World Politics 48.2 (1996) 143-179
 

The New Politics of the Welfare State

Paul Pierson *

Tables


The much-discussed crisis of the welfare state is now two decades old. The tremendous twentieth-century expansion of social programs has been a remarkable feature of advanced industrial societies. In all these countries the welfare state is a core institution, accounting for between one-fifth and one-third of GNP. Ever since the postwar economic boom ended in the early 1970s, however, social programs have faced mounting political challenges. Questions of expansion have long since given way to an acknowledgment of the limits to welfare state growth and the prospect for extended austerity. Despite this fundamental change, however, we still know stunningly little about the politics of social policy retrenchment. In contrast to our vast knowledge of the dynamics of welfare state expansion--arguably the most well-tilled subfield of comparative public policy--welfare state retrenchment remains largely uncharted terrain. 1 Theoretically informed discussion has been limited to very abstract commentaries or the rather reflexive, often implicit application of propositions derived from the study of social policy expansion.

This puzzling state of affairs results in part from the very success of earlier scholarship. The quality of historical research on the welfare state has encouraged a simple process of borrowing already developed models for the examination of a new environment. I would argue, however, that there are compelling reasons to reject such a straightforward extrapolation, that the new politics of the welfare state is instead quite different from the old. Welfare state expansion involved the enactment of popular policies in a relatively undeveloped interest-group environment. By contrast, welfare state retrenchment generally requires elected [End Page 143] officials to pursue unpopular policies that must withstand the scrutiny of both voters and well-entrenched networks of interest groups. It is therefore not surprising that variables crucial to understanding the former process are of limited use for analyzing the latter one.

This essay seeks to lay the foundations for an analysis of welfare state retrenchment. I emphasize the critical constraints on reform resulting from the role of supportive interest groups and, ultimately, voters. 2 The growth of the welfare state itself transforms the politics of social policy. As a result, the welfare state has proved to be far more resilient than other key components of national political economies and far more durable than existing theories of the welfare state would lead one to expect. The argument is presented in four stages. Section I highlights the characteristic qualities of retrenchment politics. Section II discusses in more detail the principal theories of welfare state expansion and suggests why the distinctiveness of retrenchment makes a straightforward application of these arguments to the contemporary welfare state problematic. Section III explores the dynamics of retrenchment in four cases: Great Britain, the United States, Germany, and Sweden. Section IV builds on this theoretical and empirical analysis to offer some basic propositions about retrenchment politics.

I. Why the Politics of Retrenchment Is Different

This essay's central claim is that because retrenchment is a distinctive process, it is unlikely to follow the same rules of development that operated during the long phase of welfare state expansion. There are two fundamental reasons for this. First, the political goals of policymakers are different; second, there have been dramatic changes in the political context. Each of these points requires elaboration.

There is a profound difference between extending benefits to large numbers of people and taking benefits away. 3 For the past half century, expanding social benefits was generally a process of political credit claiming. Reformers needed only to overcome diffuse concern about tax rates (often sidestepped through resort to social insurance "contributions") and the frequently important pressures of entrenched interests. Not surprisingly, the expansion of social programs had until recently [End Page 144] been a favored political activity, contributing greatly to both state-building projects and the popularity of reform-minded politicians. 4

A combination of economic changes, political shifts to the right, and rising costs associated with maturing welfare states has provoked growing calls for retrenchment. At the heart of efforts to turn these demands into policy have been newly ascendant conservative politicians. Conservative governments have generally advocated major social policy reforms, often receiving significant external support in their effort, especially from the business community. 5 Yet the new policy agenda stands in sharp contrast to the credit-claiming initiatives pursued during the long period of welfare state expansion. The politics of retrenchment is typically treacherous, because it imposes tangible losses on concentrated groups of voters in return for diffuse and uncertain gains. Retrenchment entails a delicate effort either to transform programmatic change into an electorally attractive proposition or, at the least, to minimize the political costs involved. Advocates of retrenchment must persuade wavering supporters that the price of reform is manageable--a task that a substantial public outcry makes almost impossible.

Retrenchment is generally an exercise in blame avoidance rather than credit claiming, primarily because the costs of retrenchment are concentrated (and often immediate), while the benefits are not. That concentrated interests will be in a stronger political position than diffuse ones is a standard proposition in political science. 6 As interests become more concentrated, the prospect that individuals will find it worth their while to engage in collective action improves. Furthermore, concentrated interests are more likely to be linked to organizational networks that keep them informed about how policies affect their interests. These informational networks also facilitate political action.

An additional reason that politicians rarely get credit for program cutbacks concerns the well-documented asymmetry in the way that voters react to losses and gains. Extensive experiments in social psychology [End Page 145] have demonstrated that individuals respond differently to positive and negative risks. Individuals exhibit a negativity bias: they will take more chances--seeking conflict and accepting the possibility of even greater losses--to prevent any worsening of their current position. 7 Studies of electoral behavior, at least in the United States, confirm these findings. Negative attitudes toward candidates are more strongly linked with a range of behaviors (for example, turnout, deserting the voter's normal party choice) than are positive attitudes. 8

While the reasons for this negativity bias are unclear, the constraints that it imposes on elected officials are not. When added to the imbalance between concentrated and diffuse interests, the message for advocates of retrenchment is straightforward. A simple "redistributive" transfer of resources from program beneficiaries to taxpayers, engineered through cuts in social programs, is generally a losing proposition. The concentrated beneficiary groups are more likely to be cognizant of the change, are easier to mobilize, and because they are experiencing losses rather than gains are more likely to incorporate the change in their voting calculations. Retrenchment advocates thus confront a clash between their policy preferences and their electoral ambitions.

If the shift in goals from expansion to cutbacks creates new political dynamics, so does the emergence of a new context: the development of the welfare state itself. Large public social programs are now a central part of the political landscape. As Peter Flora has noted, "Including the recipients of [pensions,] unemployment benefits and social assistance--and the persons employed in education, health and the social services--in many countries today almost 1/2 of the electorate receive transfer or work income from the welfare state." 9 With these massive programs have come dense interest-group networks and strong popular attachments to particular policies, which present considerable obstacles to reform. To take one prominent example, by the late 1980s the American Association of Retired People (AARP) had a membership of twenty-eight million and a staff of thirteen hundred (including a legislative [End Page 146] staff of more than one hundred). 10 The maturation of the welfare state fundamentally transforms the nature of interest-group politics. In short, the emergence of powerful groups surrounding social programs may make the welfare state less dependent on the political parties, social movements, and labor organizations that expanded social programs in the first place. Nor is the context altered simply because welfare states create their own constituencies. The structures of social programs may also have implications for the decision rules governing policy change (for example, whether national officials need the acquiescence of local ones) and for how visible cutbacks will be. "Policy feedback" from earlier rounds of welfare state development is likely to be a prominent feature of retrenchment politics. 11

In short, the shift in goals and context creates a new politics. This new politics, marked by pressures to avoid blame for unpopular policies, dictates new political strategies. 12 Retrenchment advocates will try to play off one group of beneficiaries against another and develop reforms that compensate politically crucial groups for lost benefits. Those favoring cutbacks will attempt to lower the visibility of reforms, either by making the effects of policies more difficult to detect or by making it hard for voters to trace responsibility for these effects back to particular policymakers. 13 Wherever possible, policymakers will seek broad consensus on reform in order to spread the blame. Whether these efforts succeed may depend very much on the structure of policies already in place. As I argue in the next section, theoretical discussions of the welfare state have failed to take this point sufficiently to heart.

II. The Limits of Existing Theories

Each of the main theories of welfare state expansion--arguments about economic development, arguments about the power of the left, and arguments about institutions--has been loosely appropriated for discussions of the contemporary welfare state. In this section I make explicit the kinds of claims that an inversion of existing theories of expansion might suggest for a theory of retrenchment. While each perspective [End Page 147] contains insights, my main contention is that a convincing analysis of retrenchment must move beyond these formulations.

A New "Logic of Industrialism"?

The earliest well-developed theory of welfare state growth, associated particularly with the work of Harold Wilensky, stressed the relationship of welfare state expansion to processes of economic growth. 14 Welfare state effort was correlated with affluence, suggesting that "strong economies produce strong welfare states." Yet while this argument explained broad social policy differences between rich and poor nations, it proved to be far less helpful in explaining differences within the club of rich democracies.

Nonetheless, versions of economic determinism have been prominent in discussions of the contemporary welfare state. The claim that a new "logic of industrialism" encourages a convergence of national social policy models is based on asserted consequences of global economic change. 15 One possibility is that the globalization of capital markets removes crucial economic policy tools from national governments and constrains social policy options. However, as Garrett and Lange have persuasively argued, while monetary and fiscal policies may face new restrictions, there is little evidence that supply-side policies of transfers and services need be constrained, so long as tax bases remain sufficient to insure relatively low deficits. 16 Some social policies may in fact contribute to economic growth, for example, by encouraging the formation of human capital and certain kinds of labor market flexibility.

A related argument is that heightened economic integration will result in what Europeans call "social dumping." The term refers to the possibility that firms operating where "social wages" are low may be able to undercut the prices of competitors, forcing higher-cost firms either to go out of business, or to relocate to low social wage areas, or to pressure their governments to reduce social wage costs. In extreme scenarios these actions could fuel a downward spiral in social provision, eventually producing very rudimentary, lowest-common-denominator national welfare states. 17 The evidence for a process of social dumping [End Page 148] remains limited. The social wage is only one factor in investment decisions, and firms will not invest in countries with low social wages unless other factors (for example, infrastructure, worker productivity) justify such investments. 18 Neoclassical trade theory suggests that countries with high social wages should not face problems so long as overall conditions allow profitable investment. 19

A further, critical objection to the economic determinist argument has received less attention. Even if a high social wage were associated with poor economic performance, the assumption that downward adjustments in social policy would necessarily follow is unjustified. As North has recently reminded us, countries have maintained widely different levels of economic performance for very long periods of time, without marked pressures toward convergence. 20 In a world where particular arrangements have been institutionalized and where local actors have adopted strategies that succeed in local terms, there is no automatic or necessary movement toward any particular definition of efficiency--market dictated or otherwise. Politicians in democratic systems generally worry first and foremost about getting elected. Helping improve the economy may make that easier, but not if it requires hugely unpopular policies, and not if the economic benefits are likely to appear at some point in the distant (that is, postelection) future.

