Welfare-State Retrenchment Revisited:
Entitlement Cuts, Public Sector Restructuring, and Inegalitarian Trends in Advanced Capitalist Societies

Richard Clayton and Jonas Pontusson *

Tables

The goal of this article is to provide a balanced assessment of what has happened to the welfare state in advanced capitalist societies since 1980. What is one to make of recent spending cuts and welfare reforms? Do they represent a fundamental rollback of the welfare activities of the state and thus a belated victory for Thatcherite-Reaganite ideas? Or should these changes rather be seen as incremental adjustments of mature welfare states, proven to be more enduring than their critics?

Already in the 1970s many observers concluded that the welfare state had reached its outer limits and began to speak of a crisis of the welfare state. The rhetoric of crisis was inspired by the idea, shared by neo-Marxists and neoliberals, that the redistributive logic of the welfare state was contradicted by the logic of capitalism, and that the welfare activities of the state would have to be rolled back or reconfigured so as to conform to the needs of capitalism. With neoliberal ideas gaining ascendancy in both the United Kingdom and the United States at the onset of the 1980s, the fate of the welfare state appeared to be sealed. In retrospect, however, it is clear that Thatcher's and Reagan's achievements in the realm of welfare reform fell short of what their rhetoric promised. Against this background, the resilience of the welfare state has emerged as a prominent theme in the scholarly literature of the 1990s, with Paul Pierson's widely cited World Politics article of [End Page 67] 1996 providing, we think, the clearest and most compelling presentation of the case for welfare-state resilience. ¹ Using this article as a foil, we seek to examine some of the conventional wisdom in the literature and to sketch an alternative approach to the study of welfare states in transition.

Basing his argument on aggregate OECD statistics as well as on case studies of Germany, Sweden, the U.K., and the U.S., Pierson contends that welfare cutbacks and reforms have been strictly limited in scope. At the same time, he observes, "the power of organized labor and left parties has shrunk considerably in many advanced industrial societies." ² Together, these observations pose a challenge for the power-resource model developed by Walter Korpi and others to explain cross-national variations in welfare-state development. ³

The politics of welfare-state retrenchment appear to be fundamentally different from the politics of welfare-state expansion. Pierson's notion of a "new politics of the welfare state" yields three specific arguments to explain welfare-state resilience. First, "the welfare state now represents the status quo, with all the political advantages that this status confers." ⁴ Especially in countries where different institutions share power, radical reform is inherently difficult. Second, Pierson argues, welfare cutbacks tend to be associated with high electoral costs for the simple reason that basic welfare programs enjoy widespread popular legitimacy. Third, he attributes resilience to successful mobilization by well-organized groups representing the interests of consumers of welfare benefits (such as retirees), as well as employees of the welfare state. The combination of these factors yields a politics of blame avoidance in which cutbacks can take place only through incremental and surreptitious mechanisms or during moments of extraordinary fiscal stress and political consensus. [End Page 68]

The literature on welfare-state retrenchment raises the thorny question of how to distinguish radical change from incremental adjustments. For instance, writing about Sweden in the early 1990s, Pierson states that conditions were uniquely favorable to a "complete overhaul of social policy," but even so "there was no sign that the welfare state would be radically restructured." ⁵ Exactly what, then, would a "complete overhaul of social policy" or a "radical restructuring of the welfare state" entail? And should we not allow for some outcomes that are neither "incremental adjustments" nor "complete overhauls"? Without resurrecting the crisis rhetoric of the 1970s and its functionalist premises, the following analysis shows that major changes have indeed occurred in the scope and organization of public welfare provision not only in the U.K. and the U.S., but across the OECD area more generally.

The presentist concerns of the retrenchment literature exacerbate the conceptual problem of distinguishing incremental adjustments and radical change. Do cuts in social spending represent a long-term trend or simply a response to transitory macroeconomic conditions? Like most of the retrenchment literature to date, Pierson's discussion deals primarily with the experience of the 1980s. As we update his quantitative and qualitative evidence, we gain some analytical leverage on trajectories of change and find that the resilience thesis becomes less compelling.

Going beyond updating, this article seeks to broaden the discussion of welfare-state retrenchment and, at the same time, to promote a more careful consideration of measurement issues. Exemplified by Pierson's work, the existing literature tends to focus on the efforts by politicians to enact entitlement changes or, more precisely, on the significance of the entitlement changes that have been enacted. This way of thinking about welfare-state retrenchment is too narrow. First, recent cutbacks and welfare reforms must be situated in the context of rising social inequality and insecurity. Since the late 1970s the dynamics of advanced capitalism have been undoing some of the postwar achievements of welfare states. Increased welfare effort would have been required to maintain these achievements. Moreover, the rise of mass employment and the decline of employment opportunities for unskilled workers affect the way welfare states work, irrespective of whether governments cut or reform social programs. Even in the Scandinavian welfare states, celebrated for their universalism, the system of social insurance has remained closely tied to employment. Since about the mid-1980s the number of people who do not have access to these universalistic programs [End Page 69] and who must instead rely on means-tested social assistance has increased considerably in proportion to the total population.

We argue in a similar vein that measuring the size of the welfare state in terms of social spending as a percentage of GDP, as virtually all of the literature does, is problematic because such measures fail to take account of changes in societal welfare needs. The alternative measures that we propose show that the rapid growth of social spending in the 1960s and 1970s came to an end in the 1980s and that public services were more affected by the deceleration of growth than transfer programs. Measuring the welfare state in terms of the absolute size of the public sector labor force, we find quite a few instances of actual welfare-state shrinkage in recent years.

By and large, the retrenchment literature tends to ignore the question of changes in the delivery of social services or, in other words, the question of how the public sector is organized. While Pierson does discuss health care, most of the entitlement programs that he considers are based on transfer payments. At least in Sweden and the U.K., however, it is in the realm of public services that we find the most significant cutbacks and market-oriented reforms. Related to this, finally, we argue that summing up changes in individual social programs does not provide the basis for an adequate assessment of what has happened to welfare states since 1980. We must also consider changes in the overall configuration of welfare spending, that is, how the allocation of resources among individual programs might have changed. Thus we propose to explore not only the extent of welfare-state retrenchment, but also forms of welfare-state restructuring.

The importance that we assign to the public sector as a site of service production follows from Gösta Esping-Andersen's well-known and much-admired comparative analysis of welfare-state development. ⁶ As Esping-Andersen points out, the Scandinavian welfare states are distinguished by their reliance on the direct provision of services. Yet state-produced services constitute a crucial dimension of the public provision of social welfare in virtually all advanced capitalist societies these days. And, to the extent that it involves nonprofit production and allocation of output according to political criteria, it is this dimension of the welfare state that most directly contradicts the logic of capitalism. [End Page 70]

Population aging and international capital mobility constitute sources of pressure on contemporary welfare states across the OECD area. ⁷ Like most students of comparative welfare-state development, we believe that politics very much affects how these pressures play themselves out. In our conception of politics, however, societal interests play a more important role than they do in Pierson's politician-constituent frame. In Western Europe, at least, mass unemployment has become an institutionalized feature of labor markets, with long-term unemployment and early labor-force exit among unskilled workers coexisting with continued employment security for other segments of the labor force. While public opinion continues to favor core welfare programs, this situation has rendered the maintenance of welfare-state universalism politically more precarious. Also, the antiservice bias of the ongoing restructuring of the welfare state can be seen at least in part as an expression of the interests of export-oriented coalitions of private sector employers and labor.

