group-based behavior does not re­ quire the invocation of cultural norms. Fourth, the cultural argument conjures up a vision of groups spontane­ ously espousing a common set of views or values, whether they are driven Protestants, amoral familists of southern Italy, civic-minded Britons, or citizens of East Asian countries steeped in Confucian values. The out­ come is spontaneous in the sense that a collective perspective simply emerges " independently of any institutional midwifery of any kind" (Ken­ dall 1 94 1 , 1 2 8 - 29 ) . Again, the vision is not compelling, since we have known for a considerable time that, absent a specified institutional frame­ work, a single distribution of preferences among a population can gener­ ate a variety of different sets of collective or group preferences ( Arrow r 9 5 1 ) . There is thus no basis for expecting groups reflexively to express a given set of values or preferences. We have accordingly argued for an account in which political actors optimize within a set of institutional constraints and against a cultural or social capital interpretation of political behavior. We have made the case in two complementary ways. In Part I of the book ( chaps. 2 through 4 ) , w e examined i n detail the empirical bases for the cultural perspective and found it wanting. In Part II ( chaps. 5 and 6), we showed that key aspects of mass political behavior are more profitably explained in terms of the incentive patterns generated by different institutional configurations. Let us briefly recapitulate our analysis. Our critique of the cultural account in Part I has three components. In chapter 2, we examined the Weberian thesis linking Protestant (especially Calvinist) theology to the capitalist spirit. Weber's thesis is, of course, the

190

B E F O RE NOKMS

routinely invoked intellectual godfather of the cultural account, but evalu­ ating the thesis is also important because Weber's analysis additionally serves as a precursor to the methodological problems manifested in more recent treatments. Weber to the contrary, a close examination of patterns of economic growth within Europe since the early Middle Ages reveals no qualitative shift in economic activity in post-Reformation Europe, so that the rise of Protestantism during the latter period cannot be invoked to explain the rise of capitalism. Even if we leave to one side the " rise of capitalism " and concentrate instead on the emergence of a " capitalist spirit," risk-taking behavior (surely part of any such spirit) hardly originated with the Refor­ mation. Recent studies that either directly or indirectly invoke Weber's thesis in either its original or pastoral form to j ustify claims about alleged consequences of cultural values or social capital for political and eco­ nomic outcomes are thus appealing to an argument with no empirical foundation. The Protestant ethic thesis is also noteworthy on methodological grounds, since the analysis suffers from severe selection effects . As we noted, Weber framed his analysis with a statistical study of Baden as of r 89 5 carried out by his student, Martin Offenbacher. Among other things, this study reported that Protestants were more likely to attend high school than Catholics, to be educated in the modern languages and science, and to be wealthier than Catholics. Taking this flawed description of Baden at the end of the nineteenth century as his springboard, Weber leapt quickly backward to a discussion of Benjamin Franklin, and then further on back to the preacher Richard Baxter of Cromwell's England, and so on, search­ ing for sources of this apparent religious difference and finding connec­ tions along the way. To borrow Brandel's ( r 9 8 2, 5 67) term, such a mode of analysis is most kindly described as " retrospective, " because it does not begin by specifying the relevant cultural attributes then tracing out their subsequent impact on outcomes. Unfortunately, this backward approach to empirical analysis is another of Weber's enduring legacies to more recent studies of the effects of values and culture. In chapter 3 we evaluated the analyses of civic-mindedness in Italy and the United States carried out by Robert Putnam. The Italian study, Mak­ ing Democracy Work ( r 99 3 ), builds explicitly on Banfield's classic in­ quiry completed four decades ago, while Bowling A lone ( 2ooo) evokes themes commonly associated with de Tocqueville's Democracy in Amer­ ica. The Italian and American studies address different elements of the argument for civic community. The former advances the argument that the effectiveness of regional governments in Italy hinges critically on

Culture, Institutions, and Political Behavior

191

patterns o f civic engagement that can b e traced back t o a quite distant past, while the latter focuses on trends in civic engagement over the more recent past to suggest that Americans are becoming increasingly isolated from each other. In each setting, Putnam explicitly links patterns of civic virtue to political performance. These analyses are important on at least two counts. First, they can be read as the most current in an ongoing series of such studies in these two countries. Second, Putnam's analyses have struck a most responsive chord and have received an extraordinary amount of attention from scholars, social and political commentators, and the policy community. We have shown that the evidence for these arguments is weak. The Italian study is plagued by statistical problems in which, among other things, assumptions are conflated with conclusions. Further, the claim that regional differences in community involvement reach well back to early modern Italian history is fundamentally inconsistent with the rec­ ord. Even painting with a broad brush, there is no sustained evidence either that northern Italians have been distinctively virtuous citizens or that they have consistently outpaced the residents of southern Italy in this respect. The basic issue in Bowling Alone is that information is marshaled selec­ tively. For example, the claim that levels of interpersonal trust have been declining in recent years stems from an analysis of responses to a single survey item. When the analysis is broadened to incorporate other survey questions commonly used to measure trust, convincing support for the reported decline is more difficult to find. Similarly, using the best available data, we find no sustained evidence of a decay in overall levels of group membership or in particular forms of participation. Indeed, church mem­ bership has grown remarkably since the time of de Tocqueville. To be sure, PTA membership has dropped in recent years as more women have entered the labor force, and league bowling has become less popular than it once was. But such declines have been accompanied by growth in other areas, including environmental groups and parent-organized youth groups. Bowling Alone may constitute a striking metaphor, but then so too does Kicking in Groups (Lemann 1 99 6 ) . We conclude that the two most visible recent country studies fail to make the case for political culture. Along the way, we have shown that there is little evidence either for enduring regional civic virtue in Italy or for the proposition that recent years have seen a pronounced erosion of community values in the United States. Italy and the United States are key cases within which to address these issues, of course, and focusing on them in more detail has enabled us to

I92

B E F O RE NOKMS

address elements of the problem at a more disaggregated level than would otherwise be possible. At the same time, these two cases hardly exhaust the possible varieties of political cultures. Chapter 4 accordingly evalu­ ates recent and more broadly based cross-country studies claiming that values affect economic growth and democratic performance. We show that none of these analyses makes an effective case for a cultural interpre­ tation. Each of them is driven by one or more enigmatic empirical deci­ sions without which the argument does not work. These decisions, in turn, are idiosyncratic to the study being considered. First, while all the studies suggest that " culture " makes a difference, what they actually count as culture is another matter. In the case of economic growth, the shopping cart has included materialist values, achievement orientations, communitarianism, Confucianism, trust, and civic-mindedness. The studies of democratization are similarly varie­ gated, relying in turn on such values as satisfaction, support for the current order, materialism, and trust. While several studies thus claim to establish a link between culture and democracy, the operational proce­ dures they actually employ reveal that each of them means something different by the term. Our appraisal shows that the results obtained are a function of these operational idiosyncrasies . We have further shown that the different approaches employed are far from interchangeable manifes­ tations of a coherent underlying set of cultural norms. Second, following Weber's lead, these analyses betray a fundamental tendency toward ex post reasoning. Thus, Confucianism has been in­ voked as the growth stimulus for many East Asian economies, an asser­ tion that makes sense only in the light of the growth already observed in those economies. Given their earlier growth rates, such a claim for the countries involved would have seemed preposterous in, say, r 9 s o or r 9 6o, even though their cultures then were presumably no less " Confu­ cian " than they are now. And are the same Confucian values responsible for the recent economic crisis in the region? A similar ex post flavor permeates those analyses that employ cultural values measured at the end of a given period to explain growth or patterns of democratization in the preceding decades. It is difficult to take such conclusions seriously. Given these problems, the empirical research program that casts politi­ cal and other outcomes as a function of underlying values has clearly failed. This may seem a strong judgment, but studies like these convey the distinct impression that proponents of the cultural or social capital ap­ proach tailor the evidence to yield the desired results. Perhaps this ana­ lytic strategy is to be expected, since the theoretical foundations on which

Culture, Institutions, and Political Behavior

193

the cultural perspective rests are a s ad hoc as the empirical foundations we have evaluated. Throughout this book, we have argued that a more fruitful approach to the kinds of values like trust that are central to social capital and related accounts is to treat them as endogenous. Such an approach casts values as a response to the conditions within which people find them­ selves and thereby underscores the contexts (or constraints and opportuni­ ties) that they confront. A clear advantage of endogenizing values in this way is that we expect them to shift given changes in the underlying conditions that generate them. None of this, of course, is to suggest that values are inconsequential. Take the case of the Sicilian Mafia, one of particular interest because Mafia activities have often been viewed as the expression of uniquely Sicilian values that assign little importance to trust (e.g., Hess 1 9 7 3 ; Schneider and Schneider 1994 ) . But the Mafia can b e recast a s a loose cartel of families that produces and sells private protection. Gambetta ( I 99 3 ) treats this product as a surrogate for the enforcement of property rights by the state and hence as an imperfect substitute for mutual trust. Demand for such a service is clearly highest when the state cannot or will not enforce contracts, so that Mafia protection typically involves illegal activities. It is equally clear that while state enforcement of property rights constitutes a public good, the distinctive feature of the Mafia's product is that the protection offered is private. Traders can therefore distinguish insiders (other traders with Mafia protection) from outsiders (traders without it) : cheating the latter does not affect their reputation with the former. Further, the incentives for both insiders and the Mafia favor cheating outsiders. In Gambetta's formulation, (dis)trust is thus endogenized, a product of the Mafia, not its source. A singular feature of this approach is that Mafia activities are not cast as the inevitable by-product of Sicilian life. Further, the approach helps identify the conditions under which similar activities might be anticipated in other settings. Along these lines, the structure of private protection has also been shown to account for the rise of organized crime in Russia (Varese 200 1 ) . But observe that values, in the form of levels of trust, remain central to the analysis. Values also retain a core position in the analysis of ethnically divided societies. Ethnicity is an especially interesting case, since ethnic markers have so often been taken to reflect the boundaries of " primordial" groups that pose particular threats to states, the quintessential form of exoge­ nously defined cultural groups (the classic statement here is Geertz 1 9 63 ) .

I 94

B E F O RE NOKMS

The question, then, i s why ethnic conflict i s relatively uncommon, given the number of ethnically divided societies (Fearon and Laitin r 9 9 6 ) . Again, the issue is most fruitfully broached b y endogenizing trust. As Weingast ( r 9 9 7, r 99 8 ) shows, a principal source of ethnic conflict in divided societies is fear of becoming a victim. Where no single ethnic group is dominant, all groups fear being subj ect to aggression from a rival group . When a group fears victimization and has the chance to employ the state against another, it has an incentive to mount a preemptive strike against the other. Ethnic conflict thus occurs because it is much preferable to be an aggressor than a victim, and because promises not to aggress by individual groups are themselves insufficient to prevent the outbreak of conflict. In such settings, it is crucial to construct political institutions that raise the costs for any group to use the state for violent, ethnic purposes. By altering each group's incentives, effective institutions can make toleration self-enforcing, thus credibly committing each group to honor its promises and thereby minimizing the odds that any group will be victimized. In­ creasing the costs of defection in this manner provides incentives for groups and their leaders to conclude that it is neither in their interest, nor in the interest of their rivals, to strike first. Trust and cooperation are thus seen not as exogenous, but as created. Observe again the advantages of this institutional perspective. An ex­ ogenous approach to ethnic trust incorrectly implies that violent ethnic conflict is endemic to all divided societies at all times. In contrast, taking trust as endogenous leads us more accurately to expect higher volumes of such conflict in some divided societies than in others. It also helps ac­ count more directly for the temporal ebbs and flows of ethnic conflict that occur with institutional collapse and regeneration within divided societies. But values, in the form of levels of (dis)trust, again retain a key role in the framework. We provided scattered evidence throughout Part I of the book for the institutional view. In Part II, we moved beyond our critique of cultural accounts and presented the institutional alternative more directly and in greater detail. Here, we evaluated the argument in the context of mass political behavior, that is, the behavior of ordinary citizens. Following much previous research, we examined rates of political participation in the form of voter turnout in the established democracies over the last fifty years (chap. 5 ) and rates of support for extremist political parties in Western Europe over the past three decades (chap. 6 ) . O u r analysis of voter turnout starts from the premise that political institutions as embodied in electoral laws play a powerful role in shaping

Culture, Institutions, and Political Behavior

19 5

the distribution of incentives for people to participate in the political system. They do so in two complementary ways. First, institutions di­ rectly affect the incentives of individual citizens contemplating whether to vote. Compulsory voting laws, for example, typically enlarge voter turn­ out (even when the penalties for noncompliance are minor) , presumably by their direct marginal impact on the incentive structures of citizens who would otherwise be nonvoters. Second, the effect of institutions on turn­ out is often indirect, in the sense that they alter the incentives of parties and candidates to mobilize voters, which in turn influences turnout. Thus, in single-member district systems, some legislative districts may be considered so safe for incumbents that nonincumbent parties and candi­ dates move resources (e.g., time and money) away from those districts to others where they believe their resources have more potential to affect the outcome by mobilizing voters. The earliest systematic institutional account of political participation was offered by Gosnell ( r 9 3 o ), whose analysis of European political sys­ tems indicated that levels of political participation varied consistently with variations in types of electoral system. Following this line of thought, subsequent research concluded that variations in turnout rates across the industrial democracies in the r 9 6os and 1 9 70s are largely a function of institutional arrangements stemming from electoral laws and party sys­ tems. In a similar vein, students of American politics have devoted consider­ able attention to the impact on turnout of legal factors such as residency requirements and registration procedures, both of which have powerful implications for how institutions might affect political participation. The principal alternative against which we evaluated this institutional view is the cultural approach that stresses linkages between enduring values and participation. These linkages, too, have long been central to key analyses of mass political behavior, following Almond and Verba's ( 1 9 63 ) classic analysis of the relationship between culture and political participation. In participatory cultures, according to this view, citizens are more politically satisfied with their institutions (whatever the particu­ lar form of those institutions) and are therefore more politically effica­ cious. Cultures that foster such values thereby enhance participation in general and voter turnout in particular. The empirical analyses in chapter 5 provided scant support for the view that rates of political participation hinge meaningfully on cultural values. To be sure, we do find a mild suggestion that values may make a difference in our analysis of McClelland's data on need for achievement, but our other tests provide no corroborating evidence. That is, when we adopt a country-specific approach to political culture or when we employ

I96

B E F O RE NOKMS

data from the World Values Surveys, there i s no consistent evidence for a cultural interpretation of turnout. At the same time, we have provided additional evidence that levels of voter turnout reflect institutional arrangements as embodied in electoral procedures, and our model offers a plausible account of the observed patterns in the established democracies over the last half century. Specifi­ cally, we identified five institutional arrangements that systematically im­ pinge on turnout. We further argued that the lower-than-predicted rates of turnout in two cases ( Switzerland and the United States) are best explained by the distinctively high volume of electoral choices that con­ front citizens in those two countries. This, of course, is simply another institutional feature that is manifested in those two particular settings. This leads us to a final point. As we have emphasized throughout this book, cultural explanations emphasize the durability of values and thereby lead us to anticipate limited changes in political behavior over time in particular settings. The institutional perspective, by contrast, implies that changes in behavior are induced by changes in rules and incentives. As is well known, institutional arrangements of the sort we have examined are themselves typically durable. They are nonetheless modified on occasion, and when they are, our analyses show that political behavior appears also to follow suit. Further, as the case of the Swiss referenda reminds us, rates of voter participation can fluctuate considerably within a given setting (or national " culture " ) over the longer haul (more than a century), fluctua­ tions resulting substantially from electoral incentives in the form of the number of items on the ballot and the recency of the last referendum election. In chapter 6 we turned our attention to the systemic conditions that have favored parties of the extreme right in West European politics over the last three decades, parties that many had thought (hoped ? ) were crea­ tures of the past. We found evidence that electoral and party-system factors interact with each other to account for the response that these parties have generated from voters. Specifically, electoral disproportionality (through the mechanism of thresholds) increasingly dampens support for the ex­ treme right as the number of parliamentary parties expands. In parallel manner, multipartyism increasingly fosters parties of the extreme right with rising electoral proportionality. Our analyses also indicate that higher rates of unemployment provide a favorable environment for these political movements. In our view, higher rates of unemployment epitomize uneven economic performance that fosters support for the extreme right by provid­ ing the pretext for political entrepreneurs to mount the xenophobic politi­ cal appeals that characterize these political movements. Increasing unem-

Culture, Institutions, and Political Behavior

197

ployment is significant because it provides a fertile environment for such appeals. Scholars have for many years pondered the ways in which an electoral system can be revised or manipulated to achieve political and policy goals. Among the more recent of these analyses, Lijphart ( 1 9 9 4 ) has suggested that electoral disproportionality may be especially subject to modification. Unemployment rates may be the least subj ect to direct policy intervention, although there is evidence that levels of unemploy­ ment reflect political choices as different governments offer different mixes of intervention in such concrete labor market programs as adult training and recruitment subsidies. None of this is meant to suggest that the possibilities for ( and effective­ ness of) electoral engineering or other policy innovations are unlimited. For one thing, electoral laws are typically sticky, since those in positions to modify them (governments or their constituent parties) seldom have any incentive to do so (the status quo rules are, after all, those by which they achieved their positions in the first place ) . On those rare occasions when they are motivated to support electoral reform, key political actors operate in an environment of uncertainty and often miscalculate. Thus, there are limits . But it bears emphasis that since the electoral successes of extreme right-wing parties are a function in part of conditions that are themselves subject to modification, then those successes could be limited by changes in these conditions. In all, then, the analyses in this book offer no consistent support for the cultural account of political behavior. By the same token, they offer considerable support for the alternative institutional view according to which political actors optimize, given a set of constraints. This is clear from the analyses of Part II that centered on decisions of ordinary citizens whether or not to vote and whether to support parties of the extreme right. Further evidence supporting this perspective as it centers on other outcomes is scattered throughout the book. Perhaps our argument is best encapsulated in the context of a particular case. Although it originated with observations from one small village, the impact of Banfield's ( 1 9 5 8 ) study of " Montegrano " has been monumen­ tal, underpinning many of the analyses we have considered here. For ex­ ample, it played a central role in motivating the cultural arguments ad­ vanced by Harrison ( T 9 9 2 ) , Fukuyama ( T 9 9 5 ; see esp . chap. T o, " Italian Confucianism " ) , and, of course, Putnam ( 1 99 3 ) in his analysis of civic virtue in Italy. As we have already observed, the core of Banfield's argu­ ment is that the backwardness of " Montegrano " in I 9 5 4 - 5 5 stemmed directly from the " bacillus of amoral familism " infecting its inhabitants

198

B E F O RE NOKMS

(the phrase i s from Pizzorno 1 9 7 1 , 9 2 ) . Since the norm was deeply in­ grained, the prognosis was grim. It is thus striking to learn that on visiting " Montegrano " (actually Chiaramonte, a mountain village in southern Italy) in 1 9 89 as part of a " pilgrimage, " Harrison found " dramatic change. " Almost everyone was literate. Half were high school graduates . All families had television. Sixty percent had telephones. Sixty percent had automobiles. Several agricultural cooperatives had been formed. Fami­ lies are now much smaller, averaging two children. A highway was built in the 1 9 70s that cut travel time to Naples in half ( from six hours to three ) . People now travel much more, and many attend school else­ where. Many Chiaromontese have migrated to Northern Italy and other European countries to work, and a fair number of them have returned. ( 1 9 9 2 , S o )

Thus, in just thirty-five years, Banfield's village was completely trans­ formed, so that, in Harrison's words, " Chiaramonte is no longer Montegrano. " ' It is difficult to construct a consistent argument specifying Monte­ grana's backwardness observed in the middle T9 sos as a function of amoral familism that can also accommodate the changed circumstances of 1990 reported by Harrison. Banfield, after all, cast amoral familism not as a fleeting attitude but as a fundamental and ingrained orientation to the world, an orientation too deep-seated to allow for the complete 1. While Putnam, Fukuyama, Harrison, and other recent proponents of the cultural account have enthusiastically embraced Banfield's ( r 9 5 8 ) emphasis on durable village val­ ues, they seem to have overlooked the rapid transformation of a similarly " b ackward" Turkish village reported in the same year by Lerner in "The Grocer and the Chief: A Parable" ( ' 9 5 8 , chap. , ) . According to Lerner, the transformation of Balgat (a village close to Ankara ) took place between T9 50 and T9 54 and was occasioned by the electoral loss of Kemal Ataturk's Republican People's Party in 1 9 5 0 . In short order, the new government formed by the Democrat Party brought a new road, bus service, electrification, and running water to the village, all of which generated a profound shift in commercial activities, population growth, and access to the media (especially radio s ) . From a discussion with a group of men at a coffeehouse about the upcoming election in 1 9 5 4 , Lerner concluded that villagers had quickly internalized a key political lesson. There was . . . general agreement, at least among the older men, that it would be better to have a small margin between the major parties. When the parties are competing and need our votes, then they heed our voices-thus ran the underlying proposition of the colloquy. ( T 9 5 8 , 4 T; emphasis added) Alas, electoral margins between the major parties proved not to be small in the subsequent general elections prior to the military coup of May T 9 60 (Hale T994 ) .

