Jump to ContentJump to Main Navigation
The New Politics of the Welfare State$

Paul Pierson

Print publication date: 2001

Print ISBN-13: 9780198297567

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198297564.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 25 February 2017

Political Institutions and Welfare State Restructuring

Political Institutions and Welfare State Restructuring

The Impact of Institutions on Social Policy Change in Developed Democracies

Chapter:
(p.197) 7 Political Institutions and Welfare State Restructuring
Source:
The New Politics of the Welfare State
Author(s):

Duane Swank

Publisher:
Oxford University Press
DOI:10.1093/0198297564.003.0008

Abstract and Keywords

The first of three chapters on the implications of electoral politics and the design of political institutions for welfare state adjustment. Swank first provides an overview of two key domestic and international pressures on developed welfare states: domestic fiscal stress and international capital mobility. He then outlines the theoretical argument that democratic institutions fundamentally determine government responses to domestic and international structural change, focusing on formal and informal institutions and drawing on and fusing insights from ‘power resources’ theory, the new institutionalism, and new cultural arguments about the determinants of social policy in advanced capitalist democracies. The next two sections utilize new data on social welfare effort, national political institutions, and internationalization to provide an econometric assessment of the social policy impacts of domestic fiscal stress and capital mobility during the period 1965 to 1995, looking first at the direct impacts of rises in public sector debt and in international capital mobility on social welfare provision, and second at the welfare state effects of fiscal stress and global capital flows across nationally and temporally divergent democratic institutional contexts; the initial focus is on total social welfare effort and then the analysis is shifted to changes in cash income maintenance and social services. The conclusion assesses the implications of the arguments and findings for the future course of social policy in developed democracies, and potentially bolsters the evidence for the central assertion that domestic institutions systematically determine the direction of welfare state restructuring.

Keywords:   capital mobility, capitalist democracies, cash income maintenance, democratic institutions, developed democracies, domestic institutions, electoral politics, fiscal stress, international capital mobility, internationalization, political institutions, public sector debt, social policy, social services, social welfare effort, social welfare provision, structural change, welfare state, welfare state reform

Since the mid‐1970s, governments of advanced capitalist democracies have in varying degrees attempted to retrench the welfare state. In many nations, policy makers have reduced the generosity of benefits and tightened programme eligibility; they have also imposed mechanisms for cost control in service delivery, privatized some social services, and increased targetting of benefits. As a result, social policy has tended to move in a ‘market‐conforming’ (i.e. work and efficiency oriented) direction. Neoliberal policy changes have not been confined to right‐of‐centre governments in the Anglo democracies; even the most developed social democratic welfare states of Northern Europe have experienced some reductions in social protection (e.g. Stephens 1996; Swank 2000).

Numerous explanations for these changes exist. Synoptic overviews of new pressures on developed welfare states typically highlight the roles of post‐1960s international and domestic structural changes characteristic of most advanced capitalist democracies (e.g. Esping‐Andersen 1996a; George and Taylor‐Gooby 1996; Pierson, in this volume; Rhodes 1997b; van Kersbergen 1997). Specifically, scholars highlight domestic pressures on social policy that arise from post‐1973 economic stagnation and rising unemployment, (p.198) burgeoning public sector deficits and debt, demographic shifts (e.g. the ‘crisis of ageing’), and changes in labour market structure (e.g. the rise in female labour force participation, structural unemployment). With respect to external factors, social policy analysts emphasize the downward pressures on welfare states that stem from the notable post‐1970 rise in the international mobility of capital and the secular increase in trade openness and international competition for markets.

Yet, while neoliberal policy reform is widespread and programmatic restructuring is universal, the pace and depth of these changes (and their impacts on inequality and the quality of social protection) vary notably. In this paper, I outline and test the argument that democratic political institutions determine the depth and character of welfare state restructuring. This is so, I will argue, because national configurations of democratic institutions directly or indirectly shape the degree to which domestic and international pressures are translated into neoliberal policy reforms. Specifically, democratic institutions significantly influence the political strength and structure of opportunities of the interests ideologically opposed to neoliberal social policy reforms; institutions also shape the political strength and opportunities of the interests harmed or put at risk by domestic structural change and the heightened integration of world markets. Moreover, institutions influence levels of mass support for the welfare state and prevailing political culture (e.g. conflict and competition versus solidarity and cooperation) that impede or facilitate welfare state retrenchment. My argument is that, if anything, democratic institutions—interest representational systems, the formal structure of decision making within the polity, and welfare state structures themselves—are as important as ever in shaping the policy trajectories of welfare states.

In the following pages, I first provide an overview of two key domestic and international pressures on developed welfare states: domestic fiscal stress and international capital mobility. I then outline the theoretical argument that democratic institutions fundamentally determine government responses to domestic and international structural change. As suggested, I focus on formal and informal institutions and draw on and fuse insights from ‘power resources’ theory, the new institutionalism, and new cultural arguments about the determinants of social policy in advanced capitalist democracies. As such, this paper builds on and extends theory developed in my work on how domestic institutions mediate the domestic policy impacts of internationalization (Swank 1998b, forthcoming). Next, I utilize new data on social welfare effort, national political institutions, and internationalization to provide an econometric assessment of the social policy impacts of domestic fiscal stress and capital mobility during the period 1965 to 1995. First, I focus on the direct impacts of rises in public sector debt and in international capital mobility on social welfare provision. Then, I analyse the welfare state effects (p.199) of fiscal stress and global capital flows across nationally and temporally divergent democratic institutional contexts. I initially focus on total social welfare effort and then shift analysis to changes in cash income maintenance and social services. In concluding, I assess the implications of my arguments and findings for the future course of social policy in the developed democracies. With this design, I extend the time frame and substantive focus of my complementary work (Swank 1998b, forthcoming) and potentially bolster the evidence for my central assertion that domestic institutions systematically determine the direction of welfare state restructuring.

1. International and Domestic Pressures on Contemporary Welfare States

As noted, scholars have typically identified a broad array of contemporary domestic and international pressures on the mature welfare states in developed capitalist democracies. Indeed, contributions by Iversen, Pierson, Schwartz, and others in this volume highlight the social policy consequences of several domestic and international structural shifts attendant to the post‐industrial political economy. In the current paper, I would like to focus on two major features of domestic and international pressures on the welfare state—domestic fiscal stress and international capital mobility—and the ways in which political institutions shape the impact of these factors on welfare state restructuring.

Fiscal Stress and Welfare State Restructuring

Many observers of contemporary welfare state politics have highlighted the policy roles of rising needs, initially as sources of increases in welfare budgets and ultimately as imperatives for programmatic retrenchment. Two areas of rising needs stand out: ageing and unemployment.1 As Table 7.1 makes clear, the relative proportion of the population over 65 as well as the share of population over 80 (i.e. the ‘frail elderly’) have increased substantially in recent years. The social policy consequences of these familiar demographic trends have been widely debated (see, among others, the OECD's 1994b, 1998f synoptic overviews). In fact, the direct budgetary implications are substantial. In the relatively encompassing models of 1965 to 1995 total (p.200)

Table 7.1. Domestic Fiscal Pressures on Developed Welfare Statesa

1960–73

1979–82

1983–6

1987–91

1992–5

Ageing (% of population 65+)

11.7

13.1

13.5

14.3

14.7

Ageing II (% of population 80+)

1.8

2.6

3.0

3.3

3.5

Unemployment rate

3.2

5.9

7.7

6.7

9.0

Long‐term unemployment

23.5

31.9

32.0

32.0

Real growth per capita

4.2

1.3

2.6

1.8

2.5

Aggregate public sector debt

45.0

56.9

58.8

71.4

Ageing

Percentage of population over 65 years of age (in last year in period).

Ageing II

Percentage of the population over 80 years of age (in last year in period).

Unemployment rate

Percentage of the civilian labour work force unable to find employment (period average).

Long‐term unemployment rate

Percentage of unemployed population who are out of work for 12 months or more (period average).

Real growth per capita

Annual average (for period) of percentage change in real GDP per capita in international prices.

(a) Cell entries are 15 nation period averages of row variables. The nations are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany (unified after 1990), Italy, Japan, the Netherlands, Norway, Sweden, United Kingdom, and the United States.

Sources: see Appendix Table 7.2.

social welfare effort presented below, an increase of 1 per cent of population over 65 is directly associated with an increase in aggregate social expenditure equal to 0.65 per cent of GDP (see Table 7.5 below). Over the long term, this estimate and Table 7.1 data suggest that the 1973 to 1995 absolute change in the size of aged population is alone responsible on average for an increase in social welfare outlays equivalent to roughly 2.6 per cent of GDP.