Though we shall present some quantitative data for a larger set of OECD countries, our analysis will focus on the four countries discussed by Pierson. Although he does not explicitly discuss case selection, it seems clear that the cases were chosen as examples of Esping-Andersen's three types of welfare states, with Sweden exemplifying the social democratic type, Germany the conservative type, and the U.S. the liberal type. While the British welfare state does not fit neatly into Esping-Andersen's typology, the British case deserves our attention as a limiting case of neoliberal reform of the welfare state since 1980. Nowhere else has the neoliberal agenda been pursued more rigorously and under more favorable circumstances.

Growing Market Inequality

In this section we present evidence to support the propositions that inequality has increased and that security of employment and income has diminished for many wage earners in advanced capitalist societies since 1980. ⁸ The literature that emphasizes the resilience of the welfare state [End Page 71] tends to ignore these trends. Pierson and others seem to take the view that the growth of inequality and insecurity is relevant only to the extent that it is a direct result of spending cuts or reforms of the welfare state. In other words, they confine their discussion to the question of the extent to which the welfare state has become less redistributive or less effective in providing protection against market risks. This view fails to incorporate Esping-Andersen's crucial insight that the activities of the welfare state influence the way that labor markets operate. Moreover, the context of rising inequality and insecurity must be considered when we assess the significance of recent changes in the size and character of welfare states. For example, Swedish governments lowered the replacement rate of unemployment insurance from 90 percent to 75 percent in the first half of the 1990s. Had unemployment remained what it had been in the 1980s, these decisions might well have been described as a minor retrenchment of the welfare state. In the context of the dramatic increase of unemployment that occurred in the early 1990s, they take on a different significance.

It is commonplace to measure the distribution of income in terms of the ratio of income at the lower end of the 90th percentile (the lower end of the top 10 percent of income earners) to income at the upper end of the 10th percentile (the upper end of the bottom 10 percent). Referring to the earnings of full-time employees, Table 1 summarizes recent trends in 90-10 ratios in all OECD countries for which such data are available. For men and women combined, wage inequality increased sharply from the late 1970s to the mid-1990s in the U.K. as well as in the U.S. Most other countries experienced increases in the 1-7 percent range, but a handful of countries, most notably Germany, actually moved in the opposite direction. The trend toward increased inequality becomes more pronounced when we take gender differentials out of the picture and especially when we look at the distribution of earnings among men. The 90-10 ratio for men increased in all but two countries, Belgium and Germany. In the U.K and the U.S., it increased by more than a third, and Italy and New Zealand also registered double-digit percentage increases. In many countries, rising within-gender inequality has been offset by the continuation of the reduction of gender differentials that began in the 1960s or 1970s. ⁹

The figures in Table 1 capture only part of the tendency since 1980 for market forces to generate more inequality. Several other considerations [End Page 72] must be introduced to complete the picture. First, disparities of income from capital have undoubtedly reinforced the effects of these trends in the distribution of wage income. Second, the individual-level trends shown in Table 1 have likely been magnified by the pooling of wage income within families. For the U.S., Gary Burtless shows that the correlation between the incomes of spouses has increased very significantly (well-paid men being increasingly likely to be married to well-paid women) and that this development accounts for a large part of the growth of household inequality. ¹⁰

Third, the figures presented in Table 1 understate the rise of inequality because they are restricted to full-time employees. In fifteen out of nineteen OECD countries for which data are available, the incidence of part-time employment increased from 1983 to 1996 and in nine of these countries it increased by more than a third. ¹¹ As women [End Page 73] constitute the vast majority of part-time employees in all countries and part-time employees earn less than full-time employees on an hourly basis, the proposition that pay differentials based on gender continued to decline through the 1980s may have to be qualified in light of the growth of part-time employment. Finally, data on the distribution of income from employment fail to capture the impact of unemployment. Because unemployment tends to be concentrated among unskilled, low-paid workers, it correlates negatively with wage inequality as measured in Table 1 . ¹² As the rate of unemployment increases, low-paid workers disappear from the population used to calculate 90-10 ratios and the wage distribution becomes more compressed, but we certainly would not want to conclude from this that unemployment promotes social equality.

Unskilled workers are more likely to become unemployed than more skilled workers, and their spells of unemployment tend to be longer than those of more skilled workers. As noted by Andrew Glyn, educational disparities in labor-force participation have also become more pronounced since the late 1970s: unable to find jobs, many older unskilled workers have simply dropped out of the labor force. To capture the combined effects of employment and earnings disparities, Glyn first calculates the ratio of the employment rate for male wage earners with higher education qualifications (college graduates in American terminology) to the employment rate of male wage earners without an upper-secondary diploma (high school dropouts) and then multiplies this figure by the ratio of the average earnings of the former group to those of the latter group. With both employment and earnings differentials moving against unskilled workers, the British score on this index of educational disparities in income from employment (EDDIE) increased by 27 percent from 1979 to 1991, and the U.S. score increased by the same figure from 1979 to 1991. In Germany between 1978 and 1987 and Sweden between 1987 and 1993, relative earnings trends favoring the least educated were more than offset by relative employment trends favoring the most educated and, in each case, the EDDIE index increased by 3 percent over the time period in question. ¹³

The incidence of poverty provides another obvious indicator of social inequality and insecurity. One common measure of poverty is the percentage [End Page 74] of the population living in households with an income of less than 40 percent of the median household income. Using this definition and drawing on data from the Luxembourg Income Study, Table 2 presents estimates of the incidence of poverty before as well as after taxes and government transfers for Sweden, Germany, the U.K., and the U.S. around 1980 and 1991. In each of these countries poverty measured in terms of the distribution of "market income" (that is, the distribution of income before taxes and transfers) increased noticeably over this relatively brief period of time. The fact that the percentage of the population receiving some form of means-tested social assistance increased in fifteen out of eighteen OECD countries from 1980 to 1992 suggests that Table 2 captures a general trend, reversing the prior trend toward a reduction of poverty. ¹⁴ The pervasiveness of recent inegalitarian trends is indeed striking, especially in view of the strong tendency among students of comparative political economy to emphasize national [End Page 75] diversity. We hasten to add that common trends do not necessarily add up to cross-national convergence, for convergence requires that the most egalitarian countries experience the most rapid growth of wage inequality, poverty, and so on, and this does not appear to have been the case.

Alongside the factors typically cited by labor economists (structural unemployment, immigration, trade with low-wage countries, technology-driven shifts in demand for labor, and the slower growth of higher education), changes in labor-market institutions (deunionization and employer-initiated decentralization of wage bargaining) have contributed to the growth of wage inequality since 1980. ¹⁵ Later in this article, we will suggest that public sector restructuring has also played an important role. For now, suffice it to say that while the redistributive effects of taxation and welfare spending were broadly consistent with labor-market trends in the 1960s, labor-market conditions changed profoundly in the wake of the international recessions of the 1970s. To maintain the disposable income distribution that had been achieved by the late 1970s, a significant expansion of redistributive welfare state activities would have had to occur in the 1980s.