Thus, even if social dumping arguments proved valid--a big if--much would still depend on the balance of political forces favoring and resisting a substantial restructuring of the welfare state. Tendencies toward fragility must be weighed against the considerable sources of welfare [End Page 149] state stability. In short, policy outcomes cannot be derived directly from economic trends. The new logic of industrialism suffers from many of the defects of the old one. Expansive claims of economic determinism pay insufficient attention to the politics of policy change. The same cannot be said about power resource and institutionalist arguments, to which I now turn.

Left Power Resources and the Politics of Retrenchment

The power resources perspective, which attributes cross-national variations in social provision largely to differences in the distribution of political resources among classes, has been the leading approach in comparative politics to explaining patterns of welfare state expansion. 21 Class-based struggle over social provision occurs because social programs affect the bargaining position of workers and employers. Many social programs limit the economic vulnerability of wage earners and increase worker solidarity. According to power resource theorists, strong unions and left parties contribute to the growth of these programs. The power resources approach has had considerable success in accounting for cross-national variations in social provision during the three decades following World War II. Furthermore, the replacement of crude social expenditure data with more fine-grained distinctions among patterns of social provision has greatly improved the model's explanatory power. 22

A straightforward application of power resource arguments to retrenchment would suggest that welfare states are in deep trouble. The power of organized labor and left parties has shrunk considerably in many advanced industrial societies. As I will argue in Section III, however, there is very little evidence that this decline has had a fundamental impact on welfare states. Cutbacks in social programs have been far more moderate than the sharp drop in labor strength in many countries might lead one to expect, and there appears to be little correlation between declines in left power resources and the magnitude of retrenchment.

The shift in both context and goals that characterizes the new politics of the welfare state helps to explain why this is so. Analysis of the contemporary welfare state's supporters must include not only organized [End Page 150] labor and left parties but also the more varied constituencies of individual programs. Interest groups linked to particular social policies are now prominent political actors. The rise of these interest groups is one of the clearest examples of how policy feedback from previous political choices can influence contemporary political struggles. Groups of program beneficiaries did not build the welfare state, but the welfare state contributed mightily to the development of these groups. By the time a politics of austerity began to emerge in the mid-1970s, most welfare state programs were connected to extensive networks of social support. Most prominent were the recipients of the various benefits--pensioners, the disabled, health care consumers. The providers of public services also had a major stake in sustaining expenditure and were usually well organized. 23 Finally, there was a range of public-interest organizations seeking to protect those too weak to mobilize on their own. Indeed, even when groups of recipients are not particularly well organized, politicians may have reason to be attentive to their concerns. Cutbacks, if recognized, are likely to incense voters, and political competitors stand ready to exploit such opportunities. The possibility of exacting punishment at the polls means that the potentially mobilized influence policymakers even in the absence of ongoing organized activity. 24

Thus, there are good reasons to believe that the centrality of left party and union confederation strength to welfare state outcomes has declined. The diminished relevance of power resource arguments reflects the fact that welfare states are now mature and that retrenchment is not simply the mirror image of welfare state expansion. In a context where public social provision is just emerging, the existence of very broad organizations pushing an expansionist policy agenda may be crucial. However, the unpopularity of program cutbacks and the emergence of new bases of support will give politicians pause even where unions and left parties are weak. Equally important, maturing social programs develop new bases of organized support that have substantial autonomy from the labor movement. This shifting base of support may have consequences for the dynamics of policy development, but the weakening of organized labor need not translate automatically into a commensurate weakening of the welfare state. [End Page 151]

New Institutionalism and Welfare State Politics

Patterns of governance matter. The "new institutionalist" resurgence in political science reflects a renewed appreciation of how relatively stable, routinized arrangements structure political behavior. 25 The political institutions of different countries vary along crucial dimensions, such as the rules of electoral competition, the relationship between legislature and executive, the role of the courts, and the place of subnational governments in politics. Institutions establish the rules of the game for political struggles--influencing group identities, policy preferences, and coalitional choices, and enhancing the bargaining power of some groups while devaluing that of others. Institutions also affect government capacities--their administrative and financial resources for fashioning policy interventions.

In the field of comparative social policy, the claim that political institutions must be considered consequential structures has been developed primarily in the work of Theda Skocpol and her collaborators. 26 These researchers have raised significant questions about the dynamics of welfare state development, pointing out, for example, that the power resources approach has had little success in accounting for pre-World War II social policy developments. The new institutionalist research agenda, however, has centered on explaining "American exceptionalism"--the belated and halfhearted development of social welfare policies in the United States. Just as a focus on Sweden was central to the development of the power resources model, concentration on the United States has underscored the importance of political institutions.

Institutionalists make two broad claims about welfare state development. First, strong states are likely to produce strong welfare states, with state strength defined in terms of governmental administrative capacities and institutional cohesion. Extensive administrative and financial resources make it easier to build expansive social policies. 27 Where political authority is fragmented, entrenched minorities will often block social legislation. 28 Federalism, separation of powers, strong bicameralism [End Page 152] or reliance on referenda all may restrict welfare state development. Although much of the work on institutions and welfare states has been based on comparative historical case studies, this claim about formal institutional veto points has recently received support in an important quantitative study as well. 29

The second central institutional argument concerns policy legacies, or feedback--the consequences of previously introduced welfare state programs. Arguments about policy feedback are essentially arguments about the consequences of big government. As policy decisions have had increasingly pervasive effects on economic and social life, their impact on political processes has expanded. Given that the welfare state is at the very heart of big government, it should come as no surprise that studies of welfare state development have generated some of the most persuasive arguments about policy feedback.

The possible consequences of preexisting policy structures for welfare state politics are quite diverse. 30 Welfare state structures affect the size and orientation of various societal groups as well as patterns of interest-group formation. Programs may provide the basis for processes of social learning that affect prospects for future program expansion, whether negatively or positively. Policies can create long-term commitments--such as the pay-as-you-go intergenerational contracts common to public pension systems--that lock in particular paths of policy development. 31

How easily can these arguments be applied to the distinctive politics of retrenchment? Again, there is no reason to assume that claims for earlier periods can simply be turned on their heads. For example, there is little reason to expect that bureaucratic capacities will be particularly important in an age of retrenchment. Can we administer it? may be an important question in a discussion of new or greatly expanded public initiatives, but for advocates of retrenchment the primary goal is to dismantle existing programs. Closing offices, curtailing services, and cutting benefits do not require formidable administrative capacity.

Arguments about the consequences of governmental cohesion also need to be reappraised in this new context. 32 At first glance, one might [End Page 153] expect that systems with fewer institutional veto points would be in a stronger position to pursue an agenda of radical retrenchment. Because retrenchment is generally unpopular, however, there are compelling reasons to question this expectation. While cohesive systems concentrate authority, they also concentrate accountability. The former tendency facilitates retrenchment, but the latter impedes it. Where authority is centralized, the public knows that the government of the day can prevent groups from suffering cutbacks. Strong governments, anticipating the high political cost of retrenchment, may forgo the opportunities provided by concentrated power. Thus, the theoretical basis for believing that government cohesion facilitates retrenchment is weak. We are left with the empirical question of whether concentration of power effects outweighs concentration of accountability effects. 33

The tremendous scope of modern welfare states suggests that institutionalist arguments about policy feedback may be even more relevant to the politics of retrenchment. There are, however, two quite different ways to apply policy feedback arguments to retrenchment. First, one could develop nuanced arguments that emphasize the distinctive characteristics of individual programs and their implications for successful retrenchment. Alternatively, one could present arguments about the overarching structures of particular welfare state regimes, suggesting that they are likely to promote a particular kind of politics.

The first approach has promise but is difficult to carry out in practice. Previous policies help shape the distribution of political resources. Preexisting policy designs may influence interest-group networks, the extent of long-term commitments, the rules governing programmatic reform, and the availability of techniques to reduce the visibility of cutbacks. These kinds of arguments have been used in qualitative case studies of both welfare state expansion and retrenchment. 34 Because they often apply to idiosyncratic characteristics of individual programs, however, it is hard to use them to generate general propositions about variations across countries.

The alternative of developing broad arguments about how welfare state regimes affect contemporary politics is more easily applied to cross-national comparisons. Esping-Andersen has developed propositions that link welfare state structures to cross-national variation in occupational [End Page 154] structures and, in turn, to contemporary political cleavages. 35 He suggests, for example, that social democratic welfare states like Sweden will face growing clashes between public and private sector workers, while conservative welfare states like Germany will produce a divide between labor market "insiders" and "outsiders." Such an approach offers the tantalizing promise of linking the evident significance of large welfare state structures to clear propositions about political change. Unfortunately, the argument does not work. As I will document in the next section, the hypothesized political cleavages emerge in only muted forms and have failed to generate a sustained backlash against the welfare state. The flaw in these broad arguments about welfare state regimes is that they greatly underestimate the difficulty of assembling and sustaining proretrenchment coalitions.

New institutionalist arguments have contributed greatly to our understanding of welfare state politics. Moreover, unlike arguments about the role of organized labor, there is no reason to think that the importance of institutional structures and the legacies of previous policy choices has declined. These arguments cannot simply be transferred from one context to another, however; they must be recast to apply to the specific settings and strategic problems that are characteristic of retrenchment politics.