As to unemployment, A. Martin (1996) and Huber and Stephens (1998) have highlighted the role that low unemployment has played in sustaining the large social democratic welfare states of Northern Europe—the logic is generalizable. Increases in aggregate unemployment rates place substantial strains on social welfare budgets. Again, relying on estimates presented in Table 7.5, an increase of 1 per cent in the nominal unemployment rate is directly associated with an increase in total social spending equal to 0.33 per cent of GDP. In concrete terms, the growth of the average level of unemployment from roughly 3 to 9 per cent of the labour force between the 1960–73 and 1991–5 periods (row three in Table 7.1) generates increases in social outlays equivalent to 2 per cent of GDP. In addition, as Table 7.1 makes clear, the developed welfare states have also experienced relatively substantial increases in long‐term unemployment. Rises in ‘structural unemployment’ place additional strains not only on unemployment assistance programmes but also on active labour market policies and a broad array of social supports (p.201) and services (and, similar to rises in overall unemployment, reduce general and occupationally based taxes paid by employees). Thus, everything else being equal, most welfare states—and generous welfare states especially—face increasing pressure to reduce programme benefits, restrict eligibility, and otherwise reform policy to offset costs associated with long‐term shifts in levels of unemployment.

These familiar trends in the size and composition of major clientele groups of mature welfare states—and the fiscal pressures they exert—tell one side of the story. The second part involves economic growth (and the productivity levels that underscore it); as is commonly understood, increases in the size of claimant pools have occurred in the context of a notable slowdown in economic growth rates, or what many observers describe as the ‘end of the golden age’ of strong macroeconomic performance in developed capitalist democracies (e.g. contributions to Crafts and Toniolo 1996). Utilizing data expressed in cross‐nationally and temporally comparable international prices, Table 7.1 displays the average real growth rate of per capita GDP for the latter years of the ‘golden age’ (1960–73) and subsequent periods. The slowdown effectively amounts to a reduction by half or more of long‐term real economic growth and, in turn, a substantial reduction in the amount of tax revenues that would have been collected at ‘golden age’ era economic growth rates.

The last row of Table 7.1 displays a general and consequential manifestation of the concomitant rises in the size of welfare constituencies and economic slowdown: a substantial increase in public sector debt. Moreover, it is arguably the case that the well‐known political barriers to tax increases in the post‐1970 era (e.g. Wilensky 1981) have made it very difficult to address fiscal imbalance through statutory increases in tax rates. Overall, these economic and political factors have contributed to the universal rise in public sector borrowing in the 1980s and 1990s and the notable average increase in public sector debt from 45 to 71 per cent of GDP between the late 1970s and mid‐1990s. As to the course of contemporary social policy reform, accumulated debt and its budgetary and macroeconomic consequences (e.g. high interest rates, the prospect of future taxes) generate substantial pressure for neoliberal welfare state (and public sector) reforms. Indeed, the magnitude of public sector debt provides a parsimonious and encompassing index of the interaction of rising needs and economic slowdown and, in turn, the level of fiscal pressure on developed welfare states.

Global Capital and the Welfare State

Many observers have highlighted the role of internationalization—especially notable increases in the international mobility of capital—in contemporary efforts to retrench the welfare state. Indeed, international movements of (p.202)

Table 7.2. Internationalization of Markets in Developed Democracies, 1960–95a

1960–73

1979–82

1983–6

1987–91

1992–5

Capital

Total capital flows

6.7

18.6

26.1

32.3

46.3

Foreign direct investment

1.0

1.2

1.3

2.9

2.7

International capital markets

0.5

2.1

3.8

3.4

4.7

Trade

Total exports + imports

46.9

61.7

62.7

59.3

60.6

Exports + imports developing

8.2

8.8

8.4

7.1

8.2

Total capital flows

Total inflows and outflows of foreign direct investment, portfolio investment, and bank lending as a percentage of GDP.

Foreign direct investment

Total inflows and outflows of foreign direct investment as a percentage of GDP.

International capital markets

Total borrowing on international capital markets as a percentage of GDP.

Total exports + imports

Total exports and imports of goods and services as a percentage of GDP.

Exports + imports developing

Merchandise exports and imports to and from developing nations as a percentage of GDP (where developing nation category excludes oil‐exporting nations).

(a) Cell entries are 15 nation period averages of row variables. The nations are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany (unified after 1990), Italy, Japan, the Netherlands, Norway, Sweden, United Kingdom, and the United States.

Sources: see Appendix Table 7.2.

capital, and the potential for such movements, have increased dramatically since the early 1970s.2 Table 7.2 illustrates these trends for the fifteen focal nations of this study.3 Total transborder flows of capital increased on average from less than 10 to close to 50 per cent of GDP between the 1960s and mid‐1990s. Foreign direct investment expanded nearly 300 per cent (from roughly 1 to 3 per cent of GDP in the typical country) while borrowing on international capital markets increased on average from 0.5 to nearly 5 per cent of GDP. In absolute terms, 1995 annual foreign direct investment flows (p.203) totalled $474 billion in the developed countries and $632 billion world‐wide (United Nations Centre on Transnational Corporations 1996). Total borrowing on international capital markets (international and traditional foreign bonds, equity issues, intermediate and long‐term bank lending) totalled $732 billion for the OECD and $832 billion worldwide in 1995. Reductions in some forms of interest rate differentials across countries and markets, as well as removal of legal restrictions on capital and overall financial movements have also proceeded rapidly. Today, few formal impediments to capital inflows and outflows exist in the large majority of advanced capitalist democracies. Similar patterns are observed for overall trade openness (exports and imports of goods and services in relation to GDP), although the rate of increase is not remotely close to the rate of growth of capital flows: trade flows in relation to economic product expanded on average from roughly 45 to 60 per cent of GDP between the 1960s and mid‐1990s. The relative magnitude of trade with developing economies (excluding oil‐exporting countries), a widely debated source of competitive pressures on the welfare state, has in fact not increased during the last thirty‐five years. What are the consequences of the notable rise in international mobility of capital for the welfare state?

As noted, it is commonplace to link the dramatic increase in international capital mobility with reductions in social welfare effort. Classic political economists such as Adam Smith (1976 [1776]) as well as modern scholars such as Bates and Lien (1985) have argued that increasingly mobile capital poses substantial problems for governments that seek to raise revenues and pursue policies adverse to the economic interests and ideological orientations of (mobile) business. Contemporary neoliberal economists (e.g. McKenzie and Lee 1991), Marxian analysts (e.g. Gill and Law 1988; Ross and Trachte 1990), international relations theorists (e.g. Cerny 1996; Strange 1996), and popular analysts (e.g. Grieder 1997) use nearly identical reasoning to argue that the globalization of capital markets has effectively increased the power of capital over governments that seek to expand or maintain relatively high levels of social protection and taxation.4 This is purportedly the case because, in the presence of international capital mobility, governments must encourage internationally mobile firms to remain in the domestic economy, induce (p.204) foreign enterprises to invest, and satisfy international financial markets by reducing taxes, allaying fears of inflationary pressures, and eliminating a variety of distortions and inefficiencies associated with social welfare spending. It may also be the case because the ‘exit option’ enhances the conventional political resources of mobile businesses and the interest organizations that represent them and because greater capital mobility strengthens arguments embedded in neoliberal macroeconomic orthodoxy for market‐oriented reforms in social policy.

As I have argued elsewhere (Swank 1998b, forthcoming), the globalization thesis rests on both an economic and political logic. As to economics, theory suggests governments of all ideological and programmatic complexions not only have to consider the domestic requisites of business confidence—of actual and anticipated profitability, but in a world of few impediments to international capital movements, they also have to consider international investment climates, the relevant policies of other nations, and policy assessments of internationally mobile asset holders and markets. The strategic interaction of governments and transnationally mobile business and finance becomes a prisoner's dilemma for national policy makers. That is, in the face of inherent impediments to international policy coordination, they each face incentives to engage in competition for investment. Such competition presumably leads to a bidding war where social welfare transfers, social services, and the tax burdens that support them are progressively lowered to a ‘lowest common denominator’, or what Jessop (1996) calls a ‘Schumpetarian workfare state’. Such a welfare state tends to be residualist, increasingly organized along the lines of minimum means‐tested benefits, reliance on private insurance, work and efficiency principles, and relatively low (and distributionally neutral) tax burdens.