Culture, Institutions, and Political Behavior

199

about-face required for the pattern Harrison observed. This, after all, was an abrupt turnaround that took less than a generation and a half to complete. When, on the other hand, we focus our attention on the incentive struc­ tures faced by the residents of Chiaramonte, the observed change becomes much more explicable. Southern Italy after the Second World War was characterized by severe overpopulation problems, deforestation, and a critical shortage of arable land ( Schachter 1 9 6 5 ) . Indeed, a 1 9 5 2 report by the Association for the Industrial Development of the Italian South esti­ mated that in excess of 7 million people ( 4 2 percent of the southern popula­ tion) lived in zones "without consistent possibilities of development" (cited in Lopreato T 9 6 5 , 299 ) . Generally, agricultural village populations experienced chronic insolvency, economic precariousness, poor health, and social marginalization. Lopreato further reported ( 1 9 6 5 , 3 00) that a r 9 5 2 national survey of Italian farmhands found over two-thirds of them to be unhappy with "their actual life conditions. " The reasons cited for this unhappiness were " little work, irregular work, little money, high prices, hunger, and other economic reasons and various difficulties . " Given such an environment, the low levels of trust and the higher levels of present­ orientedness reported by Banfield in the middle 1 9 5 0s make up a sensible and understandable response. 2 Whether this warrants the j udgmental la­ beling associated with the term amoral familism is a completely separate question. After 1 9 5 0, however, major changes came to southern Italy. Most notable was the massive pattern of outmigration, fueled both by the push factors j ust described and the pull factors of economic growth in the north and elsewhere. Lopreato reported that, between 1 9 5 1 and 1 9 6 3 , approximately 2, 5 oo,ooo people migrated ( 1 9 6 5 , 29 8 ) . This number con­ stitutes a remarkable I4 percent of the total southern population of T 7,4 3 3 , 5 00 in T 9 5 T , and it obviously represents an even more significant fraction of the (potentially) economically active labor force. Over the same period, the number of employed male agricultural workers plum­ meted approximately 40 percent, from 2,689,000 to r , 6 6 r ,ooo .3 2 . This is the point emphasized by Lopreato in his pointedly titled essay " How Would You Like to Be a Peasant ? " ( r 9 6 5 ). The implicit rational choice account of peasant behavior offered by Lopreato was explicitly expanded in Popkin's ( r 979 ) well-known analysis of peasant behavior in Vietnam. 3 · None of this is meant to suggest that southern Italy had somehow converged with the rest of the country by r990. For example, southern unemployment levels for the age groups from I 4 through 29 consistently exceeded those in the rest of the country by a factor of 2 or 3 to r throughout the r 9 8os. Further, outmigration continued through the r 9 8 os, albeit at a rate lower than that of the T 9 5 0s and T 9 6os (Kostoris T 9 9 3 , tables 8 . 2 and 8 . 4 ) .

200

B E F O RE NOKMS

Among other things, these changes elicited a political response. Most notably, Lopreato observed that national political parties increasingly came to see the articulation of southern issues as a significant source of possible votes ( 1 9 6 5 , 3 0 3 ) . The new road from Chiaramonte to Naples reported by Harrison assumes new meaning in this context! In a related vein, Lopreato pointed to evidence that governments at the national level increasingly recognized a " southern problem, " which they took specific measures to address. For example, he reported passage of a law in the early 1 9 60s to encourage the consolidation of minuscule landholdings gen­ erated by laws of inheritance into larger, more viable, economic units ( 1 9 6 5 , 3 0 7 ) . There is thus extensive evidence to suggest that Banfield's amoral familism constituted no more than a reasonable response to a set of incentives that were widespread in southern Italy at the beginning of the 1 9 5 0s.4 Given the ensuing outmigration and enhanced political interest from Rome, that set of incentives changed. Hence the creation of the " new" Chiaramonte documented by Harrison, an outcome in whose gen­ eration amoral familism was wholly incidental. The cultural and institutional approaches clearly comprise distinct explanations (here, of underdevelopment) . Further, the distinctions at stake reflect real issues, not arcane and pedantic fine points. We believe that the institutional account easily outperforms the cultural interpreta­ tion. We trust that we have made that case.

4 - This point was raised repeatedly after the publication of Banfield's analysis (see, e.g., Lopreato I 9 6 5 ; Marselli I 9 6 3 ; Pizzorno I 9 7 I ; Silverman I 9 6 8 ) .

References

Abramson, Paul R., and Ronald Inglehart. 199 5 . Value Change in Global Perspec­ tive. Ann Arbor: University of Michigan Press . Adorno, T. W. , Else Frenkel-Brunswik, Daniel J . Levinson, and R. Nevitt Sanford. 1 9 50. The Authoritarian Personality. New York: Harper and Row. Ajayi, J. F. Ad e. 1 9 6 5 . Christian Missions in Nigeria: The Making of a New Elite. Evanston: Northwestern University Press . Aldrich, John H. 1 99 3 . " Rational Choice and Turnout. " American Journal of Political Science 3 7 : 24 6- 7 8 . . 1 9 9 7 . "When I s I t Rational t o Vote ? " I n Perspectives o n Public Choice, ed. Dennis C. Mueller. New York: Cambridge University Press. Alesina, Alberto. 1 9 8 9 . " Politics and Business Cycles in Industrial Democracies. " Economic Policy 4 : 5 5 - 9 8 . Almond, Gabriel A . 1 9 90. A Discipline Divided: School and Sects in Political Science. Newbury Park, CA: Sage. Almond, Gabriel A., and Sidney Verba. 1 9 6 3 . The Civic Culture. Princeton: Princeton University Press. --- . 1 9 8 0 . The Civic Culture Revisited. Boston: Little, Brown. AIt, James E. T98 5. " Party Strategies, World Demand, and Unemployment in Britain and the United States. " American Political Science Review 7 9 : I O I 6-4o. Amemiya, Takeshi. 1984. "Tobit Models: A Survey. " Journal of Econometrics 24: 3-6I. Andrews, Josephine T. 2002. When Majorities Fail: The Russian Parliament, I 9 9 0 - I 9 9 3 · New York: Cambridge University Press. Armstrong, J. Scott. 1 9 67 . "Derivation of Theory by Means of Factor Analysis or Tom Swift and His Electric Factor Analysis Machine. " American Statistician 2I: 17-21. Arrow, Kenneth. 1 9 5 1 . Social Choice and Individual Values. New Haven: Yale University Press. Austin, Granville. 1 9 66. The Indian Constitution: Cornerstone of a Nation. Ox­ ford: Clarendon. ---

20I

202

REFERENCES

Backer, Susann. 2000. " Right-Wing Extremism in Unified Germany. " In The Politics of the Extreme Right: From the Margins to the Mainstream, ed. Paul Hainsworth. London: Pinter. Baldwin, John W. I9 5 9 · The Medieval Theories of the Just Price: Romanists, Canonists and Theologians in the Twelfth and Thirteenth Centuries. Transac­ tions of the American Philosophical Society (Philadelphia), n.s., vol. 49, no. 4 · Banfield, Edward C. I 9 5 8 . The Moral Basis o f a Backward Society. New York: Free Press. . I 9 74 · The Unheavenly City Revisited. Boston: Little, Brown. Barro, Robert J. I 9 9 I . " Economic Growth in a Cross Section of Countries . " Quarterly Journal o f Economics I o6 : 4 0 7 - 4 4 . . I 9 9 7 · Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge: MIT Press. Barry, Brian. I 9 70. Sociologists, Economists, and Democracy. Chicago: Univer­ sity of Chicago Press. Bates, Robert H., Rui J. P. de Figueiredo, and Barry R. Weingast. l 99 8 . "The Politics of Interpretation: Rationality, Culture, and Transition. " Politics and Society 26: 603 - 4 2 . Bauer, Yehuda. 200 I . Rethinking the Holocaust. N e w Haven: Yale University Press. Baumgartner, Frank R., and Jack L. Walker. I 9 8 8 . " Survey Research and Mem­ bership in Voluntary Associations. " American Journal of Political Science 3 2: 908 - 2 8 . . I 990. " Measurement Validity and the Continuity of Results in Survey Re­ search" (Response to Smith ) . American Journal ofPolitical Science 3 4 : 662-70. Beck, Nathaniel, and Jonathan N. Katz. I 9 9 5 · "What to Do ( and Not to Do) with Time-Series Cross-Section Data in Comparative Politics. " American Political Science Review 8 9 : 6 3 4 - 4 7 . Bell, Daniel. I 9 6o. The E n d o f Ideology: On the Exhaustion o f Political Ideas in the Fif'ties. Glencoe, IL: Free Press. . I 9 7 3 · The Coming of' Post-industrial So ciety: A Venture in Social Fore­ casting. New York: Basic Books. Bellah, Robert N., Richard Masden, William M. Sullivan, Ann Swidler, and Steven M. Tipton. I 9 8 5. Habits of' the Heart: Individualism and Commitment in American Lif'e. Berkeley: University of California Press. Berelson, Bernard, Paul Lazarsfeld, and William McPhee. I 9 54· Voting: A Study of' Opinion Formation in a Presidential Campaign. Chicago : University of Chicago Press. Berk, Richard A. I 9 8 3 . " An Introduction t o Sample Selection Bias i n Sociological Data . " American Sociological Review 4 8 : 3 8 6- 9 8 . --- . I990. " A Primer on Robust Regression. " In Modern Methods of' Data Analysis, ed. John Fox and J. Scott Long. Newbury Park, CA: Sage. Betz, Hans-Georg. I990. " Politics of Resentment: Right-Wing Radicalism in West Germany. " Comparative Politics 23 : 4 5 - 60. . I 99 3 · "The New Politics o f Resentment: Radical Right-Wing Populist Parties in Western Europe. " Comparative Politics 2 5 : 4 I 3 - 2 7 .

---

---

---

---

---

References

203

lletz, Hans-Georg, and Stefan Immerfall, eds. I 9 9 8 . The New Politics of the Right: Neo -populist Parties and Movements in Established Democracies. New York: St. Martin's. Billig, Michael S . I994· "The Death and Rebirth of Entrepreneurism on Negros Island, Philippines: A Critique of Cultural Theories of Enterprise . " Journal of Economic Issues 2 8 : 6 5 9 - 7 8 . Black, Duncan. I 9 5 8 . The Theory o f Committees and Elections. Cambridge: Cambridge University Press. Blais, Andre. 2000. To Vote or Not to Vote: The Merits and Limits of Rational Choice Theory. Pittsburgh: University of Pittsburgh Press. Blais, Andre, and R. K. Carty. I 99 I . " The Psychological Impact o f Electoral Laws : Measuring Duverger's Elusive Factor. " British Journal of Political Sci­ ence 2 I : 7 I - 9 3 · Blais, Andre, and Louis Massicotte. I 9 9 7 · " Electoral Formulas: A Macroscopic Perspective. " European Journal of Political Research 3 2: I 0 7 - 2 9 . Bollen, Kenneth A. l 99 3 · " Liberal Democracy: Validity a n d Method Factors in Cross-National Measures. " American Journal of Political Science 3 7: I 207- 3 o. Bollen, Kenneth A., and Robert W. Jackman. I 9 8 5. " Political Democracy and the Size Distribution of Income. " American Sociological Review 5 0: 4 3 8 - 5 7 · . I 9 8 9 . " Democracy, Stability, and Dichotomies . " American Sociological Review 5 4 : 6o 5 - I 4 · --- . I 990. " Regression Diagnostics: An Expository Treatment of Outliers and Influential Cases . " In Modern Methods of Data Analysis, ed. John Fox and J. Scott Long. Newbury Park, CA: Sage. Boyd, Richard W. L 9 8 6. " Election Calendars and Voter Turnout. " American Politics Quarterly I 4 : 8 9 - I04. . I989. " The Effects of Primaries and Statewide Races on Voter Turnout. " .Journal of Politics 5 I : 7 3 o- 3 9 . Bradburn, Norman M . , and David E. Bedew. I 9 6 I . " Need for Achievement and English Industrial Growth. " Economic Development and Cultural Change I O : 8 - 20. Brass, Paul R. I994· The Politics of India since Independence. 2d ed. New York: Cambridge University Press. Braude!, Fernand. I 9 8 2 . The Wheels o f Commerce: Civilization and Capitalism, I s th- I 8th Century. Vol. 2. Berkeley: University of California Press. Brechon, Pierre, and Subrata K . Mitra. I99 3 · "The National Front in France: The Emergence of an Extreme Right Protest Movement. " Comparative Politics 2 5 : 63 - 8 2. Brehm, John, and Wendy Rahn. I 9 9 7 · " Individual-Level Evidence for the Causes and Consequences of Social Capital. " American Journal of Political Science 4 I : 9 99- I02 3 . Britnell, Richard. I 9 8 9 . " England and Northern Italy i n the Early Fourteenth Century: The Economic Contrasts. " Transactions of the Royal Historical Soci­ ety, sth ser. , 3 9 : I 67- 8 3 . --- . I 99 3 · " Commerce and Capitalism i n Late Medieval England: Problems of Description and Theory. " .Journal of Historical Sociology 6: 3 5 9 - 7 6 . ---

---

204

REFERENCES

. r 9 9 6 . The Commercialisation of bnglish Society, I o o o - Is o o . 2 d ed. Manchester: Manchester University Press. Brown, Donald E. I 99 I . Human Universals. New York: McGraw-Hill. Browning, Christopher R. 2000. Nazi Policy, Jewish Workers, German Killers. New York: Cambridge University Press. Brucker, Gene. I 9 9 9 · " Civic Traditions in Premodern Italy. " Journal of Interdisci­ p linary History 29: 3 5 7- 7 7 . Brustein, William. r 9 9 6. The Logic of Evil: The Social Origins of the Nazi Party, I925- I93 3 · New Haven: Yale University Press. Bueno de Mesquita, Bruce. r 9 7 8 . " Redistricting and Political Integration in In­ dia . " Comparative Political Studies I I : 279- 8 8 . Burkhart, Ross E . , and Michael Lewis-Beck. I 9 9 4 · " Comparative Democracy: The Economic Development Thesis. " American Political Science Review 8 8 : 9 0 3 - I O. Butler, David E. r 9 63 . The Electoral System in Britain since I9I8. 2d ed. Oxford: Clarendon. Butler, David, and Austin Ranney. I 9 9 4 · Referendums around the World: The Growing Use of Direct Democracy. Washington, D C : American Enterprise Institute. Calise, Mauro. I 9 9 3 · " Remaking the Italian Party System: How Lijphart Got It Wrong by Saying It Right. " West European Politics r 6 : 5 4 5 - 6o. Calmfors, Laers, and Ragnar Nymoen. r 990. " Real Wage Adjustment and Em­ ployment Policies in the Nordic Countries. " E conomic Policy 5 : 3 97-44 8 . Calvert, Randall L . r 9 8 6. Models o f Imperfect Information in Politics. New York: Harwood Academic. . I 9 9 5 · "The Rational Choice Theory of Social Institutions : Cooperation, Coordination, and Communication. " I n Modern Political Economy: O ld Top­ ics, New Directions, ed. Jeffrey S. Banks and Eric A. Hanushek. New York: Cambridge University Press. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. r 9 6o. The American Voter. New York: John Wiley. Carter, Elisabeth L. 2002. " Proportional Representation and the Fortunes of Right-Wing Extremist Parties . " West European Politics 2 5 : I 2 5 - 4 6. Carus-Wilson, E. M. r 9 6 7 . Medieval Merchant Venturers: Collected Studies. London: Methuen. Castles, Francis G., and Peter Mair. r 9 8 4 . " Left-Right Political Scales: Some Expert Judgements. " European .Journal of Political Research u: 7 3 - 8 8 . Chatterj ee, Samprit, and Bertram Price. I 9 9 I . Regression Analysis by Example. 2d ed. New York: Wiley-Interscience. Chwe, Michael Suk-Young. 200 I . Rational Ritual: Culture, Coordination, and Common Knowledge. Princeton: Princeton University Press. Cipolla, Carlo M. I 9 9 4 · Before the Industrial Revolution: European Society and Economy, I O O O- I 7 0 0 . 3d ed. New York: Norton. Clarke, Harold D . , and Nitish Dutt. I 9 9 I . " Measuring Value Change in Western Industrialized Societies: The Impact of Unemployment. " American Political Science Review 8 5 : 90 5 - 20.

---

---

References

20 5

Clarke, Harold D . , Nitish Dutt, and Allan Kornberg. I 99 3 · "The Political Econ­ omy of Attitudes toward Polity and Society in Western European Democra­ cies . " Journal of Politics 5 5 : 99 8 - I 02 I . Clarke, Harold D . , Nitish Dutt, and Jonathan Rapkin. I 9 9 7 · " Conversations in Context: The (Mis )measurement of Value Change in Advanced Industrial Soci­ eties. " Political Behavior I 9 : I 9 - 3 9 · Cohen, Jere. I 9 8 o. " Rational Capitalism i n Renaissance Italy. " American Journal of Sociology 8 5 : I 3 4 0- 5 5 . . I 9 8 3 . " Reply t o Holton. " American Journal o f Sociology 8 9 : I 8 I - 8 7 . . 2002. Protestantism and Capitalism: The Mechanisms o f Influence. New York: Aldine de Gruyter. Cole, Arthur H. I 9 6 I . "The Relations o f Missionary Activity t o Economic Devel­ opment. " Economic Development and Cultural Change 9: I 20- 2 7. Cortes, Juan B. I 9 6 I . "The Achievement Motive in the Spanish Economy be­ tween the Thirteenth and Eighteenth Centuries . " Economic Development and Cultural Change 9 : l44 - 63 . Costa, Dora L. 2000. "From Mill Town to Board Room: The Rise of Women's Paid Labor. " Journal of Economic Perspectives I 4 : I O I - 22 . Costa, Dora L . , a n d Matthew E. Kahn. 200 I . "Understanding the Decline in Social Capital, I 9 5 2- I 9 9 8 . " Working Paper No. w 8 29 5 . Cambridge, MA: NBER. Coupland, Reginald . I 9 4 4 · The Indian Pro blem: Report on the Constitutional Problem in India. 3 vols. New York: Oxford University Press. Cox, Gary W. I997· Making Votes Count: Strategic Coordination in the World's Electoral Systems. New York: Cambridge University Press. . I999· "Electoral Rules and the Calculus of Mobilization. " Legislative Studies Quarterly 24: 3 8 7 - 4 I 9 . . 2000. " O n the Effects o f Legislative Rules. " Legislative Studies Quarterly 2 5 : I 69 - 9 2 . Crawford, Susan, a n d Peggy Levitt. I 9 9 9 · " Social Change a n d Civic Engagement: The Case of the PTA. " In Civic Engagement in American Democracy, ed. Theda Skocpol and Morris P. Fiorina. Washington, D C : Brookings Institution. Crewe, Ivor. I 9 8 I . " Electoral Participation. " In Democracy a t the Polls: A Comparative Study of Competitive National Elections, ed. David Butler, How­ ard R. Penniman, and Austin Ranney. Washington, DC: American Enterprise Institute. Cronin, Helena. l 99 L The Ant and the Peacock: Altruism and Sexual Selection from Darwin to Today. New York: Cambridge University Press. Crummey, Robert 0 . , Robert V. D aniels, Richard Hellie, Richard Wortman, and Edward L. Keenan. I 9 8 7. " Symposium on 'Muscovite Political Folkways. "' Russian Review 4 6 : I 5 7- 209. Dahl, Robert A. I 9 6 I a. Who G overns? Democracy and Power in an American City. New Haven: Yale University Press. . I 9 6 I b. "The Behavioral Approach in Political Science: Epitaph for a Monument to a Successful Protest. " American Political Science Review 5 5 : 7 6 3 - 72.