From Theories of Expansion to Theories of Retrenchment

It is commonly maintained that our knowledge of welfare state expansion provides us with considerable insight into the new politics of the welfare state. Although this view is rarely stated explicitly, it undoubtedly explains why political scientists, who are usually preoccupied with understanding contemporary or rather recent events, have become so interested in historical studies of social policy. The presumed similarity of expansion and retrenchment is evident in some of the most important work on comparative social policy. In their sophisticated quantitative investigation of welfare state effort from 1956 to 1988, for example, Huber, Ragin, and Stephens acknowledge that their use of a pooled cross-sectional time-series model assumes "that causal processes are uniform through time." Esping-Andersen claims that "[t]he risks of [End Page 155] welfare-state backlash depend . . . on the class character of welfare states. . . . [T]he class coalitions in which the three welfare-state regime-types were founded, explain not only their past evolution but also their future prospects." More generally, he maintains that "a theory that seeks to explain welfare-state growth should also be able to understand its retrenchment or decline." 36

I see no reason to believe that this is true. Retrenchment is not simply the mirror image of welfare state expansion. Why should we assume that theories designed to explain outcomes in a particular context and involving the pursuit of particular goals will still apply once the political environment and the goals of key actors have undergone radical change? The question of whether theories of welfare state expansion offer insights into the retrenchment process is still open, but this preliminary discussion suggests that major modifications are probably required. In the following section, I draw on evidence from Great Britain, the United States, Germany, and Sweden to demonstrate the distinctiveness of retrenchment politics. Of declining importance are some formerly critical factors, such as the role of organized labor. Others, such as the design of political institutions, are of continuing significance but in new ways. Yet a crucial emerging factor is the mature welfare state itself, and its broad and deep reservoirs of public support.

III. Retrenchment Politics in Four Countries

To what extent have welfare states undergone retrenchment? What countries and programs have been most vulnerable to retrenchment initiatives and why? In this section I address these questions by reviewing the evolution of welfare states in four affluent democracies since the late 1970s. The evidence supports a number of claims. (1) There is little evidence for broad propositions about the centrality of strong states or left power resources to retrenchment outcomes. (2) The unpopularity of retrenchment makes major cutbacks unlikely except under conditions of budgetary crisis, and radical restructuring is unlikely even then. (3) For the same reason, governments generally seek to negotiate consensus packages rather than to impose reforms unilaterally, which further diminishes the potential for radical reform. And (4) far from creating a self-reinforcing dynamic, cutbacks tend to replenish support for the welfare state. [End Page 156]

Measuring retrenchment is a difficult task. Quantitative indicators are likely to be inadequate for several reasons. First, pure spending levels are rarely the most politically important or theoretically interesting aspects of welfare states. As Esping-Andersen put it in his analysis of welfare state expansion, "It is difficult to imagine that anyone struggled for spending per se." 37 In particular, rising unemployment may sustain high spending even as social rights and benefits are significantly curtailed. Second, spending estimates will fail to capture the impact of reforms that are designed to introduce retrenchment only indirectly or over the long term. Analysis must focus on qualitative and quantitative changes in programs and on prospective, long-term changes, as well as on immediate cutbacks. My investigation therefore relies on a combination of quantitative data on expenditures and qualitative analysis of welfare state reforms. 38 Rather than emphasizing cuts in spending per se, the focus is on reforms that indicate structural shifts in the welfare state. These would include (1) significant increases in reliance on means-tested benefits; (2) major transfers of responsibility to the private sector; and (3) dramatic changes in benefit and eligibility rules that signal a qualitative reform of a particular program. 39 The selection of countries to investigate was based on the desire to achieve significant variation on what the welfare state expansion literature suggests are the most plausible independent variables. The cases vary widely in the structure of political institutions, the extent of shifts in the distribution of power resources, the design of preexisting welfare states, and the severity of budgetary crisis.

Beginning with the quantitative evidence, aggregate measures provide little evidence that any of the four welfare states have undergone dramatic cutbacks. From 1974 through 1990 the expenditure patterns across the four cases are quite similar, despite widely different starting [End Page 157] points. As Tables 1 and 2 show, social security spending and total government outlays as a percentage of GDP are relatively flat over most of the relevant period. The exception is the recent surge in Swedish expenditures, which will be discussed below. There is a slight upward trend overall, with fluctuations related to the business cycle. Table 3, which tracks public employment, reveals a similar pattern (although the expansion of Swedish public employment from an already high base stands out). For none of the countries does the evidence reveal a sharp curtailment of the public sector.

Table 4 offers more disaggregated indicators of shifts in social welfare [End Page 158] spending among the four countries; spending patterns are reported for what the OECD terms "merit goods" (primarily housing, education, and health care) as well as for various income transfers. The figures suggest a bit more divergence among the cases, with the United States and Germany emerging as somewhat more successful in curbing spending. A very few program areas--notably British housing and German pensions--experienced significant reductions. Nonetheless, similarities across countries remain more striking than differences. None of the cases show major rises or declines in overall effort, and there are few indications of dramatic change in any of the subcategories of expenditure.

The data suggest a surprisingly high level of continuity and stability. 40 These figures must be treated with caution, however, since major changes in the spending for particular programs could be occurring within these broad categories. Policy reforms could have imposed lagged cutbacks that do not show up in spending figures. Furthermore, many other features of programs, not just spending levels, are of significance. To investigate these issues and to get a better sense of the processes that generated these aggregate outcomes, we turn to a more detailed investigation of the four cases.

Great Britain

By the mid-1970s Britain had developed a midsize welfare state with a mix of fairly modest income transfer programs and relatively extensive [End Page 159] [Begin Page 161] and interventionist policies in health care and housing. The past fifteen years of Conservative rule represent a powerful test of a number of claims about retrenchment. There are few if any cases within the OECD of such a radical swing to the right in the distribution of power resources. A balanced two-party system gave way to Conservative hegemony, as reflected in four consecutive electoral victories. 41 A fairly strong labor movement suffered repeated defeats in the face of high unemployment and changes in industrial relations law; union density declined from over 55 percent in 1979 to 35 percent in 1993. 42 Furthermore, Britain's Westminster political system concentrates political authority. If a strong state with few internal veto points is the key to reforming the contemporary welfare state, the British Conservatives were in an unusually favorable position. In short, a simple inversion of welfare state expansion arguments would suggest that Britain should have been fertile ground for social policy cutbacks. The evidence does not support such a conclusion. 43 While there are individual instances of significant retrenchment, notably in housing and pensions policy, these are the exception rather than the rule. The British welfare state, if battered, remains intact. As Table 4 indicates, social expenditure (merit goods plus transfers) as a share of GDP remains almost unchanged after more than a decade of Conservative governance.

It is worth starting with the two cases of retrenchment success, because these experiences are instructive. One of the triumphs of the Thatcher government was the privatization of public council housing. The sale of roughly 1.5 million homes to tenants laid the foundation for a major expansion of home ownership and sharp cutbacks in subsidies for those who remained in the public sector. 44 Housing reform was a uniquely popular case of retrenchment--in many respects, the exception that proves the rule. The liquidation of extensive public sector assets created a rare opportunity for credit claiming rather than blame avoidance, offering considerable benefits for both home purchasers and taxpayers. Public sector tenants, who under other circumstances would have fought the government, became disinterested or divided. [End Page 161]

The government's pension reforms were also relatively successful. 45 A less generous indexation rule for the basic pension produced gradual but considerable budgetary savings. The government also introduced significant cutbacks in the State Earnings-Related Pension Scheme (SERPS), while introducing a private personal pension alternative. The government's initial proposal--to abolish SERPS with a very limited phase-in period--met with overwhelming opposition, even from usual supporters like the Confederation of British Industry (CBI). Forced into an embarrassing retreat, the government introduced more modest but nonetheless substantial changes in SERPS, which will gradually expand private pensions while reducing public sector benefits. These changes, though criticized, failed to generate the kind of outcry that often led the government to back off from other reforms. Offering the carrot of personal pensions diminished the pain of the cuts in public pensions (although the required tax incentives also diminished the government's budgetary savings). More important, the government's use of incremental and seemingly technical reforms limited the emergence of opposition.

Careful political planning in the context of programs that were either weakly institutionalized (SERPS) or offered rare opportunities for credit claiming (council housing) protected the Thatcher government. These cases were exceptions, however. On the whole, the British Conservatives found the welfare state to be a political minefield, as popular support for social spending remained strong. In fact, British public opinion regarding social policy reveals the same pattern found elsewhere: a modest decline of support for the welfare state preceded the arrival of the Thatcher government but rebounded at the first hint of serious retrenchment. Public opinion in Britain has run strongly, and increasingly, in favor of maintaining or even expanding social provision. 46

Thus while the government was clearly hostile to much of the welfare state and had the political authority to implement dramatic changes if it chose to do so, the fear of being held accountable for unpopular initiatives held it in check. Preserving the National Health Service, for example, became far and away the Labour Party's best issue in the 1980s. Initial exploration of various options for privatization provoked public outrage. The Thatcher government repeatedly backed off [End Page 162] from plans for fundamental restructuring of the system. 47 By the end of the decade the government's repeated promise had become "the NHS is safe with us."

While the government achieved some nontrivial incremental cutbacks in various programs, radical retrenchment efforts failed, often at considerable political cost. The universal Child Benefit, frequently rumored to be in jeopardy, survived (although at a somewhat reduced level), due in part to well-organized lobbying efforts. A massive and highly publicized social security review promising to offer a "new Beveridge" ended up making modest adjustments to programs for low-income groups. 48 Sickness Pay, touted as an instance of successful privatization, remains a thinly disguised public sector benefit. 49 Furthermore, attempts to foster retrenchment arguably caused Thatcher's ultimate downfall. The tremendously unpopular poll tax--designed to cripple the social spending of local governments by cutting their financial legs out from under them--was a major source of the downward spiral in the polls that forced Thatcher's resignation. Her successor, John Major, followed a more moderate course. While attempting to consolidate Thatcher's least unpopular reforms, he launched few initiatives of his own and retreated on unpopular issues like cutbacks in Child Benefit.