Politically, the conscious political action of large enterprises with mobile assets as well as the interest associations that represent them may also produce substantial policy impacts. At a minimum, authoritative policy makers within the state have consistently given substantial weight to the explicit policy preferences of mobile capital in formulating substantive policies and in designing institutions (Bates and Lien 1985). At most, mobile asset holders and their interest associations are able to consistently and credibly use the ‘exit option’ as an implicit (or explicit) threat in legislative, centralized bargaining, and executive branch policy making forums, enhancing the conventional political resources that are commonly brought to bear in efforts to shape policy (e.g. Block 1987; Kurzer 1993; Schmidt 1995). Moreover, the impact of international capital mobility on domestic policies in recent decades may in part be channelled through the increasing acceptance of neoliberal macroeconomic orthodoxy. Specifically, the widespread ascendance of neoliberal macroeconomic ideas, which call for roll‐backs in government intervention and highlight market distortions of social welfare programmes (p.205) and redistributive taxation, provides a supportive theoretical framework for appeals for market‐oriented reforms in social and tax policies to enhance trade competitiveness and business climate; arguments about the adverse impacts of moderate to high levels of social welfare provision and taxation on a nation's international economic performance lend further weight to neoliberal claims of general welfare state inefficiencies and calls for market‐oriented reforms (see Evans 1997; Hay 1998; Mishra 1993; Singh 1997).

A Note on Trade Openness

It is important to point out that some of the work on globalization and the welfare state highlights trade openness and increasing competitiveness in international markets as sources of pressure for policy change (e.g. Pfaller, Gough, and Therborn 1991; for critical assessments, Pierson, in this volume; Schwartz, in this volume). In this familiar argument, policy makers in increasingly open economies may encounter pressures for reduced (social security and other) tax burdens on domestic producers in order to lower labour costs and promote price competitiveness of exports; increases in trade openness may also push governments to reduce social outlays in order to lessen a variety of market distortions (e.g. perceived work disincentives). However, despite these considerations, I exclude well‐developed analysis of the question of trade impacts in the present paper. I do so for several reasons. First, a full analysis of trade impacts is a complex enterprise and would require a second paper. Second, as Huber and Stephens (1998) forcefully argue, the large welfare states of Northern Europe developed in the context of highly open markets for goods and services and played important roles in the process of structural adjustment to international competition; trade openness is not likely to be the central international mechanism generating pressure for welfare state retrenchment.5

I now turn to the explication of an alternative explanation of the dynamics of welfare state restructuring. My central proposition is that the welfare state pressures generated by the expanding national debt and the economic and political logics of globalization will be fundamentally conditioned and shaped by national configurations of democratic political institutions. Specifically, the general character of systems of interest representation, the organization of decision making within the polity, and the programmatic structure of welfare states should determine national policy responses to domestic and international pressures.

(p.206) 2. Democratic Institutions and the Welfare State

Democratic institutions are central to the determination of contemporary social welfare reform for three reasons.6 First, national institutions provide (restrict) opportunities for resistance to unwanted policy change by those ideologically opposed to the common neoliberal responses to domestic and international pressures; institutions also provide (restrict) opportunities for seeking compensatory policies by those who are adversely affected by domestic structural shifts and internationalization. Second, national institutions also influence directly and indirectly the relative political strength (weakness) of affected groups (e.g. their votes, seats, organization, cohesion) and the relative strength of traditional welfare state constituencies and coalitions. Finally, political institutions promote or impede certain constellations of values important to social welfare policy change: some institutions foster cooperation and consensus as well as support for (and confidence in) the welfare state specifically, and the efficacy of state intervention generally; other institutions tend to promote competition and conflict and pro‐market orientations. In sum, national political institutions shape the overall political capacity of relevant social aggregates to oppose neoliberal policy responses to domestic and international change. Who are these groups?

With regard to domestic socio‐economic change, political institutions are an important determinant of the political capacity of increasingly large populations of elderly to defend their social policy interests. They are also crucial for those social aggregates who are disfavoured by post‐industrial structural change (e.g. groups facing long‐term unemployment). This change, in part, is associated with the expansion of the service sector and knowledge class; it is also associated with the decline of large‐scale, geographically concentrated manufacturing of homogenized products and the ascent of specialized, flexible, small‐scale, and spatially diffused production of high value‐added products. Each of these changes—post‐industrial occupational transition and post‐Fordist production—contributes to a general climate of economic uncertainty for those in traditional occupations and industries. Marginalized groups consist of workers in low‐skill positions within both the manufacturing and service sectors. Semi‐ and unskilled production and clerical personnel as well as less educated and less experienced workers are increasingly at risk in terms of stagnant and declining incomes and higher (p.207) probabilities of unemployment, particularly structural and long‐term unemployment (see Esping‐Andersen 1990; OECD 1995d). Moreover, focusing on the overarching process of de‐industrialization, the new economic uncertainties may extend to many skilled workers and middle‐class strata as well (see Iversen, in this volume).

With respect to globalization, the internationalization of markets also affects economic and political interests in concrete ways. Specifically, as Rodrik's (1997) review of the literature on the impacts of globalization on economic interests suggests (1997: 4; cf. OECD 1994c):

reduced barriers to trade and investment accentuate the asymmetry between groups that can cross international borders . . . and those that cannot. In the first category are owners of capital, highly skilled workers, and many professionals, who are free to take their resources where they are most in demand. Unskilled and semi‐skilled workers and most middle managers belong in the second category.

Moreover, in addition to diverse effects across class and occupational strata, international financial integration produces differential short‐run effects on domestic economic sectors. As Jeffry Frieden has argued (1991: 426): ‘in the developed world, financial integration favors capitalists with mobile or diversified assets and disfavors those with assets tied to specific locations and activities such as manufacturing and farming.’ The capacities of those actors whose interests are threatened by globalization (in terms of both real and perceived threats) to press for compensation of losses and insurance against new risks is determined by the character of national political institutions.

Finally, national institutions shape the political capacity of those social aggregates who may oppose neoliberal policy reforms on ideological grounds. These groups, many of whom also have concrete material interests in preservation of extant levels of social protection, consist of trade union movements and labour‐based political parties, communitarian Catholic parties and groups, far‐left and left‐libertarian groups and parties as well as many social aggregates that form the core constituencies for welfare state programmes. The following dimensions of national political institutions are most relevant.

Institutions for Interest Representation

The basic character of democratic institutions for interest aggregation and representation should matter quite a bit for conditioning the ways in which domestic and international structural changes affect welfare state policy trajectories. I focus on the system of interest group representation, or the degree to which the interest group system is social corporatist as opposed to pluralist, and the system of electoral‐party interest representation, or the degree to which the electoral rules and party structure fosters ‘inclusive’ as opposed to ‘exclusive’ representation of societal interests.

(p.208) Social Corporatist Versus Pluralist Systems of Interest Representation

First, the character of the interest group system, especially the degree to which the system is social corporatist, should facilitate the extent to which affected actors can press their claims against adverse policy changes in the face of fiscal stress and internationalization. Corporatist institutions, particularly economy‐wide bargaining in which broadly organized and centralized labour movements have regularly exchanged wage restraint for full employment commitments and improvements in social protection have been important in the development of welfare states in capitalist democracies (e.g. Hicks and Swank 1992; Katzenstein 1985; Lehmbruch 1984). Despite new pressures on such institutions (e.g. Kurzer 1993), we should expect that the continued existence of extensively organized and relatively highly centralized trade union movements as well as the surprising persistence of (political and economic) forms of corporatist bargaining in some nations (e.g. Hoefer 1996; Visser and Hemerijck 1997; Wallerstein and Golden 1997) will be important. Indeed, as Rhodes (in this volume) argues, national bargains or ‘social pacts’ on social policy reform have, if anything, become more common as nations have attempted to forge politically acceptable national policy strategies for equity and efficiency.7 Generally, social corporatism provides an institutional mechanism whereby factions of labour affected adversely by different aspects of globalization and domestic change can articulate preferences and press claims on national policy makers to maintain (or in limited cases expand) social protection. Moreover, given the economic interests and ideological orientations of labour and labour's relatively powerful position in highly organized systems of interest representation, social corporatism should constitute a general political barrier to neoliberal policy change.