The Employment Rate and the Welfare State

Before we proceed to a discussion of social policy changes, let us briefly expand on the point that employment conditions affect how welfare states work. The Swedish case illustrates this point most clearly. ¹⁶ From a comparative perspective, Esping-Andersen is undoubtedly correct to emphasize the universalism of the Swedish welfare state, but the benefits provided by many of Sweden's social insurance programs are in fact tied to employment. Three of the major programs are truly universalistic in the sense that they are available to every citizen (or resident): family allowances, health care, and the basic, flat-rate pension. By contrast, sick pay and parental leave insurance represent government spending on social welfare that, as a matter of definition, only benefits people who are employed. In effect, though not by law, the same goes for subsidized public child care. Finally, entitlements and benefit levels in the supplementary pension (ATP) system, which accounted for 60 [End Page 76] percent of pension benefits paid out by the government in 1994, are determined by years of employment and income from employment.

In large measure, the universalism that distinguished the Swedish welfare state in the 1970s and 1980s derived from the universalism of employment in Sweden. While men's labor-force participation remained constant, with employed men representing 88.4 percent of the male population between the ages of fifteen and sixty-four in 1990, women's labor-force participation increased from 59.4 percent in 1970 to 83.2 percent in 1990. With unemployment at less than 2 percent, the combined employment rate for women and men stood at 84.4 percent in 1990 (as compared with 67.5 percent for the OECD as a whole). While unemployment rose sharply in the first half of the 1990s to reach 10 percent in 1996, the labor force participation of men and women alike fell precipitously. By 1996 the overall employment rate had fallen to 72.7 percent (as compared with 66.5 percent for the OECD as a whole). ¹⁷ In the first half of the 1990s Sweden's nonworking population of working age nearly doubled; this represents a major increase in the number of people with only limited access to universalistic social programs. Badly in need of public support, many of the people who are no longer employed have ended up on some form of means-tested social assistance. Without any policy change, the employment crisis has thus shifted the balance between "universalism" and "residualism" within the Swedish welfare state.

Sweden's employment rate remains well above the OECD average. What is unique about the Swedish experience is the fact that the advent of mass unemployment was delayed for so long and then occurred so abruptly. As welfare benefits are even more closely tied to employment in other countries, the argument about the implications of mass unemployment for the public provision of social welfare sketched here surely applies more generally.

Measuring Welfare Effort

For the four countries which he surveys, Pierson presents aggregate OECD statistics on the evolution of social security transfers as a percentage of GDP, total government outlays as a percentage of GDP, and government employment as a percentage of total employment. In Pierson's words, the quantitative data show "a surprisingly high level of continuity [End Page 77] and stability," and "for none of the countries does the evidence reveal a sharp curtailment of the public sector." ¹⁸ This is indeed a judicious assessment of the data presented, but the data are problematic on several counts.

While two of Pierson's time series end in 1990 and the other ends in 1992, "total government outlays" is obviously too broad and "social security transfers" too narrow a measure of the size of the welfare state. More importantly, government spending as a percentage of GDP provides a useful measure of cross-national differences at any point in time, but the GDP denominator of this measure makes it difficult to interpret change over time. In the time series presented by Pierson, two things are changing--the amount of money spent by the government and the size of GDP--and it is impossible to separate one from the other. Still more importantly, Pierson's quantitative measures do not take socioeconomic and demographic changes into account. At any given level of entitlement provisions, an increase in the number of unemployed, poor, or retired people automatically generates increased social spending by the government. Indeed, increased spending might be associated with a reduction of entitlements. To use government spending as a proxy for "welfare effort," we must somehow control for these variables.

Table 3 is the upshot of an alternative approach to the problem of measuring welfare effort or, in a sense, the size of the welfare state. A [End Page 78] recently available OECD data set provides year-to-year observations of total social spending as a percentage of GDP, enabling us to avoid the unhappy choice between total outlays and social security transfers. Using these figures and the amount of GDP, expressed in U.S. dollars at 1990 prices and exchange rates, we can compute the amount of total social spending at constant prices. The poverty rates reported in Table 2 in turn allow us to divide this figure by the number of people living in households with a market income of less than 40 percent of the median household income. Though the figures for total social spending as a percentage of GDP are available for the entire period 1960-93, the availability of poverty data restricts the time period over which we can observe change in this measure. Table 3 reports the annual growth rate of real social spending per poor person in the 1980s for Sweden, Germany, the U.K., and the U.S., and compares these figures with the annual growth rate of real GDP per capita. In all four countries the growth of real social spending per poor person failed to keep up with the growth of GDP per capita, and, except for the U.K., the differential between these growth rates was quite considerable.

Table 4 reports the results of replicating the exercise described above with a denominator made up of the sum of the number of unemployed and the number of people above sixty-four years of age for 1960, 1980, and 1993. In all four countries, the growth rate of real social spending measured in this fashion was much lower in 1980-93 than in 1960-80. In the U.S. the growth rate in the latter period was just about half of the growth rate in the former period, and in Sweden and Germany it was less than a quarter of the growth rate in the former period. As in Table 3, the U.K. stands out as the country in which real social spending per person has grown most rapidly since 1980. In view of the anti-welfare [End Page 79] rhetoric of Mrs. Thatcher and her governments, this is surely a striking finding. Two observations regarding this puzzle should be made. First, real social spending per poor person (and per unemployed or aged person) was lower in the U.K. than in the other three countries at the end as well as the beginning of the 1980s. Second, the figures in Table 2 show that the ratio of disposable-income poverty to market-income poverty increased sharply in the U.K. in the 1980s, despite the growth of real social spending per poor person. ¹⁹ More so than in any of the other three countries, the effectiveness of social spending in combating poverty appears to have declined in the U.K. The most obvious explanation would be that an increasing share of social spending has been allocated to people who are not poor.

Using data reported by Evelyne Huber and John Stephens in a recent paper, Table 5 illustrates another way to measure changes in the size of the welfare state relative to societal needs. The measure here is public health spending as a percentage of total health spending. From 1980 to 1993 this figure fell in ten of the seventeen OECD countries for which data are available and essentially remained unchanged in another three countries. Even in those countries where the government's share of total health spending continued to increase, the rate of increase was, on an annual basis, much slower in this period than it had been in the previous two decades. One might perhaps argue that the growth of private health spending is primarily a matter of satisfying "frivolous" health care needs, such as cosmetic surgery, and that public programs continue to provide adequately for "basic needs." But waiting lines have become longer and the quality of health services provided by the public sector has deteriorated in at least some countries. As health care needs change, moreover, it is far from obvious where to draw the line between frivolous and basic services.

The Size of the Public Sector

Table 6 summarizes the real growth of final consumption expenditure (that is, the costs of goods and services produced by the public sector) in seventeen OECD countries over the period 1960-94. Like total government outlays, this measure encompasses a range of government activities that have little or nothing to do with the provision of social welfare, but it speaks more directly to the size of the public sector. As [End Page 80] noted earlier, services provided by the government represent an important component of the welfare state not only in Scandinavia but across the OECD countries. To conceive this component simply in terms of health care and social services in the narrow sense of the term (as the OECD does for the purpose of computing total social spending) seems unduly narrow. Child care, education, retraining programs, and a great many other services promote social welfare in general, and at least some of these services benefit low-income groups in particular.