Parts of the Conservative record demonstrate that major retrenchment is not impossible. Where a government can obscure the consequences of reform--or better yet, turn reform into a source of tangible benefits--the welfare state may be vulnerable. Had the entire British welfare state resembled council housing, the Conservative agenda would have succeeded. Yet the government's record reveals more cases of incremental adjustments or failed attempts to restructure programs. Overall social expenditure remains almost unchanged; housing and pensions aside, the government paid a significant price for the cutbacks it was able to impose. This outcome stands in sharp contrast to the Thatcher government's remarkably successful efforts in other policy areas, such as its overhaul of Britain's industrial relations system and its privatization of publicly owned industries. [End Page 163]

The United States

Many have seen the United States as another likely candidate for a dismantling of the welfare state. Like Britain, the United States experienced a significant rightward shift in power resources during the 1980s. Republicans captured the presidency in 1980 (as well as the Senate until 1986). Union power--never that extensive to begin with--continued a seemingly inexorable decline. The American welfare state was already fairly modest in scope, suggesting to some observers that public support for maintaining it would be weak. Unlike Thatcher, however, retrenchment advocates in the United States operated in a context of severely fragmented political authority. The combination of weak parties, separation of powers, and federalism created an institutional environment that was in many respects the polar opposite of that in Britain.

Yet at least through 1994 the American story reads like the British one, minus most of Thatcher's sporadic successes. Reagan's first year was the exception. Riding the antitax wave that had helped to elect him, Reagan was able to cobble together a loose coalition of southern Democrats and Republicans in his first year to pass some cuts in social programs, especially those affecting the poor. A decade-long expansion of low-income housing programs was rapidly reversed. Significant cuts were introduced in the main program for the poor, AFDC, and in Unemployment Insurance. The Reagan administration successfully exploited the fact that responsibility for these two programs was shared with fiscally strapped state governments. Both fared poorly in the 1980s, although many of the cutbacks occurred at the state level. 50

This first-year record has shaped many appraisals of the Reagan revolution. 51 As in other countries, however, popular support for retrenchment dissipated rapidly. The Reagan assault petered out in 1982, when further budget cuts were overwhelmingly rejected. Reagan's single major reform initiative, the New Federalism proposal, would have transferred responsibility for AFDC and food stamps to the states (in return for the federal government's assumption of complete control over [End Page 164] Medicaid). The proposal was so unpopular, however, that it could not find even a single congressional sponsor and died without ever being introduced. 52 By Reagan's second term incremental expansions of various social programs for the poor were back on the agenda, and Congress passed modest increases in food stamps and Medicaid and a dramatic expansion of the earned income tax credit for poor working families. 53

Middle-class programs also weathered the storm. Trust fund difficulties forced significant reforms of Social Security, and there was a series of efforts to trim Medicare's exploding costs (mostly through cuts in provider compensation). In all these efforts, however, Republicans fearful of the electoral consequences of retrenchment refused to move forward in the absence of bipartisan agreement. The need for consensus in turn assured that reforms would be acceptable to program defenders within the Democratic Party. This was most clearly evident in the 1983 Social Security amendments, where the reliance on a bipartisan, quasi-corporatist commission assured that radical reforms would be rejected. 54 It was also true, however, of the series of broad budget packages that sought cutbacks in the Medicare system. 55

Although it is generally argued that the residual nature of the welfare state in the United States creates a narrow political base, the initial backlash against the welfare state was short-lived. Declining support for social programs preceded Reagan's election. From 1982 onward--that is, immediately following the first round of budget cuts--polls revealed growing support for the welfare state. 56 As is true elsewhere, support for means-tested programs was far lower than that for middle-class programs, but the same public opinion pattern of modest declines in support followed by rapid recovery is evident for both targeted and universal programs. Reagan became much more hesitant as popular enthusiasm for retrenchment faded. In any event, the fragmented nature [End Page 165] of American political institutions assured that plans for further cutbacks met a hostile reception in Congress.

To be sure, with both the emergence of a large structural budget deficit during the 1980s and resistance to tax increases, little room was left for social policy expansion. The American welfare state moved into a zero-sum era, in which gains for some programs often came at the expense of others. This atmosphere continued when the Democrats returned to the White House in 1992, although social spending for the poor was a prime beneficiary of the 1993 budget agreement. From the late 1980s through 1994 the situation was one of reallocation within an essentially stagnant budget. 57

The 1994 elections, which ended forty years of Democratic predominance in Congress, precipitated the most vigorous challenge yet to the American welfare state. Republicans had learned important lessons from past defeats. Social Security was taken off the table. The issue was carefully framed as a matter of controlling the deficit and bringing government "closer to the people." Efforts at institutional change (the line-item veto and balanced-budget amendment) preceded (though with limited success) attacks on social programs. To diminish visibility, most of the major cuts were scheduled for the year 2000 and beyond. If fully implemented, the Republican budget proposals introduced in 1995 would represent a fundamental reform of American social policy. It will, however, be several years before the outcome of this latest battle is clear. Republican cutbacks are only now being formulated, and it will take some time for their impact to register with the electorate. While the current political environment poses a major test of the resilience of the welfare state, both American precedent and the experience elsewhere cast doubt on the proposition that Republicans will discover a deep reservoir of public tolerance for sharp cuts in social programs.

Germany

Germany, like Sweden, has a very extensive welfare state, though the German system is geared toward transfer payments rather than public services, and toward redistribution over the life cycle rather than across income groups. As in Britain and the United States, there has been a considerable swing to the right in elections during the period of austerity. A right-of-center coalition has been in office since 1982. In recent [End Page 166] years, however, the Social Democrats, with their majority in the Bundesrat, have gained considerable influence. And although Germany's powerful unions have been under pressure, compared with unions in other countries their organizational strength has held up quite well. 58 Thus, Germany represents a case of moderately diminished left power resources and relatively fragmented political authority.

The German welfare state is based, not on maximizing employment, but on providing subsidies to the "outsiders," who are encouraged to leave the labor market to those who are highly productive. Esping-Andersen has speculated that the result is likely to be an "insider-outsider" conflict in which the employed (along with employers) increasingly balk at the cost of subsidizing a large and growing "surplus population." 59 Indeed, within the recent wave of commentary about the contribution of high wages and extensive social protection to Europe's economic problems, Germany's huge wage costs (including steeply rising payroll taxes) have received particular attention. 60 Adding to the stress has been the cost of unification, as the West German welfare state was extended to cover the far less productive East Germany.

The fiscal pressures facing Germany are evident and are unlikely to go away. Demographic shifts will increase costs even if expenses related to unification begin to subside. There is, however, little sign that these pressures will translate into a sharp insider/outsider conflict. This line of potential cleavage is based largely on age, with by far the largest and most expensive group of outsiders consisting of former insiders: pensioners and early retirees. There are formidable barriers to the development of political cleavages along generational lines. 61 Where costs associated with aging are the main source of budgetary pressure, insiders will have to recognize that in the future they too will be outsiders. This is likely to temper any tendency toward a polarization between "them" and "us." 62

Far from revealing a sharp new political cleavage, the reform of pensions in the late 1980s fits the general cross-national pattern for retrenchment [End Page 167] in popular social programs. 63 With projections indicating enormous long-term deficits as a result of demographic trends, the need for pension reform was widely recognized. While some critics pressed for a rethinking of the entire scheme, the Kohl government instead searched for consensus among experts and the social partners, and eagerly sought support from the opposition Social Democrats. Union and business representatives submitted joint statements on pension reform. All parties except the Greens supported the Pension Reform Act passed in November 1989. 64 The resulting plan incorporated familiar pruning techniques: slightly lower replacement rates, an increase in the retirement age, and increased contributions. Combined with earlier cutbacks, these reforms have generated substantial budgetary savings. The basic structure of Germany's generous pension system remained unaltered, however. In the case of health care reform as well, corporatist accommodation of entrenched interests and a search for cross-party consensus has been the rule. 65

The one indication of a distinctly conservative cast to retrenchment initiatives prior to 1989 came in the pattern of benefit cuts. Most affected were welfare and unemployment insurance benefits that could be considered a hindrance to labor market flexibility. While these benefits were relatively well protected under the SPD-led coalition of the late 1970s, they experienced disproportionate reductions after 1982. 66 Still, the differences were limited and did not overturn the basic reality that social welfare policy operated within a relatively consensual framework. As Offe has argued, moderate cutbacks were carefully designed by a "de facto bipartisan coalition" and orchestrated to prevent a political outcry. 67

The strains of the postunification period raised the possibility of more dramatic reform. Germany's worsening fiscal situation, combined with concern about industrial competitiveness, generated growing criticism [End Page 168] of the welfare state. Yet the response to unification emphasized continuity: East Germans were brought into the West German social policy regime on extremely generous terms. In response to budgetary pressures since then, a series of cuts have been introduced in major social programs, and more cuts are probably on the way. 68 Again, however, the pattern has been to trim benefit levels rather than challenge the basic structure of programs.