Indirectly, social corporatism has been an important element in the political success and policy strategies of social democratic parties. Indeed, in the post‐World War II era, the relationships across time and space in the advanced democracies between the measure of social corporatism developed below and the percentage of votes and seats won by parties of the left is highly significant (Pearson correlations of 0.630 and 0.637, respectively (N=435) for the years 1960–92 in the fifteen focal nations). In addition, as Katzenstein's seminal work has shown, social corporatism is dependent upon, and in operation reinforces, an ‘ideology of social partnership’ in which policy emerges from the cooperative and consensus‐oriented routines of repeated interactions among peak associations of labour and business. Similarly, Visser and Hemerijck (1997) argue that corporatist exchanges are analogous to networks of engagement (see Putnam 1993) where norms (p.209) of reciprocity, trust, and a sense of duty to other social partners and the common interest are cultivated. Net of other forces, social policy change in such an environment will typically involve slow, marginal, negotiated changes in which all interests are accounted; relatively quick and non‐trivial retrenchments of the welfare state in response to domestic pressures and internationalization are unlikely.

Inclusive Versus Exclusive Systems of Electoral Institutions

In addition, different features of formal constitutional structures and associated institutions that facilitate inclusive forms of representation through the electoral and party systems should matter. Important in this regard is recent work by Crepaz and Birchfield (no date) who have argued that welfare state retrenchment should, at a minimum, proceed more slowly in consensus democracies. This is so because consensus democracies have institutional mechanisms which guarantee that potential ‘losers’ in the processes of domestic structural change and internationalization are represented and that policy change occurs with incorporation of their interests. In related work, Birchfield and Crepaz (1998) have highlighted particular institutional features associated with consensus democracy that are important to those who pursue maintenance of the welfare state and egalitarian ends. Specifically, Birchfield and Crepaz argue that it is the ‘collective veto points’ afforded by proportional representation and the number of effective legislative parties, in particular, that insures some protection of these interests and egalitarian policy goals. Indeed, PR and multi‐party systems are more likely to afford workers and regional and sectoral economic interests relatively potent institutional mechanisms whereby to resist adverse policy changes and to pursue compensation; ideological interests opposed to market‐conforming policy changes (e.g. communitarian Christian democratic parties, environmentalists) will also be relatively advantaged. Similar institutional opportunities are arguably weaker in polities with exclusive electoral systems (e.g. single‐member plurality rules) and two‐party dominant systems.

The existence of what might be labelled ‘inclusive electoral institutions’ has also historically advantaged those groups and parties that constitute the core ideological support for the welfare state. As to political parties, social democratic and Christian democratic parties have been important in welfare state development (e.g. Esping‐Andersen 1990; Hicks and Swank 1992; Hicks 1999; Huber, Ragin, and Stephens 1993). Examining long‐term relationships, the correlations between the index of electoral inclusiveness developed below and shares of votes and national legislative seats for left and Christian democratic parties are highly significant (Pearson correlations for 1960–92 data (N=435) across the focal nations are 0.500, 0.477, 0.402, 0.407, respectively). In addition, inclusive electoral institutions tend to exist in broader constellations of consensus democratic institutions (p.210) (Lijphart 1984) and tend to involve repeated interactions between a multiplicity of social interests within decision making institutions such as lower chambers of parliaments and coalition governments. As such, norms of co‐operation, reciprocity, and consensus‐building may be fostered and adopted policies may enjoy significant legitimacy (Crepaz and Birchfield, no date; Birchfield and Crepaz 1998). As in the case of social corporatism, relatively rapid and non‐trivial roll‐backs in social welfare provision are, on balance, unlikely in this context.

Organization of Authoritative Decision Making Within the Polity

It is not only national institutional structures of collective interest representation that matter; the organization of formal policy making authority in the political system should also influence the political capacity of focal groups and, in turn, shape the ways in which national policies respond to international and domestic change (see Bonoli, in this volume, for a complementary analysis). Indeed, work on the welfare state (e.g. Hicks and Swank 1992; Huber, Ragin, and Stephens 1993; M. Schmidt 1996b) has highlighted the way in which the ‘dispersion of policy making authority’ has shaped welfare state development. Federalism, bicameralism, and presidentialism constitute potentially important ‘institutional’ veto points that allow conservative interests the opportunity to oppose welfare policy development and slow policy change.8 These institutional veto points may accord welfare state constituencies and coalitions similar opportunities to impede or otherwise shape neoliberal policy changes to their best advantage. However, not all ‘veto players’ are created equal. In the case of social corporatism and inclusive electoral systems, institutions create collective veto points (through encompassing representation) that advantage pro‐welfare state clienteles, groups, and coalitions. Both sets of institutional structures also promote directly and indirectly the political strength of these actors and foster values and policy making routines that enhance the ability of these interests to blunt radical neoliberal reforms. On the other hand, while dispersion of policy making authority creates potential institutional veto points, it also creates notably weak welfare state clienteles, groups, and coalitions as well as broad policy making orientations and climates that are conducive to retrenchment. These negative impacts on pro‐welfare state actors and policy making orientations are particularly pronounced where dispersion of authority (e.g. federalism) (p.211) is historically embedded, as in Australia, Canada, Switzerland, and the United States. They are less pronounced where contemporary patterns of dispersion are relatively short‐lived, as in post‐World War II Germany.

Specifically, formal fragmentation of policy making authority has large effects on the political capacities of social interests that have traditionally supported the welfare state. Focusing on the decentralization of authority, Noble's (1997: esp. 28–34) survey of the literature suggests that decentralization of policy making authority has generally tended to mobilize socially heterogeneous forces, undercut progressive and egalitarian political forces at the national level (e.g. trade union movements, social democracy), and favour local economic and political elites (also see F. G. Castles 1998; Pierson 1995; Stephens 1979). Moreover, federalism accentuates differences between rich and poor regions and, in turn, generates conflicts over the content of policy, fiscal equalization, and the distribution of financial burdens of national expenditure (e.g. contributions to de Villiers 1994; Banting 1997). In addition, institutional structures that disperse policy making responsibility tend to undercut the formation of coherent national policy strategies by groups and parties; the organization of parties and groups tends to be concentrated at regional and local levels and politics tends to be focused on narrow distributional issues within regional and local jurisdictions (e.g. on Canada, see Bradford and Jenson 1992; on the United States, see Piven 1992 and Skopol 1995). In fact, the long‐term relationships between institutional fragmentation, on the one hand, and the electoral strength of left parties and social corporatism, on the other, are strong. The correlations across time and space (1960–95 for the fifteen focal nations) between the index of ‘decentralization of policy making authority’ developed below (federalism and bicameralism) and left parties' votes and seats are −0.468 and −0.522 (N=465); for decentralization of authority and social corporatism the correlation is −0.513.

In addition, to the extent that dispersion of authority is associated with low levels of welfare state development, it will create weak welfare state programmatic constituencies. Indeed, I present evidence below that ‘decentralization of policy making authority’ is one of the largest factors—a substantial negative influence on social welfare provision—among those forces that directly shape welfare effort across time and space. Moreover, the climate for welfare state retrenchment may be enhanced by prevailing political culture in fragmented polities. That is, the dispersion of policy making power tends to complement and reinforce the competition and conflict of pluralist politics; norms of cooperation, reciprocity, and consensus‐building, potentially conducive to defence of the welfare state against substantial and rapid cuts, will tend to be weaker in fragmented polities than in centralized polities. Overall, to the extent that dispersion of authority favours conservative forces and produces a small welfare state, pro‐welfare coalitions, (p.212) programme‐specific alliances, and pools of social programme and service personnel will be relatively weak in systems of dispersed authority and relatively stronger in centralized polities. Indeed, one might argue that the institutional features of welfare states themselves (e.g. universalism), particularly those that cultivate strong constituencies and supportive public opinion and value orientations, should be considered as a separate set of institutional characteristics that condition the policy effects of external socio‐economic and political pressures for retrenchment.

The Institutional Structures of the Welfare State

Pierson (1994, 1996) outlines the general logic of welfare state retrenchment by noting that, unlike the politics of welfare expansion, welfare retrenchment involves taking away concentrated, politically popular benefits from organized constituencies for the promise of future, diffuse benefits; institutional characteristics of programme structure constrain various retrenchment strategies. Esping‐Andersen (1996a; 1996b) makes a similar point by arguing that institutional legacies of welfare states shape the impact of the variety of social, economic, and political changes on welfare state transformation. As Esping‐Andersen's (1990) seminal work on ‘worlds of welfare capitalism’ argues, contemporary welfare states may be classified as social democratic, conservative‐corporatist, and liberal (cf. F. G. Castles and Mitchell 1993). The core structural characteristics of these welfare states consist of universalism, occupationally based social insurance, and selectivity. As Rothstein (1998b) notes, universal welfare states consist of programmatic structures of extensive or full coverage of target populations, egalitarian benefits, and widely accessible social services; liberal welfare states are characterized by disproportionate reliance on means‐tested and private benefits.