In all but two countries the average growth rate of real government final consumption was lower in 1973-79 than in 1960-73 and, again, in all but two countries, the average rate in 1979-89 was lower than in 1973-79. In a handful of countries, the growth of government final consumption rose in the early 1990s, but all of these countries had comparatively low growth rates in the 1980s. For every single country, the growth rate of 1989-94 was lower than that of 1973-79 and of [End Page 81] 1960-73. It would appear that as overall government spending has slowed down while the costs of social assistance and social security entitlements have continued to grow, the service components of the welfare state have been squeezed.

The size of the public sector might also be measured in terms of employment, but we must beware of the denominator problem. From 1990 to 1994 the public-sector labor force in Sweden declined by nearly 12 percent. As total employment declined even more, however, government employment as a percentage of total employment actually increased slightly. To avoid this problem, Table 7 tracks the evolution of the public sector labor force, measured in absolute terms (people) rather than in relative terms (percentage of total employment). For nine of these countries the continuous deceleration story of Table 6 becomes a story of outright shrinkage of the public sector. Most remarkably, the size of the public sector labor force declined by nearly 30 percent in the U.K. from 1988 to 1994. ²⁰ [End Page 82]

Given that only some government employees are engaged in welfare-related activities, to what extent do the employment cuts documented in Table 7 pertain to the public provision of social welfare? In Sweden employment in public health and hospitals fell by 7 percent, employment in care for children and the elderly fell by 10 percent, and employment in education and defense each fell by 4 percent from 1990 to 1993. ²¹ As Table 8 shows, the shrinkage of the public sector labor force has been highly differentiated in the British case. The largest employment cuts by far have been sustained by central government agencies other than the National Health Service (NHS) and the military, and the number of people employed by local government authorities in social services actually increased from 1985 to 1994 (along with the number of people employed in the police forces). All of this might be taken as evidence that overall employment data overstate the extent of welfare-state shrinkage. The category "other central government," [End Page 83] however, encompasses welfare-related activities, and employment in both education and the NHS did shrink significantly over this period.

It may be that public sector productivity growth, notoriously difficult to measure, has made it possible to provide the same level of services with fewer employees, but it is hard to believe that labor-force reductions on the scale observed in Australia, Finland, and Sweden, let alone in the U.K., do not translate into less public-welfare provision. And, again, increased societal needs must be taken into account. In view of the societal changes discussed earlier, the figures presented in Tables 6-7 would appear to represent a rather broad-based reduction of service-based welfare effort. Be that as it may, these tables call into question the way that Pierson and other proponents of the welfare-state resilience thesis seem to conflate continuity with stability. Except in a couple of instances, final government expenditure on consumption has continued to grow in real terms, and thus we might say spending patterns have been stable, but if we compare growth rates in the 1980s and 1990s with those of the 1970s and 1960s, discontinuity is a prominent feature of Table 6.

The deceleration of government spending growth might be viewed as a result of the maturation of welfare states, that is, a result of the fact that they now provide for basic needs and have perhaps also reached the [End Page 84] limits of politically acceptable taxation. ²² Clearly, the growth of government spending as a percentage of GDP must inevitably slow down over time. However, there is hardly any correlation, on a cross-national basis, between annual growth of the public sector labor force in 1989-94 and 1990 levels of government employment, measured as a percentage of total employment (r=-.142). The correlation of annual growth rates of final government consumption in 1979-89 and 1985 levels is also weak (r=-.226). It is not the case that countries with large public sectors have experienced more deceleration (or greater public sector cutbacks) than other countries.

Assessing Entitlement Changes

In this section we review and update Pierson's four case studies. We argue that Pierson's own account of British social policy developments belies his claim that the Thatcher governments of the 1980s failed to achieve a significant rollback of the welfare state. For the 1980s the other cases support Pierson's emphasis on welfare-state resilience. However, the benefits provided by social-assistance programs deteriorated steadily in the U.S., and more recently these programs have been dismantled or profoundly restructured. In the Swedish case and, to a lesser extent, the German case as well, the politics of austerity in the 1990s has been accompanied by significant benefit cutbacks and programmatic reforms.

Pierson's account of welfare-state retrenchment in the U.K. focuses on three primary issue-areas--pensions, income support, and housing --but it also encompasses two supplemental issue-areas--sickness/
disability support and health care. Summarizing his analysis, Pierson characterizes retrenchment in pensions and housing as "high," retrenchment in income-support programs and health care as "low," and retrenchment in sickness/disability benefits as "low/moderate." ²³ Thus the Thatcher regime apparently achieved significant retrenchment in only two of the five areas of welfare-state provision on which Pierson reports. However, the criteria behind this coding of retrenchment outcomes are not spelled out very clearly, and the coding strikes us as dubious in several instances.

In particular, Pierson's characterization of retrenchment in British income support seems to have little to do with the actual level of benefits across the types of transfers that he covers. Pierson acknowledges [End Page 85] that the income replacement provided by public unemployment insurance was sharply reduced, that eligibility for unemployment benefits was restricted, and that that the real value of the universal child benefit fell by 14 percent in the 1980s. ²⁴ The expansion of the means-tested Family Credit, he suggests, offsets these cuts. However, in the context of mass unemployment, it seems reasonable to discount the impact of tax credits in offsetting cuts in support for children, the unemployed, and the poor. In any case, the expansion of the Family Credit allowance clearly constitutes a shift of resources toward means-tested welfare provision.

Pierson correctly points out that public health care spending has steadily increased in real terms since 1981 and that the 1990 reform of the NHS preserved publicly financed care, but there is a lot more to the story of British health care. Most informed observers argue that, despite real growth, NHS spending fell well short of increases in demand for health care in the 1980s, resulting in substantial shortages of care in some parts of the country. While fees and charges in the NHS increased from 1.9 percent of total spending in 1979 to 3.2 percent in 1994, the role of private enterprise within the health care sector increased considerably under Thatcher and Major. This combination of underfunding, increased fees, distorted priorities (for example, a huge increase of managerial staff within the NHS), and creeping privatization warrants a less sanguine assessment than Pierson offers.

Turning to the U.S. experience, the absence of any significant cuts in social security and Medicare benefits lends support to the resilience thesis, but Pierson's characterization of income support as an instance of low retrenchment is again questionable. Jens Alber's analysis shows that the maximum food stamps benefit stagnated in real terms and fell in relation to earnings during the Reagan and Bush years. Alber also shows that the percentage of poor households receiving AFDC benefits fell sharply in the early 1980s and never regained its previous levels, while the maximum benefit for such a family of three fell from 67 percent to 36 percent of the poverty level. ²⁵

The drive to reduce antipoverty spending in the U.S. was capped by the welfare reform bill signed by President Clinton in July 1996. This bill cut food stamp benefits by $24 billion, denied immigrants eligibility for food stamps, restricted social security eligibility for disabled immigrants and children, and cut federal funding for social services and [End Page 86] child nutrition programs. Most importantly, however, it replaced AFDC with block grants to the states, with no minimum requirements pertaining to either eligibility or benefit levels, but mandating a five-year lifetime limit on receipt of benefits and requiring 50 percent of each state's caseload to be working for benefits by 2002. Though cushioned by economic growth, the distributive consequences of this reform are still considerable. At the time that the reform was passed, the Urban Institute estimated that it would result in some income loss for 11 million families, and move 2.6 million people, including 1.1 million children, into poverty. ²⁶

While the resilience of more universalistic government programs (middle-class entitlements) remains a conspicuous feature of the American case, changes in the realm of employer-organized welfare schemes also deserve to be noted here. According to Esping-Andersen, "Private coverage in health and pensions has declined steadily during the 1980s, particularly among young and low-wage workers." ²⁷

In contrast to the U.S., the benefit cuts introduced by Swedish governments since 1990 have been spread more or less evenly across the entire range of entitlement programs. Through a series of piecemeal changes, implemented by both social democratic and nonsocialist governments, the earnings replacement provided by sick pay, parental leave, and unemployment insurance has been reduced from 90 percent to 80 percent and waiting days have been introduced for unemployment benefits as well as sick pay. ²⁸ While the real value of pension benefits has been cut through changes in indexation, the qualifying conditions for early retirement have been tightened. The general child allowance, supplementary allowances for families with more than one child, and need-based housing subsidies have also been cut on the order of 10-15 percent.