The recent ambitious expansion of long-term care covering in-home and nursing-home services is clear grounds for skepticism about the prospects for radical retrenchment or a generational backlash against the German welfare state. 69 The system will relieve the sickness funds and local social assistance budgets of responsibility for long-term care expenditures. While the scheme partly amounts to fiscal relief for strapped Länder (state) governments, it involves significant new benefits as well. Its introduction at a time of budgetary stress and widespread discussion of high social wage costs indicates the continuing political attractiveness of social programs, as well as the electoral clout of the elderly. 70

There is a possibility that groups more easily defined as problem cases, such as the long-term unemployed (especially disaffected youth, immigrants, or East Germans) could become the targets for political attacks. Concern about mounting unemployment rates and labor market inflexibilities point in the same direction. The Kohl government's recent proposals for cuts in social expenditures do indeed concentrate on unemployment and welfare benefits. Sharp cuts, however, will provoke a strong challenge from the SPD and unions, since the conditions of the unemployed have significant implications for workers. Furthermore, the force of demographic demands is such that even these politically difficult reforms would have only a limited effect on the long-term budgetary situation. 71 The recent history of retrenchment suggests that most governments are likely to be skittish about pursuing [End Page 169] big social policy battles unless they promise to yield substantial budgetary gains.

Demographic and budgetary pressures assure that an atmosphere of austerity will continue to surround the German welfare state. Indeed, the German government has been quite successful in holding the line on spending. Yet, as Offe aptly puts it, it has been a period of "smooth consolidation." A fundamental rethinking of social policy seems a remote possibility. The structure of political institutions--both constitutional rules and corporatist policy networks--puts a premium on consensus. The SPD's significant gains in the October 1994 elections, where Kohl's coalition government retained only a razor-thin majority, are unlikely to invigorate a governing coalition that has never shown much appetite for radical initiatives.

Sweden

Sweden combines a tremendously successful social democratic party, powerful unions, and one of the world's most extensive and redistributive welfare states. With the breakdown of centralized collective bargaining, union power has weakened in the past decade, and the political hegemony of the SAP has given way to periodic fluctuations between SAP governance and coalitions of the bourgeois parties. 72 Swedish institutions, like British ones, provide for a considerable concentration of political authority.

As with Germany, an analysis of trends in Swedish social policy must distinguish two periods--the cyclical patterns that Sweden shared with other countries prior to 1990 and the dramatic worsening of economic and budgetary circumstances over the past four years. Despite the distinctiveness of Swedish politics and the Swedish welfare state, the dynamics of reform prior to 1990 bears many similarities to the processes discussed in the three preceding cases. Regardless of who was in office, there was only occasional trimming of social programs; no government showed much enthusiasm for a major revamping of the Swedish welfare state.

Esping-Andersen has tried to link the prospects for retrenchment to policy feedback from the structure of Sweden's welfare state, positing that Sweden's system of extensive service provision and high public sector employment promotes a cleavage between workers in the public and [End Page 170] private sectors. 73 In the 1980s both private sector unions and employers criticized public employees. This occurred in part because of the way Sweden's centralized industrial relations system created linkages between public and private sector pay, and in part simply because of the high tax costs of massive public sector employment expansion. 74

It might be expected that the events of the past few years would exacerbate this tension. The Swedish economy came under extraordinary pressure after 1990. The fall in output and the rise in unemployment seem to have had relatively little to do with the welfare state itself. 75 Yet if the welfare state did not cause the economic crisis, it is clearly implicated in Sweden's current predicament. Unemployment, including those participating in government-financed active labor market schemes, rose from 3 percent in 1989 to over 12 percent in 1993. 76 Designed to operate at full employment, the Swedish welfare state cannot tolerate burdens of this magnitude. Sweden's generous and expensive social benefits led to immediate fiscal trouble when rising unemployment produced falling revenues and higher outlays. In only four years Sweden went from having the largest surplus in the OECD to operating the largest deficit. Government outlays reached the extraordinary level of 73 percent of GDP in 1993, and the deficit stood at over 14 percent of GDP. 77 Given existing levels of taxation, the adjustment burden was bound to fall heavily on the expenditure side.

The emergence of a true fiscal crisis after 1990 required extensive reforms. Furthermore, between 1991 and 1994 Sweden was governed by a bourgeois coalition--and one whose center of gravity was considerably to the right of the one that governed between 1976 and 1982. Conditions thus might have seemed uniquely favorable to a complete overhaul of social policy. What is striking is that even under these extraordinary circumstances there was no sign that the welfare state would be radically restructured. Instead, considerable care was taken to operate within the structure of existing programs. 78 Prominent examples [End Page 171] include an expansion of waiting days for unemployment and sickness benefits, a lowering of replacement rates for unemployment benefits (from 90 percent to 80 percent), a freeze on adjustments in child allowances, and a raise in the retirement age, from sixty-five to sixty-six.

Ironically, the breakdown of other components of the Swedish economic model also diffused some of the pressure on the divide between public and private sector workers. The end of centralized bargaining has made it easier for private sector workers to decouple pay settlements from the public sector. 79 Higher unemployment made the option of cutting the public sector workforce less attractive and shifted attention to transfer programs, where the provision of generous benefits has always depended on keeping the cohort of recipients small. Both the crisis package accepted by the unions and the SAP in late 1992 and the government's 1993 budget were geared toward cuts in transfer programs (sick pay, unemployment insurance, pensions). 80 Thus, while economic problems created added pressure for the welfare state, they did so in a way that encouraged a more dispersed and incremental politics of austerity, rather than a polarization between public and private sector workers.

The government showed little inclination to use the opening for radical restructuring of the welfare state. Coping with the pain of administering austerity weakened the bourgeois coalition's enthusiasm for drastic measures. In any event, only the Conservatives within the governing coalition expressed serious doubts about Sweden's universalist welfare state. Polling data from 1992 indicated overwhelming support for social spending, including increases since 1986 for the major programs. 81 The coalition carefully followed the principal rules of blame avoidance. It sought SAP and union support for crisis packages of reform, purchasing political cover at the price of forgoing opportunities for radical change. 82 A similar approach was taken to pension reform, paralleling the process followed in Germany and the United States. The pension reform introduced in June 1994, based on extensive consultations among the major parties and representative of labor and capital, [End Page 172] seeks to get Sweden's public pension system on a stable, long-term footing without challenging its basic principles. 83

Far from using its electoral breakthrough and the onset of a budgetary crisis as a golden opportunity to remake the welfare state, Sweden's bourgeois coalition chose a cautious course. Throughout, the coalition studiously avoided initiatives that might lend credence to the accusation that it sought to dismantle the welfare state. The October 1994 elections vindicated the coalition's concerns, but in a way that revealed the unpopularity of even modest assaults on social programs. In the face of Sweden's economic difficulties, the SAP triumphed, posting one of its highest vote totals ever. Moreover, the victory was achieved on a platform that stressed the SAP's intention to reduce the deficit largely through tax increases rather than budget cuts.

IV. The New Politics of the Welfare State

Economic, political, and social pressures have fostered an image of welfare states under siege. Yet if one turns from abstract discussions of social transformation to an examination of actual policy, it becomes difficult to sustain the proposition that these strains have generated fundamental shifts. This review of four cases does indeed suggest a distinctly new environment, but not one that has provoked anything like a dismantling of the welfare state. Nor is it possible to attribute this to case selection, since the choice of two prototypical cases of neoconservatism (Britain and the United States) and two cases of severe budgetary shocks (Germany and Sweden) gave ample room for various scenarios of radical retrenchment. Even in Thatcher's Britain, where an ideologically committed Conservative Party has controlled one of Europe's most centralized political systems for over a decade, reform has been incremental rather than revolutionary, leaving the British welfare state largely intact. In most other countries the evidence of continuity is even more apparent. 84 [End Page 173]

To be sure, there has been change. Many programs have experienced a tightening of eligibility rules or reductions in benefits. On occasion, individual programs (such as public housing in Britain) have undergone more radical reform. In countries where budgetary pressures have been greatest, cuts have been more severe. Over the span of two decades, however, some changes in social policy are inevitable; even in the boom years of the 1960s specific social programs sometimes fared poorly. What is striking is how hard it is to find radical changes in advanced welfare states. Retrenchment has been pursued cautiously: whenever possible, governments have sought all-party consensus for significant reforms and have chosen to trim existing structures rather than experiment with new programs or pursue privatization.

This finding is striking, given that so many observers have seen the post-1973 period as one of fundamental change in modern political economies. A harsher economic climate has certainly generated demands for spending restraint. Additional pressures have stemmed from the maturation of social programs and adverse demographic trends. Yet compared with the aspirations of many reformers and with the extent of change in fields such as industrial relations policy, macroeconomic policy, or the privatization of public industries, what stands out is the relative stability of the welfare state.

I have suggested that to understand what has been happening requires looking beyond the considerable pressures on the welfare state to consider enduring sources of support. There are powerful political forces that stabilize welfare states and channel change in the direction of incremental modifications of existing policies. The first major protection for social programs stems from the generally conservative characteristics of democratic political institutions. The welfare state now represents the status quo, with all the political advantages that this status confers. Nondecisions generally favor the welfare state. Major policy change usually requires the acquiescence of numerous actors. Where power is shared among different institutions (for example, Germany, the United States), radical reform will be difficult.

As the British and Swedish cases show, radical change is not easy even in a situation of concentrated political power. A second and crucial source of the welfare state's political strength comes from the high electoral costs generally associated with retrenchment initiatives. Despite scholarly speculation about declining popular support for the welfare state, polls show little evidence of such a shift, and actual political struggles over social spending reveal even less. On the contrary, even halting efforts to dismantle the welfare state have usually exacted a high [End Page 174] political price. Recipients of social benefits are relatively concentrated and are generally well organized. They are also more likely to punish politicians for cutbacks than taxpayers are to reward them for lower costs. Nowhere is there evidence to support the scenario of a self-reinforcing dynamic, with cutbacks leading to middle-class disenchantment and exit, laying the foundation for more retrenchment. Instead, the recurrent pattern in public-opinion polls has been a mild swing against the welfare state in the wake of poor economic performance and budgetary stress, followed by a resurgence of support at the first whiff of significant cuts.