Opportunity Structures and Welfare State Retrenchment

These principal institutional features of existing programme structures can play significant roles in impeding or facilitating retrenchment of the welfare state in response to the economic and political pressures attendant internationalization and domestic fiscal stress. While universal and liberal programme structures centralize policy formation and implementation in the hands of governmental institutions and bureaucracies, conservative corporatist welfare states decentralize authority to networks of quasi‐public administrative bodies comprised of constituency and professional groups, labour, business, and government.9 As such, conservative‐corporatist welfare (p.213) states provide notable opportunities for constituency groups and their allies to resist adverse policy change (see Clasen and Freeman 1994 on Germany); such opportunities are, on balance, more restricted in universal and liberal welfare programme structures.

Political Support for the Welfare State

Institutional features of welfare states may also strengthen (weaken) constituency groups and their allies. Central to the politics of welfare retrenchment is the extent to which welfare state institutions promote large, unified constituencies and unify (fragment) the panoply of interests that may mobilize to resist roll‐backs in social protection (Pierson 1994). Universal welfare states, most notably, tend to create large cohesive constituency groups organized around relatively generous, universal programmes of social welfare provision; conservative‐corporatist and liberal welfare states tend to fragment programme constituencies (on the basis of occupational status in the case of the former and on the basis of social class in the case of the latter). However, the political division of constituencies in corporatist conservative welfare states may be mitigated by the relatively generous social insurance protections accorded both working‐ and middle‐class groups and other features of contemporary occupationally based welfare programmes (see Swank, forthcoming).

Universalism may also generate high levels of mass political support and foster the development of broad political coalitions that support the welfare state. Specifically, Moene and Wallerstein (1996) have recently highlighted the importance of universal versus means‐tested benefits in maintaining the support of median voters for national systems of social protection: the disproportionate use of universalism cultivates the political support of median voters for the welfare state, since median voters face relatively high probabilities of benefiting from universally structured social programmes; means‐tested programmes do not generate such support, since median voters face low odds of benefiting from means‐tested, targeted programmes. Supporting this rational calculus, historical analyses of welfare state development (e.g. Esping‐Andersen 1990) suggests national programme structures providing high levels of universal coverage, basic security, and income replacement have effectively fused the interests of working‐ and middle‐class strata in the development and maintenance of the universal welfare state (also see Rothstein 1998b). While some have suggested that the recent growth in tax preferences for upscale groups and private occupational schemes will contribute to growing fissures within this traditional welfare state coalitions (Ervik and Kuhnle 1996), the persistence of universalism should significantly blunt the policy impact of domestic and international retrenchment pressures when compared to liberal welfare systems.

(p.214) Politics, Values, and Social Welfare Provision

Bo Rothstein's (1998b) important and insightful study of the universal welfare state of Northern Europe extends the previous analyses of political dynamics and offers an argument for the importance of the ‘moral logic of the welfare state’ in retrenchment politics. According to Rothstein, support for the welfare state hinges on the ‘contingent consent’ of strategically self‐interested and moral citizens. In turn, this consent is dependent on citizens' appraisals of the substantive, procedural, and distributional fairness of the welfare state. The principles of equal respect and concern embodied in programme structure, broadly targeted universal benefits, carefully adapted delivery organizations, and participatory administrative processes achieve relatively high levels of citizen contingent consent. Solidarity, trust, and confidence in state intervention are promoted. In liberal welfare states, problems related to substantive justice (e.g. conflicts over defining the ‘deserving poor’), procedural justice (perceptions of bureaucratic aggrandizement and waste), and a fair distribution of burdens (e.g. constituency fraud) are endemic. In sum, Rothstein's work suggests that net of other forces, we should observe notably different magnitudes and paces of welfare policy reform across universalistic and liberal welfare states.10

Overall, substantial theory suggests that roll‐backs of social protection in response to domestic fiscal stress and internationalization will be notably more difficult in universal than in liberal welfare states: the relative political strength of welfare state constituencies, and pro‐welfare state coalitions as well as high levels of mass political approval and supportive value orientations will constitute a significant barrier to social policy retrenchments. The reverse may be true in liberal welfare states where institutional features may facilitate neoliberal restructuring. In the case of conservative‐corporatist welfare states, the picture is mixed. While occupationally based systems often provide institutional opportunities for resistance to retrenchment, the fragmentation of constituency groups and of potential coalitions by programmatic structure may facilitate welfare state roll‐backs (see Swank, no date, for a discussion of features of occupationally based welfare states that may blunt retrenchment).

I now turn to an empirical exploration of the paper's central questions. Are there systematic, direct relationships between domestic fiscal stress and internationalization of capital markets, on the one hand, and retrenchment of the welfare state, on the other? Or, alternatively, do institutional (p.215) mechanisms for interest representation, the formal organization of policy making authority in the polity, and extant welfare state structures determine patterns of social welfare policy response to domestic and international pressures for retrenchment?

3. Democratic Institutions and Welfare State Restructuring: Econometric Analysis

Methodology

In the subsequent analysis, I examine the direct effects of the two features of domestic and international pressure on contemporary welfare states highlighted above—fiscal stress and international capital mobility. In 1965 to 1995 (and 1979/80 to 1995) empirical models of social welfare provision, I initially focus on an aggregate measure of welfare state size: total social welfare expenditures as a share of GDP.11 To deepen the analysis and to provide a check on the robustness of findings for total social welfare effort, I also examine the effects of fiscal stress and capital mobility on cash income maintenance (for families and sick days, and for the elderly, disabled, unemployed) and on social services (for families, the elderly, and the disabled). (On precise operationalizations of all variables and on data sources, see the Appendix to this chapter.) I utilize three central indicators to measure focal dimensions of domestic and international pressures. For domestic fiscal stress, I use lagged gross public sector debt as a percentage of GDP; for capital mobility, I use two central aspects of cross‐border capital flows: inflows and outflows of direct foreign investment and borrowing on international capital markets. These measures are standardized by (p.216) a nation's GDP and, to smooth occasionally volatile annual movements, are operationalized as three‐year moving averages (lags 1–3).12

To estimate the magnitude and significance of the direct policy effects of these facets of domestic and international change, I utilize empirical models of annual 1965 to 1995 (or 1979/80 to 1995) social welfare benefits (as a percentage of GDP) in the fifteen focal nations discussed in Note 3 above. I initially use 1965 to 1995 models to ascertain general welfare state effects of political institutions (and other factors), to test for differences in institutional effects across the contemporary era of welfare state retrenchment (i.e. 1979/80 to the present) and previous periods of welfare expansion, and to anchor contemporary models in a general model of social welfare provision. I then focus on 1979/80 to 1995 models of total and disaggregated social welfare effort.13 These models incorporate core socio‐economic determinants highlighted in theory and the extant literature. For all models, exogenous variables include (one‐year) lagged annual economic growth, unemployment, inflation rates, and the relative size of the elderly population; the partisan character of the national executive; and the level of affluence defined as per capita GDP in international prices.14 Finally, as noted above, I incorporate in empirical models a measure of trade openness, or total imports and exports as percentages of national GDP.

To ascertain direct welfare state effects of democratic institutions and to offer ready tests of the extent to which democratic institutions shape the impacts of domestic fiscal stress and international capital mobility, I include (p.217) in all empirical models indicators of institutional structures of interest representation—social corporatism and inclusive electoral institutions—as well as measures of the ‘dispersion of national policy making authority’ developed below (i.e. decentralization and presidentialism). In subsequent models, I test individually for the roles of welfare state structure in mediating the impacts of domestic and international change. Specifically, I use measures (circa 1980) of programmatic attributes of welfare states (Esping‐Andersen 1990) in models of social welfare spending to test the proposition that impacts of domestic and international pressures vary across welfare institutions. (See below on precise measurement and related details.)