---

---

---

---

---

206

REFERENCES

. I 9 66. "Patterns of Opposition. " In Political Opposition in Western De­ mocracies, ed. Robert A. Dahl. New Haven: Yale University Press. . I 9 7 I . Polyarchy: Participation and Opposition. New Haven: Yale Univer­ sity Press. Dalton, Russell J. 2002. Citizen Politics: Public Opinion and Political Parties in Advanced Industrial Democracies. 3d ed. New York: Chatham House. Davidson, Chandler, and Bernard Grofman, eds. I 9 9 4 · Quiet Revolution in the South: The Impact of the Voting Rights Act, I96 5 - I 9 9 0 . Princeton: Princeton University Press. Davies, Christie. I 9 9 2 . "The Protestant Ethic and the Comic Spirit of Capital­ ism. " British Journal of Sociology 4 3 : 4 2 I - 4 2 . Davis, James A., a n d Tom W. Smith. I 99 3 · General Social Surveys, I 9 7 2 - I 9 9 3 : Cumulative Codebook. Chicago: National Opinion Research Center. . I999· General Social Surveys, I 9 7 2 - I 9 9 8 : Cumulative File. 2d ICPSR Release, April. Ann Arbor: Interuniversity Consortium for Political and Social Research. Dawkins, Richard. I 9 8 9 . The Selfish Gene. 2d ed. New York: Oxford University Press. Day, John. I 9 8 7. The Medieval Market Economy. Oxford: Blackwell. Delacroix, Jacques, and Fran<;:ois Nielsen. 200 I . "The Beloved Myth: Protestant­ ism and the Rise of Industrial Capitalism in Nineteenth-Century Europe. " Social Forces So: 5 09- 5 3 · D e Roover, Raymond. I 9 4 2 . "The Commercial Revolution o f the Thirteenth Century. " Bulletin of the Business Historical Society I 6: 3 4 - 3 9 . --- . l 9 5 8 . "The Concept o f the Just Price: Theory and Economic Policy. " Journal of Economic History I 8 : 4 I 8- 3 4 · --- . I 9 74· Business, Banking, and Economic Thought in Late Medieval and Early Modern Europe. Chicago: University of Chicago Press. Donaghy, Peter J., and Michael T. Newton. I 9 8 7 . Spain: A Guide to Political and Economic Institutions. Cambridge: Cambridge University Press . Downs, Anthony. I 9 5 7 · An Economic Theory o f Democracy. New York: Harper and Row. Draper, John William. I 8 74· History of the Conflict between Religion and Sci­ ence. London: Kegan Paul, Trench. Duby, Georges . I 9 74· The Early Growth of the European Economy: Warriors and Peasants from the Seventh to the Twelfth Century. Ithaca: Cornell Univer­ sity Press. Duch, Raymond M., and Michaell A. Taylor. I 99 3 · " Postmaterialism and the Economic Condition. " American Journal of Political Science 3 7 : 7 4 7 - 7 9 . Dunteman, George H. I 9 8 9 . Principal Components Analysis. Newbury Park, CA: Sage. Du Plessis, Johannes. I 9 I I . A History of Christian Missions in South Africa. London: Longmans Green. Durlauf, Steven N. 2002. " B owling Alone: A Review Essay. " .Journal of Eco­ nomic Behavior and Organization 4 7 : 2 5 9 - 7 3 . ---

---

---

References

207

Duvall, Louis M. I 9 2 8 . Baptist Missions in Nigeria. Richmond, VA : Southern Baptist Convention. Duverger, Maurice. I 9 6 3 . Political Parties: Their Organization and Activity in the Modern State. New York: John Wiley and Sons. . I 9 8 6. "Duverger's Law: Forty Years Later. " In Electoral Laws and Their Political Consequences, ed. Bernard Grofman and Arend Lijphart. New York: Agathon. Easton, David. I 9 5 7 · " An Approach to the Analysis of Political Systems. " World Politics 9 : 3 8 3 -400. Eatwell, Roger. 2000. "The Rebirth of the 'Extreme Right' in Western Europe ? " Parliamentary Affairs 5 3 : 40 7 - 2 5 . Eckstein, Harry. I 9 9 2 . " Rationality and Frustration. " I n Regarding Politics: Es­ says on Political Theory, Stability, and Change, ed. Harry Eckstein. Berkeley: University of California Press. Edgeworth, F. Y. I 8 9 8 . " Miscellaneous Applications of the Calculus of Probabili­ ties, IV. " Journal of the Royal Statistical Society, ser. A, 6 l : 53 4- 44· Ekelund, Robert B., Robert F. Hebert, Robert D. Tollison, Gary M. Anderson, and Audrey B. Davidson. I 9 9 6 . Sacred Trust: The Medieval Church as an Economic Firm. New York: Oxford University Press. Epstein, Lee, and Jack Knight. 2000. "Toward a Strategic Revolution in Judicial Politics: A Look Back, a Look Ahead. " Political Research Quarterly 5 3 : 62 5 - 6 I . Epstein, Steven A . I 9 9 I a. Wage Labor and Guilds in Medieval Europe. Chapel Hill: University of North Carolina Press. --- . l 9 9 l b . "The Theory and Practice of the Just Wage. " Journal of Medieval History I 7 : 5 3 - 69 . Fearon, James, and David Laitin. I 9 96. " Explaining Interethnic Cooperation. " American Political Science Review 90: 7 I 5 - 3 5 . Featherstone, Kevin. I 9 8 7. " Elections and Voting Behavior. " In Political Change in Greece: Before and After the Colonels, ed. Kevin Featherstone and Dimi­ trious K. Katsoudas. London: Croom Helm. Ferej ohn, John, and Debra Satz. I 9 9 5 · "Unification, Universalism, and Rational Choice Theory. " Critical Review 9: 7 I - 84 . Festinger, Leon. I 9 5 7· A Theory o f Cognitive Dissonance. Stanford: Stanford University Press. Finke, Roger, and Rodney Stark. I992. The Churching of America, I 7 7 6- r 9 9 0 : Winners and Losers in O u r Religious Economy. New Brunswick: Rutgers University Press. Finke, Roger, and Laurence R. Iannaccone. I 9 9 3 · " Supply-Side Explanations for Religious Change. " Annals 5 2 7 : 2 7- 3 9 . Finkelstein, Norman G . , and Ruth Bettina Birns. I 9 9 8 . A Nation o n Trial: The Goldhagen Thesis and Historical Truth. New York: Henry Holt. Firebaugh, Glenn, and Brian Harley. I 9 9 I . "Trends i n U. S . Church Attendance: Secularization and Revival, or Merely Lifecycle Effects. " Journal for the Scien­ tific Study of Religion 3 0 : 4 8 7- 500. ---

208

REFERENCES

Flury, Bernhard. I 9 8 8 . Common Principal Components and Related Multivariate Methods. New York: Wiley-Interscience. Friedman, Jeffrey, ed. I 99 5. " Special Issue: Rational Choice Theory and Politics. " Critical Review 9: I - 3 oo. Friedrich, Robert J. I 9 8 2 . " In Defense of Multiplicative Terms in Multiple Regres­ sion Equations. " American .Journal of Political Science 26: 797- 8 3 3 . Fukuyama, Francis. I 9 9 5 . Trust: The Social Virtues and the Creation of Prosper­ ity. New York: Free Press. Furnham, Adrian. I990. The Protestant Work E thic: The Psychology of Work­ Related Beliefs and Behaviors. New York: Routledge. Gabel, Matthew J., and John D. Huber. 2000. "Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions from Party Manifestos Data . " American .Journal o f Political Science 4 4 : 94- I 0 3 . Gallagher, Michael. I 9 9 I . " Proportionality, Disproportionality, and Electoral Sys­ tems . " Electoral Studies I 0: 3 3 - p . . L 9 9 2 . " Comparing Proportional Representation Electoral Systems : Quo­ tas, Thresholds, Paradoxes, and Maj orities . " British .Journal of Political Sci­ ence 22: 4 69 - 9 6 . Gamhetta, Diego. I 99 3 · The Sicilian Mafia: The Business of Private Protection. Cambridge: Harvard University Press. Gans, Herbert J. I 9 67. The Levittowners: Ways of Life and Politics in a New Suburban Community. New York: Pantheon. Gautam, Ram Sakha. I 9 8 I . Indian National Congress and Constitutional Changes in India, r 8 8 J - I 797· New Delhi: Chetana. Geddes, Barbara. l99 L "How the Cases You Choose Affect the Answers You Get: Selection Bias in Comparative Politics. " Political Analysis 2: I3 I- 50. . I 9 94· Politician's Dilemma: Building State Capacity in Latin America. Berkeley: University of California Press. Geertz, Clifford. I 9 6 3 . "The Integrative Revolution: Primordial Sentiments and Civil Politics in the New States. " In O ld Societies and New States, ed. Clifford Geertz. New York: Free Press. Gentile, Pierre, and Hanspeter Kriesi. I 9 9 8 . " Contemporary Radical Right Par­ ties in Switzerland: History of a Divided Family. " In The New Politics of the Right: Nco-populist Parties and Movements in Established Democracies, ed. Hans-Georg Betz and Stefan Immerfall. New York: St. Martin's. Gies, Frances, and Joseph Gies. I 9 9 4 · Cathedral, Forge, and Waterwheel: Technol­ ogy and Invention in the Middle Ages. New York: HarperCollins. Gilchrist, J. I 9 69 . The Church and Economic Activity in the Middle Ages. New York: St. Martin's. Gimpel, Jean. I 9 8 8 . The Medieval Machine: The Industrial Revolution of the Middle Ages. 2d ed. Aldershot, UK: Wildwood House. Goheen, John, M. N. Srinivas, D. G. Karve, and Milton Singer. I 9 5 8 . " India's Cultural Values and Economic Development: A Discussion. " Economic Devel­ opment and Cultural Change 7= I - I 2. Goldberg, Ellis. I 9 9 I . " Smashing Idols and the State: The Protestant Ethic and ---

---

References

209

Egyptian Sunni Radicalism. " Comparatiue Studies in Society and History 3 3 : 3-35· . I 996. "Thinking about How Democracy Works. " Politics and Society 24: 7- I S . Golder, Matt. 200 3 a . "Explaining Variation i n the Success o f Extreme Right Parties in Western Europe. " Comparatiue Political Studies 3 6: 4 3 2- 66. . 200 3 b . " Electoral Institutions, Unemployment, and Extreme Right Par­ ties: A Correction . " British .Journal of Political Science 3 3 : 5 2 5 - 3 4 · Goldhagen, Daniel ] . I 9 9 6 . Hitler's Willing Executioners: Ordinary Germans and the Holocaust. New York: Knopf. Gorski, Philip S. I 99 3 · "The Protestant Ethic Revisited: Disciplinary Revolution and State Formation in Holland and Prussia. " American .Journal of Sociology 99 : 2 6 5 - } I 6. . I 9 9 9 · " Calvinism and State-Formation in Early Modern Europe. " In State/Culture: State-Formation after the Cultural Turn, ed. George Steinmetz. Ithaca: Cornell University Press. Gosnell, Harold F. I 9 3 0. Why Europe Votes. Chicago: University of Chicago Press. Granato, Jim, Ronald Inglehart, and David Leblang. I 9 9 6. "The Effect of Cul­ tural Values on Economic Development: Theory, Hypotheses, and Some Em­ pirical Tests . " American .Journal of Political Science 40: 607- 3 I . Greeley, Andrew M . I 9 8 9 . Religious Change in America. Cambridge: Harvard University Press. Green, Donald P. , and Ian Shapiro. I994· Pathologies of Rational Choice Theory: A Critique of App lications in Political Science. New Haven: Yale University Press. Greif, Avner. I 9 94· " Cultural Beliefs and the Organization of Society: A Histori­ cal and Theoretical Reflection on Collectivist and Individualist Societies . " .Journal of Political Economy I 0 2 : 9 I 2- 5 0 . Hagen, Everett E. I 9 67 . " British Personality and the Industrial Revolution: The Historical Evidence. " In Social Theory and Economic Change, ed. Tom Burns and S . B. Saul. London: Tavistock. Hainsworth, Paul, ed. I 9 9 2a. The Extreme Right in Europe and the USA. Lon­ don: Pinter. . I 9 92b. "The Cutting Edge: The Extreme Right in Post-War Western Europe and the USA. " In The Extreme Right in Europe and the USA, ed. Paul Hainsworth. London: Pinter. , ed. 2000. The Politics of the Extreme Right: From the Margins to the Mainstream. London: Pinter. Hale, William. I 9 9 4 · Turkish Politics and the Military. New York: Routledge. Hamilton, Richard F. I 9 8 2 . Who Voted for Hitler? Princeton: Princeton Univer­ sity Press. Hardin, Russell. I 99 5 · One for A ll: The Logic of Group Conflict. Princeton: Princeton University Press. Harrison, Lawrence E. I 9 8 5 . Underdeuelopment Is a State of Mind: The Latin

---

---

---

---

---

2IO

REFERENCES

American Case. Lanham, MD: Harvard Center for International Affairs and University Press of America. . I 9 9 2 . Who Prospers ? H ow Cultural Values Shape Economic a nd Politi­ cal Success. New York: Basic Books. . I997· The Pan-American Dream: D o Latin American Cultural Values Discourage True Partnership with the United States and Canada? New York: Basic Books. Hatcher, John. I 9 8 6. " Mortality in the Fifteenth Century: Some New Evidence. " Economic History Review, 2d ser. , 3 9 : I 9 - 3 8 . . I994· " England i n the Aftermath of the Black Death. " Past and Present I44: 3 - 3 5 · Hechter, Michael. I 9 8 7 . Principles o f Group Solidarity. Berkeley: University of California Press. Hedlund, Stefan. I 9 9 9 · Russia 's "Market'' Economy: A Bad Case of Predatory Capitalism. London: UCL Press. Herlihy, David. l 9 9 7 · The B lack Death and the Transformation of the West. Cambridge: Harvard University Press. Hermens, Ferdinand A. I 94 I . Democracy o r Anarchy? A Study of Proportional Representation. Notre Dame: University of Notre Dame Press . Hess, Henner. I 9 7 3 . Mafia a n d Mafiosi: The Structure o f Power. Lexington, MA: Lexington Books. Heston, Alan, Robert Summers, and Bettina Aten. 2002. Penn World Tables, Version 6. I. Philadelphia: Center for International Comparisons at the Univer­ sity of Pennsylvania. Hibbs, Douglas A., Jr. l 9 8 7 . The Political Economy of Industrial Societies. Cam­ bridge: Harvard University Press. Highton, Benjamin, and Raymond E. Wolfinger. I 9 9 8 . " Estimating the Effects of the National Voter Registration Act of I 99 3 · " Political Behavior 20: 79- I 04 . Hindess, Barry. I 9 8 8 . Choice, Rationality, and Social Theory. London: Unwin Hyman. Hirschman, Albert 0. I 9 8 4 . "A Dissenter's Confession: 'The Strategy of Eco­ nomic Development' Revisited. " In Pioneers in Development, ed. Gerald M. Meier and Dudley Seers. New York: Oxford University Press. Holton, Robert J. I 9 8 3 . " Max Weber, 'Rational Capitalism,' and Renaissance Italy: A Critique of Cohen . " American Journal of Sociology 8 9 : I 66- 8 o. . I 9 8 5 . The Transition from Feudalism to Capitalism. London: Macmillan. Hout, Michael, and Andrew M. Greeley. l 9 8 7 . "The Center Doesn't Hold: Church Attendance in the United States, I940- I 9 8 4 . " American Sociological Review 5 2: 3 2 5 - 4 5 . Huber, John D . I 9 8 9 . "Values and Partisanship i n Left-Right Orientations: Mea­ suring Ideology. " European Journal of Political Research I 7 : 5 99- 6 I I . Huber, Peter J. I 9 8 1 . Robust Statistics. New York: Wiley-Interscience. Huitt, Ralph K. I 9 6 I . "The Outsider in the Senate: An Alternative Role. " Ameri­ can Political Science Review 5 5 : 5 66- 7 5 . Huntington, Samuel P. I 9 6 8 . Political Order i n Changing Societies. New Haven: Yale University Press. ---

---

---

---

References

2II

Huppert, George. I 9 9 8 . After the 13 lack Death: A Social History o f Jl,arly Modern Europe. 2d ed. Bloomington: Indiana University Press. Husbands, Christopher. I 99 2 . " The Other Face o f I 9 9 2 : The Extreme-Right Explosion in Western Europe. " Parliamentary Affairs 4 5 : 267- 8 4 . Iannaccone, Laurence R. I 99 L " The Consequences o f Religious Market Struc­ ture: Adam Smith and the Economics of Religion. " Rationality and Society 3 : I s 6- n . . I99 5 . "Voodoo Economics ? Reviewing the Rational Choice Approach to Religion. " Journal for the Scientific Study of Religion 3 4 : 7 6- 8 9 . Iannaccone, Laurence R., Roger Finke, and Rodney Stark. I997· "Deregulating Religion: The Economics of Church and State . " Economic Inquiry 3 5 : 3 5 0 - 64 . Ignazi, Piero. I 9 9 2 . " The Silent Counter-Revolution: Hypotheses o n the Emer­ gence of Extreme Right-Wing Parties in Europe. " European Journal of Politi­ cal Research 2 2 : 3 - 3 4 . Inglehart, Ronald. I 9 8 8 . " The Renaissance o f Political Culture . " American Politi­ cal Science Review 8 2 : L 203 - 3 0. . I990. Culture Shift in Advanced Industrial Society. Princeton: Princeton University Press. . I992. "Democratization in Advanced Industrial Societies. " Paper pre­ sented at the Annual Meeting of the Midwest Political Science Association, Chicago. . I997· Modernization and Postmodernization: Cultural, Economic, and Political Change in Forty-three Societies. Princeton: Princeton University Press. . 200 3 . World Values Surveys and European Values Surveys, r 9 8 r - 84, 1 9 9 0 - 9 3 , a n d 1 9 9 5 - 9 7 (ICPSR Version) . Ann Arbor: Interuniversity Consor­ tium for Political and Social Research. Inglehart, Ronald, Miguel Basafiez, and Alejandro Moreno. I 9 9 8 . Human Values and Beliefs: A Cross-Cultural Sourcebook. Ann Arbor: University of Michigan Press. Inkeles, Alex, and Daniel J. Levinson. I') 54· "National Character: The Study of Modal Personality and Social Systems. " In Handbook of Social Psychology, ed. Gardner Lindzey. Cambridge, MA: Addison-Wesley. . I 9 6 9 . "National Character: The Study of Modal Personality and Socio­ cultural Systems . " In Handbook of Social Psychology, ed. Gardner Lindzey and Elliot Aronson, vol. 4· 2d ed. Reading, MA: Addison-Wesley. Inkeles, Alex, and David Horton Smith. I974· Becoming Modern: Individual Change in Six Developing Countries. Cambridge: Harvard University Press. Jackman, Richard, Christopher Pissarides, and Savvas Savouri. I990. " Labour Market Policies and Unemployment in the OECD . " Economic Policy 5 : 449-90. Jackman, Robert W. I 9 7 3 · " O n the Relationship of Economic Development to Democratic Performance." American Journal of Political Science I7: 6 I I - 2 r . . I98o. " A Note on the Measurement of Growth Rates in Cross-National Research. " American .Journal of Sociology 8 6: 6o4 - I 7. . I 9 8 7 . " Political Institutions and Voter Turnout in the Industrial Democra­ cies . " American Political Science Review 8 I : 40 5 - 2 3 . ---

---

---

---

---

---

---

---

2I2

REFERENCES

Jackman, Robert W. , and Ross A. Miller. I99 5 . " Voter Turnout in the Industrial Democracies during the l 9 8 o s . " Comparative Political Studies 2 7 : 4 67 - 9 2 . --- . I 9 9 6 . " A Renaissance of Political Culture ? " American Journal o f Politi­ cal Science 4 0 : 6 3 2 - 5 9 . Jackman, Simon. I994· "Measuring Electoral Bias: Australia, I949 - 9 3 . " British Journal of Political Science 24: 3 I 9 - 5 7 . Jahoda, Gustav. I 99 8 . '"Ordinary Germans' before Hitler: A Critique of the Goldhagen Thesis. " Journal of Interdisciplinary History 29 : 69 - 8 8 . Jones, Bryan D . I 9 9 9 · "Bounded Rationality. " Annual Review o fPolitical Science 2: 297- 3 2 1 . Joyce, John M . I 9 8 4 . "The Old Russian Legacy. " Foreign Policy 5 5 : I 3 2- 5 3 · Kahn, Herman. I 9 7 9 · World Economic Development: 1 9 7 9 a n d Beyond. Boul­ der: Westview. Kalberg, Stephen. I 9 9 4 · Max Weber's Comparative Historical Sociology. Chi­ cago: University of Chicago Press. Keenan, Edward L. l 9 8 6. "Muscovite Political Folkways. " Russian Review 4 5 : I I S- 8 I . Kelman, Herbert C . , and V. Lee Hamilton. I 9 8 9 . Crimes of Obedience: Toward a Social Psychology of Authority and Responsibility. New Haven: Yale Univer­ sity Press. Kendall, M. G., and A. Stuart. I 9 5 0 . "The Law of the Cubic Proportion in Election Results. " British Journal of Sociology I : I 8 3 - 9 6 . Kendall, Willmoore. I94 I . john Locke and the Doctrine of Majority Rule. Ur­ bana: University of Illinois Press. Kennedy, Peter. l 99 8 . A Guide to Econometrics. 4th ed. Cambridge: MIT Press. Kim, Jae-On, and Charles W. Mueller. I 9 7 8 . Factor Analysis: Statistical Methods and Practical Issues. Newberry Park, CA: Sage. Kinder, Donald R., and D. Roderick Kiewiet. I 9 8 1 . " Sociotropic Politics: The American Case . " British Journal of Political Science I I: I 29- 6 I . King, Gary, Ori Rosen, Martin Tanner, and Alexander F. Wagner. 2002. "The Ordinary Election of Adolf Hitler: A Modern Voting Behavior Approach. " Cambridge: Department of Government, Harvard University. Mimeo. Kitschelt, Herbert. I 9 8 8 . " Left-Libertarian Parties: Explaining Innovation in Competitive Party Systems. " World Politics 40: I94- 2 3 4 . --- . I 9 9 5 · The Radical Right i n Western Europe: A Comparative Analysis. Ann Arbor: University of Michigan Press. Kmenta, Jan. l 9 8 6. Elements of Econometrics. 2d ed. New York: Macmillan. Reprint, Ann Arbor: University of Michigan Press, I 9 9 7 · Knack, Stephen, and Philip Keefer. I 9 9 7 · " Does Social Capital Have an Economic Payoff? A Cross-Country Investigation. " Quarterly Journal of Economics I I 2 : I2)I-88. Knight, Jack. I 9 9 2 . Institutions and Social Conflict. N e w York: Cambridge Uni­ versity Press. Kobach, Kris W. I994· " Switzerland. " In Referendums around the World: The Growing Use of Direct Democracy, ed. David Butler and Austin Ranney. Washington, D C : American Enterprise Institute Press. Kornhauser, William. T 9 5 9 · The Politics of Mass Society. Glencoe, I L : Free Press.