Nor does the welfare state's political position seem to have been seriously eroded--at least in the medium term--by the decline of its key traditional constituency, organized labor. Only for those benefits where unions are the sole organized constituency, such as unemployment insurance, has labor's declining power presented immediate problems, and even here the impact can be exaggerated. 85 The growth of social spending has reconfigured the terrain of welfare state politics. Maturing social programs produce new organized interests, the consumers and providers of social services, that are usually well placed to defend the welfare state.

The networks associated with mature welfare state programs constitute a barrier to radical change in another sense as well. As recent research on path dependence has demonstrated, once initiated, certain courses of development are hard to reverse. 86 Organizations and individuals adapt to particular arrangements, making commitments that may render the costs of change (even to some potentially more efficient alternative) far higher than the costs of continuity. Existing commitments lock in policymakers. Old-age pension systems provide a good example. Most countries operate pensions on a pay-as-you-go basis: current workers pay "contributions" that finance the previous generation's [End Page 175] retirement. Once in place, such systems may face incremental cutbacks, but they are notoriously resistant to radical reform. 87 Shifting to private, occupationally based arrangements would place an untenable burden on current workers, requiring them to finance the previous generation's retirement while simultaneously saving for their own.

Over time, all institutions undergo change. This is especially so for very large ones, which cannot be isolated from broad social developments. The welfare state is no exception. But there is little sign that the last two decades have been a transformative period for systems of social provision. As I have argued, expectations for greater change have rested in part on the implicit application of models from the period of welfare state expansion, which can be read to suggest that economic change, the decline in union power, or the presence of a strong state creates the preconditions for radical retrenchment. I find little evidence for these claims.

This preliminary investigation still leaves us some distance from a coherent comparative theory of retrenchment politics. It does, however, suggest some of the building blocks for such a theory. The pressures of a shifting global economy, which have long been at the center of discussions of the contemporary welfare state, continue to deserve major (if more nuanced) attention. What needs more consideration is what happens when these considerable pressures collide with popular, deeply institutionalized public policies. I have emphasized that politicians are likely to pursue strategies that will not damage their chances for reelection. The centrality of electoral considerations, combined with the general unpopularity of welfare state cutbacks, suggests some plausible hypotheses about the political preconditions for significant reform. These hypotheses are only tentative and would need to be subjected to sustained comparative scrutiny. Each, however, is compatible with the analysis of retrenchment outlined here and with the evidence presented.

First, radical retrenchment may be facilitated when there is significant electoral slack, that is, when governments believe that they are in a strong enough position to absorb the electoral consequences of unpopular decisions. 88 Thus, one reason for Thatcher's relative (though still limited) success may have been the division among her opponents [End Page 176] within a first-past-the-post electoral system. This may have given her more room to pursue unpopular policies that would have been beyond the reach of a government in a precarious electoral position. However, calculating electoral slack ex ante is a tricky business, and most governments are likely to proceed cautiously. As I have indicated, even the Thatcher government generally retreated when confronted with widespread opposition.

Second, moments of budgetary crisis may open opportunities for reform. Advocates of retrenchment will try to exploit such moments to present reforms as an effort to save the welfare state rather than destroy it. Framing the issue in this manner may allow governments to avoid widespread blame for program cutbacks. Making the claim of crisis credible, however, generally requires collaboration with the political opposition. In turn, the need for consensus makes it difficult to utilize crises to promote radical restructuring. Thus, while the appearance of fiscal stress encourages downward adjustments in social programs, it is far less clear that it provides a platform for a radical overhaul of social policy.

Third, the success of retrenchment advocates will vary with the chances for lowering the visibility of reforms. Those seeking retrenchment will try to avoid political outcries by diminishing the visibility of their cutbacks or by trying to hide their own responsibility for unpopular outcomes. Success in these efforts, I have argued, depends partly on the design of political institutions. Whether political authority is concentrated or not helps to structure the choices available to retrenchment advocates. Where authority is concentrated (as in Britain and Sweden), governments will be hard-pressed to avoid blame for unpopular decisions, but they will have a greater capacity to develop and implement strategies that obscure cutbacks. Governments in more fragmented systems must fashion strategies that minimize the need to force multiple policy changes through institutional veto points. However, they may find it easier to duck accountability for unpopular policies. Federalism, for example, opened up considerable possibilities for Reagan to shift the blame for cuts in some programs, a tactic that is central to the current efforts of congressional Republicans.

Finally, the prospects for changing institutions (the rules of the game) may be of great significance. If retrenchment advocates can restructure the ways in which trade-offs between taxes, spending, and deficits are presented, evaluated, and decided, they may be able to shift the balance of political power. So far, these institutional shifts have been rare, but several instances may be of growing relevance. In Europe [End Page 177] the increasing policy significance of the EC may alter the terrain for struggles over the welfare state. If reforms can be presented as legally required or economically imperative because of the single market or moves toward monetary union, national governments may be freed from some blame for welfare-state cutbacks. 89 In the United States the new Republican majority in Congress deferred efforts to cut programs until after a strong (but only modestly successful) push to change the rules of the game. The intent of the rule changes was to increase the salience of taxes and create a more favorable climate for attacking social spending.

If this analysis suggests some plausible sources of cross-national variation, it also highlights the need to disaggregate welfare states and consider variations across programs. Indeed, in the four cases considered outcomes often differed more across programs (for example, the contrast between council housing and the NHS in Britain) than across countries. It is commonly argued that the crucial distinction across programs will be between universal programs and those that target the poor, and that the latter will be especially vulnerable. 90 Yet the current investigation does not support this assertion. Rather, variations in outcomes across programs in the four countries do not generally track the universal/targeted divide. Among the likely reasons that targeted programs have generally not proved more vulnerable: cuts in these programs tend to yield only minimal budgetary savings, and conservative governments interested in radical reform object the most to the universal programs that require high tax rates and compete with plausible private alternatives. Instead, the current investigation suggests that a promising area of research concerns the features of programs that allow governments either to obscure the impact of retrenchment on voters or to diminish their own accountability for unpopular reforms. 91 Programs that are poorly indexed, for example, make it easier to pursue a low-visibility strategy of allowing inflation to gradually erode the value of benefits.

All of these hypotheses build on the core argument of this essay: that frontal assaults on the welfare state carry tremendous electoral risks. [End Page 178] The contemporary politics of the welfare state is the politics of blame avoidance. Governments confronting the electoral imperatives of modern democracy will undertake retrenchment only when they discover ways to minimize the political costs involved. But as I emphasize, such techniques are hard to come by. While this analysis suggests some of the possible keys to variation in policy outcomes, the most significant finding concerns not variation but commonality. Everywhere, retrenchment is a difficult undertaking. The welfare state remains the most resilient aspect of the postwar political economy.

Understanding why this is so requires that old arguments be rethought and recast to address the exigencies of a new setting. At a time when historical institutionalism has become fashionable, this conclusion has broad implications for the study of comparative politics. The strong calls to incorporate historical analysis into the study of contemporary politics are compelling. 92 Yet we must remain cognizant of the hazards of drawing on history in the wrong way. There are significant dangers in using historical analogies to study contemporary social politics, since the goals of social policy reformers and the context in which they operate have undergone profound change. Instead, historically grounded analysis should emphasize that social policy change is a process that unfolds over time. My focus on the impact of inherited social policy structures draws on this precise point. The growth of the welfare state has transformed the politics of social policy. A historical perspective should stress that today's policymakers operate in an environment fundamentally shaped by policies inherited from the past, rather than suggesting that current politics will echo the conflicts of a previous era.

Paul Pierson is Associate Professor of Government at Harvard University. He is the author of Dismantling the Welfare State? Reagan, Thatcher and the Politics of Retrenchment (1994) and coeditor of European Social Policy: Between Fragmentation and Integration (1995).

Notes

* I am grateful to the Russell Sage Foundation for financial and administrative support and to Miguel Glatzer for considerable research assistance, as well as helpful comments.

1 On welfare state expansion, see Gösta Esping-Andersen, The Three Worlds of Welfare Capitalism (Princeton: Princeton University Press, 1990); and Evelyne Huber, Charles Ragin, and John D. Stephens, "Social Democracy, Christian Democracy, Constitutional Structure, and the Welfare State," American Journal of Sociology 99 (November 1993).

2 In this sense, my analysis parallels Verdier's recent effort to interpret foreign trade policy as "the outcome of a process influenced by voters." Daniel Verdier, Democracy and International Trade: Britain, France and the United States, 1860-1990 (Princeton: Princeton University Press, 1994), xv.

3 R. Kent Weaver, "The Politics of Blame Avoidance," Journal of Public Policy 6 (October-December 1986).

4 Peter Flora and Arnold J. Heidenheimer, eds., The Development of Welfare States in Europe and America (New Brunswick, N.J.: Transaction, 1982).

5 As recent research has suggested, it would be wrong to treat business as always and everywhere opposed to welfare state programs. For illuminating studies of the United States, see, for example, Colin Gordon, New Deals: Business, Labor, and Politics in America, 1920-1935 (Cambridge: Cambridge University Press, 1994); and Cathie Jo Martin, "Nature or Nurture? Sources of Firm Preference for National Health Reform," American Political Science Review 89 (December 1995). Nonetheless, it is clear that most business organizations in all the advanced industrial democracies have favored--often vehemently--cutbacks in the welfare state over the past fifteen years.

6 Mancur Olson, The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge: Harvard University Press, 1965); James Q. Wilson, Political Organizations (New York: Basic Books, 1973), 330-37.

7 Daniel Kahneman and Amos Tversky, "Prospect Theory: An Analysis of Decision under Risk," Econometrica 47 (March 1979); idem, "Choices, Values and Frames," American Psychologist 39 (April 1984).