Measuring Democratic Political Institutions

To develop a measure of social corporatism, I utilize new cross‐nationally and temporally varying data on the organization of trade unions and the labour and industrial relations systems in the advanced industrial democracies. Following the lead of Golden (1997), I employ correlation and principal components analysis of theoretically relevant dimensions of corporatism—union density, inter‐confederal concentration, confederal power (e.g. control over strike funds, involvement in wage bargaining), and the level of wage bargaining—to develop a core empirical indicator. As presented in Table 7.3, three characteristics of unions and bargaining systems consistently cohere (i.e. factor loadings in access of 0.80): union density, confederal power, and the level of wage bargaining. A weighted standard‐score index of these three factors—an index varying across time and space and weighted by each component's factor loading—is used as the principal indicator of social corporatism. Confirmatory analysis suggests that while the focal indicator inherently stresses economic features of social corporatism, it is highly correlated across time and space with measures of the incorporation of peak functional associations in the national policy process.15

In order to develop indicators of electoral institutions and the organization of policy making authority, I follow the precedent established by Lijphart's (1984) seminal work on dimensions of democratic political institutions and conduct analyses of the relationships between aspects of formal democratic institutions and their close correlates: the major institutional dimensions highlighted in the literature discussed above are the focus. Using operationalizations outlined in the notes to Table 7.3, I employ principal components analysis of cross‐nationally and temporally varying measures of the degree of proportionality in the electoral system, the number of (p.218)

Table 7.3. Principal Components Analysis for National Political Institutions

Social corporatism

Factor 1

Factor 2

National political institutions

Factor 1

Factor 2

Factor 3

Union concentration

0.03

−0.98

Proportional representation

0.82

−0.04

0.47

Union density

−0.86

−0.18

Number of effective parties

0.93

0.13

−0.37

Confederal power

−0.85

0.05

Federalism

0.09

−0.78

−0.37

Level of bargaining

−0.85

0.28

Bicameralism

0.09

−0.92

−0.05

Presidentialism—separation of powers

0.02

−0.14

−0.54

Union concentration

Herfindahl index of inter‐confederal concentration, or the probability that any two union members are members of the same national confederation.

Union density

Percentage of employed labour force who are members of unions.

Confederal power

Index of largest confederation's involvement in wage‐setting process, power of appointment, veto over wage agreements, veto over strikes, and maintenance of strike of funds.

Level of bargaining

Four‐level scale for centralization of wage bargaining where 0 is plant level, 1 is industry level, 2 is sectoral level without sanctions, and 3 is sectoral level with sanctions.

Proportional representation

Three‐level scale (0 = single member district with plurality rules, 1 = quasi‐proportional, 2 = proportional) of degree of proportional representation.

Effective number of legislative parties

Laasko and Taagepera index as presented in Lijphart (1984: 120).

Federalism

Three‐level scale where 0 = no, 1 = weak, and 2 = strong federalism.

Bicameralism

Three‐level scale where 0 = no or symbolic second chamber, 1 = weak bicameralism, and 2 = strong bicameralism.

Presidentialism

Two‐level scale where 0 = parliamentary and 1 = presidential government.

Sources: See Appendix Table 7.2.

(p.219) effective legislative parties, presidentialism, bicameralism, and federalism.16 As Table 7.3 reveals, and as prefigured in the literature, PR and the number of effective legislative parties strongly cohere. In addition, consistent with expectations about facets of the organization of decision making authority, federalism and bicameralism form a distinct institutional dimension. Presidentialism does not ‘load’ on this dimension. Thus, in subsequent analysis I focus on two dimensions of the ‘dispersion of policy making power’: the vertical separation of powers (i.e. presidentialism) and the horizontal dispersion of power (i.e. decentralization). For empirical indicators, weighted standard‐score indices of PR and effective legislative parties and of federalism and bicameralism are used to measure ‘inclusive electoral institutions’ and ‘decentralization of policy making authority’, respectively.17

To measure attributes of welfare systems, I rely on Esping‐Andersen's rankings of welfare states on the dimensions of ‘socialism’ or what might simply be called universalism (i.e. universal coverage with high benefit equality), ‘conservatism’ (i.e. occupationally stratified welfare programmes and special public employee plans), and ‘liberalism’ (i.e. means‐testing and reliance on private pensions and health insurance). Clearly, ‘universalism’, ‘conservatism’, and ‘liberalism’ provide general indicators of the degree to which welfare states are characterized by those factors emphasized in the theoretical discussion above. The scores of the fifteen focal nations on these three dimensions of welfare state structure as well as the 1965–1994/5 average annual scores for social corporatism, inclusive electoral institutions, and dispersion of authority are listed in Table 7.4.

Statistical Estimation

Estimation of empirical models is conducted by Ordinary Least Squares regression with corrections for first‐order autoregressive errors, panel correct standard errors (i.e. heteroskedastic‐consistent variance‐covariance matrices for panel data), and unit dummies (see Beck and Katz 1995). Given that inclusion of a full set of unit dummy variables produced R2‐deletes approaching 0.99 for institutional variables, a conventional fixed effects model is not used; country dummies to adjust for unit effects (and thus obviate potentially (p.220)

Table 7.4. Dimensions of National Political Institutions: Country Positions, 1965–95

Country

Universalism

Conservatism

Liberalism

Social corporatism

Electoral inclusiveness

Decentralization

Presidentialism

Sweden

8

0

0

1.26

0.25

−0.58

0

Norway

8

4

0

1.02

0.30

−0.66

0

Denmark

8

2

6

0.74

0.76

−0.66

0

Finland

6

6

4

0.67

0.89

−0.66

1

Netherlands

6

4

8

0.08

0.73

−0.11

0

Austria

2

8

4

0.33

−0.10

−0.15

0

Belgium

4

8

4

0.08

1.28

−0.08

0

France

2

8

8

−0.72

−0.27

−0.66

1

Germany

4

8

6

−0.56

0.06

1.47

0

Italy

0

8

6

0.17

0.47

−0.11

0

United States

0

0

12

−1.02

−1.23

1.47

1

Canada

4

2

12

−0.78

−1.08

0.37

0

Japan

2

4

10

−0.55

−0.42

−0.11

0

Australia

4

0

10

0.26

−0.56

0.95

0

Britain

4

0

6

−0.42

−1.15

−0.66

0

Universalism

Esping Anderson's (1990) score for degree of universalism and benefit equality in social welfare programmes, circa 1980.

Conservatism

Esping‐Andersen's (1990) score for degree welfare is occupationally stratified and public employees have special programmes, circa 1980.

Liberalism

Esping‐Andersen's (1990) score for degree welfare state relies on means‐testing and degree of private pensions and health care, circa 1980.

Social corporatism

Weighted standard score index of union density, union peak association power, and level of collective bargaining, annual average, 1965–94.

Electoral inclusiveness

Weighted standard score index of degree of proportional representation and number of effective legislative parties, annual average, 1965–95.

Decentralization

Weighted standard score index of degree of federalism and bicameralism, annual average, 1965–95.

Presidentialism

Dichotomous variable where presidential systems scored 1.00 and parliamentary systems scored 0.00.

Sources: see Appendix Tables 7.1 and 7.2 for descriptions of data and sources.

(p.221) biased estimates of social policy effects of focal variables) are included when t‐statistics for country dummies exceeded 1.00. Within the context of OLS models, I test for direct linear effects of political institutions, fiscal stress and international capital mobility, and the control variables of the general model. To assess core hypotheses about how welfare state effects of fiscal stress and international capital mobility vary by political institutional context, I employ interaction analysis in which multiplicative terms involving public sector debt, direct foreign investment, and borrowing on international capital markets, on the one hand, and political institutional variables, on the other, are added to the general empirical models of social welfare effort.18

4. Fiscal Stress, International Capital, and Political Institutions: Results of Empirical Tests

Total Social Welfare Effort

The findings from the estimation of basic models of total social welfare provision are presented in Table 7.5. In the first two columns, I presents tests for direct linear effects of flows of direct foreign investment, four central facets of national democratic institutions, and exogenous factors composing the general model. As noted above, the models are estimated initially with 1965–95 data for the fifteen focal nations; I then shift the focus to the contemporary era (1979–95) and examine direct welfare state effects of fiscal stress, both measures of international capital mobility, political institutions, and the variety of socio‐economic and political factors of the general model.