References

213

Korpi, Walter. 199 1 . "Political and Economic Explanations for Unemployment: A Cross-National and Long-Term Analysis . " British .Journal of Political Sci­ ence 2 1 : 3 1 5 - 4 8 . Kostoris, Fiorella P. S. 19 9 3 . Italy: The Sheltered Economy. Oxford: Clarendon. Kuper, Adam. 1 9 9 9 . Culture: The Anthropologists ' Account. Cambridge: Har­ vard University Press. Laakso, Markku, and Rein Taagepera. 1 9 7 9 . '"Effective' Number of Parties: A Measure with Application to West Europe. " Comparative Political Studies 1 2 : 3 - 27 . Ladd, Everett C. 1 9 9 9 . The Ladd Report. New York: Free Press. Laitin, David D. 1 9 9 5 . " The Civic Culture a t Thirty. " American Political Science Review 8 9 : 1 68 - 7 3 . Lane, Robert E. 2000. The Loss of Happiness in Market Democracies. New Haven: Yale University Press. LaPalombara, Joseph. 1 99 3 . " Review of Making Democracy Work. " Political Science Quarterly w 8 : 549- 5 0. Lasswell, Harold D. 1 9 3 0. Psychopathology and Politics. Chicago: University of Chicago Press. . 1 9 3 6. Politics: Who Gets What, When, How. New York: McGraw-Hill. Laver, Michael, and Norman Schofield. 1 990. Multiparty Government: The Poli­ tics of Coalition in Europe. Oxford: Oxford University Press. Reprint, with a new preface, Ann Arbor: University of Michigan Press, 1 9 9 8 . Layard, Richard, Stephen Nickell, and Richard Jackman. 199 1 . Unemployment: Macroeconomic Performance and the Labour Market. Oxford: Oxford Univer­ sity Press. Lemann, Nicholas. 1 9 9 6. " Notes and Comments: Kicking in Groups. " A tlantic Monthly 277: 2 2 - 2 6. Lerner, Daniel. 19 5 8 . The Passing of Traditional Society: Modernizing the Middle East. Glencoe: Free Press. Levine, Ross, and David Renelt. 1 9 9 2 . " A Sensitivity Analysis o f Cross-Country Growth Regressions. " American Economic Review 8 2: 94 2- 6 3 . Lewis-Beck, Michael S . , and Glenn E . Mitchell II. 199 3 . " French Electoral Theory: The National Front Test. " Electoral Studies 1 2 : 1 1 2- 2 7 . Lichbach, Mark Irving. 1 99 5 . The Rebel's Dilemma. Ann Arbor: University of Michigan Press. . 1 9 9 6. The Cooperator's Dilemma. Ann Arbor: University of Michigan Press. Lijphart, Arend. 1 9 8 4 . Democracies: Patterns of Maioritarian and Consensus Government in Twenty-One Countries. New Haven: Yale University Press. . 1994. Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, I 9 4 5 - 1 9 9 0 . New York: Oxford University Press. . 1997. "The Difficult Science of Electoral Systems. " Electoral Studies 1 6: 7 3 - 77 · . 1999. Patterns o f Democracy: Government Forms a n d Performance in Thirty-Six Countries. New Haven: Yale University Press. Lipset, Seymour Martin. 1 9 60. Political Man: The Social Bases of Politics. New York: Doubleday. ---

---

---

---

---

214

REFERENCES

Little, Lester K . 1 9 7 8 . Religious Poverty and the Profit bconomy in Medieval Europe. Ithaca: Cornell University Press. Livi-Bacci, Massimo. 1 9 9 2 . A Concise History of World Population. Cambridge: Blackwell. Lopez, Robert S . 1 9 7 6 . The Commercial Revolution of the Middle Ages, 9 5 0 I J S O . New York: Cambridge University Press. Lop rea to, Joseph. 1 9 6 5· " How Would You Like to Be a Peasant ? " Human Orga­ nization 24 : 29 8 - 3 07 . Lowi, Theodore J. 1 9 9 2 . "The State i n Political Science: How We Became What We Study. " American Political Science Review 8 6: I - 7 . Lubbers, Marcel, Merove Gij sberts, and Peer Scheepers. 2002. " Extreme Right­ Wing Voting in Western Europe. " European Journal of Political R esearch 41: 345-78. Lupia, Arthur W., and Mathew D . McCubbins. 1 9 9 7 . The Democratic Dilemma: Knowledge, Deception, and the Foundations of Choice. New York: Cam­ bridge University Press. Luther, Kurt Richard. 2000. " Austria: A Democracy under Threat from the Freedom Party ? " Parliamentary Affairs 5 3 : 4 2 6 - 4 2 . Macaulay, Thomas Babington. 1 84 9 . The History of England from the Accession ofJames II. New York: Hurst. Macintyre, Alasdair C. 1 9 8 1 . After Virtue: A Study in Moral Theory. London: Duckworth. . 1 9 8 8 . Whose .Justice? Which Rationality? London: Duckworth. Mackie, Thomas T. , and Richard Rose. 199 1 . The International Almanac of Electoral History. 3 d ed. Washington, D C : Congressional Quarterly Press. . 1997. " A Decade of Election Results: Updating the International Alma­ nac . " Centre for the Study of Public Policy. Glasgow: University of Strathclyde. MacKinnon, Malcolm H. 1 9 8 8 . "Part I. Calvinism and the Infallible Assurance of Grace: The Weber Thesis Reconsidered " and " Part II. Weber's Exploration of Calvinism: The Undiscovered Provenance of Capitalism. " British .Journal uf Sociology 3 9 : 1 4 3 - 2 1 0 . --- . 1994. "Believer Selectivity in Calvin and Calvinism. " British .Journal of Sociology 4 5 : 5 8 5 -9 5 · Maddison, Angus. 199 1 . Dynamic Forces in Capitalist Development: A Long­ Run Comparative View. New York: Oxford University Press. Mair, Peter, and Francis G. Castles. 1 9 9 7 . " Reflections: Revisiting Expert Judge­ ments. " European .Journal of Political Research 3 l : l 5 0- 5 7 . Maltzman, Forrest, James F . Spriggs I I , and Paul J. Wahl beck. 2000. Crafting Law on the Supreme Court: The Collegial Game. New York: Cambridge University Press. Manchester, William. 1 99 3 . A World Lit Only by Fire. Boston: Little, Brown. Mansbridge, Jane J., ed. 1990. Beyond Self-Interest. Chicago: University of Chi­ cago Press. March, James G., and Johan P. Olsen. 1 9 8 4 . "The New Institutionalism: Organiza­ tional Factors in Political Life . " American Political Science Review 7 8 : 7 3 4- 4 9 . Marselli, Gilberta A. 1 9 6 3 . " American Sociologists a n d Italian Peasant Society: With Reference to the Book of Banfield. " Sociologia Ruralis 3 : 3 T 9- 3 8 . ---

---

References

215

Matthews, Donald R . 1 9 60. U. S. Senators and Their World. Chapel Hill: Univer­ sity of North Carolina Press. Mayer, Nanna, and Pascal Perrineau . 1 9 9 2 . "Why Do They Vote for LePen ? " European journal o f Political Research 22: 1 2 3 - 4 1 . Mayhew, Nicholas. 199 s a . " Modelling Medieval Monetisation . " In A Commer­ cia/ising Economy: England, r o 8 6 to c. I 3 0 0 , ed. Richard H. Britnell and Bruce M. S. Campbell. Manchester: Manchester University Press. . 1 9 9 5 b . " Population, Money Supply, and the Velocity of Circulation in England, 1 3 00- 1 700. " Economic History Review, 2d ser., 4 8 : 2 3 8 - 5 7 . McClelland, David C. 1 9 6 r . The Achieving Society. Princeton: D. Van Nostrand. . 1 9 6 3 . "National Character and Economic Growth in Turkey and Iran . " I n Communications and Political Development, e d . Lucian W. Pye. Princeton: Princeton University Press. McClelland, David C., and David G. Winter. 1 9 69 . Motivating Economic Achieve­ ment. New York: Free Press. McDonald, Michael P., and Samuel L. Popkin. 200 L " The Myth of the Vanishing Voter. " American Political Science Review 9 5 : 9 6 3 - 74 . McNeill, William H . 1 9 7 7 . Plagues and Peoples. New York: Anchor Books. Merkl, Peter, and Leonard Weinberg, eds. 1 9 9 2 . Encounters with the Contempo­ rary Radical Right. Boulder: Westview. Merriam, Charles E. [ 1 9 3 1] 1 9 66. The Making of Citizens: A Comparative Study of Methods of Civic Training. New York: Teachers College Press, Columbia University. Messmore, J. H. 1 9 0 3 . The Life of Edwin Wallace Parker, D.D., Missionary Bishop of Southern Asia, Forty-one Years a Missionary in India. New York: Eaton and Mains. Miller, Edward, and John Hatcher. 199 5. Medieval England: Towns, Commerce, and Crafts, r o 8 6- r3 4 8 . New York: Longman. Minkenberg, Michael. 1 9 9 2 . " The New Right in Germany. " European Journal of Political Research 22: 5 s - 8 r . Mishler, William, and Detlef Pollack. 200 3 . " O n Culture Thick and Thin: To­ ward a Neo-Cultural Synthesis . " In Political Culture in Post-Communist Eu­ rope, ed. Detlef Pollack and Jorg Jacobs. Aldershot, UK: Ashgate. Mishler, William, and Richard Rose. 200 1 . "What Are the Origins of Political Trust? Testing Institutional and Cultural Theories in Post-communist Soci­ eties. " Comparatiue Political Studies 3 4 : 3 0- 62. Mitchell, J. Murray. l 8 9 9 · In Western India: Recollections of My Early Mission­ ary Li{e. Edinburgh, Scotland: D. Douglas. Mitra, Subrata. 1 9 8 8 . "The National Front i n France: A Single-Issue Move­ ment ? " In Right-Wing Extremism in Western Europe, ed. Klaus von Beyme. London: Frank Cass. Money, Jeannette. 1 9 9 9 . Fences and Neighbors: The Political Geography of Immi­ gration Control. Ithaca: Cornell University Press. Monroe, Kristen Renwick. 1 9 9 6 . The Heart of A ltruism: Perceptions ol Common Humanity. Princeton: Princeton University Press. Moore, Barrington. 1 9 66. Social Origins o{ Dictatorship and Democracy. Bos­ ton: Heacon. ---

---

2I6

REFERENCES

Morgan, James N. I 9 6 4 . "The Achievement Motive and Economic Behavior. " Economic Development and Cultural Change n : 24 3 - 67 . Morgan, Philip. I 99 5 · Italian Fascism, I 9 I 9 - I 9 4 _) . New York: S t . Martin's. Morris-Jones, W. H. I 9 8 8 . "The Politics of the Indian Constitution. " In Constitu­ tions in Democratic Politics, ed. Vernon Bogdanor. Brookfield, VT: Gower Publishing. Mudde, Cas. 2000. The Ideology of the Extreme Right. Manchester: Manchester University Press. Mudde, Cas, and Joop Van Holsteyn. 2000. "The Netherlands: Explaining the Limited Success of the Extreme Right. " In The Politics of the Extreme Right: From the Margins to the Mainstream, ed. Paul Hainsworth. London: Pinter. Mueller, Dennis C. 200 3 . Public Choice III. New York: Cambridge University Press. Muir, Edward. I999· "The Sources of Civil Society in Italy. " .Journal of Interdisci­ plinary History 29: 3 79-406. Muller, Edward N., and Mitchell A. Seligson. l 9 9 4 · " Civic Culture and Democ­ racy: The Question of Causal Relationships . " American Political Science Re­ view 8 8 : 63 5 - 5 2. Munch, Paul. I 99 3 · "The Thesis before Weber: An Archaeology. " In Weber 's Protestant E thic: Origins, Evidence, Contexts, ed. Hartmut Lehmann and Guenther Roth. New York: Cambridge University Press. Nagler, Jonathan. I 9 9 I . "The Effect o f Registration Laws and Education o n U. S. Voter Turnout. " American Political Science Review 8 5 : I 3 9 3 - I 4 0 5 . Newell, James L . 2000. " Italy: The Extreme Right Comes i n from the Cold. " Parliamentary Affairs 5 3 : 4 69- 8 5 . Nicholson, Stephen P. 2003 . "The Political Environment and Ballot Proposition Awareness. " American .Journal of Political Science 4 7 : 4 0 3 - I O . Nickell, Stephen. I 9 9 7 · " Unemployment and Labor Market Rigidities: Europe versus North America. " .Journal of Economic Perspectives I I: 5 5 - 7 4 · . I 99 8 . "Unemployment: Questions and Some Answers. " Economic .Jour­ nal I o 8 : 8o2- I 7 . Nipperdy, Thomas. I 99 3 · " Max Weber, Protestantism, and the Context o f the Debate around I 9 oo . " I n Weber's Protestant E thic: Origins, Evidence, Con­ texts, ed. Hartmut Lehmann and Guenther Roth. New York: Cambridge Uni­ versity Press. Noonan, John T. I9 57· The Scholastic Analysis of Usury. Cambridge: Harvard University Press. NPR (National Public Radio) . 2000. " On a Roll. " Weekend Edition, April 29. Oakes, Guy. I 99 3 · "The Thing That Would Not Die: Notes on Refutation. " In Weber's Protestant Ethic: Origins, Evidence, Contexts, ed. Hartmut Lehmann and Guenther Roth. New York: Cambridge University Press. OECD (Organisation for Economic Cooperation and Development ) . I 99 I . OECD Economic Outlet: Historical Statistics. Paris: Organisation for Eco­ nomic Cooperation and Development. O'Hearn, Denis. I 9 9 8 . Inside the Celtic Tiger: The Irish Economy and the Asian Model. London: Pluto Press. ---

References

217

Park, Richard L . , and Bruce Bueno d e Mesquita. 1 9 7 9 . india's Political System. Englewood Cliffs, NJ : Prentice-Hall. Paxton, Pamela. 1 9 9 9 . " Is Social Capital Declining in the United States ? A Multi­ ple Indicator Assessment. " American Journal of Sociology 1 0 5 : 8 8 - 1 27. Persson, Karl Gunnar. 1 9 8 8 . Pre-industrial Economic Growth: Social Organiza­ tion and Technological Progress in Europe. Oxford: Blackwell. . 1 99 3 . "Was There a Productivity Gap between Fourteenth Century Italy and England ? " Economic History Review, 2d ser. , 4 6: 1 0 5 - 1 4 . Pettinico, George . 1 9 9 6 . " Civic Participation Alive and Well i n Today's Environ­ mental Movement. " Public Perspective 7 : 27- 3 0. Phillips, Derek L. 1 99 3 . Looking Backward: A Critical Appraisal of Communi­ tarian Thought. Princeton: Princeton University Press. Philpott, Daniel. 2000. "The Religious Roots of Modern International Rela­ tions. " World Politics 5 2 : 206-4 5 . Pirenne, Henri. 1 9 3 7 . Economic and Social History o f Medieval Europe. New York: Harcourt Brace. Pizzorno, Alessandro. 1 9 7 1 . " Amoral Familism and Historical Marginality. " In European Politics: A Reader, ed. Mattei Dogan and Richard Rose. Boston: Little, Brown. Poggi, Gianfranco. 1 9 8 3 . Calvinism and the Capitalist Spirit: Max Weber's Protes­ tant Ethic. London: Macmillan. Popkin, Samuel L. 1 9 7 9 . The Rational Peasant: The Political Economy o f Rural Society in Vietnam. Berkeley: University of California Press. Partes, Alej andro. 1 9 7 6 . " On the Sociology of National Development: Theories and Issues. " American Journal of Sociology 8 2 : 5 5 - 8 5 · Posner, Richard A . 2oo r . Public Intellectuals: A Study o f Decline. Cambridge: Harvard University Press. Powell, G. Bingham, Jr. 1 9 8 6a. " American Voter Turnout in Comparative Perspec­ tive . " American Political Science Review So: 1 7-44. . 1 9 8 6b. " Extremist Parties and Political Turmoil: Two Puzzles. " Ameri­ can Journal of Political Science 3 0: 3 5 7 - 7 8 . Przeworski, Adam, and Henry Teune. 1 9 7 0 . The Logic o f Comparative Social Inquiry. New York: Wiley-Interscience. Putnam, Robert D. 1 9 9 3 . Making Democracy Work: Civic Traditions in Modern Italy. Princeton: Princeton University Press. . 1 9 9 5 3 . " Bowling Alone: America's Declining Social Capital. " Journal of Democracy 6: 6 5 - 7 8 . . 1 9 9 5 b . "Tuning In, Tuning Out: The Strange Disappearance of Social Capital in America. " PS: Political Science and Politics 2 8 : 664- 8 3 . . 2000. B owling A lone: The Collapse and Revival of American Commu­ nity. New York: Simon and Schuster. Putnam, Robert D., and Lewis M. Feldstein. 200 3 . Better Together: Restoring the American Community. New York: Simon and Schuster. Pye, Lucian W. 1 9 8 5 . Asian Power and Politics: The Cultural Dimensions of Authority. Cambridge: Harvard University Press. . 1 9 9 1 a . "Political Culture Revisited. " Political Psychology 1 2: 4 8 7 - s o 8 . ---

---

---

---

---

---

218

REFERENCES

. 1 9 9 1 b . China: An introduction. 4th ed. New York: HarperCollins. . 2000. '"Asian Values ' : From Dynamos to Dominoes ? " In Culture Matters: How Values Shape Human Progress, ed. Lawrence E. Harrison and Samuel P. Huntington. New York: Basic Books. Radcliff, Benj amin. 1 9 9 6 . "Turnout a n d t h e Economy. " Comparative Political Studies 29 : 7 1 8 - 2 3 . Rae, Douglas W. 1 9 7 1 a . The Political Consequences o f Electoral Laws. 2d ed. New Haven: Yale University Press. . 1 9 7 1 b . " An Estimate for the Decisiveness of Election Outcomes. " In Social Choice, ed. Bernhardt Lieberman. New York: Gordon and Breach. Rice, Tom W. , and Jan L. Feldman. 1 9 9 7 . " Civic Culture a n d Democracy from Europe to America. " Journal of Politics 5 9 : 1 1 4 3 - 72 . Riedwyl, Hans, a n d Jiirg Steiner. 199 5 . "What I s Proportionality Anyhow ? " Comparative Politics 27: 3 5 7- 69 . Riesman, David, with Reuel Denney and Nathan Glazer. 1 9 5 0 . The Lonely Crowd: A Study of the Changing American Character. New Haven: Yale University Press. Riker, William H. 1 9 8 2 . "The Two-Party System and Duverger's Law: An Essay on the History of Political Science . " American Political Science Review 7 6 : 7 5 3 - 66. . 19 90. "Political Science and Rational Choice. " In Perspectives on Posi­ tiue Political Theory, ed. James E. Alt and Kenneth A. Shepsle. New York: Cambridge University Press. Riker, William H., and Peter C. Ordeshook. 1 9 7 3 . An Introduction to Positiue Political Theory. Englewood Cliffs, NJ : Prentice-Hall. Ross, Andrew. 1 9 8 6. John Philip, I 7 7J - I 8J I : Missions, Race, and Politics in South Africa. Aberdeen, Scotland: Aberdeen University Press. Russell, Josiah Cox. 198 5. Late Ancient and Medieval Population Control. Phila­ delphia: American Philosophical Society. Rustow, Dankwart A. 1 9 50. " Some Observations on Proportional Representa­ tion. " Journal of Politics 1 2 : 1 0 7 - 2 7· Sabetti, Filippo. 1 9 9 6. "Path Dependency and Civic Culture: Some Lessons from Italy about Interpreting Social Experiments. " Politics and Society 24 : 1 9 - 4 4 . Samuelson, Robert J . 1 9 9 6 . "Join the Club . " Washington Post National Weekly Edition, April I s - 2 1 , 5 . Samuelsson, Kurt. 1 9 6 1 . Religion and Economic A ction. London: Heinemann. Sandel, Michael J. l 9 9 6 . Democracy's Discontent: America in Search of a Public Philosophy. Cambridge: Harvard University Press. --- . 1 9 9 8 . Liberalism a n d the Limits o fJustice. 2d e d . New York: Cambridge University Press. Sartori, Giovanni. 1 9 68 a . " Political Development a n d Political Engineering . " In Public Policy, vol. 1 2, ed. John D. Montgomery and Albert 0. Hirschman. Cambridge: Harvard University Press. . 1 9 68b. " Representational Systems. " In International Encyclopedia of the Social Sciences, vol. 1 3 , ed. David L. Sills. New York: Macmillan. . 1 9 7 6. Parties and Party Systems. Cambridge: Cambridge University Press. . T 9 8 6. "The Influence of Electoral Systems: Faulty Laws or Faulty

---

---

---

---

---

-----

References

2I9

Method ? " I n Jilectoral Laws and Their Political Consequences, e d . Bernard Grofman and Arend Lijphart. New York: Agathon. . I 994· Comparative Constitutional Engineering: An Inquiry into Struc­ tures, Incentives, and Outcomes. New York: New York University Press. Schachter, Gustav. I96 5. The Italian South: Economic Development in Mediterra­ nean Europe. New York: Random House. Schain, Martin A. I 9 8 7 . "The National Front in France and the Construction of Political Legitimacy. " West European Politics ro: 229 - 5 2. . I 9 8 8 . " Immigration and Changes in the French Party System. " European Journal of Political Research I 6: 5 9 7 - 6 2 r . Schneider, Jane. I990. " Spirits and the Spirit of Capitalism. " In Religious Ortho­ doxy and Popular Faith in European Society, ed. Ellen Badone. Princeton: Princeton University Press. Schneider, Jane, and Peter Schneider. I994· " Mafia, Antimafia, and the Question of Sicilian Culture. " Politics and Society 22: 2 3 7- 5 8 . Schuessler, Alexander A. 2000. A Logic of Expressive Choice. Princeton: Prince­ ton University Press. Sears, David 0 . , and Carolyn L. Funk. I990. " Self-Interest in Americans' Political Opinions. " In Beyond Self-Interest, ed. Jane J. Mansbridge. Chicago: Univer­ sity of Chicago Press. Seligson, Mitchell A. 2002. "The Renaissance of Political Culture or the Renais­ sance of the Ecological Fallacy ? " Comparative Politics 3 4 : 273 - 9 2. Seton-Watson, Christopher. I 9 8 3 . "Italy. " In Democracy and Elections: Elec­ toral Systems and Their Political Consequences, ed. Vernon Bogdanor and David Butler. Cambridge: Cambridge University Press. Shepsle, Kenneth A. I 9 7 9 · " Institutional Arrangements and Equilibrium in Multi­ dimensional Voting Models. " American Journal of Political Science 2 3 : 2 7- 5 9· . I 9 8 9 . " Studying Institutions: Some Lessons from the Rational Choice Approach. " Journal of Theoretical Politics I : I 3 I - 4 7 · Shugart, Matthew S. I 9 9 2 . " Electoral Reform i n Systems o f Proportional Repre­ sentation. " European Journal of Political Research 2 I: 207- 24. Sidjanski, Dusan. I979· "Turnout, Stability, and the Left-Right Dimension. " In Switzerland at the Po lls: The National Elections of 1 9 79, ed. Howard Penni­ man. Washington, D C : American Enterprise Institute. Silverman, Sydel F. I 9 6 8 . " Agricultural Organization, Social Structure, and Values in Italy: Amoral Familism Reconsidered. " American Anthropologist 70: I - 20. Simon, Herbert A. l98 5 . "Human Nature in Politics: The Dialogue of Psychology with Political Science. " American Political Science Review 79 : 29 3 - 3 04 . --- . I 9 8 6. " Rationality in Psychology and Economics. " I n Rational Choice: The Contrast between Economics and Psychology, ed. Robin M. Hogarth and Melvin W. Reder. Chicago: University of Chicago Press. Singer, Milton. I 9 5 6. " Cultural Factors in India's Economic Development. " An­ nals of the American Academy of Political and Social Science 3 o 5 : 8 I - 9 r . Skocpol, Theda. I999· " How Americans Became Civic. " I n Civic Engagement in American Democracy, ed. Theda Skocpol and Morris P. Fiorina. Washington, D C : Brookings Institution. Smelser, Neil J. , and Richard Swedberg. T 9 9 4 · "The Sociological Perspective on