8 Howard S. Bloom and H. Douglas Price, "Voter Response to Short-Run Economic Conditions: The Asymmetric Effect of Prosperity and Recession," American Political Science Review 69 (December 1975); Samuel Kernell, "Presidential Popularity and Negative Voting: An Alternative Explanation of the Midterm Congressional Decline of the President's Party," American Political Science Review 71 (March 1977); and Richard R. Lau, "Explanations for Negativity Effects in Political Behavior," American Journal of Political Science 29 (February 1985).

9 Peter Flora, "From Industrial to Postindustrial Welfare State?" Annals of the Institute of Social Science, special issue (Institute of Social Science, University of Tokyo) (1989), 154.

10 Christine L. Day, What Older Americans Think: Interest Groups and Aging Policy (Princeton: Princeton University Press, 1990), 25-26.

11 Gösta Esping-Andersen, Politics against Markets: The Social Democratic Road to Power (Princeton: Princeton University Press, 1985); Paul Pierson, "When Effect Becomes Cause: Policy Feedback and Political Change," World Politics 45 (July 1993).

12 Weaver (fn. 3); Paul Pierson, Dismantling the Welfare State? Reagan, Thatcher and the Politics of Retrenchment (Cambridge: Cambridge University Press, 1994), chap. 1.

13 R. Douglas Arnold, The Logic of Congressional Action (New Haven: Yale University Press, 1990).

14 Wilensky, The Welfare State and Equality (Berkeley: University of California Press, 1975).

15 One could fill a small library with books and articles that make this claim. For a good recent example, see Herman Schwartz, "Small States in Big Trouble: State Reorganization in Australia, Denmark, New Zealand, and Sweden in the 1980s," World Politics 46 (July 1994).

16 Geoff Garrett and Peter Lange, "Political Responses to Interdependence: What's Left for the Left," International Organization 45 (1991).

17 For a classic treatment of this dynamic, see William L. Cary, "Federalism and Corporate Law: Reflections on Delaware," Yale Law Journal 83 (March 1974).

18 Peter Lange, "The Politics of the Social Dimension," in Alberta Sbragia, ed., Euro-politics (Washington, D.C.: Brookings Institution, 1992). For evidence casting doubt on the proposition that systems of social protection have significant negative effects on economic performance, see Rebecca M. Blank, ed., Social Protection versus Economic Flexibility: Is There a Trade-off? (Chicago: University of Chicago Press, 1994).

19 "Unraveling" could occur in a more subtle way, as heightened capital mobility strengthens the bargaining position of business, leading to the gradual erosion of "tightly coupled" systems of industrial relations and, perhaps, welfare states. For a good example of this kind of argument, see Wolfgang Streeck, "From Market-Making to State-Building? Reflections on the Political Economy of European Social Policy," in Stephan Leibfried and Paul Pierson, eds., European Social Policy: Between Fragmentation and Integration (Washington, D.C.: Brookings Institution, 1995). While such a scenario cannot be ruled out, the evidence for it remains quite limited. These arguments appear to be more popular among those, like Streeck, who focus on industrial relations systems. See, for example, Jonas Pontusson and Peter Swenson, "Labor Markets, Production Strategies, and Wage Bargaining Institutions: The Swedish Employer Offensive in Comparative Perspective," Comparative Political Studies (forthcoming); and Kathleen Thelen, "West European Labor in Transition: Sweden and Germany Compared," World Politics 46 (October 1993). Industrial relations systems, however, seem more fragile than welfare state structures. Welfare states, I will suggest, have considerably broader bases of support, which promote the restoration of equilibrium and inhibit the kind of unraveling that occurs in some industrial relations systems.

20 Douglass C. North, Institutions, Institutional Change, and Economic Performance (Cambridge: Cambridge University Press, 1990).

21 John D. Stephens, The Transition from Capitalism to Socialism (London: Macmillan, 1979); Walter Korpi, The Democratic Class Struggle (London: Routledge, 1983); Esping-Andersen (fn. 11); Walter Korpi, "Power, Politics, and State Autonomy in the Development of Social Citizenship: Social Rights during Sickness in Eighteen OECD Countries since 1930," American Sociological Review 54 (June 1989).

22 Esping-Andersen (fn. 1); Huber, Ragin, and Stephens (fn. 1).

23 In this respect, organized labor (public employee unions) continues to be of significance, although not in the way posited by power resource theorists. Union interests are now linked primarily to the employment-generating effects of specific public programs rather than to the broad consequences of generous public provision for the bargaining position of workers.

24 On this point, see Arnold (fn. 13).

25 Peter Hall, Governing the Economy (Oxford: Oxford University Press, 1986); James March and Johan Olsen, Rediscovering Institutions (New York: Free Press, 1989); North (fn. 20).

26 See, for example, Margaret Weir, Ann Shola Orloff, and Theda Skocpol, eds., The Politics of Social Policy in the United States (Princeton: Princeton University Press, 1988); and Theda Skocpol, Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States (Cambridge: Harvard University Press, 1992).

27 Hugh Heclo, Modern Social Politics in Britain and Sweden (New Haven: Yale University Press, 1974).

28 Keith G. Banting, The Welfare State and Canadian Federalism, 2d ed. (Montreal: McGill-Queen's University Press, 1987); Ellen Immergut, Health Politics: Interests and Institutions in Western Europe (Cambridge: Cambridge University Press, 1992).

29 Huber, Ragin, and Stephens (fn. 1).

30 Pierson (fn. 11).

31 For examples of each of these arguments, see, respectively, Esping-Andersen (fn. 11); Skocpol (fn. 26); and Paul Pierson, "'Policy Feedbacks' and Political Change: Contrasting Reagan and Thatcher's Pension-Reform Initiatives," Studies in American Political Development 6 (Fall 1992).

32 This paragraph relies on Paul Pierson and R. Kent Weaver, "Imposing Losses in Pensions Policy," in R. Kent Weaver and Bert Rockman, eds., Do Institutions Matter? Government Capabilities at Home and Abroad (Washington, D.C.: Brookings Institution, 1993).

33 What seems more likely is that the structure of formal institutions will influence the strategies of retrenchment advocates. I return to this point in the conclusion.

34 On expansion, see Heclo (fn. 27); and Skocpol (fn. 26). On retrenchment, see Pierson (fn. 12).

35 Esping-Andersen (fn. 1); Gösta Esping-Andersen, "Postindustrial Cleavage Structures: A Comparison of Evolving Patterns of Social Stratification in Germany, Sweden, and the United States," in Francis Fox Piven, ed., Labor Parties in Postindustrial Societies (Oxford: Polity Press, 1991); and idem, "The Emerging Realignment between Labour Movements and Welfare States," in Marino Regini, ed., The Future of Labour Movements (London: Sage, 1992).

36 Huber, Ragin, and Stephens (fn. 1), 733; Esping-Andersen (fn. 1), 33, 32. Of course, Esping-Andersen has also emphasized that the growth of the welfare state affects welfare state politics.

37 Esping-Andersen (fn. 1), 21.

38 A recent draft paper by Stephens, Huber, and Ray presents the first sophisticated statistical analysis of retrenchment, utilizing newly assembled data that allow investigation of fairly detailed programmatic changes over a large number of countries. There are important limitations: much of the programmatic data end in 1986 or 1987; many programs are not covered; and the still-small sample allows the statistical testing of only a few broad hypotheses (essentially, the impact of partisanship) about the politics of program change. The results reported strongly support most of the analysis presented here, although they view Thatcher as more successful than I do. John D. Stephens, Evelyne Huber, and Leonard Ray, "The Welfare State in Hard Times" (Paper presented at the conference on the "Politics and Political Economy of Contemporary Capitalism," University of North Carolina, Chapel Hill, September 1994).

39 Establishing what constitutes "radical" reform is no easy task. For instance, it is impossible to say definitively when a series of quantitative cutbacks amounts to a qualitative shift in the nature of programs. Roughly though, that point is reached when because of policy reform a program can no longer play its traditional role (e.g., when pension benefits designed to provide a rough continuation of the retiree's earlier standard of living are clearly unable to do so).

40 This broad conclusion is echoed for a much larger number of cases in Stephens, Huber, and Ray (fn. 38).

41 Geoff Garrett, "The Politics of Structural Reform: Swedish Social Democracy and Thatcherism in Comparative Perspective," Comparative Political Studies 25 (January 1993).

42 The 1979 figure is from Robert Price and George Sayers Bain, "Union Growth in Britain: Retrospect and Prospect," British Journal of Industrial Relations 21 (March 1983), 47. The 1993 figure is from Derek Bird and Louise Corcoran, "Trade Union Membership and Density, 1992-93," Employment Gazette (June 1994), 193.

43 Pierson (fn. 12), chaps. 3-6.

44 Ray Forrest and Alan Murie, Selling the Welfare State: The Privatization of Public Housing (London: Routledge, 1988).

45 Alan Walker, "Thatcherism and the New Politics of Old Age," in John Myles and Jill Quadagno, eds., States, Labor Markets, and the Future of Old Age Policy (Philadelphia: Temple University Press, 1991); Pierson (fn. 31).

46 Peter Taylor-Gooby, "The Future of the British Welfare State: Public Attitudes, Citizenship and Social Policy under the Conservative Governments of the 1980s," European Sociological Review 4 (May 1988).

47 Rudolf Klein, The Politics of the National Health Service, 2d ed. (London: Longman, 1989); Edwin Griggs, "The Politics of Health Care Reform in Britain," Political Quarterly 62 (October-December 1991).

48 Nicholas Deakin, The Politics of Welfare (London: Methuen, 1987).

49 Nicholas Barr and Fiona Coulter, "Social Security: Solution or Problem?" in John Hills, ed., The State of Welfare: The Welfare State in Britain since 1974 (Oxford: Oxford University Press, 1990), 281; Michael O'Higgins, "Social Welfare and Privatization: The British Experience," in Sheila B. Kamerman and Alfred J. Kahn, eds., Privatization and the Welfare State (Princeton: Princeton University Press, 1989).