As the table reveals, three of the four dimensions of democratic political institutions have significant effects on total social welfare effort. In fact, social corporatism, inclusive electoral institutions, and decentralization of policy making authority all have substantively large effects. As theorized, the first two institutional factors have strong positive effects while decentralization has a large negative effect on total social welfare effort. For instance, net of other forces, the difference in social welfare provision between nations at −1.00 and 1.00 on the index of social corporatism (roughly the difference (p.222)

Table 7.5. The Impact of Political Institutions on Total Social Welfare Effort

(1) 1965–95

(2) 1965–95

(3) 1979–95

(4) 1979–95

Democratic political institutions

Social corporatism

0.8591* (0.3767)

0.9076* (0.3588)

2.2813* (0.4618)

2.2316* (0.4408)

Inclusive electoral institutions

0.8549* (0.3027)

0.8678* (0.3007)

0.7130* (0.3413)

0.7018* (0.3417)

Decentralization of policy making authority

−2.7485* (0.7545)

−3.0616* (0.5760)

−4.2301* (0.7457)

−4.1621* (0.7378)

Presidentialism/separation of powers

−0.8524 (0.9330)

Domestic and international pressures

Public sector debt

−0.0189* (0.0114)

−0.0209* (0.0117)

Direct foreign investment flows

0.0770 (0.1304)

0.0930 (0.1265)

0.0079 (0.1047)

International capital market flows

0.0393 (0.0672)

General model

Left government

−0.0029 (0.0078)

−0.0025 (0.0076)

−0.0019 (0.0067)

−0.0017 (0.0065)

Christian democratic government

−0.0044 (0.0132)

−0.0041 (0.0132)

−0.0061 (0.0130)

−0.0065 (0.0130)

Elderly population

0.6536* (0.1744)

0.6145* (0.1493)

0.3896* (0.1705)

0.4005* (0.1705)

Unemployment rate

0.3350* (0.0560)

0.3381* (0.0551)

0.3558* (0.0655)

0.3538* (0.0642)

Inflation rate

0.0422* (0.0231)

0.0414* (0.0234)

−0.0201 (0.0450)

−0.0201 (0.0455)

Level of affluence

0.8367* (0.1048)

0.8421* (0.1038)

0.8881* (0.1408)

0.8774* (0.1385)

Economic growth rate

−0.1426* (0.0232)

−0.1429* (0.0232)

−0.2026* (0.0490)

−0.2034* (0.0451)

Trade openness

0.0122 (0.0146)

0.0127 (0.0146)

−0.0203 (0.0182)

−0.0204 (0.0185)

Intercept

−2.5996

−2.3596

3.3446

3.3032

Buse R2

0.7973

0.7934

0.9157

0.9166

Note: Each model is estimated with 1965–95 (1979–95) data by Ordinary Least Squares; equations are first‐order autoregressive. The table reports OLS unstandardized regression coefficients and panel correct standard errors. All models include nation‐specific dichotomous variables (if t > 1.00) and dichotomous variables to control for series breaks in the dependent variable.

(*) indicates significance at the 0.05 level or below.

between 1965–94 mean levels for the United States and Norway) is 1.7 per cent of GDP (0.8591*2). A similar difference of 2.00 on the index of electoral inclusiveness (e.g. the difference between mean levels of electoral inclusiveness in Canada and Finland) is associated with an equivalent (p.223) difference in social protection of 1.7 per cent of GDP (0.8549*2). The social welfare impact of decentralization of policy making authority is even larger: the difference in social welfare effort between a nation at 1.00 on the index at a particular point in time (e.g. roughly the 1965–95 mean level for Australia) and a nation scoring −0.66 (e.g. the mean level for Denmark) is approximately 4.4 per cent of GDP (2.7485*1.61). On the other hand, presidentialism is not significantly associated with total social welfare outlays. Because of this initial finding and the absence of consistent direct or indirect effects in subsequent analysis, I drop this institutional dimension from the analysis. Column 2 of Table 7.5 displays the basic model after deletion of presidentialism; all other institutional effects retain or increase their substantive effects and significance.

These long‐term general models of social welfare effort are also useful for testing for differences in social policy effects of political institutions in shaping welfare effort during the latter years of welfare expansion (i.e. mid‐1960s to mid‐1970s) and the post‐1970s period of welfare state retrenchment. Using the basic model of column 2 and interaction terms for a post‐1978 dummy variable and the three institutional variables, I estimated the effects of social corporatism, electoral inclusiveness, and decentralization of policy making authority during periods 1965–79 and 1979–95 (and tested for significance in the difference in magnitude of institutional effects across the periods). The effects of social corporatism are largely confined to the post‐1979 period: its effect on total social welfare effort is a statistically insignificant 0.3763 in the pre‐1979 era and a substantively large and significant 1.5185 in the post‐1979 era. The effects of electoral inclusiveness are significant in both periods (regression coefficients of 0.6298 and 1.0919, respectively), although the post‐1979 effect on social welfare effort is statistically and substantively larger. A similar pattern holds for decentralization: its effects on social welfare effort (−2.6376 in the pre‐1979 era and −3.7532 in the post‐1979 period) are significant in both eras but greater in the contemporary era of accelerating pressures on the welfare state. In addition, focusing on columns 3 and 4 of Table 7.5, estimations of institutional effects during the 1979–95 era confirm the conclusion that national political institutions continue to be very important (if not more important) in shaping social policy change in the contemporary era.

The second panel of Table 7.5 displays the direct social welfare effects of public sector debt and international capital mobility. With respect to capital mobility, neither rises in the level of flows of foreign direct investment nor borrowing on international capital markets is associated with declines in social welfare provision. However, the magnitude of public sector debt is related to social welfare effort (although the effect is moderate in substantive magnitude). Controlling for political institutions, economic openness, and the variety of socio‐economic and political pressures on contemporary welfare states entailed in the general model, increases in government debt are (p.224) negatively and significantly related to total social provision. Examining column 4 of Table 7.5, one can see that, net of other forces, an increase in debt equivalent to 10 per cent of GDP, would produce a decline in social welfare effort equivalent to 0.2 per cent of GDP; an increase in debt of 40 per cent of GDP (the mean change for our nations from 1978 to 1995) is associated with an absolute decline in social protection of 1 per cent of GDP. However, as I show below, where increases in debt are above the mean and where national institutional structures are conducive to welfare retrenchment, the impacts of debt are larger.

A Comment on the Combined Effect of Fiscal Stress and Global Markets

It is useful to note that in related work (Swank 1998b, forthcoming), I have examined the 1965–93 welfare effects of a variety of alternative measures of capital mobility (e.g. total capital flows, capital market liberalization, interest rate differentials, and various categories of capital flows standardized by gross domestic investment), plausible non‐linear effects of capital mobility, and the interactions of each dimension of international capital mobility with trade openness. None of these alternative specifications produced findings of significant downward pressures of international capital on social welfare provision. Finally, in the present analysis, I explored the possibility that capital mobility would have significant negative effects on the welfare state at high levels of deficits and debt. Specifically, Garrett (1998a) has suggested the hypothesis that international capital markets impose a high interest rate premium on governments that run large deficits. That is, in the presence of high debt and deficits, capital markets anticipate higher inflation (and other adverse outcomes), engender currency depreciation, and hence necessitate higher interest rates in the fiscally imprudent nation. Ultimately, one might speculate that the interaction of high debt and deficits, on the one hand, and high capital mobility, on the other, will place serious downward pressures on the welfare state. I find (through a test of the interaction of public debt and the capital mobility variables) that, net of other forces, the social welfare impact of international capital mobility becomes significant and negative only when gross government debt exceeds 100 per cent of GDP (e.g. Belgium in the 1980s and Italy in the 1990s).19

(p.225) Turning to the bottom panel of the table, I present results for the general model of total social welfare effort. As to the other forces shaping social welfare provision, the size of the aged population, unemployment, inflation, and affluence are all positively and significantly associated with total social welfare outlays; the economic growth rate is negatively and significantly related to social welfare provision. Trade openness, inconsistent with the classic arguments of Cameron (1978) and Katzenstein (1985b) and the embedded liberalism thesis, is not significantly related to social provision in the models of Table 7.5.

A Note on Party Effects

Finally, in the presence of nation‐specific dummy variables and controls for social corporatism, electoral institutions, and the decentralization of authority—all of which are related to partisan strength in direct and indirect ways (see above), the relatively short‐term direct social welfare impacts of both left and Christian democratic government control are statistically insignificant in the basic models of Table 7.5 (and in the models of cash income maintenance and social services presented below). However, it is important to note that in the presence of unit dummies and institutional variables, levels of multicollinearity for partisan variables are high and thus it is difficult to draw conclusions about their exact relevance. In fact, if one deletes unit dummies in the column 3 equation, allowing institutional, partisan, and other variables to absorb cross‐sectional variance, government control by Christian democratic governments is positively and significantly related to total social protection (e.g. the regression coefficient for Christian democratic government—per cent of cabinet portfolios held by these parties—is 0.0684 with a t‐statistic of 4.803); left government is also marginally significant in this specification (regression coefficient of 0.0151; t=1.472). In other formulations (e.g. cumulative years of left and Christian democratic government from 1950), left (and Christian democratic) government is clearly significant in Table 7.5 models in the absence of unit effects. (See Kitschelt, in this volume, for a systematic analysis of the strategic environment for partisan welfare state retrenchment/defence.)