As of 1994 the funding of public health care in Sweden has been partially shifted to employee-paid payroll taxes. At the same time, fees charged for doctor's visits and copayments for medications have increased sharply over the last ten years. Also, the sick pay reform of 1991 shifted responsibility for the first fourteen days of sick pay to employers, [End Page 87] creating an incentive for employers to reduce absenteeism by improving the workplace environment and/or by monitoring employees to prevent abuse of sick-pay provisions. Reinforcing workers' dependence on their employers, this reform must be considered a step in the direction of "recommodification." Most importantly, the principles of pension reform agreed to by the major political parties provide for shifting the financing of pension benefits toward employee contributions (the current system being based entirely on employer contributions) and the introduction of privately managed individual retirements accounts (to receive roughly 10 percent of total contributions).

Of the four countries surveyed by Pierson, Germany provides the strongest and most consistent support for his emphasis on welfare-state resilience. Like the Swedish welfare state, the German welfare state remained intact through the 1980s. As Alber shows, the real value of both pension benefits and social assistance grants fell in the late 1970s and early 1980s; yet each program subsequently recovered lost ground and, in marked contrast to U.S. income-support programs, benefit levels grew in line with average earnings from the mid-1980s through the mid-1990s. ²⁹ The process of German unification initially served to boost social spending, but its long-term fiscal consequences, combined with the Maastricht criteria for monetary union, precipitated significant cuts in welfare entitlements in 1994-96. ³⁰

Summarizing some of these entitlement changes, Figure 1 plots the evolution of the "social wage" from the early 1960s through the mid-1990s in the four countries. As operationalized in this figure, based on OECD data, the "social wage" refers to the gross income replacement provided by the welfare state to an unemployed person over five years, and encompasses social assistance and guaranteed-income schemes, as well as unemployment insurance benefits. ³¹ In all four countries the five-year replacement rate increased in response to initial employment problems in the 1970s. In the Swedish case this upward movement continued through most of the 1980s, when the five-year replacement rate reached 30 percent, but a decline of several percentage points occurred in the first half of the 1990s. In Germany the five-year replacement declined steadily from about 30 percent in 1979 to less than 27 [End Page 88] percent in 1995. In the U.S. the rate fell from an all-time high of 15.5 percent in 1977 to 11 percent in 1987, and has subsequently remained more or less constant. The long and precipitous decline of the social wage in the British case, from more than 27 percent in the late 1960s to about 18 percent in the mid-1990s, represents the most outstanding feature of Figure 1. [End Page 89]

The Changing Composition of Social Spending

From the analytical perspective of the welfare-state literature, summing up changes in individual social programs hardly provides an adequate basis for assessing the extent to which welfare states have changed, for the central concern of this literature is with relations among social programs, in other words, with the overall configuration of welfare states. One interpretation of Esping-Andersen's work holds that all welfare states consist of three basic components--a universalistic component providing benefits as a matter of citizen rights, a social-insurance component linking benefits to employment, and a means-tested social-assistance component --and that the types of welfare state are essentially distinguished from each other by the relative weight that they assign to these three components. The question becomes whether there have been significant shifts in the mix of welfare-state components over the last fifteen or so years. Showing the distribution of total social spending by type of spending in 1980 and 1993, Table 9 represents a first stab at this question.

In the Swedish and British cases alike, we observe two important changes: first, a shift of social spending from services to transfer payments and, second, a shift of spending on transfers from social insurance schemes to social assistance. The former shift is particularly pronounced in the Swedish case, and the latter is most pronounced in the British case. In the case of the U.S., we observe that the relative importance of public health spending rose at the expense of social insurance, suggesting that rising health care costs have been a major factor behind the continued growth of social spending. In a European context the increased relative importance of health spending would represent an increase in the service intensity of the welfare state, but in the U.S., of course, public health spending primarily takes the form of transfer payments. Consistent with our earlier discussion, Germany stands out in Table 9 as a case of remarkable stability.

The spending figures in Table 9 do not fit Esping-Andersen's conceptual categories perfectly. While transfer payments may be provided on a universalistic basis rather than being tied to employment, services may be provided on a means-tested basis. Recognizing the limitations of these data, Table 6 lends at least some support to the idea that the Swedish welfare state has become institutionally more like the German welfare state and that the British welfare state has become more like the American welfare state. ³² Based on this limited sample of OECD countries, [End Page 90] it would appear that universalistic service-based welfare states have undergone more far-reaching changes than social-insurance or residualist welfare states. Yet the trajectory of change in the Swedish and British cases is clearly different, suggesting that partisan politics still matters.

Reforms of the Public Sector

As noted earlier, government spending on social security transfers has tended to grow more rapidly than final consumption spending in most countries since the late 1970s. Across the OECD area we also observe important changes in the organization of the public sector and how it has delivered services to its "customers" over the last decade or so. In this section we briefly review public sector reforms in Sweden and the U.K. and consider their distributive implications. We are interested both in how these reforms affect the quality of welfare services provided by the state and access to services for different categories of the population and in what they imply for employment conditions and wage distribution in the public sector.

It should be noted at the outset that public sector reforms have not specifically targeted welfare activities in either country. Rather, organizational changes within health care, elderly care, social work, and so on have been part of a broader process of restructuring the public sector. [End Page 91] In this area, as in many others, it is difficult to discuss the welfare state in isolation, as if it were a detachable appendix of the modern state.

In the early 1980s the Swedish social democrats established a new ministry to oversee public sector reform. Originally oriented toward promoting democratic participation, public sector reform has increasingly emphasized cost reduction. Increased autonomy for administrative agencies and local governments has been accompanied by new forms of accountability to the central government, which has increasingly stipulated desired outputs and budgetary constraints while leaving it to managers to determine how to deploy the organization's resources. To encourage long-term improvements in efficiency, the government also began to allow unit managers to retain a share of user fees in the 1980s. Exemplifying a general trend among OECD countries, the administrative mechanisms of the Swedish welfare state increasingly mimic those of private corporations. ³³

While a number of state agencies were transformed into state-owned corporations in 1985-95 and thus moved off the government budget altogether, many state-owned corporations have been at least partially privatized. As a result of rationalization as well as privatization, employment in the Swedish state enterprise sector fell from 330,000 in 1980 to 210,000 in 1994. ³⁴ The social democratic approach to privatization in the 1980s clearly excluded welfare-related services, but during the bourgeois coalition government of 1991, a number of new legislative enactments promoted private alternatives to state-provided services, and these measures were not reversed when the social democrats returned to power in 1994. State subsidies are now available on a restricted basis for private child care and education, and the so-called house doctor system has introduced an element of private practice into health care. From 1988 to 1994 private employment increased from 1 percent to 5 percent in child care, and from 6 percent to 7 percent in health care. ³⁵