50 In the case of AFDC, which is not indexed, this happened largely because state governments failed to index benefits to inflation. Given this structural feature of the program, "nondecisions" allowed quiet retrenchment. This trend predated Reagan's arrival in office. Indeed, cuts in real benefits were greater during Carter's presidency (when inflation was high) than under Reagan.

51 See, for example, two frequently cited studies: Sara A. Rosenberry, "Social Insurance, Distributive Criteria and the Welfare Backlash: A Comparative Analysis," British Journal of Political Science 12 (October 1982); and John Palmer and Isabel Sawhill, eds., The Reagan Experiment (Washington, D.C.: Urban Institute Press, 1982).

52 Tim Conlan, New Federalism: Intergovernmental Reform from Nixon to Reagan (Washington, D.C.: Brookings Institution, 1988), 95-238.

53 Paul E. Peterson, "The Rise and Fall of Special Interest Politics," Political Science Quarterly 105 (Winter 1990-91); Robert Greenstein, "Universal and Targeted Approaches to Relieving Poverty," in Christopher Jencks and Paul Peterson, eds., The Urban Underclass (Washington, D.C.: Brookings Institution, 1991).

54 Paul Light, Artful Work: The Politics of Social Security Reform (New York: Random House, 1985).

55 Joseph White and Aaron Wildavsky, The Deficit and the Public Interest (Berkeley: University of California Press, 1989).

56 Fay Lomax Cook and Edith J. Barret, Support for the American Welfare State (New York: Columbia University Press, 1992); Robert Y. Shapiro and John T. Young, "Public Opinion and the Welfare State: The United States in Comparative Perspective," Political Science Quarterly 104 (Spring 1989). In line with my general argument, Shapiro and Young's research indicates similar patterns in other countries.

57 On recent spending trends, see House Ways and Means Committee, Background Material on Programs within the Jurisdiction of the Committee on Ways and Means (Washington, D.C.: GPO, March 1994).

58 Thelen (fn. 19).

59 Esping-Andersen (fn. 35, 1991 and 1992).

60 Combined employer and employee social insurance contributions were 26.5% of gross income in 1970, 32.4% in 1980, and are forecast to hit 39.2% in 1994. Financial Times, July 2, 1993, p. 13.

61 Hugh Heclo, "Generational Politics," in John L. Palmer, Timothy Smeeding, and Barbara Boyle Torrey, eds., The Vulnerable (Washington, D.C.: Urban Institute Press, 1988).

62 Esping-Andersen (fn. 35, 1992) rightly suggests that the development of private pension schemes could encourage such a polarization, since it would allow current workers to sever the link between their own retirement situation and that of the preceding generation. Yet the enormous institutional and political barriers to any radical change in a mature, pay-as-you-go pension system make a major development along these lines highly unlikely. See Pierson (fn. 31).

63 Karl Hinrichs, "Public Pensions and Demographic Change: Generational Equity in the United States and Germany," Centre for Social Policy Research Working Paper no. 16/93 (Bremen: University of Bremen, 1993); Winifried Schmähl, "Die Finanzierung der Rentenversicherung im vereinten Deutschland," Wirtschaftsdienst, no. 1 (1992).

64 Friedbert Rüb and Frank Nullmeier, "Altersiccherungspolitik in der Bundesrepublik," in Berhard Blanke and Hellmut Wollmann, eds., Die alte Bundesrepublik: Kontinuität und Wandel (Opladen: Westdeutscher Verlag, 1991).

65 Douglass Webber, "Krankheit, Geld und Politik: Zur Geschichte der Gesundheitsreform in Deutschland," Leviathan 16 (1988).

66 Jens Alber, "The West German Welfare State in Transition," in R. Morris, ed., Testing the Limits of Social Welfare (Hanover: Brandeis University Press, 1988); Claus Offe, "Smooth Consolidation in the West German Welfare State: Structural Change, Fiscal Policies, and Populist Politics," in Piven (fn. 35).

67 Offe (fn. 66), 140.

68 Jürgen Plaschke, "Die Sozialpolitik in den Monaten November, Dezember 1993 und ein sozialpolitischer Rückblick auf das Jahr 1993," Nachrichtendienst, no. 2 (1994).

69 Jens Alber, "The Debate over Long-Term Care Insurance in Germany" (Contribution to an OECD seminar on the Care of the Elderly, Paris, OECD, 1994); Ulrike Götting, Karin Haug, and Karl Hinrichs, "The Long Road to Long-Term Care Insurance in Germany: A Case Study in Welfare State Expansion" (Paper presented at the World Congress of Sociology, Bielefeld, July 1994).

70 Tyll Necker, president of the Association of German Industry (BDI), described the original proposal (which did not include the reduction of one paid holiday as an offset) as an official declaration of war against German industry. Alber (fn. 69), 17.

71 By 2040 pensions are expected to account for 61% of German social expenditure, compared with 40% in Britain and 44% in Sweden. OECD, Aging Populations: The Social Policy Implications (Paris: OECD, 1988).

72 A bourgeois coalition government was in power from 1976 to 1982 and, following the SAP's worst showing in decades in the "earthquake" election of 1991, from 1991 to 1994.

73 Esping Andersen (fn. 1); Esping-Andersen (fn. 35, 1992); see also Peter Swenson, "Labor and the Limits of the Welfare State," Comparative Politics 23 (July 1991).

74 Schwartz (fn. 15); Pontusson and Swenson (fn. 19).

75 OECD, Economic Survey: Sweden (Paris: OECD, 1994); Evelyne Huber and John Stephens,"The Future of the Social Democratic Welfare State" (Paper presented at the International Sociological Association Meeting, Oxford, September 1993), 7-11.

76 OECD (fn. 75), 36.

77 Ibid.

78 John D. Stephens, "The Scandinavian Welfare States: Development and Crisis" (Paper presented at the World Congress of Sociology, Bielefeld, Germany, July 1994).

79 Pontusson and Swenson (fn. 19).

80 Evelyne Huber and John Stephens, "The Swedish Welfare State at the Crossroads," Current Sweden, no. 394 (January 1993), 1-7; OECD (fn. 75).

81 Bo Rothstein, "The Crisis of the Swedish Social Democrats and the Future of the Universal Welfare State," Governance 6 (October 1993). Rothstein himself argues that weakening middle-class attachment to the welfare state may have contributed to the SAP's 1991 election defeat, but his article provides no real evidence for this claim and the polling data he presents suggest otherwise.

82 Huber and Stephens (fn. 80).

83 Stephens (fn. 78), 19. Ministry of Health and Social Affairs, Pension Reform in Sweden: A Short Summary (Stockholm: Cabinet Office, 1994).

84 Schwartz (fn. 15) argues that there has been major change in the four small states he studies: Sweden, Denmark, Australia, and New Zealand. His study focuses on the internal organization of public service provision, rather than on the level and quality of services actually provided, and it does not even discuss the transfer payments that account for the majority of welfare-state spending. Even on its own narrow terms, however, Schwartz's study provides remarkably little evidence that the changes he catalogs add up to radical reform rather than the continuous tinkering common in all modern public sectors. The evidence looks credible only for New Zealand, a tiny country on the periphery of the world economy, which clearly faced severe adjustment problems in light of its long (and unusual) tradition of protectionism. It seems far more reasonable to treat this case as an outlier than to view it as the pacesetter in a global march toward radical reform of the welfare state. See Stephens, Huber, and Ray (fn. 38).

85 Indeed, a cross-national comparison of unemployment programs provides further support for this analysis. The OECD has measured replacement rates for UI (benefits as a percentage of previous income) over time in twenty countries, with data through 1991. This data thus permit, for one program, a recent quantitative appraisal of program generosity rather than simply spending levels. In the majority of cases (twelve out of twenty), replacement rates were higher in 1991 than the average rate for either the 1970s or the 1980s, while most of the other cases experienced very marginal declines. Organization for Economic Cooperation and Development, The OECD Jobs Study: Facts, Analysis, Strategies (Paris: OECD, 1994), chart 16, p. 24.

86 See Paul David, "Clio and the Economics of QWERTY," American Economic Review 75 (May 1985); and W. Brian Arthur, "Competing Technologies, Increasing Returns, and Lock-In by Historical Events," Economic Journal 99 (March 1989), 116-31. For good extensions to political processes, see Stephen A. Krasner, "Sovereignty: An Institutional Perspective," in James A. Caporaso, ed., The Elusive State: International and Comparative Perspectives (Newbury Park, Calif.: Sage Publications, 1989); and North (fn. 20).

87 Thus in Germany, Sweden, and the United States the maturity of existing schemes limited policymakers to very gradual and incremental reforms of earnings-related pension systems. More dramatic reform was possible in Britain because the unfunded earnings-related scheme was far from maturity, having been passed only in 1975. Pierson (fn. 31).

88 For an example of this argument, see Garrett (fn. 41).

89 For an argument about how EC institutions may allow blame-avoiding behavior on the part of member state governments, see Andrew Moravcsik, "Why the EC Strengthens the State" (Manuscript, 1994).

90 Robert Kuttner has called this "the most fundamental principle in the political economy of social spending." Kuttner, "Reaganism, Liberalism, and the Democrats," in Sidney Blumenthal and Thomas Byrne Edsall, eds., The Reagan Legacy (New York: Pantheon, 1988), 113. For a critique, see Pierson (fn. 12), 6, 170.

91 See Pierson (fn. 12), 17-26, 169-75.

92 Sven Steinmo, Kathleen Thelen, and Frank Longstreth, eds., Structuring Politics: Historical Institutionalism in Comparative Analysis (Cambridge: Cambridge University Press, 1992).

.