---

---

---

220

REFERENCES

the Economy. " In The Handbook of JC,conomic Sociology, ed. Neil J. Smelser and Richard Swedberg. Princeton: Princeton University Press. Smith, Adam. [ I 7 7 6] I 9 76. An Inquiry into the Nature and Causes of the Wealth of Nations. Chicago: University of Chicago Press. Smith, Tom W. I990. "Trends in Voluntary Group Membership: Comments on Baumgartner and Walker. " American Journal of Political Science 34: 64 6- 6 I . Sobel, Joel. 2002. " Can We Trust Social Capital ? " Journal o f Economic Litera­ ture 40: I 3 9 - 5 4 · SOPEMI. Various years. Trends i n International Migration. Paris: Organisation for Economic Cooperation and Development. Spiro, Melford E. I 9 8 7. Culture and Human Nature. Chicago: University of Chicago Press. Spitz, Lewis W. I 9 8 5 . The Protestant Reformation, I S I 7- I 5 5 9 · New York: Harper and Row. Spruyt, Hendrik. I 9 9 4 · The Sovereign State and Its Competitors: An Analysis of Systems Change. Princeton: Princeton University Press. Squire, Peverill, Raymond E. Wolfinger, and David P. Glass. I 9 8 7 . " Residential Mobility and Voter Turnout . " American Political Science Review 8 I : 4 5 - 6 5 . Stark, Oded. I 99 5 · Altruism and Beyond: A n Economic Analysis o f Transfers and Exchanges within Families and Groups. New York: Cambridge University Press. Stark, Rodney. I 9 96. The Rise of Christianity: A Sociologist Reconsiders History. Princeton: Princeton University Press. Stark, Rodney, and Roger Finke. woo. Acts of Faith: Explaining the Human Side of Religion. Berkeley: University of California Press. Stengel, Richard. l99 6. " Bowling Together: Civic Engagement in America Isn't Disappearing hut Reinventing Itself. " Time, July 22, 3 5 - 3 6. Stewart, Robert. I 8 9 9 · Life and Work in India: An Account of the Conditions, Methods, Difficulties, R esults, Future Prospects, and Reflex Influence of Mis­ sionary Labor in India. Philadelphia : Pearl Publishing. Stimson, James A. I 9 8 5. " Regression in Space and Time: A Statistical Essay. " American Journal of Political Science 29 : 9 I 4 - 4 7 . Stolzenberg, Ross M., Mary Blair-Loy, a n d Linda J . Waite. I 9 9 5 · " Religious Participation in Early Adulthood: Age and Family Life Cycle Effects on Church Membership. " American Sociological Review 6o: 84- I 0 3 . Strang, David. I 9 90. " From Dependency to Sovereignty: An Event History Analysis of Decolonization, I 8 70- I 9 8 7 . " American Sociological Review 5 5 : 8 4 6- 6o. Straussman, Jeffrey D. I 9 7 5 · "What Did Tomorrow's Future Look Like Yester­ day ? " Comparative Politics 8: I 66- 8 2 . Street, John. I 9 9 4 · " Political Culture: From Civic Culture t o Mass Culture. " British Journal o f Political Science 24: 9 5 - I I 4 . Strom, Kaare. I 9 9 0 . Minority Government and Majority Rule. New York: Cam­ bridge University Press. Strom, Kaare, Ian Budge, and Michael J. Laver. I 9 9 4 · " Constraints on Cabinet Formation in Parliamentary Democracies . " American Journal of Political Sci­ ence 3 8 : 3 03 - 3 5 . Summers, Robert, and Alan H eston. T 99 T . "The Penn World Table ( Mark 5 ) : An

References

221

Expanded Set of International Comparisons. " Quarterly journal of liconom­ ics w6: 3 2 7 - 6 8 . Swank, Duane. 1 9 9 6 . " Culture, Institutions, and Economic Growth. " American Journal of Political Science 40: 660- 7 9 . Swanson, Guy E. 1 9 7 1 . " An Organizational Analysis of Collectivities . " American Sociological Review 3 6: 607 - 24 . Swyngedouw, Marc. 1 99 8 . "The Extreme Right i n Belgium: Of a Non-existent Front National and an Omnipresent Vlaams Bloc . " In The New Politics of the Right: Neo-populist Parties and Movements in Established Democracies, ed. Hans-Georg Betz and Stefan Immerfall. New York: St. Martin's. Swyngedouw, Marc, and Gilles Ivaldi. 200 1 . " The Extreme Right Utopia in Belgium and France: The Ideology of the Flemish Vlaams Blok and the French Front National. " West European Politics 24: 1 - 22. Taagepera, Rein. 1 99 8 . "Effective Magnitude and Effective Threshold. " Electoral Studies 1 7 : 3 9 3 -404. Taagepera, Rein, and Matthew S . Shugart. l 9 8 9 . Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven: Yale University Press. Taggart, Paul. 1 99 5 . "New Populist Parties in Western Europe. " West European Politics 1 8 : 3 4 - p . Tarrow, Sidney. 1 9 9 6. "Making Social Science Work across Space and Time: A Critical Reflection on Robert Putnam's Making Democracy Work. " American Political Science Review 90: 3 8 9 - 9 7 . Tawney, Richard H. 1 9 2 6 . Religion a n d the Rise o f Capitalism. New York: Harcourt Brace. Taylor, Charles. l9 8 5 . Philosophy and the Human Sciences. New York: Cam­ bridge University Press. . 1 9 8 9 . Sources of the Self: The Making of the Modern Identity. Cam­ bridge: Harvard University Press. Taylor, Charles L., and Michael C. Hudson. 1 9 7 2 . World Handbook of Political and Social Indicators. 2d ed. New Haven: Yale University Press. Taylor, Michael, and V. M. Herman. 1 9 7 1 . "Party Systems and Government Stability. " American Political Science Review 6 5 : 28 - 3 7 · Teixeira, Ruy A. 1 99 2 . The Disappearing American Voter. Washington, D C : Brookings Institution. Tingsten, Herbert. 1 9 3 7 . Political Behavior: Studies in Election Statistics. Lon­ don: P. S. King and Son. Tobin, James. l 9 5 8 . " Estimation of Relationships for Limited Dependent Vari­ ables. " Econometrica 26: 24- 3 6. Tocqueville, Alexis de. [ 1 8 3 5 - 40] 1 94 5 . Democracy in America. New York: Knopf. Tong, James. 1 9 8 8 . " Rational Outlaws: Rebels and Bandits in the Ming Dynasty, 1 3 6 8 - 1 64 4 . " In Rationality and Revolution, ed. Michael Taylor. New York: Cambridge University Press. Trivers, Robert. 1 9 7 1 . "The Evolution of Reciprocal Altruism. " Quarterly Re­ view of Biology 4 6: 3 5 - 5 7 . Tsebelis, George. 1 9 9 0 . Nested Games: Rational Choice i n Comparative Politics. Berkeley: University of California Press. ---

222

REFERENCES

Tversky, Amos, and Daniel Kahneman. I 9 7 4 · "Judgment under Uncertainty: Heuristics and Biases. " Science l8 5: H 24 - 3 L . I 9 8 6. " Rational Choice and the Framing of Decisions. " In Rational Choice: The Contrast between Economics and Psychology, ed. Robin M. Hogarth and Melvin W. Reder. Chicago: University of Chicago Press. Valentine, Charles A. I 9 6 8 . Culture and Poverty: Critique and Counter-Propo­ sals. Chicago: University of Chicago Press. Varese, Federico. 200 I . The Russian Mafia: Private Protection in a New Market Economy. New York: Oxford University Press. Verba, Sidney. I 9 6 5 . " C onclusion: Comparative Political Culture. " In Political Culture and Political Development, ed. Lucian W. Pye and Sidney Verba. Princeton: Princeton University Press. Verba, Sidney, Norman H. Nie, and Jae-on Kim. I 9 7 8 . Participation and Political Equality: A Seven-Nation Comparison. New York: Cambridge University Press. Voerman, Gerrit, and Paul Lucardie. l992. "The Extreme Right in the Nether­ lands. " European fournal of Political Research 22: 3 s - 5 4 · Von Beyme, Klaus. I 9 8 8 . Right- Wing Extremism i n Western Europe. London: Frank Cass. Walzer, Michael. I 9 6 5. The Revolution of the Saints. Cambridge: Harvard Uni­ versity Press. Warner, R. Stephen. I 9 9 3 · "Work in Progress toward a New Paradigm for the Sociological Study of Religion in the United States. " American Journal of Sociology 9 8 : I 044-9 3 · Warwick, Paul. l992a. " Ideological Diversity and Government Survival in West­ ern European Parliamentary Democracies. " Comparative Political Studies 2 5 : 3 3 2- 6 I . . I 9 9 2b. " Economic Trends and Government Survival i n West European Parliamentary Democracies . " American Political Science Review 8 6: 8 7 5 - 8 7 . Wattenberg, Martin P. 2002. Where Have All the Voters Gone? Cambridge: Harvard University Press . Weber, Max. I9 5 0 . General Economic History. Glencoe, IL: Free Press. . I 9 5 I . The Religion of China: Confucianism and Taoism. Glencoe, IL: Free Press. . [ I 90 5 ] I 9 5 8a. The Protestant Ethic and the Spirit of Capitalism. New York: Scribner's. . l 9 5 8 b . The Religion of India: The Sociology of Hinduism and Bud­ dhism. Glencoe, IL: Free Press. --- . I 9 7 8 . Economy and Society: An Outline o f Interpretive Sociology. Berke­ ley: University of California Press. Wedeen, Lisa. 2002. " Conceptualizing Culture: Possibilities for Political Sci­ ence . " American Political Science Review 9 6: 7 I 3 - 2 8 . Weiner, Myron. I 9 5 7· Party Politics i n India: The Development o f a Multi-Party System. Princeton: Princeton University Press. , ed. I 9 66. Modernization: The Dynamics of Growth. New York: Basic Books. ---

---

---

---

---

---

References

223

I989. The indian Paradox: t:ssays in indian Politics. New Delhi: Sage. Weingast, Barry R. l 9 9 7 · "The Political Foundations of Democracy and the Rule of Law. " American Political Science Review 9 I : 24 5 - 6 3 . . I 9 9 8 . " Constructing Trust: The Political and Economic Roots o f Ethnic and Regional Conflict. " In Institutions and Social Order, ed. Karol Soltan, Eric M. Uslaner, and Virginia Haufler. Ann Arbor: University of Michigan Press. Weingast, Barry R., and William ]. Marshall. I 9 8 8 . "The Industrial Organization of Congress: Or, Why Legislatures, Like Firms, Are Not Organized as Mar­ kets. " Journal of Political Economy 9 6: I 3 2- 6 3 . Westle, Bettina, and Oskar Niedermayer. I 9 9 2 . " Contemporary Right-Wing Ex­ tremism in West Germany. " European Journal of Political Research 22: 8 3 - IOO. White, Andrew Dickson. I 89 6 . A History of the Warfare of Science with Theol­ ogy in Christendom. New York: D. Appleton. White, Halbert. l 9 8o. " A Heteroskedasticity-Consistent Covariance Matrix Esti­ mator and a Direct Test for Heteroskedasticity. " Econometrica 4 8 : 8 I 7- 3 8 . White, Lynn, Jr. I 9 7 2 . "The Expansion o f Technology, 5 00- I 5 0o. " I n The Fon­ tana Economic History of Europe, vol. I. London: Fontana. . I 9 7 8 . Medieval Religion a n d Technology. Berkeley: University o f Califor­ nia Press. White, Peter. I 9 9 2 . Predestination, Policy, and Polemic: Conflict and Consensus in the English Church from the Reformation to the Civil War. New York: Cambridge University Press. White, Stephen. l 9 8 4 . "Political Culture in Communist States. " Comparative Political Studies I 6: 3 5 I - 6 5 . Whittam, John. I 9 9 5 . Fascist Italy. Manchester: Manchester University Press. Willems, Emilio. I 9 5 5 · " Protestantism as a Factor of Culture Change in Brazil. " Economic Development and Cultural Change 3 : 3 2 I - 3 3 · Wilson, Edward 0 . I 9 7 8 . On Human Nature. Cambridge: Harvard University Press. Wittman, Donald A. I 99 5. The Myth of Democratic Failure: Why Political Insti­ tutions Are Efficient. Chicago: University of Chicago Press. Wolfinger, Raymond E., and Steven ]. Rosenstone. I 9 8 o. Who Votes ? New Haven: Yale University Press. Wuthnow, Robert. I 9 9 8 . Loose Connections: Joining Together in America's Frag­ mented Communities. Cambridge: Harvard University Press. Zagare, Frank C. I990. " Rationality and Deterrence. " World Politics 4 2 : 23 8 - 6o. Zaret, David. I 9 9 2 . " Calvin, Covenant Theology, and the Weber Thesis . " British Journal of Sociology 4 3 : 3 69 -9 I . Zeitlin, Irving. I990. Ideology and the Development of Modern Sociological Theory. 4th ed. Englewood Cliffs, NJ: Prentice-Hall. Zeller, Richard A., and Edward C. Carmines. I 9 8o. Measurement in the Social Sciences: The Link between Theory and Data. New York: Cambridge Univer­ sity Press.

---

---

Index

No te: Page numbers in italics indicate figures or tables. Abramson, Paul R., I O i n. 3, I 0 2 achievement motivation, I 0 3 , I 9 2; economic growth and, I 0 3 - 4 ; Granato, Inglehart, Leblang and, I o i ; McClelland on progress and, I, 6, 3 3 - 3 4 ; McClelland's I 9 5 0 data on, I 0 2- 3 , r o 3 , I 5 6n. 2 2 ; po­ litical culture approach and, l07; principles of, I 07; voter turnout and, I 5 6- 5 8 , r 5 7; Weber's entre­ preneurial values vs. McClelland's, 4 011. 5 Adorno, T. W., I 69 Advantage: correlations among Trust, Helpfulness, and, 7 6 - 7 7 , 77; GSS measures of, 7 5 - 76. See also Trust aggregate properties of societies, political culture and, 8 - 9 Ajayi, J . F . Ade, I I on. I O Aldrich, John H., I 9 , 2011. I 8 , I 4 Ill. 4 Alesina, Alberto, I 8 6 Almond, Gabriel A . : behavioral ap­ proach of, I4; civic cultures and de­ mocracies studies by, 4 , 6, 8 6, I 5 711. 2 3; on civic political culture, 7; Mer­ riam and, I 5n. I 2 ; on motivation for political behavior, I 6; on politi­ cal culture and electoral participa-

tion, I 3 8 , I 9 5 ; on political culture vs. rational choice, 29; on preva­ lence of participatory orientations in societies, 9 9 · See also Verba, Sidney Alt, James E., I 8 6 altruism: rational behavior and, 293 0, l 8 8 - 89; reciprocal, l 9 - 2o, l 9 2on. I 7 Amemiya, Takeshi, I 8 o, I 8 on. 2 I American Youth Soccer Organization, 9 4 . 9411· 4 5 . 95 amoral familism, I 9 7 - 9 8 ; Banfield on progres s and, I, 6, 3 3 - 3 4 , I 9 8 - 99 ; medieval communities and, 70; southern Italian villages of T 9 s os and, I 9 9 Andrews, Josephine T. , I 7 antisystem parties, I 64 , I 6 s - 66. See also extreme right-wing parties Aquinas, Thomas, 5 I Armstrong, J. Scott, 63 11. 2 Arrow, Kenneth, I ) n. I 2, I 7, I 8 9 Association for the Industrial Development of the Italian South, I99 Ataturk, Kemal, I 9 8n. I Aten, Bettina, I oon. 2

225

226 Austin, Granville, o n Congress Party of India, 2 7n. 2 5 Austria, extreme right-wing party in, I 6 5 , I 65n. 2, I 74n. Io, r 7 6 Austrian Freedom Party, I 6 5 n. 2 authoritarian regimes: Inglehart on overall life satisfaction and, 1 26, I 27 - 2 8 Backer, Susann, I 67 Baden, Offenbacher's statistical study of, 5 4 - 5 5 , I 9o; high school enroll­ ment data, s 6; Samuelsson's cri­ tique of, 5 6 - 5 8 ; taxable wealth data, 5 7 Baldwin, John W. , 5 2 Balgat, Turkey, transformation of, I 9 8n. I Banfield, Edward C . , 2oon. 4; on amoral familism and progress, I, 6, 3 3 - 3 4 , I 9 8 - 9 9; on conservative im­ plications of political culture ap­ proach, I 3 ; on durability of politi­ cal culture, 9, I 9 8 n. I; institutional factors and amoral familism thesis of, 200; "Montegrano " study by, I 9 7 - 9 8 ; Putnam's Italian study and, I9o; on sense of community, 5 9 ; on significance of political cul­ ture to other outcomes, I 2; on trust, 7 ; Weber's Protestant ethic thesis and, 3 3 Barra, Robert ]. , I oon. I , I o 3 ; endoge­ nous growth model of, I o i ; Knack and Keefer's models and, I I 2, 1 13 , I I 4 - I 5 ; national achievement lev­ els data of, l 0 2n. 4 Basaiiez, Miguel, I 5 8 Bates, Robert H., I 6 Bauer, Yehuda, 2n. I Baumgartner, Frank R., 8 2 , 8 2n. 26, 8 3 Baxter, Richard, 5 5 , I 9 0 Beck, Nathaniel, I 4 7n. 1 4 behavioral approach, I 4 ; political cul­ ture approach and, 3 0. See also po­ litical behavior

INDEX Belgium, extreme right-wing party in, l 6 8 , l 74n. l O, L 7 6 Bell, Daniel, I 6 3 , I 8 5 Bellah, Robert N., 69n. I I Berelson, Bernard, I 6 Berk, Richard A., I 02n. 5, I 2 3 n. 2 3 , I 8o Bedew, David E . , 6n. 7 Better Together (Putnam and Feld­ stein), 72n. I 2 Betz, Hans-Georg, I 67n. 4 , I 68 - 69 , I 68n. 5 , I 72, I 7 6n. I 4 Bharatiya Janata Party ( BJP) o f India, 28n. 26 Billig, Michael S . , 5 5 , I 3 on. 2 8 Birns, Ruth Bettina, 2 n . l Black, Duncan, I 5 n. I 2 Black Death, economic effects of, 4 8 -49 Blair-Loy, Mary, 8 5 n. 3 1 Blais, Andre, 26, I 4 3 n. 7, I 7 I Bollen, Kenneth A . , I 2o, I 2 I , I 24n. 2 5 , I 29 bookkeeping, Weber on rational, 4 8 bounded rationality, 20 Bowling Alone ( Putnam) , 6o, 9 596, I 90- 9 I ; group membership analysis of, 8 3 n. 27; group member­ ship thesis of, 77n. 20; selection bias in, I ')I ; social trust analysis in, 7 5. See also group membership in U. S. bowling leagues, 9 6; Putnam on, 72, 9I; quality of, 9 I n. 40. See also group membership in U. S . Boyd, Richard W. , I 5 I Bradburn, Norman M., 6n. 7 Bradley, Bill, 7 2 - 7 3 Brass, Paul R., 2 4 , 2 7 n . 2 5 Braude!, Fernand: o n Calvin's beliefs about usury, s o- p ; on Weber de­ bate with Marxism and historical materialism, 3 6; on Weber's growth pattern thesis, 4 3 ; on Weber's retro­ spective analysis, 54, 5 5 , I 90 Brh:hon, Pierre, I 67

Index Brehm, John, 7 5 Britain, marginalization and extrem­ ism in, I 69. See also England British Commonwealth: missionary ac­ tivities and economic growth in, no- n ; single-member districts in, 26 Britnell, Richard, 4 3 -44, 4 6, 4 9 Brown, Donald E . , I 3 on. 2 8 Browning, Christopher R., 2 n . I Brucker, Gene, 70 Brustein, William, 3, I 70 bubonic plague, economic effects of, 48-49 Budge, Ian, 2 3 Bueno de Mesquita, Bruce, 27, 28n. 26 Bureau of Labor Statistics, labor force participation data, 8 9 - 90n. 3 7 Burkhart, Ross E., I 29 Bush ( George W. ) administration, 7 3 , 7 3 nn. I S - I 6 business separation from household, Weber on, 4 8 Butler, David E., 66n. 8, l 8 6n. 2 8 Calise, Mauro, I 8 7n. 2 9 Calmfors, Laers, I 8 6 Calvert, Randall L., I 4 n. I O, 20, 22, 30 Calvin, John, 3 8 , 3 9- 4 0, 4on. 4 , so5 I. See also Protestant ethic thesis Calvinism, pastoral teachings vs. dogma of, 3 9 -40, 50 Campbell, Angus, I4n. n , I 6 Carmines, Edward C . , 67n. 9 Carter, Elisabeth L . , l 8 o Carty, R . K . , I 7 I Cams-Wilson, E . M . , 4 6, 4 7 Castles, Francis G . , I 64 , I 74- 7 5 , I 74n. I I Centre Party (Netherlands ) , r 7 6 change: alternative behaviors and, 23, n; political culture approach and, 3 o. See also electoral engineer­ ing; institutional approach