In comparison with the British case, Swedish privatization has been very limited in its scope. In the U.K. the labor force of nationalized industry fell from 1.8 million in 1979 to less than half a million in 1997. ³⁶ [End Page 92] In contrast to Sweden, the effort to shrink and marketize the state led Britain's conservative governments to increase central government control over local governments and other units. At the same time the civil service has been subject to a series of reforms intended to reduce staffing levels, to increase managerial flexibility, and to make departmental cost centers more accountable for financial performance. Of more immediate relevance to our present concerns, the Tories introduced market principles into the National Health Service and the state education system for England and Wales in the late 1980s. In both areas authority for budgetary and personnel decisions was devolved to unit-level managers. NHS hospitals were converted to trusts and now have the status of public corporations, with the ability to borrow independently for capital expenditure and to retain operating surpluses. General practitioners may now opt out of the NHS and establish a fund-holding practice to which the central government allocates a budget. Fund-holding gps, hospital trusts, and health authorities are expected to contract independently for the services they require. ³⁷ Schools have also been given the possibility of opting out of local government control to become grant-maintained institutions, receiving funding directly from the central government. While school principals have gained responsibility for personnel and budgetary decisions, parents have gained the right to choose schools for their children. School and hospital budgets are now largely determined by the number of students and patients they attract, increasing the incentives for unit managers to economize on costs.

The Tories also sought to increase private provision of education and health services by subsidizing tuition for private school and premiums for private health insurance. From 1979 to 1991 the number of private health insurance policyholders increased from one million to three million. ³⁸ By compelling unit-level managers to subject stipulated contracts to an open-bidding procedure referred to as compulsory competitive tendering, the government encouraged private competition in hospital cleaning, catering, and laundry services, as well as in local government services. From 1979 to 1994 private employment increased from 15.9 percent to 39.6 percent of total health-related employment. ³⁹

In Sweden and the U.K. alike, public sector restructuring has involved a sustained effort to decentralize wage-setting mechanisms within the public sector so as to allow unit managers to respond more [End Page 93] effectively to local labor-market conditions and to incorporate productivity-enhancing incentives into their pay systems. ⁴⁰ The significance of the decentralization of wage setting within the public sector should be seen against the background of public sector wage bargaining in both countries taking on a particularly solidaristic cast from the late 1960s onward. With public sector wage compression putting pressure on private employers to raise the relative wages of unskilled workers, especially women, the expansion of public sector employment contributed to the decline of overall wage inequality in this period. ⁴¹

As Table 10 indicates, aggregate wage dispersion remained more or less constant in the British public sector, while it rose steadily in the private sector from 1984 to 1995. However, significant dispersion occurred among both men and women working in the public sector. In the course of the 1980s the dynamics of public sector wage setting became distinctly less solidaristic, and there is every reason to believe that this change is, in part, attributable to market-oriented reforms. At the same time, of course, public sector employment cutbacks and privatization have served to shift labor from the public sector (with more compressed wage differentials) to the private sector (with less compressed wage differentials). At least in the British case, not only has the welfare state failed to keep up with rising market inequality in recent years, but welfare-state restructuring has itself been a source of rising market inequality.

The question of how market-oriented reforms of the public sector have affected quality of services and equality of access is of central importance, but there is very little systematic evidence on this score and it is virtually impossible to separate the effects of marketization from the effects of cutbacks. While consumer choice has been the focus of much of the rhetoric in support of public sector reforms, critics of Thatcherism argue that marketization has undermined universal access to high-quality services by generating both regional and status tiers within the welfare state. The fact that an increasing number of middle-class Britons opt for private alternatives over free public services is indicative of the deterioration in quality and, at the same time, signals a source of inequality. According to the well-respected King's Fund Institute, patients of fund-holding general practitioners routinely receive preferential [End Page 94] treatment in the NHS system. ⁴² In a similar vein, Swedish research shows that the reorganization of elderly care "has affected working class elderly much more than it has others." ⁴³

Conclusion

Though the willingness of citizens to pay taxes obviously varies with the political circumstances, there is surely something to the idea that the overall level of taxation has reached its upper limits in many OECD countries. At the same time the combination of sluggish productivity growth and mass unemployment since the late 1970s has meant that government revenues at a given rate of taxation have grown more slowly than they did in the 1950s and 1960s, and the internationalization of financial markets has constrained the ability of government to engage in long-term deficit spending. Together, these factors have put downward pressure on overall government spending.

The fact that the British and Swedish welfare states have become less service-oriented can partly be explained in terms of demographic changes and the maturation of social insurance systems (the extension of full pension benefits to all retired people). In these and other OECD countries final government consumption expenditure continued to grow (in real terms) through the 1980s; it simply grew less rapidly than spending on social security transfers. However, it would also appear to [End Page 95] be the case that governments have preferred cutting the public sector to cutting entitlement programs, and it is first and foremost the service components of welfare states that have been reformed according to market principles. This pattern of retrenchment and restructuring does not seem to accord well with Pierson's assessment of the political risks entailed when politicians challenge entrenched interests, following from the common public choice argument that concentrated interests generally prevail over diffuse interests, for public sector employees constitute the entrenched pro-welfare constituency par excellence.

What, then, accounts for the antiservice bias of welfare-state retrenchment? Several arguments seem plausible. Consistent with Pierson's emphasis on the politics of blame avoidance, one might argue that the effects of cutting the public sector are less immediate and less tangible (or less visible) than the effects of cutting entitlement programs. Public sector cutbacks will likely result in a deterioration in quality and this in turn might result in middle-class opt out, but such deterioration will not necessarily be proportionate to spending cuts and no one knows at which point middle-class opt out becomes a serious problem. Even social democratic politicians are likely to find the risks involved here more palatable than those entailed in cutting pension or unemployment benefits.

Second, the popular legitimacy of programs based on the social insurance principle might be invoked to explain the antiservice bias of recent cutbacks. As Esping-Andersen argues with reference to the welfare states of continental Europe, the consensual manner in which such programs were developed and the sense of entitlement that the insurance system produces make it very difficult to reform these transfer programs. ⁴⁴ Third, the preference for social-insurance welfarism may reflect anxieties about further European integration, insofar as EU legislation outlaws discrimination on the basis of nationality. Since social-insurance benefits are typically based on income from employment, such programs sidestep the political problem of foreigners taking advantage of generous benefits. ⁴⁵

Most importantly perhaps, the patterns of retrenchment we have documented might be seen as a response to political pressure from a cross-class coalition of employers and workers in the export and multinational sectors. Both Peter Swenson and Herman Schwartz argue that [End Page 96] with increased openness and intensified international competition, workers and employers in exposed sectors become acutely concerned with containing the upward pressure on domestic costs generated by large public sectors. ⁴⁶ In this context a new political-economic cleavage between sheltered and exposed sectors opens up and the exposed sector coalition exerts increasing pressure for public sector reform. It is important to recognize that this coalition is based on compromise among its constituent units, rather than on a complete convergence of interests. Left to their own devices, export-oriented employers would probably have favored across-the-board cuts in the welfare state, but the maintenance of basic social insurance entitlements is a condition for private sector unions to support public sector cutbacks and reforms.