227 Chatterj ee, Samprit, I 4 7n. I4 Chaucer, Geoffrey, 47, 47n. l 2 , 54n. I5 Chirac, Jacques, I 6 5 , I 6 5n. I Chong, Dennis, I 6n. I 3 Christian Democratic Party (Portugal) , I76

churches. See religious participation Chwe, Michael Suk-Young, 22n. 2 I Cipolla, Carlo M . , 4 4 , 4 6n. 9 , 47n. I0, 48 civic community, I 9 2 ; institutional performance and, 66- 6 8 , 67; medi­ eval characteristics of, 70; Putnam on Italian, 6o- 7 r . See also group membership in U. S . ; Making De­ mocracy Work Civic Culture, The (Almond and Verba ) , 3 0, 8 6 civil rights groups i n U. S . , 8 2 Clarke, Harold D . , I 0 2 , I 2on. 20, I 27 Clinton, Bill, 72 coefficient estimates: on support for extreme right-wing parties, l 7 8 - 8 4 , r 8 r; for voter turnout analysis, I47-49 Cohen, Jere, 3 8 , 3 8n. 3 Cole, Arthur H., I I I " Common Interpretation, " pastoral Weherianism and, 4 I n. 6 communal loyalties, substantive ratio­ nality and, I 6 communitarianism, 69 , I 9 2 ; Confu­ cianism and, I o 8 - 9 . See also civic community communitarian norms, Swank on cul­ ture and economic growth links as, I07- 8 competitive elections, national: institu­ tions and voter turnout for I 9 ) 02000 and, r 4 7; mixed systems and measures of, I 4 5 ; voter turnout and, I 3 9- 4 I configurations o f attitudes, political culture and, 8

228 Confucianism, I 9 2 ; communitarian­ ism and, l 0 8 - 9 ; economic growth in failed Protestant colonies and, I I I - I 2; Swank on economic growth and, I 09 , I 09n. 9, I 3 0- 3 I Congress Party of India, 2 3 - 2 8 ; elec­ toral engineering by, 28n. 2 6; single-member districts and, 2 s , 27- 2 8 ; votes and legislative seats, I9 s 2- 9 9 , 24, 2 6. See also India conservatism of cultural values, I 3 ; change due to alternative behaviors and, 2- 3 , I I construct validation, 67n. 9 consumer behavior, procedural ratio­ nality and, l 9 consumer groups in U. S . , 8 2 cooperation vs. collusion, political cul­ ture and, I 8 9 corporatism: communitarianism and, I o 8 ; Confucianism, economic growth, and, I I O, I I O Cortes, Juan B . , 6n. 7 Costa, Dora L., 9 I n. 3 9 country dummy variables: institu­ tional approach USe of, I S s - s 6, I s sn. 2 I ; measuring extreme right­ wing party support and, I 8 3 - 84 ; Radcliff o n , I S S ; for Switzerland, I 4 8 , I S 3 - S 4 , I J 3 , I )4n. I 9 ; for U. S . , I 4 8 Coupland, Reginald, 2 7 , 27nn. 24- 2 S Cox, Gary W. : o n electoral laws and voter preferences, 22; on electoral laws in single-member districts, 24; on electoral proportionality and close elections, l 4 ln. 3 ; on legisla­ tive rules or procedures and legisla­ tion, 22; on mobilizing U. S. voters, I so; on multipartyism and electoral proportionality, I 7 In. 6; on voter turnout and political competition structure, I 4 0 Crawford, Susan, 8 9 , 9 0 Cresson, Edith, I 66- 67 Crewe, Ivor, I 4 6n. I 2, I S O, I 6I

INDEX Cronin, Helena, I 8 Crummey, Robert 0 . , 9n. 8 cube law, Congress Party of India and, 2 7 - 2 8n. 2 S cultural differences, political behavior and, I cultural values: alternative behaviors and conservative change due to, 23; causal priority of, 2; Inglehart on economic growth and, 99- I oo; as outcome of conditions, 2; Putnam on tensile strength of, 70, 7 I . See also political culture approach Dahl, Robert A., I 4 , I 4n. I I , I ) n. I 2, l 2 0- 2 l , l 4 2n. 6 Dalton, Russell ]. , I 3 7 Davidson, Chandler, 3 0 Davies, Christie, 3 s Davis, James A., 7 4, 7 s, 7 sn. I 8 , 7 8 Dawkins, Richard, I 8 Day, John, 4 7 - 4 8 , 47n. Io, 49, ) I de Figueiredo, Rui ]. P., I 6 Delacroix, Jacques, 54 democracy, evolution in meaning of, I 20- 2 I , I 2 I n . 2 I Democracy i n America (Tocqueville ) , 60, 69 , I90-9 I democratic performance, I I 6-29, I 3 0, I 9 2 ; cross-country studies of cultural values, economic growth and, I 29- 3 o; Ingle hart on overall life satisfaction levels and, I 26- 2 7, I 2 7n. 26; Ingle hart on successful, I 20- 2 I ; Inglehart's indicators of po­ litical culture and, I I 6- I 7, I I9 w ; Inglehart's political culture indi­ cators and, I I 6- 2o; Muller and Seligson on changes in, I 2 I - 2 S ; postmaterialism, economic develop­ ment and, I 2 I n . 22; product­ moment correlations of Inglehart's indicators of political culture and, I I 7- I 8 , I I 7nn. I ) - I 6, II7. See also economic conditions; economic growth, values and

Index democratization, I 9 2; degree in I 9 70s vs. l 9 8os of, l 22- 2 3 , 1 2 2 , 1 2 3 ; dif­ ferences in estimation of, I 2 3 - 24 , I 2 4 n . 24; influences on process of, I 2 I - 22, I 2 2 ; measuring changes in, I 2 8 - 29 , I 2 8; modified specifica­ tion for degree of, in I 9 70s vs. I 9 80s, I 24 - 2 ) , I 24n. 2 5 , I 2 J; po­ litical culture and, 4- 5 Democrat Party (Turkey ) , I 9 8n. I Denmark, extreme right-wing party in, I 74n. IO, I 7 6 dependency theory, political culture ap­ proach vs., 29 dependent variables: for extreme right­ wing parties analysis, l7 4 - 7 8 , 1 7 8; selecting observations for analysis based on, I I I - I 2, I I In. I 2 ; in World Values Surveys, I ) 9 , I 59n. 24 Derksen, Wilfried, q4n. 8, I 7 6n. I 5 de Roover, Raymond, 4 6- 4 7 , 4 7n. I O, 5 I , 5 2 Diermeier, Daniel, I 6n. I 3 Donaghy, Peter ]., q 6n. I 2 Downs, Anthony, l 8 ; on framing choices and citizens' preferences, 2on. I 8; on irrationality in rational choice model, I 7; on multipartyism and government formation, I 4 I , I 4 2 ; o n uncertainty and rational choice model, 20 Draper, John William, 3 4 Duby, Georges, 44- 4 5 , 4 7n. I O Duch, Raymond M . , I 0 2 , I 2 In. 2 2 dummy variables for countries. See country dummy variables Dunteman, George H., 62 Du Plessis, Johannes, I I On. I O durability o f political culture, 9- I I ; Putnam on, 69, 8 8n. 3 4 ; rules changes and, I 6 I - 62, I 9 6 Durlauf, Steven N . , 7 7 , 84n. 2 8 Dutt, Nitish, I 02, I 2on. 2 0 , I 2 7 Duvall, Louis M., I I on. I O Duverger, Maurice, 2 4 , I 4 2 , I 7 I - 72, I 7 In. 6

229 Easton, David, I 8 8 Eatwell, Roger, l 64 Eckstein, Harry, 2 I , 2 I n. 20 economic conditions: democratic performance and, I 29 ; extreme right­ wing parties and, I 7 3 ; political cul­ tuu an� I I 7- I 9 , I I 8 - I 9n� I 7I 9 , n 9 ; postmaterialism and, I 2 In. 22 economic growth: Barra's endogenous model of, I o i ; in British Common­ wealth, I I O- I I ; in Confucian coun­ tries with failed Protestant colonies, I I I- I 2; continuous compounding of, 44, 4 5; cross-country studies of political culture, democratic perfor­ mance, and, I 29 - 3 0; decisive pat­ tern from Weber's thesis of, 4 2- 4 3 ; i n England during Middle Ages, 4 5 - 4 6; in Europe during Middle Ages, 44-4 5 ; Knack and Keefer's endogenous model of, I I 4 - I 5 , I I 6; overall life satisfaction and short-term, I 27n. 27; Pye on circum­ stances of the day and, l 09n. 9 economic gro'W1:h, values and, 99I I 6, I 3 o, I 9 2; Granato, Inglehart, and Leblang's model of, I O I - 7; Inglehart's correlation of, 99- I oo; Knack and Keefer's model of, I I 2I 6; Swank's model of, I 07 - I 2; thrift and, I 04 , I 04n. 6. See also democratic performance Edgeworth, F. Y. , 2 7 - 2 8n. 2 5 Egyptian Sunnis, Protestant work ethic among, 3 5 Ekelund, Robert B., 5 3 elections: closeness of, I 3 9n. I , I 4 In. 3, I 9 8n. I ; frequency in U. S. and Switzerland, I 5o; lower legislative house, I 4 3 - 4 6 . See also propor­ tional representation (PR) systems; single-member districts; single­ member plurality ( SMP) systems Elections around the World Web site (Derksen) , I 4 4n. 8, I 7 6n. I 5

230 electoral disproportionality: effective threshold and, l 7 6- 7 7 , l 7 6-77nn. I 6- I 7; extreme right-wing parties and, I 9 6; formula for, I 4 5 - 4 6; in­ stitutions and voter turnout for I9 50- 2000 and, 147, I 4 8 ; minor parties and, I 7 I ; multipartyism and, I 7 2 . See also electoral proportionality electoral district structure, 1 4 8 - 4 9 . See also competitive elections, na­ tional; single-member districts electoral engineering: by Congress Party of India, 2 8n. 26; extreme right-wing parties and, I 8 5 - 8 7, l97 electoral laws: Congress Party of India and, 24; voter preferences and, 22; voter turnout and, I 6 I ; votes-seats relationship in single-member dis­ tricts and, 2 5 , 2 5 - 26n. 2 3 . See also institutional approach electoral participation: extreme right­ wing parties and barriers to, I 7072; motivation for, l 6. See also voter turnout electoral proportionality: extreme right-wing parties and, I7o; Liberal Party in United Kingdom and, I 8 6n. 2 8 ; multipartyism, voter turnout, and, I 4 4 - 4 6, 144; policy interven­ tion and, I 8 5, I 8 6; of political par­ ties vs. individuals, I 8 6n. 26; voter turnout and, I 4 r. See also electoral disproportionality Electoral Studies, I4 5n. 8 endogenized values, l } l - 3 2; achieve­ ment motivation and, I 04; ethni­ cally divided societies and, I 9 3- 94; Sicilian Mafia and, I93; trust as, I 3 2- 3 3 , I94 endogenous growth model: Barro's, I o i ; Knack and Keefer's, I I 4 - I 5 , II6 England, economic growth in Middle Ages in, 4 5 - 46 . See also Britain

INDEX environmental groups in U. S . , 8 2 , 9 I 92, 9 6 Epstein, Lee, 2 2 Epstein, Steven A . , 4 8 , 49 ethnic divisions : ethnic conflict and, I 9 3 - 94; exclusionism by extreme right-wing parties and, I 64 . See also immigration Eurobarometer studies, 99 European Community (EC), unem­ ployment in, I 72, I 7 3 n. 7 , I 7 3 n. 8 . See also specific countries European Free Trade Association (EFTA), unemployment in, I 7 2- 7 3 , I 7 3 nn. 7- 8 European Journal of Political Re­ search, I4 5n. 8 Europe during Middle Ages: bubonic plague in, 4 8 - 4 9 ; commercial revolu­ tion in, 4 6- 4 7 , 4 8 ; economic growth in, 44-4 5; merchant venturers in, 4 7, 4 7n. I I; Protestant ethic thesis and capitalism in, 4 , 5, 29; techno­ logical innovation in agriculture, 4 6; wage labor of guild system in, 4 9 · See also Protestant ethic thesis ex post explanations, I 3 0- 3 I , I 84n. 25, I 9 2 extreme right-wing parties, I 63 - 8 7 ; Belgian cultural changes and, I 68 ; characteristics of, I 64 - 67; coeffi­ cient estimates on support for, I 7 8 8 4 , r 8 r; data sources on, I 7 4 - 7 8 , 1 7 8 ; dummy variables in measuring support for, I 8 3 - 84 ; Dutch na­ tional security and, I 67 - 6 8 ; eco­ nomic conditions and, l 7 2- 7 3 ; elec­ toral engineering and, I 8 5- 87; ex­ planations for, I 67-74; immigra­ tion and, I 7 3 - 74 ; institutional con­ ditions favoring, I 9 6- 9 7 ; mass soci­ ety argument for, I 6 8 - 7o, I 74n. I I , I 7 5 nn. I 2- I 3 ; measuring de­ gree of multipartyism and, I 7 7 ; measuring effective threshold for, I 7 6- 77; measuring support for,

Index I 74 - 7 6, I 74n. I o; multipartyism and, l 8 2; political conditions and, I 70 - 7 2 ; sources of support for, I 9 70- 2ooo, r S r ; unemployment measures and, I 7 7 - 7 8 , I 77n. I 8 fair wage, pre-Reformation Church on, 5 2n. I4 faith-based initiative, Bush administra­ tion's, 73 Farmer, John S . , 54n. I 5 fascism, extreme right-wing parties and, I 69 . See also neofascism Fearon, James, I 9 4 Featherstone, Kevin, I 4 5n. 9 Feldman, Jan L., l O- l l Feldstein, Lewis M., 72n. I 2 Ferej ohn, John, I 6n. I 3 , I 8 Ferrarotti, Franco, 3 5n. I Festinger, Leon, I 09 Finke, Roger: on procedural rational­ ity, I ') ; o n religious participation in Sweden, 8 7n. 3 3; on religious par­ ticipation in U. S. since I 776, 848 5 , 8 4 , 84n. 29, 8 5n. 3 0 ; on sub­ stantive rationality, I 6 Finkelstein, Norman G., 2n. I Finland, extreme right-wing parties in, I 74n. I O Fiorina, Morris, I 6n. I 3 Firebaugh, Glenn, 8 5 n . 3 I Flemish Block ( Belgium) , q 6 Flury, Bernhard, 62 Fortuyn, Pim, I 6 8 . See also Pim Fortuyn List framing choices, 2on. I 8 France: extreme right-wing party in, I 74n. IO, 1 7 6; extremist parties and mainstream political agenda in, I 66- 67; marginalization and ex­ tremism in, I 69 ; underprediction of turnout in, I44; voter turnout in na­ tional elections of, I 3 7 Franklin, Benj amin, 5 5 , I ') O fraternal groups, membership declines in U. S . , 7 9 - 8 0

23 I Free Democrats ( Germany) , I 67 Freedom House, l 2 8 Freedom Party (Austria), 1 7 6 French National Front, I 6 5 , I 65n. I Friedman, Jeffrey, I 6n. I 3 Friedrich, Robert ]. , I 8 2 Froza Italia, I994 general election and, I 8 7n. 29 Fukuyama, Francis: Banfield's durable village values and, I ') 8n. I; Ban­ field's " Montegrano " study and, I 9 7 ; on political culture as social capital, 7 ; on significance of politi­ cal culture to other outcomes, I 2 ; o n trust in economic performance, 4; Weber's Protestant ethic thesis and, 3 3 Funk, Carolyn L., I ') n . I 6 Furnham, Adrian, 3 5 Gabel, Matthew J., I 74n. I I Gallagher, Michael, 1 4 5 -4 6, 1 4 5 n. Io, I 77n. I 6 Gallup polls, 90 Gambetta, Diego, l 04 , l93 Gans, Herbert ]., 5 9 Gautam, Ram Sakha, 2 7 n . 2 5 Geddes, Barbara, 2 3 , 5 5 , I I I , I 2 I , I 8o Geertz, Clifford, I 9 3 - 94 General Social Survey ( GSS ) , 7 5 ; data availability, 7 5 n . I 8 ; on group mem­ bership, 7 8 , 7 8n. 2 I , 8 2- 8 3 , 8 2n. 2 6; Putnam's use of, 74 - 7 7 Gentile, Pierre, I 6 7 Germany: cultural traits a n d Nazism in, l, l 64 ; extreme right-wing par­ ties in, I7 4n. I O , 1 7 6; marginali­ zation and extremism in, I 69; pro­ portional electoral institutions of, I 64 ; Republikaner support in, I 70 Gies, Frances, 4 6, 4 6n. 9 Gies, Joseph, 4 6, 4 6n. 9 Gijsberts, Merove, I 7o, I 7 3 , I 8 4n. 2 5 Gilchrist, J. , 5 I , 5 2 Gimpel, Jean, 4 6n. 9

23 2 Glass, David P. , I 3 S Goheen, John, l 09n. 9 Gold berg, Ellis, 3 5 , 6 I Golder, Matt, I 7 5 , I 7 5 11. I 3 , I S 21m. 2 3 - 24 Goldhagen, Daniel ]., I , 2, 2n. I , 3 , I 64 Gorski, Philip S . , 3 5 Gosnell, Harold F. , I 3 S , I 4 I , I 4 3 , I S O, I 9 5 Granato, Jim, Ronald Ingle hart, and D avid Leblang: on achievement mo­ tivation, I04, I07, I 3 o; on cluster of attitudes across and within coun­ tries, I 0 5 - 7, r o 6; Confucian coun­ tries and need for achievement com­ ponents of, I I In. I I; dependent variables used by, I I In. I 2; on eco­ nomic growth and thrift, I 04n. 6; McClelland's I9 50 achievement need data and, I 02- 3 , r o 3 ; on role of postmaterialist values and achievement, I o i ; Swank and, I o9 ; Weber's Protestant ethic thesis and, 3 3 ; World Values Surveys used by, I O in. 3 Greece: extreme right-wing parties in, I 7411. Io; OECD " standardized " un­ employment rates and, I 7911. 20; unicameralism in, I 4 6n. I 2; voter turnout in, I 62, I 62n. 2 5 Greeley, Andrew M . , S s , S s n. 3 I Green, Donald P. , I 6, I 6n. I 3 , I S , 29 Greif, Avner, 2211. 2 I Grofman, Bernard, 3 0 group membership i n U. S . , 7 7 - 9 5 , l 9 0- 9 l ; average levels by year, I 9 74 - 9 4 , 8o; bowling leagues, 72, 9 I , 9 I n. 40, 9 6; coding for, 7Sn. 22, 7911. 24; control for education and, S I - S2, 8 I n. 2 5 , 8 r ; declines in, 7 S - 8 3 , 7 Sn . 2 3 ; fraternal organi­ zations, 79- So; growth and decline in, 94 - 9 5 , 9411. 4 6; GSS question on, 7 8 , 7Sn. 2 I , 8 2- 8 3 , S 2n. 26; literary organizations, 8o; PTA, S 3 ,

INDEX 8 S - 9 I ; Putnam o n civic association and, 7 2 - 7 4 ; religious participation trends, 84- 8 5 , 84, 8411. 29, 8 6; se­ lection bias and, 9 I - 9 5 , 9 6, I 9 I ; service organizations, So; trends by group type, I 9 74 - 9 4 , 79; youth par­ ticipation in, 9 2 - 9 4 , 9 6; Youth Soc­ cer, 9 3 , 94, 94nn. 4 4 - 4 5 growth. See economic growth guild system, medieval, 49, 5 2, 69- 7 0 Hagen, Everett E., 9S-99 Haider, Jorg, I 6 5n. 2 Hainsworth, Paul, I 64 , I 7 6n. I 4 Hale, William, I 9 8n. I Hamilton, Richard F., l 7 0 Hamilton, V. Lee, 3 11 . 2 Hardin, Russell, I 6 Harley, Brian, 8 5 n . 3 I Harrison, Lawrence E . : Banfield's " Montegrano" study and, I 9 7 - 9 S , 200; o n durability o f political cul­ ture, 9 , I 9 8n. I; on political culture and social stability, 7, 29; on politi­ cal culture significance to other out­ comes, I 2; on trust and economic performance, 4; Weber's Protestant ethic thesis and, 3 3 Hatcher, John, 49 Hechter, Michael, I 6 Hedlund, Stefan, 911. 8 Helpfulness: correlations among Trust, Advantage, and, 7 6 - 7 7 , 7 7; GSS measures of, 7 5 - 76. See also Trust Herlihy, David, 49 Herman, V. M., l 6 s - 66 Hermens, Ferdinand A., I 64 Hess, Henner, I 9 3 Heston, Alan, I oon. 2, I T ) , I I 8n. I 7 heteroskedasticity, i n extreme rightwing parties support model, I 79 · See also country dummy variables Heywood, John, 5 411. I 5 Hibbs, Douglas A., Jr. , I S 6 Highton, Benj amin, I 3 S

Index Hindess, Barry, I 6 Hindu asceticism, economic growth and, I 09n. 9 Hirschman, Albert 0 . , I , 3 3 - 3 4 , I 04 historical materialism, Weber on, 3 6 Holton, Robert ]., 3 8 , 3 8n. 2 Hout, Michael, 8 sn. 3 I Huber, John D . , I 74n. I I Huber, Peter ]., I02- 3 n. 5 Hudson, Michael C . , I I 7n. I S humor, Protestant ethic and patterns m, 3 5 Huntington, Samuel P., 6 - 7 Huppert, George, 49 Husbands, Christopher, I 64 Iannaccone, Laurence R., I 6, I 9 , 8 5 , 8 s n. 3 o, 87n. 3 3 ideology, right-wing parties and, H ) 3 Ignazi, Piero, I 64 , I 68n. 5 , I 7 5 - 7 6 Immerfall, Stefan, I 7 6n. I 4 immigrants, political culture durability and, I O- I I , I 3 immigration: extreme right-wing par­ ties and, r 8 2- 8 3 , 1 83 ; extreme right-wing parties on, I 7 3 - 74 ; mea­ sures of, I 7 8 incentive structures, behavioral change and, 3, 3 0 India: Hindu asceticism theory o f eco­ nomic growth in, I 09n. 9; propor­ tional representation in, 2 6 - 2 7 . See also Congress Party of India individual attitudes, political culture and, 8 individual behavior, shared group val­ ues and accountability for, 3, 3 n. 2 influential cases, data analysis and, I 3 I Inglehart, Ronald: on adaptation to political condition changes, I I ; cor­ relation of economic growth and cul­ tural values by, 99- I oo, I oon. 2; cultural effects of democratic perfor­ mance analysis by, I 29 ; on cultural values, economic performance, and democratic stability, 4, 8; on demo-

233 eratic performance and overall life satisfaction levels, r 2 6- 27, r 27n. 2 6, I 3 0, I 3 I - 3 2, I 3 2n. 29; On du­ rability of political culture, I o; on economic growth and thrift, I 3 o; on overall life satisfaction levels in authoritarian regimes, I 2 6, I 272 8 ; on political culture and demo­ cratic performance indicators, I I 6I7, I I 9 - 20; on political culture and historical materialism, 29; po­ litical culture approach of, I 8 8 ; on postindustrial!postmaterial political organizations forms, I 8 5; product­ moment correlations of political cul­ ture and democratic performance, I I 7- I 8 , I I 7nn. I S - I 6, I I 7; right­ wing party growth and scholarly claims of, I 6 3; on significance of po­ litical culture to other outcomes, I 2; on social choice and optimiza­ tion of economic utility, I I n. 9 ; studies o n mass values i n several countries by, 99; on successful democratic performance, r 2o- 2 l ; variables used by, I 5 9 , I 5 9n. 2 4 ; on variation in cultural values over time, I 0 2 . See also Granato, Jim Inkeles, Alex, 5, 6, 8 - 9 , 9 8 institutional approach, I 9 4 - 2oo; Banfield's "Montegrano" study and, I 9 7- 2oo; Congress Party of ln­ dia as illustration of, 2 3 - 28 ; devel­ opment of, I4- I 5; extreme right, electoral disproportionality, and, I 9 6- 9 7 ; incentives structures and constraints, 2 l ; optimizing-actors premise of, I 5- 2 I ; political culture approach vs., 29- 3 o; political par­ ticipation analysis using, I 9 4 - 9 6; procedural rationality and, I 6- I 8 ; robustness of, I 4 8 -49; rules changes and durability of, I 9 6; sub­ stantive rationality and, I S - I 6. See also electoral laws; extreme right­ wing parties; voter turnout

234 institutional performance: Putnam's analysis of, 6 l - 62; Putnam's com­ posites of political culture and, 6 66 8 , 67; Putnam's measurement of, 63-65 institutions: ethnic division toleration and, I 9 4 ; incentives for voting and, I37-38, I6I International Institute for Democracy and Electoral Assistance Web site, 144, I 44n. 8 Ireland: economic growth in, I 09n. 9; extreme right-wing parties in, I 74n. IO Italian Social Movement, I 6 5n. 2, l 7 5n. l 2, 1 7 6 Italy: church attendance by region of, 8 7, 8 8 , 8 8n. 3 5; durability of civic engagement values in, 4; extreme right-wing parties in, I 6 5 , I 6 5n. 2, I 74n. Io, 1 7 6; Mussolini regime in, 70- 7 I ; National Alliance I994 suc­ cess in, I 8 7n. 29; nineteenth­ century landownership in, 7 I ; outmigration from southern, l 9 9 , I 99n. 3 ; political changes after population shifts in, 2oo; rise of capitalism in, 4 6; voter turnout in national elections of, 66, 6 6n. 8 , I 3 7, I 54n. 2 0 . See also Making l)e�ocracy \\fork Ivaldi, Gilles, I 64 Jackman, Richard, I 7 3 n. 7, I 8 6 Jackman, Robert W. , 6 5 n. 7 ; on com­ pulsory voting laws, I 4 3 ; on eco­ nomic conditions and democratic performance, I 29 ; on effects of abol­ ishing mandatory voting, I 62; insti­ tutional model development by, 1 49; on institutions a n d electoral participation, I 3 8 ; on scientific methodology, I 2 I , I 24n. 2 5 ; on uni­ cameralism, I 4 2 ; voter turnout analysis by, I 3 9 , I 4 3 n. 7; on voter turnout and political competition structure, T 3 9

INDEX Jackman, Simon, 2 5 Jahoda, Gustav, 2n. l Janata Party of India, 28n. 2 6 Jones, Bryan D . , 2 0 Jospin, Lionel, I 6 5 Joyce, John M., 9 n . 8 j udicial institutions, public policy and, 22 j ust price: pre-Reformation Church on, 5 3 ; social hierarchy of guild sys­ tem and, 5 2 Kahn, Herman, I 09 Kahn, Matthew E., 9 In. 3 9 Kahneman, Daniel, 2on. I 8 Kalberg, Stephen, 3 6 Katz, Jonathan N., I 4 7n. I 4 Keefer, Philip. See Knack, Stephen Keenan, Edward L., 9 Kelley, Stanley, I 6n. I 3 Kelman, Herbert C . , 3n. 2 Kendall, M. G., 27- 28 n . 2 5 Kendall, Willmoore, I 8 9 Kennedy, Peter, I 5 6 Kicking in Groups ( Lemann) , l 9 l Kiewiet, D. Roderick, I 7 3 Kim, Jae-On, 64, I 3 7 Kinder, Donald R., I 7 3 King, Gary, 3 , I 70 Kitschelt, Herbert, I 63 , I 66, I 7 5n. I 2, I8 5 Kmenta, Jan, I 8 o, I 8 on. 2 I Knack, Stephen, and Philip Keefer: Barra's endogenous growth model used by, I 14 - I 5 , n 6; on cultural approach to economic growth, I I 2, 1 1 3 , l 3 0; measurement problems of, I I 3 - I 4 , I I 4n. I 3 , I I 5n. I 4 , I3I Knight, Jack, 2 I - 2 3 Kobach, Kris W., I S O, I 5 2n. I 7 Kornberg, Allan, I02, I 2on. 20 Kornhauser, William, I 69 Korpi, Walter, I 8 6 Kostoris, Fiorella P. S . , I99n. 3 Kriesi, Hanspeter, I 67 Kuper, Adam, 4n. 3

Index Laakso, �arkku, I 4 6, I 4 6n. I I Ladd, Everett C., 8 3 , 84n. 2 8 , 90, 9 5 Laitin, David D . , 5 , I94 Lane, Robert E., 69n. I I LaPalombara, Joseph, I 3 Lasswell, Harold D . , I 6, I 8 8 Laver, �ichael J., 2 3 , I 66, I 74n. I I Layard, Richard, I 7 3 n. 7 , I 8 6 Lazarsfeld, Paul, I 6 least-absolute value regression, I 8 I n . 22 Leblang, David. See Granato, Jim left-wing parties, postmaterialism and, I 63 legislation, legislative rules or proce­ dures and, 22 legislative house elections, lower, I 4 3 Lemann, Nicholas, 72, 72n. I 3 , 7 3 , 94, I9I L e Pen, Jean-�arie, I 6 5 , I 6 5n. I Lerner, Daniel, I 9 8n. I Levine, Ross, I O I Levinson, Daniel J . , 5 , 6 , 8 - 9 Levitt, Peggy, 8 9 , 90 Lewis-Beck, �ichael S . , 1 29 , 1 70, I 73n. 9 Lichbach, �ark Irving, 2 I n. I 9 life satisfaction, overall: democratic performance and, I I 6; Inglehart on authoritarian regimes and, I 26, I 27 - 2 8 ; Inglehart on democratic performance and, I 2 6-27, I 27n. 2 6; short-term economic growth and, I 27n. 27; voter turnout and, I 5 9 - 60, r 6 o Lijphart, Arend: o n effective number of parliamentary parties, 1 8 6; on ef­ fective thresholds and dispropor­ tionality, I 76-77; on electoral dis­ proportionality formulas, I 4 5 - 4 6 , q sn. Io; o n electoral dispropor­ tionality modification, I 9 7 ; on elec­ toral formulas in proportional repre­ sentation systems, I 4 on. 2; on elec­ toral institutions and voter turnout, I 4 2 ; on electoral laws and voter preferences, 22; on electoral laws in

23 5 single-member districts, 24; on multiparty ism and electoral propor­ tionality, I 7 I , I 7 In. 6; on unicam­ eralism measures, I 4 6 Lipset, Seymour �artin, I 4n. I I , I 69 literary organizations in U.S., S o Little, Lester K . , p , 5 2 Little League Baseball (U. S . ) , 9 2 - 9 4 , 93

Little League Softball (U. S . ) , 9 3 Livi-Bacci, �assimo, 49 Lohmann, Susanne, I 6n. I 3 Lopez, Robert S . , 4 5 , 47n. I o , 49 Lopreato, Joseph, I 9 9 , I99n. 2, 200, 200n. 4 Lowi, Theodore J. , I 6 Lubbers, �arcel, I 70, I 7 3 , I 84n. 2 5 Lucardie, Paul, I 67n. 4 Lupia, Arthur W. , 2on. I 8 Luther, Kurt Richard, I 6 sn. 2 Luther, �artin, 3 8 �acaulay, Thomas Babington, 3 4 �acintyre, Alasdair C . , 69n. 1 1 �ackie, Thomas T., I 44n. 8 , I 4 5 , I 4 5n. 9, I 4 6, I 7 6n. I 5 �acKinnon, �alcolm H . , 3 6, 3 8 - 3 9 �addison, Angus, I 63 mailing list organizations, social capi­ tal and, 9 2 �air, Peter, I 64 , I 74 - 7 5 , I 74n. I I Making Democracy Work (Putnam), 9 5 - 9 6, I 9 0- 9 I ; civic culture mea­ surement in, 6 5 - 66; culture and per­ formance linkages in, 66- 6 8 ; data for, 6o- 6 1 ; historical origins of Ital­ ian regional differences, 6 8 - 7 I , 9 6; institutional performance measure­ ments for, 63 - 6 5 ; principal compo­ nents analysis in, 62- 63 ; purpose for, 6o; regression of institutional performance in, 6 I - 62; statistical problems in, I 9 I. See also Italy Making of Citizens, The (Merriam) , 5 �altzman, Forrest, 22 M anchester, William, 5 3

INDEX Mansbridge, Jane J., I 8 March, James G., l 5 , l 5 n. l 2 marginalization, extreme right-wing parties and, I 68 - 69 Marselli, Gilberta A., 2oon. 4 Marshall, William ]. , 22 martyrdom, procedural rationality and, I 9 Marx, Karl, 4 2 , 4 3 -44 Marxism, Weber's debate with, 3 6 Massicotte, Louis, 2 6 materialism, historical, political cul­ ture approach vs., 29. See also postma terialism Mayer, Nonna, I 67n. 4 Mayhew, Nicholas, 4 6 McClelland, David C . : on achievement motivation as progress pre­ cursor, I, 9 8 ; on cultural values and political participation, I9 5 ; Inglehart and, 9 9 ; national achieve­ ment levels data for I9 50 by, I02, I 02n. 4 ; pastoral Weberianism of, 4 I; on political values and need for achievement, 6, l 5 7n. 2 3; on Protes­ tantism as mental virus and modern­ ization, 4 I n. 7; Weber's entrepre­ neurial values vs. achievement moti­ vation of, 4on. s; Weber's Protes­ tant ethic thesis and, 3 3 . See also achievement motivation McCubbins, Mathew D . , 2on. I 8 McCulloch, J. R., 4 McCurry, Mike, 7 2n. I 4 McDonald, Michael P. , 7 4 McNeill, William H., 49 McPhee, William, l 6 Merkl, Peter, I 7 6n. I4 Merriam, Charles E., 5 , I 4 Messmore, J . H., I I On. I O Miller, Edward, 49 Miller, Ross A., 6 5n . 7 , I 3 9 , I 4 9 Minkenberg, Michael, I 67n. 4 Mishler, William, 2 3 n. 22, I 3 2 Mitchell, Glenn E . , II, I 70, I 7 3 n. 9 Mitchell, J. Murray, n on. I O

Mitra, Subrata K . , I 67, I 70, I 7 2 Mobutu Sese Seko, l 8n. l 5 modal personality structure, 8 - 9 Money, Jeannette, I 74 monopolies, pre-Reformation Church and, 5 2- 5 3 Monroe, Kristen Renwick, I 8 " Montegrano" study, Banfield's, I 9 7- 200 Moore, Barrington, I I, I 3 Moreno, Alejandro, I 5 8 Morgan, James N., 6n. 7 Morgan, Philip, 7 I Morris-Jones, W. H., 27n. 2 5 Mothers Against Drunk Driving (MADD ) , 9 l , 9 l n. 4 l Mudde, Cas, I 64 , I 67, I 68 Mueller, Charles W. , 64 Mueller, Dennis C . , I 6n. I 3 Muir, Edward, 70 Muller, Edward N. : on democratiza­ tion process influences, I 2 I - 2 2, I z z ; gradualism of culture values thesis by, I 24 - 2 5 , I 3 0; robustness of change in democracy evaluation by, I 2 3 - 24 , I 24n. 24 multipartyism: effective threshold sup­ port for extreme right-wing parties and, I 8 I, r 8 3 , I 84 ; electoral dispro­ portionality, voter turnout, and, I44- 4 6, I44; extreme right-wing parties and, I 6 5 , I 70, I 7 I - 72, I 8 2, I 9 6; institutions and voter turnout for I9 50- 2000 and, I47• q 8 ; mea­ sures of, I 4 6, I 77 , I 77n. I 7; policy intervention and, I 8 5; voter turnout and, l 4 l - 4 2 Miinch, Paul, 3 4 Muslim League o f India, 2 7 Mussolini, Benito, 70- 7 I , I 7 sn. I 2 Nagler, Jonathan, I 3 8 National Action/Vigilantes (Switzer­ land) , I 7 6 National Alliance Party (Italy ) , I 6 5n. 2, I 7 5n. I 2, I 8 7n. 29

Index National Democratic Party ( Ger­ many), 1 76 National Front (France), I 66- 67, I ?O, I 72, I 7 3 � 9 , I 7 6 National Front (United Kingdom) , I76 National Political Union ( Greece) , I 7 6 nationhood, ethnic conception of, I 64 Nazism, German, I , 3 , I ?O neofascism, I 64 ; Golder on populist parties vs. , I7 5' I7 5n. I 3 See also extreme right-wing parties Netherlands : abolition of mandatory voting in, I 62; extreme right wing and national security in, I 67 - 6 8 ; extreme right-wing party in, 1 6 5 , I 65n. 2, r 7 6; marginalization and extremism in, I 69; measuring ex­ treme right-wing party support in, I 74n. I O New Democratic Party ( Sweden ) , I 7 6 Newell, James L . , I 6 5 n . 2 Newton, Michael T., I 4 6n. I 2 Nicholson, Stephen P., I 5 I Nickell, Stephen, 1 7 3 n. 7, 1 8 6 Nie, Norman H., I 3 7 Niedermayer, Oskar, I 67n. 4 , I 69, I 70 Nielsen, Fran�;ois, 5 4 Nipperdy, Thomas, 3 4 Noonan, John T. , p , 5 2 Nordsieck, Wolfram, I 4 4 - 4 5 n . 8 Northern League ( Italy ) , I 76, I 8 7n. 29 Norway, extreme right-wing party in, I 74n. I O , I 7 6 Nymoen, Ragnar, 1 8 6 0

OECD (Organisation for Economic Cooperation and Development) , I 7 8n. I 9 , I 79n. 2 0 Offenbacher, Martin, 5 4 - 5 5 , I 9 0 officeholding, policy concerns v s . , I 8 , I 8n. I 5 O'Hearn, Denis, I 09n. 9 Olsen, Johan P., I 5 ' I 5n. I 2

237 omnibus referenda, i n Switzerland, 1 5 0, 1 5 2- 5 3 , 1 5 2 , 1 5 3 opportunity choices with constraints, 2I optimizing-actor premise, framing choices, bounded rationality, and, 2on. I 8 . See also rational choice theory optimizing alternatives, 2, I 4 , I 5 - 2 I Ordeshook, Peter C . , 20 overall life satisfaction. See life satisfac­ tion, overall Park, Richard L., 27 Parties and Elections in Europe Web site (Nordsieck) , 144-4 5n. 8 Party of the Progressives ( Greece) , q6 pastoral Weberianism, 40-42, 4 2n. 8; " Common Interpretation" and, 4 I 11. 6 Paxton, Pamela, 7 5 peasant animism, Protestant ethic and, 42 Penn World Tables, 1 1 3 Perrine au, Pascal, I 67n. 4 Persson, Karl Gunnar, 4 6, 4 6n. 9 Pettinico, George, 9 2 Phillips, Derek L . , 69 Philpott, Daniel, 3 5 Pim Fortuyn List (Netherlands ) , I 6 5n. 2, I 68 Pirenne, Henri, 4 5 Pissarides, Christopher, I 8 6 Pizzorno, Alessandro, I 9 8 , 2oon. 4 Poggi, Gianfranco, 3 6 polarization, extreme right-wing par­ ties and, I 6 5 political agenda, extreme right-wing parties and mainstream, H )? political behavior: change and alterna­ tive, 2- 3 , I I ; cultural differences and, I; motivation for, I 5 - I 6, I 8 8 - 89 . See also behavioral ap­ proach; consumer behavior; voting behavior

INDEX political culture: cross-country studies of democratic performance, eco­ nomic growth, and, I 29 - 3 0; eco­ nomic conditions and, I I 7- I 9 , I I 8 - I 9 nn. I 7 - I 9 , I I 9; Inglehart on renaissance of, 9 9 ; Inglehart's democratic performance indica­ tors on, I I 6- I 7, I I9 - 2o; post­ materialism, economic development and, I 2 In. 22; product-moment correlations of Ingle hart's demo­ cratic performance indicators on, I I 7 - I 8 , I I 7n. I S , I I 7n. I 6, I I 7; Radcliff on country-specific terms for, I 5 5; reflexive expression of, 1 89 political culture approach, 4- s; adap­ tation to, I I ; aggregate properties of societies and, 8 - 9 ; configuration of attitudes and, 8; country studies using, 97, I 9 2 ; distinctive features of, 8- I 3 ; durability of attitudes and, 9- I I ; ex post explanations for, I 3 0- 3 I , I 9 2 ; to extreme right­ wing parties, 1 69; idealism of, 1 8 8 8 9 ; influential cases and data analy­ sis for, I 3 I ; institutional approach vs., 29 - 3 0; shortcomings of, I 9 2 9 3 ; significance to other outcomes of, I 2- I 3 ; studies on, 5 - 8 ; as value clusters within countries, I 0 7 . See also group membership in U. S . ; Making Democracy Work; Protes­ tant ethic thesis political discussion levels: democratic performance and, I I 6; voter turn­ out and, 1 59 - 60, 1 60 political entrepreneurs, extreme right­ wing parties and, I 84 , I 84n. 2 5 Political Order i n Changing Societies (Huntington), 6- 7 political parties, numbers of. See multipartyism political stability, extreme right-wing parties and, I 6 5 - 66 Pollack, Detlef, 2 3 n. 22

Popkin, Samuel L., 2 I , 2 I n . I 9 , 74, 199n. 2 Popular Alliance (now Popular Party) ( Spain ) , q 6 populist parties, Golder o n neofascism VS . , I 7 5 , I 7 5n. I 3 Portes, Alej andro, I 2 , I 3 , 29 Portugal: extreme right-wing party in, I 74n. Io, I 7 6; OECD " standard­ ized " unemployment rates and, I 79n. 20; unicameralism in, I 4 6n. I 2; voter turnout in, I 62, I 62n. 2 5 Posner, Richard A . , 5 9 , 69n. I I , 84n. 28 postindustrialism, right-wing parties and, 1 63 postmaterialism: democratic perfor­ mance and, I I 6- I 7; economic secu­ rity, economic development, and, I 2 In. 22; right-wing parties and, I 6 3 ; role of values and achievement in, I o i ; shift to communitarianism after, I o8n. 8. See also materialism, historical Powell, G. Bingham, Jr. , 66; on an­ tisystem parties and political stabil­ ity, I 66; on district level electoral systems, I4o; on electoral propor­ tionality and close elections, I 4 In. 3; on institutions and electoral par­ ticipation, I 3 8 ; on nationally com­ petitive districts, I4 5; on voter turn­ out and close elections, I 3 9n. I; on voter turnout and political competi­ tion structure, I 3 9; voter turnout measures by, I 4 3 n. 7, I 6 I predestination, Calvinistic, Weber on dogma vs. teachings about, 3 8 -40 President's Council on Service and Civic Participation, 7 3 n. H ) press, on Protestant work ethic, 3 sn. I prevalence of attitudes, political cul­ ture and, 8 Price, Bertram, I 4 7n. I 4 pricing system, pre-Reformation Church on market-based, 5 2