This coalitional account works better for some countries than for others. In the British case electoral support for Thatcherism certainly had a broad cross-class character, but Mrs. Thatcher's efforts to restructure the public sector did not involve the cooperation of private sector unions, and business support came mainly from the financial sector rather than from export-oriented manufacturers. The British case would have to be couched in terms of a cross-class coalition centered on financial interests, including the financial interests of working-class homeowners and shareholders, and forged electorally rather than organizationally.

Thinking about public sector restructuring in these terms, it should be noted that market-oriented reforms have had different consequences for different segments of the public sector labor force. In the British case the ratio of median public sector to private sector wages for blue-collar workers fell from 1.05 in 1984 to .98 in 1995, but the public-private ratio for white-collar workers remained stable at 1.09-1.10. ⁴⁷ This divergence of fortunes helps explain why public sector unions were not able to mobilize more effective resistance to market-oriented reforms.

In his analysis of Thatcherism and Reaganism, Pierson distinguishes between "programmatic" and "systemic" retrenchment. While programmatic retrenchment refers to reductions in spending or benefit levels, the introduction of means testing, and so on, systemic retrenchment entails long-term changes in the political environment that make the welfare state vulnerable to future attacks. Largely on the basis of public-opinion polls, which show that popular support for core components of the welfare state remains strong, Pierson argues that further, more [End Page 97] drastic welfare-state curtailments are unlikely in the British case. ⁴⁸ However, the contrast between the Labour Party's dismal electoral performance in the 1980s and early 1990s, when it stood stoutly by traditional welfarism, and its recent electoral triumph as a revamped party committed to fiscal rectitude and low taxes suggests that the meaning of public opinion polls may be more ambiguous than Pierson allows. From the perspective adopted here, it is striking that Pierson's discussion of systemic retrenchment in the U.K. does not consider the impact of Thatcherite policies on public sector unions. In the 1960s and 1970s unionized public sector employees emerged as a key political constituency for the Labour Party and the welfare state. The cutbacks and restructuring described above have clearly weakened the market power of public sector unions and their ability to mobilize politically. In some instances, market-oriented reforms have also created incentives for them to eschew "political unionism" in favor of a more economistic or professional orientation. ⁴⁹

As public sector employees constitute a key constituency of parties on the left in most countries, the idea that public sector reform represents a form of systemic retrenchment as well as programmatic retrenchment would seem to be more broadly applicable. Despite the absence of politicians attacking the welfare state in Thatcher fashion, political and economic conditions in the second half of the 1990s are surely more favorable to welfare-state cutbacks and restructuring than they were in the second half of the 1980s. In Western Europe this is also a result in some measure of the constraints imposed by the Maastricht criteria for monetary union. Whether the current contractionary environment represents a transitional phase or a more permanent condition remains an open question, with important implications for the future of the welfare state.

Whereas Pierson essentially accepts the class-power model associated with Korpi as a valid explanation of postwar welfare-state expansion and argues that the politics of the welfare state have fundamentally changed, we believe that the coalitional approach suggested here sheds light on the past as well as the present. Societal interests and coalitional alignments have changed, but there is no need to invent a new set of analytical categories in order to explain current patterns of welfare-state retrenchment and restructuring.

Richard Clayton is a Ph.D. candidate in the Government Department at Cornell University. His dissertation, "Public Workers in Privatizing States," analyzes the causes of public sector reorganization and its consequences for the politics of public sector unions, with special emphasis on the British experience.

Jonas Pontusson is Associate Professor of Government at Cornell University. His publications include The Limits of Social Democracy: Investment Politics in Sweden (1992) and, as coeditor, Bargaining for Change: Union Politics and Intra-Class Conflict in Western Europe and North America (1992) and Unions, Employers and Central Banks: Wage Bargaining and Macroeconomic Regimes in an Integrating Europe (forthcoming). His current research focuses on the interface of social policy regimes and labor markets.

* For comments on previous drafts of this paper, we wish to thank Geoffrey Garrett, Alex Hicks, Paul Pierson, Martin Rhodes, Herman Schwartz, Duane Swank, Kees van Kersbergen, three anonymous reviewers, and the members of the Political Economy Research Colloquium at Cornell University. We are also indebted to Lane Kenworthy and Duane Swank for their help with data collection.

Notes

1. Pierson, "The New Politics of the Welfare State," World Politics 48 (January 1996). Pierson's article builds on his book, Dismantling the Welfare State? (New York: Cambridge University Press, 1994), which analyzes the Thatcher and Reagan experiments in detail. Other comparative analyses emphasizing the resilience of welfare states include Geoffrey Garrett, Partisan Politics in the Global Economy (New York: Cambridge University Press, 1998); John Stephens, Evelyne Huber, and Leonard Ray, "The Welfare State in Hard Times," in Herbert Kitschelt et al., eds., Continuity and Change in Contemporary Capitalism (New York: Cambridge University Press, forthcoming); and Duane Swank, "Social Democratic Welfare States in a Global Economy," in Robert Geyer, Christine Ingebritsen, and Jonathon Moses, eds., Globalization, Europeanization and the End of Scandinavian Social Democracy? (London: MacMillan, forthcoming). For a detailed critique of Pierson's work, see Jens Alber, "Selectivity, Universalism, and the Politics of Welfare Retrenchment" (Paper presented at the annual meetings of the American Political Science Association, San Francisco, 1996); and for a comprehensive review of recent welfare-state literature, see Kees van Kersbergen, "The Declining Resistance of National Welfare States to Change?" (Manuscript, School of Public Affairs, University of Nijmegen, 1997).

2. Pierson (fn. 1, 1996), 150.

3. E.g., Korpi, The Democratic Class Struggle (London: Routledge and Kegan Paul, 1983).

4. Pierson (fn. 1, 1996), 174.

5. Ibid., 171.

6. Esping-Andersen, The Three Worlds of Welfare Capitalism (Princeton: Princeton University Press, 1990). The importance of services relative to transfer payments constitutes the most obvious basis for distinguishing among "institutional welfare states" (say, between Sweden and the Netherlands). Esping-Andersen's concept of "decommodification" partly captures but also blurs this distinction. Scoring very high on Esping-Andersen's decommodification index, the Scandinavian welfare states have traditionally promoted labor-force participation and strengthened the power of wage earners as sellers of labor power (rather than reducing their dependence on the sale of labor power).

7. While everyone agrees that demographic pressures are important, the claim that international capital mobility exerts pressure on welfare states is contested by both Garrett (fn. 1) and Swank (fn. 1). It is neither possible nor necessary to develop and support this claim here. Suffice it to say that the question of whether capital mobility exerts downward pressure on welfare states should not be conflated with the question of whether capital mobility produces convergence among welfare states. The evidence presented by Garrett and Swank speaks primarily to the latter question.

8. It is sometimes argued that the fundamental purpose of the welfare state is to provide for social security and that only some welfare states (in the first instance, the Scandinavian welfare states) have had redistributive ambitions as well. While welfare states clearly vary in their redistributive effects, we find this argument somewhat dubious: since some groups are far more insecure than others in a capitalist society, the public provision of social security is itself a redistributive activity.