Index principal components analysis : four­ factor estimates, 6 sn . 6, l 0 5 - 7 , r o 6; institutional performance indi­ cators of Italian regional govern­ ments, 64- 6 5 , 64n. 5, 64; Putnam's use of, 62-64, 6 3 nn. 3 -4 ; subse­ quent analysis from, 6 5n . 7 ; unidimensional outcomes from, 63 n . 2, 6 5 procedural rationality: analytical scope using, I 9 ; cultural impact of, 2o- 2 I ; institutional approach and, I 6- I 8; substantive rationality vs., I 8n. I 4 , 2o- 2 I productivity, siestas and, 3 sn. I Progress Party (Denmark) , 1 7 6 Progress Parry (Norway) , r 7 6 proportional representation (PR) systems: large parties and, I 4 In. 5 ; single-member plurality vs ., I 3 9 40. See also electoral dispropor­ tionality; electoral proportionality Protestant Ethic and the Spirit of Capi­ talism, The (Weber) , 3 3 Protestant ethic thesis: ambiguity in, 3 5 , 3 6- 3 7; as basis of modern theo­ ries, 3 3 - 3 4; Confucianism vs ., I o 8 ; doctrinal and teaching distinctive­ ness of, 50- 5 3 , 5 4 ; economic break­ through after Reformation and, 425 o, 5 3 - 54; empirical foundation of, I 89 - 9 0; European capitalism and, 4 , 5, 29 ; general purpose of, 3 8 -4 2 ; growth pattern implied by, 4 2 - 4 3 ; methodological flaws in, I 9 o; pasto­ ral transcriptions of, 40-42; as sec­ tarian thesis, 3 5, 3 7 - 3 8; before Weber, 3 4 Protestantism: denominational mem­ bership shifts in, 8 6; economic growth in Confucian countries with failed colonies of, I I I - I 2; mainline denominations of, 8 6n. 3 2 Protestants, democratic performance and proportion of, I I 6, I I ?, I I 7n. IS

239 Protestant work ethic, 3 5 Przeworski, Adam, l 5 6 Psychopathology and Politics (Lasswell) , I 6 PTA (Parents and Teachers Associa­ tion) membership in U. S . , 8 3 , 8 8 9 I , 9 I n. 3 9 , 9 6, I 9 I ; data on, 8 9 9 0n. 3 7 ; social capital and, 8 9 ; women's labor force participation and, 89-90, 9on. 3 8 Putnam, Robert D . : acclaim for, s ; Banfield's durable village values and, I 9 8n. I; Banfield's "Monte­ grana" study and, I 9 7 ; on civic community, 7; on civic virtue in U. S . , 72-9 s; on durability of Italian civic engagement values, 4, IO, I 3 ; o n economic determinism, I 2 ; on political culture and electoral partici­ pation, I 3 8; on political culture and historical materialism, 29; on social capital in U. S . , 72- 74; on social norms and trust in U. S . , 7 4 - 7 7 ; on social participation in U. S . , 3 0; on USA Freedom Corps, 7 3 n. l 7 ; on U. S . group membership, 77-9 s; on voter turnout in U. S . , 7 3 , I 3 7; Weber's Protestant ethic thesis and, 3 3· See also Bowling Alone; group membership in U. S . ; Making De­ mocracy Work; principal compo­ nents analysis Pye, Lucian W. : on Confucianism and Western techno-industrial culture, I o 8 ; on failed Protestant colonies, I I I ; on motivation for political be­ havior, l 6; on political culture stud­ ies, s, s - 6n. 6; on political culture vs. rational choice, 29 racial politics, voting behavior in American South and, 3 0 Radcliff, Benj amin, I S ) , I 5 6 Rae, Douglas W. : o n electoral decisive­ ness, I 4 2n. 6; on electoral institu­ tions and voter turnout, I 4 2 ; on

INDEX Rae, Douglas W. ( continued) electoral la ws and voter preferences, 22; on electoral laws in single­ member districts, 24; on electoral proportionality, I 4 I ; on multi­ partyism and electoral proportional­ ity, I 7 In. 6 Rahn, Wendy, 7 S Rallying the Armies of Compassion, 7 3 n. I S Ranney, Austin, 66n. 8 Rapkin, Jonathan, I02, I 2on. 20 rational choice theory, I 8 ; irrationality in, I 7 ; political culture vs., 293 0; uncertainty in, 20. See also optimizing-actor premise reciprocal altruism, I 9 - 2o, I 9 - 20n. I7 reclaiming/restoring community, 69 Reformation. See Protestant ethic thesis regression procedures: institutional model, I47-4 8n. I 4 ; median estima­ tor (least-absolute value), I 8 In. 22; median or absolute value, l 2 3 n. 2 3 ; robust, I 0 2 - 3 n . s ; Tobit procedure, I 8o- 8 I , I 8 on. 2 I Religion o f China, The (Weber ) , I o 8 religious participation, 8 3 - 8 8 ; de­ nominational shifts in U. S . , 8 6; so­ cial capital and, 8 6- 8 8 , 8 7 - 8 8 nn. 3 4- 3 s , 9 6; trends in U. S . , 84- 8 s , 84, 84n. 29, I 9 I Renelt, David, I O I Republican People's Party (Turkey) , I 9 8n. I Republikaner party ( Germany), l70, I 7 2, r 7 6

Rice, Tom W. , I O- I I Riedwyl, Hans, I 8 6n. 2 6 Riesman, David, I 68 right-wing parties, I 64 Riker, William H . : on constraints in choices, 2 I ; on electoral laws in single-member districts, 24; on multipartyism and electoral propor-

tionality, I 7 I n. 6 ; on procedural ra­ tionality, 20; on rational choice model, I 7; on substantive vs. proce­ dural rationality, I 8n. I4 Root, Robert K., s 4 n. I S Roper Center, 9 0 Rose, Richard, I 3 2, I44n. 8 , I 4 S , I 4 sn. 9 , I 7 6n. I S Rosenstone, Steven J. , I 3 8 Ross, Andrew, I I On. I O RPR ( Rassemblement pour I a republique ) party (France ) , I 66- 67 Russell, Josiah Cox, 49 Russia, organized crime in, I 9 3 Rustow, Dankwart A., I 64 Sabetti, Filippo, 7 I sacred values, substantive rationality and, I 6 Samuelson, Robert J., 72n. I 3 , 78n. 2 3 Samuelsson, Kurt, s I , s 2 , s 6 - s 8 Sandel, Michael J., 69n. I I Sartori, Giovanni, I 8 S , I 8 6n. 2 6; on antisystem parties and political sta­ bility, l 66; on electoral system manipulability, I 8 6, I 8 6n. 27; on multipartyism and electoral propor­ tionality, I 7 I n. 6; on proportional representation schemes and multipartyism, I 7 I satisfaction. See life satisfaction, overall Satz, Debra, I 6n. I 3 , I 8 Savouri, Savvas, I 8 6 Schachter, Gustav, I99 Schain, Martin A., I 66, I 67n. 4 Scheepers, Peer, l 70, l 7 3 , l 84n. 2 s Schneider, Jane, 4 2 , I 9 3 Schneider, Peter, I 9 3 Schofield, Norman, I 6n. I 3 , I 66, I 74n. I I school activities, U. S . parental partici­ pation in, 90 school enrollment: Knack and Keefer emphasis on, I I s , I I sn. I 4 ; regres­ sion of need for achievement, I990,

Index on prior growth and ratios of, I 045, �05 Schuessler, Alexander A . , I9 scientific methodology: Samuelsson's critique of Weber's, 5 6- 5 8 ; Weber's retrospective, 5 5 - 5 6. See also princi­ pal components analysis; regression procedures scientific progress, dogmatic religions and, 4 2n. 8 Sears, David 0 . , I 9n. I 6 selection bias: for group membership in U. S . , 9 I - 9 5 , 9 6, I 9 I ; THRESH calculation and, I 8 o self-interest, public opinion and, I 9n. 16 Seligson, Mitchell A . : on cultural val­ ues measurement inconsistencies, I 07n. 7 ; on democratization pro­ cess influences, I 2 I - 2 2, I 2 2 ; grad­ ualism of cultural values thesis by, I 24 - 2 5 , I 3 o; robustness of change in democracy evaluation by, I 2 3 2 4 , I 24n. 24 service organizations in U. S . , 8 o Seton-Watson, Christopher, I 4 3 Shapiro, Ian, I 6, I 6n. I 3 , I 8 , 2 9 Shepsle, Kenneth A., I 4 n . Io, I ) , I 6n. I 3 , 22, 2 3 Shugart, Matthew S., 22, 24, I 8 6 Sicilian Mafia, I 9 3 Sidjanski, Dusan, I 6 I Sierra Club, 9 2 significance o f political culture to other outcomes, I 2- I 3 Silverman, Sydel F. , 2oon. 4 Simon, Herbert A., 1 8n. 1 4 , 20, 2on. I 8 Singer, Milton, I 09n. 9 single-member districts, 24; develop­ ment in India of, 2 7 ; votes-seats rela­ tionship in, 2 5 , 2 5 - 26n. 2 3 ; voting incentives in, I 3 8, I9 5 single-member plurality ( SMP) sys­ tems, proportional representation systems vs., I 3 9 - 4 0

Skocpol, Theda, 9 I n. 4 I , 9 4 n . 4 6 Smelser, Neil J., 1 6 Smith, Adam, 8 5 n. 3 0 Smith, David Horton, 98 Smith, Tom W., 74, 7 5 , 7 5 n. I 8 , 7 8 , 8 2n. 2 6 social capital: Fukuyama on political culture as, 7; historical record in Italy of, 70- 7 I ; Knack and Keefer on trust and, I I 5 - I 6; mailing list organizations and, 9 2 ; marginali­ zation, extremism and, I 69; PTA membership and, 8 9 , 90; religious participation and, 8 6- 8 8 , 8 7- 8 8nn. 3 4 - 3 5 ; women's labor force partici­ pation and, 9 1 n. 3 9 ; youth partici­ pation in sports and, 94, 9 6 Social Democrats (Germany) , I 67 social institutions: cultural variations in, 22- 2 3 ; cultures vs., 22n. 2 I ; po­ litical institutions vs., 22; rules for, 2 I - 22 social order support, democratic per­ formance and, I I 6 social participation (U. S . ) , 3 0 . See also group membership in U. S. SOPEMI ( Systeme d'observation per­ manente des migrations interna­ tionales ) ( OECD ) , I 7 8 , I 7 8n. I 9 Spain: extreme right-wing party in, I 74n. IO, r 7 6; voter turnout in, I 62, I 62n. 2 5 Spiro, Melford E . , I } On. 2 8 Spitz, Lewis W., 5 3 Spriggs, James F. , I I , 22 Spruyt, Hendrik, 70 Squire, Peverill, 13 8 Stark, Oded, I 9 Stark, Rodney: o n procedural rational­ ity, I 9 ; on religious participation in Sweden, 8 7n. 3 3 ; on religious par­ ticipation in U.S. since I 776, 8 4 8 5 , 8 4 , 8 4 n . 29, 8 5 n. 3 0; on sub­ stantive rationality, I 6 state formation, Protestant ethic and modern, 3 5

INDEX Steiner, J iirg, I 8 6n. 26 Stengel, Richard, 72n. l 3 , 93 Stewart, Robert, I I On. IO Stimson, James A., I 79 Stolzenberg, Ross M., 8 sn. 3 I Strang, David, 2 6 Straussman, Jeffrey D . , I 63 Street, John, 29 Strom, Kaare, 2 3 , I 4 2n. 6 Stuart, A., 27-28n. 2 5 substantive rationality: procedural ra­ tionality vs., I 8n. I4, 20- 2 I ; utility maximization and, I 6 Summers, Robert, I oon. 2 , I I J , I I 8n. I7 Swank, Duane: on Confucianism and economic growth, I09, I 09n. 9, I 3 0 - 3 I; on corporatism, Confucian­ ism, and economic growth, IO')- I I , I I o; on culture and economic growth links as communitarian norms, I 07- 8 , I 3 o; dummy vari­ ables used by, I 5 5 ; extreme values selection of dependent variable by, l l l-l2 Swanson, Guy E., I 5 6- 5 7 , I 5 6n. 22 Swedberg, Richard, I 6 Sweden: abolition o f mandatory vot­ ing in, I 62; extreme right-wing party in, I 74n. Io, I 7 6; religious participation in, 8 7, 8 7n. 3 3 , 87 Switzerland: dummy variable for, q 8 , I 5 3 - 5 4 , I 5 3 , I 5 4n. I9; elections or referenda frequency in, I 5o- 5 I , I 5 2- 5 3 , T ) 2 ; eligible voters in, I 5 3 n . I 8 ; extreme right-wing party in, l7 4n. lO, 1 7 6; government for­ mation in, I 4 9 - so; institutions and voter turnout for I 9 5 0- 2ooo in, I 4 7 - 4 8 , I 4 7; mainstream political agenda and right-wing parties in, I 67 ; referendum items number and turnout in, I 5 3 - 5 4 , I 5 3 , I 5 3 - 54n. I 9 , I 6 I ; unemployment data for, I 77n. I 8 ; voter turnout in national elections of, I 3 7 , I 62, I ') 6 Swyngedouw, M arc, T 64 , T 68

Taagepera, Rein, 22, 24, I 4 6, I 4 6n. l l , l77 Taggart, Paul, I 7 5n. I 2, I 7 6n. I 4 Tarrow, Sidney, 6 I , 7 I Tawney, Richard H., 4 I - 4 2 , 5 2 Taylor, Charles, 69n. I I Taylor, Charles L . , I I 7n. I S , I 2 I n . 22 Taylor, Michael, I 6s - 66 Taylor, Michael! A., I 0 2 Teixeira, Ruy A., I 3 8 , I49n. I S television, Putnam on U. S. civic participation and, 92 Teune, Henry, I 5 6 THRESH (effective threshold ) : addi­ tive model, I 8o- 8 2 , I 8 I ; calculation of, l 79 - 8o; electoral dispropor­ tionality and, I 7 6- 7 7 , I 7 6- 77nn. I 6- I 7; nonadditive model, I 8 I , I 8 I, I 8 2; support for extreme right­ wing parties and, I 8 T, I 8 3 , I 84 Tingsten, Herbert, I 4 3 Tobin, James, I 8 on. 2 I Tobit procedure, I 8 o- 8 I , I 8 on. 2 I , I 8 In. 22 Tocqueville, Alexis de, sn. 5 , 3 0, 59· See also Democracy in America Tong, James, 2 I nn. I ') - 20 traditional values, extreme right-wing parties and, I 69 Trivers, Robert, I ')n. I 7 Trust, I 9 2 ; correlations among Helpful­ ness, Advantage, and, 7 6- 77, 7 7; democratic performance and inter­ personal, I I 6; economic perfor­ mance and, 4; GSS measures of, 7 5 7 6, 76; Knack and Keefer emphasis on, l l 5 ; Mishler and Rose on cul­ tural patterns and political, I 3 2 - 3 3 ; Putnam o n social norms i n U.S. and, 7 4 - 7 7 ; Sicilian Mafia and, I 9 3 ; voter turnout and, I 5 9 - 6o, I 6 o Tsebelis, George, I 8 Tversky, Amos, 2on. I 8 UDF (Union pour l a democratic fran­ <;ais) party, I 66- 67 unemployment: economic conditions,

Index extreme right-wing party support, and, l 84 - 8 5 ; effective threshold support for extreme right-wing par­ ties and, r 8 I , r 83 , I 8 4 ; extreme right-wing parties and, I 7 2- 7 3 , I 8 2, r 8 3 , I 84 , I 9 6- 9 7 ; interper­ sonal trust as indicator of demo­ cratic performance and, I I 8 - I 9 ; measures of, I 77 - 7 8 , I 7 7n. I 8 ; OECD " standardized" rates of, I 79n. 20; policy intervention and, I 8 5; political culture and, r r 9; postmaterialism and, I 2on. 20 unicameralism: institutions and voter turnout for I 9 5 0- 2ooo and, I 4 7 • l 4 8 -4 9 ; measures of, l 4 6; voter turnout and, I 4 2 United Kingdom: extreme right-wing party in, I 74n. w, r 7 6; Liberal Party and proportional representa­ tion in, I 8 6n. 2 8 ; religious participa­ tion in, 8 7 , 8 7 � 3 3 , 87 United States: civic virtue in, 7 2 - 9 5; dummy variable for, I 4 8 ; elections frequency and voter turnout in, I 5 0- p , I 5 4 , I 6 I ; institutions and voter turnout for I 9 5 0- 2ooo in, I 4 7 - 4 8 , r 4 7; social capital in, 7274; social norms and trust in, 7477; voter registration in, I 4 9 , I 49n. I 5; voter turnout in national elec­ tions of, I 3 7, I 9 6. See also group membership in U. S. USA Freedom Corps, 73, 7 3 nn. I 6- I ? usury: Calvin's belief vs. pre­ Reformation Church on, s o- 5 2, 5 3 ; evolution of term, 5 I utility maximization, substantive ratio­ nality and, I 6 Valentine, Charles A., I 3 value enactment approach to develop­ ment, I 2 values and culture, studies o n the ef­ fects of, 5 5 Van Holsteyn, Joop, T 67, T 6 8

243 Varese, Federico, I 9 3 Verba, Sidney: behavioral approach of, I 4 ; civic cultures and democra­ cies studies by, 4 , 6, 8 6, I 5 7n. 2 3 ; o n civic political culture, 7 ; on dura­ bility of political culture, I I; on po­ litical culture and electoral participa­ tion, I 3 8 , I 9 5 ; on political partici­ pation, I 3 7; on prevalence of par­ ticipatory orientations in societies, 99· See also Almond, Gabriel A. Vlaams Bloc ( Belgium) , I 68 Voennan, Gerrit, I 67n. 4 von Beyme, Klaus, I 68n. 5 voter eligible population: age eligibil­ ity and, l 4 7n. l 3 ; registered voters VS . , I 4 3 voter registration, i n U . S . v s . Switzer­ land, I 49n. I 5 voter rolloff, I p n. I 6 voter turnout: coefficient estimates, I 4 7 - 4 9 ; compulsory voting laws and, I 4 3 ; elections frequency and, I 5 0- 5 4 ; electoral dispropor­ tionality, multipartyism, and, l444 6, I 44; electoral laws as incentives for, I 94 - 9 5; electoral proportional­ ity and, I 4 I ; institutional decisive­ ness and, I4 2n. 6; institutional inde­ pendent variables for I9 50- 2000 and, r 4 7; institutional influences for I 9 5 0- 2ooo on, I 3 9 -49; institu­ tions and, I 3 7- 62; Italian refer­ enda and, I 5 4n. 20; measurement of, 66, 6 6n. 8, I 4 3 - 4 6, I 4 3 n. 7 ; multipartyism and, I 4 I - 42; i n na­ tionally competitive elections, l 3 94I; need for achievement and, I 5 65 8 , I J 7, I 9 5 ; political culture and, I 5 4 - 60; political culture vs. institu­ tional analysis of, I 60 - 6 I , I9 5 - 9 6 ; Swiss omnibus referenda and, I 5 3 5 4 , I 5 3 n . I 8 , I J J ; Swiss referenda and, I 5 2, rsz; theories on, I 3 73 9 ; unicameralism and, I 4 2 ; in U. S . , 7 3 - 74 ; World Values Surveys on post- T 990, T 5 8 - 6o, T 9 6

244 voting age regulations, 14 7n. I 3 voting behavior: electoral laws and, 22; information costs for U. S . vot­ ers and, I ) O; marginalization, ex­ tremism, and, I 69; procedural ratio­ nality and, I 9; racial politics in American South and, 3 0; wasted vote phenomenon, I 7 I voting laws, compulsory: measures of, I 4 6 ; turnout and, I 4 3 , I 4 9 , I 6I , I 9 5 Wahl beck, Paul J . , 2 2 Waite, Linda J., 8 s n . 3 I Walker, Jack L., 8 2 , 8 2n. 26, 8 3 Walzer, Michael, 3 3 Warner, R. Stephen, 8 sn. 3 0 Warwick, Paul, I 66 wasted vote phenomenon, I 7 I Wattenberg, Martin P. , I 5 0 , I 5 In. I 6 Weber, Max: Almond and Verba's study and, 6; on Confucianism vs. Puritanism, Io8; diffusion of entre­ preneurial attitudes, 9; on free labor for capitalism, 49n. I 3 ; Hindu as­ ceticism theory of economic growth and, I 09n. 9; on industrialization preconditions, I; Inglehart and, 99; on j ust price and social hierarchy of guild system, 5 2; on medieval mer­ chant venturers, 47n. I I ; Offen­ bacher's statistical study of Baden used by, 54- 5 5 ; on Protestant ethic and European capitalism, 4, 5, 29; retrospective methodology of, 5 5 5 6 . See also Protestant ethic thesis; Samuelsson, Kurt

INDEX Wedeen, Lisa, 2 I n. I 9 Weinberg, Leonard, l 7 6n. l 4 Weiner, Myron, 27, 2 7 n . 2 5 , 4 I n. 7 Weingast, Barry R., I 6, 22, I 9 4 Westle, Bettina, I 67n. 4 , I 69, I 7 0 White, Andrew Dickson, 4 2n. 8 White, Halbert, I4 7n. I 4 White, Lynn, Jr. , 46, 4 6n. 9 White, Peter, 3 8 White, Stephen, 9n. 8 Whittam, John, 7 I Willems, Emilio, I I I Wilson, Edward 0 . , I 9 - 20n. I 7 Wittman, Donald A., 2on. I 8 Wolfinger, Raymond E., I 3 8 women's labor force participation, PTA membership in U. S. and, 8 9 90, 9on. 3 8 World Values Surveys, 99; on church attendance for six countries, 8 6, 8 7; data periods covered by, I o In. 3 ; as data source, I 5 8 ; Granato, Ingle­ hart, and Leblang's use of, I o i ; Knack and Keefer's use of, I I 2 , u 4 ; variables used by, l ) 9 , l 5 9n. 24; voter turnout comparisons us­ ing, I 5 9 - 6o, r 6 o Wuthnow, Robert, 9 I Youth Soccer, U. S . , �1 3 -94, 94nn. 44-4 5 . 95 Zagare, Frank C . , I 8n. 14 Zaret, David, 3 8 Zeitlin, Irving, 3 6 Zeller, Richard H., 67n. 9