9. Cf. OECD, Employment Outlook (Paris: OECD, July 1996), chap. 3; and Pontusson, "Wage Distribution and Labor-Market Institutions," in Torben Iversen, Jonas Pontusson, and David Soskice, eds., Unions, Employers and Central Banks (New York: Cambridge University Press, forthcoming).

10. Burtless, "Widening U.S. Income Inequality and the Growth of World Trade" (Manuscript, Brookings Institution, 1996).

11. OECD, Employment Outlook (Paris: OECD, July 1997), 177.

12. See David Rueda and Jonas Pontusson, "Wage Inequality and Varieties of Capitalism" (Working paper, Institute for European Studies, Cornell University, 1997).

13. Glyn, "Unemployment and Inequality," Oxford Review of Economic Policy 11 (Spring 1995), 10-11. As these figures refer to men only, the problem of factoring part-time employment into the calculation of employment rates can safely be ignored.

14. See Ian Gough et al., "Social Assistance in OECD Countries," Journal of European Social Policy 7 (February 1997), 24-27. Of course, the percentage of the population receiving social assistance might also increase because of political decisions to broaden the coverage of social assistance programs or to cut the benefits provided by more universalistic welfare programs. By the same token, poverty must be measured in terms of market income as well as disposable income in order to distinguish the effects of market forces from the effects of changes in social policy. Finally, it should be noted that the figures in Table 3. refer to relative poverty, as distinct from the absolute measures, such as the official U.S. poverty line. The percentage of the U.S. population living in households with a disposable income below the official poverty line fell from 22 percent in 1960 to less than 12 percent in 1973-79, and then began to rise, reaching 16-17 percent in the early 1990s; see Rebecca Blank, It Takes a Nation (Princeton: Princeton University Press, 1997), 55.

15. Pontusson (fn. 9); and Rueda and Pontusson (fn. 12). For comparative analyses of wage inequality trends by labor economists, see various contributions to Richard Freeman and Lawrence Katz, eds., Differences and Changes in Wage Structures (Chicago: University of Chicago Press, 1995).

16. The basic argument of this section is suggested by Sune Sunesson et al., "The Flight from Universalism," European Journal of Social Work 1, no. 1 (1998).

17. In 1996 the rate of male labor-force participation was 81.6 percent, and the rate for females was 76.3 percent. OECD, Historical Statistics (Paris: OECD, 1997); and OECD, Employment Outlook (Paris: OECD, July 1997), statistical appendix.

18. Pierson (fn. 1, 1996), 158-59. Cf. also Esping-Andersen, "After the Golden Age?" in Esping-Andersen, ed., Welfare States in Transition (London: Sage, 1996), 10-11.

19. The U.K. ratio of disposable-income poverty to market-income poverty increased from .16 to .21 over the time period covered by Table 3. By contrast, the German ratio increased from .15 to .17, the U.S. ratio held steady at .56, and the Swedish ratio declined from .38 to .18.

20. As state-owned corporations are not included in the OECD measure of government employment, these figure reflect the government agencies in corporate form but not the privatization of state-owned corporations.

21. Margareta Ringqvist, Om den offentliga sektorn (Stockholm: Fritzes, 1996), 111-13.

22. Cf. Stephens, Huber, and Ray (fn. 1).

23. Pierson (fn. 1, 1994), 142-46.

24. Pierson (fn. 1, 1996), 109.

25. Alber (fn. 1).

26. Peter Edelman, "The Worst Thing Bill Clinton Has Done," Atlantic Monthly (March 1997), 46.

27. Esping-Andersen (fn. 18), 16.

28. Social insurance replacement levels were cut from 90 percent to 80 percent by the bourgeois coalition government of 1991-94. The social democrats cut them further to 75 percent in 1995 but subsequently restored these cuts. This summary of Swedish changes draws primarily on Joakim Palme and Irene Wennemo, Swedish Social Security in the 1990s (Stockholm: Ministry of Health and Social Affairs, 1998). Cf. also Stephens, "The Scandinavian Welfare States," in Esping-Andersen (fn. 18).

29. Alber (fn. 1).

30. See Joyce Mushaben, "Restructuring the German Sozialstaat" (Paper presented at the annual meeting of the American Political Science Association, Washington, D.C., September 1997).

31. The data points in Figure 1 are the average replacement rate for workers at 66 percent and 100 percent of the average production worker's income in three different family situations (single, married with a dependent spouse, and married with a working spouse). For more details on this measure, see OECD, The OECD Jobs Study: Evidence and Explanations (Paris: OECD, 1994), chap. 8.

32. For a broader discussion of Swedish convergence on the German model, see Pontusson, "Between Neo-liberalism and the German Model: Swedish Capitalism in Transition," in Colin Crouch and Wolfgang Streeck, eds., Political Economy of Modern Capitalism (London: Sage, 1997).

33. Cf. Herman Schwartz, "Small States in Big Trouble: State Reorganization in Australia, Denmark, New Zealand, and Sweden in the 1980s," World Politics 46 (July 1994). See also Colin Fudge and Lennart Gustafsson, "Administrative Reform and Public Management in Sweden and the United Kingdom," Public Money and Management 9 (Summer 1989); and Rune Premfors, "The 'Swedish Model' and Public Sector Reform," West European Politics 14 (July 1991).

34. Ringqvist (fn. 21), 78.

35. Ibid., 106. Cf. Sven Olsson, Social Policy and Welfare State in Sweden (Lund: Arkiv, 1993), chap. 4.

36. Trevor Colling and Anthony Ferner, "Privatization and Marketization," in Paul Edwards, ed., Industrial Relations (Oxford: Blackwell, 1995), 493.

37. David Winchester and Stephen Bach, "The State," in Edwards (fn. 36), 311-14.

38. Economist, June 6, 1992.

39. Central Statistical Office (London), Economic Trends, no. 495 (1995), 17.

40. See Lois Wise, "Whither Solidarity? Transitions in Swedish Public-Sector Pay Policy," British Journal of Industrial Relations 31 (March 1993); and R. F. Elliott and K. Duffus, "What Has Been Happening to Pay in the Public Service Sector of the British Economy?" British Journal of Industrial Relations 34 (March 1996).

41. Cf. Pontusson (fn. 9); and Rueda and Pontusson (fn. 12).

42. David McKie, The Guardian Political Almanac 94/95 (London: Fourth Estate, 1994).

43. Sunesson et al. (fn. 16), 25.

44. Esping-Andersen, "Welfare States without Work," in Esping-Andersen (fn. 18).

45. Cf. Wolfgang Streeck, "Neo-Voluntarism: A New European Social Policy Regime?" in Fritz Scharpf, Philippe Schmitter, and Wolfgang Streeck, eds., Governance in the European Union (London: Sage, 1996).

46. Schwartz (fn. 33); and Swenson, "Labor and the Limits of the Welfare State," in Miriam Golden and Jonas Pontusson, eds., Bargaining for Change (Ithaca, N.Y.: Cornell University Press, 1992).

47. National Statistical Office, New Earnings Survey (London, various editions).

48. Pierson (fn. 1, 1994), 146-61.

49. See Richard Clayton, "Confronting the Market: Public Sector Unions and Market-Oriented Reform in British Health Services and Education" (Paper presented at the conference on Distributive Dimensions of Political Economy, Center for European Studies, Harvard University, March 1-3, 1996).

http://muse.jhu.edu/journals/world_politics/v051/51.1clayton.html.