Table 7.5A presents tests for the central argument of the paper: the social welfare impacts of domestic fiscal stress and international capital mobility should be notably shaped by configurations of national political institutions. As the table indicates, this indeed appears to be the case. Of the eighteen possible interactions between political institutions and domestic and international pressures, fourteen are correctly signed and moderately significant; a dozen are clearly statistically significant at conventional levels (0.05) and beyond. The strongest pattern pertains to the degree to which domestic political institutions shape international pressures. The character of the interest representational system, the inclusiveness of electoral institutions, and the (p.226)

Table 7.5A. The Mediation of Domestic and International Pressures on Welfare States by Political Institutions

Public sector debt

Direct foreign investment

International capital markets

Political institutions

 Social corporatism*domestic or international factor of column

0.0146a (0.0108)

0.4222* (0.1556)

0.2236* (0.0823)

 Electoral inclusiveness*domestic or international factor

−0.0089 (0.0075)

0.1887* (0.0765)

0.3275* (0.0778)

 Decentralization*domestic or international factor

−0.0173a (0.0116)

−0.3666* (0.1409)

−0.3647* (0.0903)

Welfare state programme structure

 Universalism*domestic or international factor

0.0055* (0.0027)

0.2173* (0.0569)

0.0561* (0.0278)

 Conservatism*domestic or international factor

−0.0026 (0.0022)

0.0317 (0.0270)

0.0689* (0.0216)

 Liberalism*domestic or international factor

0.0024 (0.0024)

−0.0964* (0.0314)

−0.0342* (0.0133)

Note: Interactions are estimated in the focal equations of Table 7.5. Estimation is by OLS with AR1 autoregressive parameters and panel correct standard errors. Direct effects of Table 7.5 factors remain universally unchanged in terms of significance levels and substantive effects and, hence, full table results are not reported. Tables are available from the author.

(*) Significant at 0.05 level.

(a) Significant at the 0.10 level.

See Appendix Table 7.3 for full results of interaction analysis.

organization of policy‐making authority, as well as universalism and liberalism in welfare programme structures, are all important in shaping the magnitude of impacts of international capital mobility. Computations of exact impacts of rises in international capital mobility at low, medium, and high levels of these five institutional dimensions suggest that direct investment and exposure to international capital markets exert systematic downward pressures on welfare states where decentralization and liberalism are high and where social corporatism, electoral inclusiveness, and universalism are low. At moderate to high levels of social corporatism, electoral inclusiveness, and universalism (and low levels of decentralization and liberalism), rises in international capital mobility are, net of other forces, either unrelated to social welfare protection or associated with small positive increases. (See Appendix Table 7.3 for estimates of the components of interactions. This information and procedures outlined above can be used to derive specific welfare state impacts of domestic and international pressures in particular institutional contexts.)

A similar but weaker pattern obtains for the case of domestic fiscal stress: three of the six interactions are correctly signed and modestly significant (although see below for evidence of stronger impacts of debt on cash income (p.227) maintenance). The largest institutional effect—indicative of the broader pattern of findings for domestic and international pressures—occurs with regard to universalism: at high levels of universalism (Sweden, Norway, and Denmark), the effect of an increase of public sector debt equivalent to 1 per cent of GDP is 0.00. At moderate levels of universalism it is −0.023 and at low levels of universalism it is −0.045 (see information in Appendix Table 7.3). The welfare impact of an increase in government debt of 10 per cent of GDP in a nation where programmatic attributes of welfare states entail little universalism (e.g. the United States) is −0.45.

A Comment on Big Welfare States, Increasing Debt, and Global Markets

The above findings can be further illustrated with reference to the experience of Nordic countries. These welfare states possess highly universalistic programme structures, strong social corporatism and electoral inclusiveness, and are relatively centralized polities. Indeed, the experience of the Swedish welfare state in the early and mid‐1990s, an experience grounded in rising public sector financial stress and international capital mobility, can serve to underscore and enrich the statistical findings reported above.20

In the early 1990s, Sweden experienced a dramatic fall in growth rates (for instance, real GDP per capita growth was only 0.5 per cent in 1990 and −1.7 and −2.1 per cent in 1991 and 1992, respectively). It also experienced a commensurate rise in unemployment, deficits, and public debt. For instance, total public sector debt increased from 44.3 per cent to 80.1 per cent of GDP between 1990 and 1995. The 1991–4 bourgeois government enacted two main ‘crisis packages’ in 1992 (largely with the cooperation of the opposition Social Democrats (SAP)). During 1992, the ‘base amount’ utilized to calculate social benefits was trimmed 3 per cent, sickness benefits (heretofore paid at 90 per cent with no waiting days) were reduced to 65 per cent for the second and third days, 80 per cent for the remainder of the first year of sickness, and 70 per cent of pay after that; one waiting day for sickness benefits was imposed and, to encourage better policing of the system, employers became responsible for the first two weeks of benefits. In addition, the system of work injury benefits was brought in line with sick pay, and unemployment and social assistance benefits were modestly reduced; a waiting period of five days (and new limits on duration) was introduced for unemployment benefits and new employee contributions for unemployment insurance were initiated. Perhaps the most important change was the initiation of major alterations in the earnings‐related pension: in the future, the ATP would be based on lifetime contributions and not 30 years of contributions for full pension with the ‘best 15 years’ determining pension amounts; new social (p.228) insurance contributions were planned where the employee would ultimately contribute 9.25 per cent of pay to fund ATP. (A new 40‐year residence rule, rather than citizenship, was enacted to determine eligibility for the basic, flat‐rate pension.)21 Finally, many new initiatives were taken in 1992 and 1993 to encourage the implementation of market mechanisms in social service delivery and to foster the establishment of private providers (e.g. doctors, child care businesses); additional user fees for a variety of social services were also implemented.

However, at the time of these entitlement cuts, the government increased outlays for Active Labour Market Policies (e.g. Stephens 1996). Indeed, expenditure on training, placement, and related services (as well as temporary public sector employment) increased from 1.63 per cent of GDP in 1989 to 3.11 per cent in 1994. Moreover, budgetary policy also emphasized (particularly from 1993–4 on) modest revenue increases; some of the early tax cuts (see note 21) were temporarily rescinded and moderate new revenue increases proposed (e.g. Gould 1996). The Social Democrats were re‐elected in 1994 and preserved many of the roll‐backs in social benefits and social services. However, some of the liberal provisions instigated by the bourgeois government concerning private providers were eliminated; social rights to services for the handicapped, elderly, and children were extended, and, again, the 1994 budget emphasized tax increases over further spending cuts.

Generally, while notable fiscal pressure engendered some retrenchment in social insurance benefits and eligibility, cuts in social entitlements were balanced with social support of the unemployed and with revenues. From the perspective of the mid‐ and late 1990s, most observers have noted that, while there have been cuts and modifications in the Swedish welfare state since the late 1970s, the basic elements of the social democratic model remain largely intact in Sweden (e.g. Sainsbury 1996; Clausen and Gould 1995; Gould 1996), albeit restructured in the direction of a more ‘productivist’ welfare state (Esping‐Andersen, 1996a).

Income Maintenance and Social Services

Table 7.6 reports the results of the supplementary analysis of the direct effects of political institutions, fiscal stress, and international capital mobility on 1980–95 cash income maintenance and social services. Overall, the findings reported above for the effects of these factors on total social welfare effort are reproduced in the case of these two important categories (p.229)

Table 7.6. The Impact of Political Institutions on Cash Income Maintenance and Social Welfare Services, 1980–95

(1) Cash benefits

(2) Cash benefits

(3) Social services

(4) Social services

Democratic political institutions

Social corporatism

0.8256* (0.3338)

0.8188* (0.3260)

0.2994* (0.1165)

0.2869* (0.1153)

Inclusive electoral institutions

0.6022* (0.2540)

0.5960* (0.2507)

0.0252 (0.0529)

0.0291 (0.0510)

Decentralization of policy making authority

−4.1062* (0.5238)

−4.0547* (0.5148)

−0.5050* (0.0814)

−0.4948* (0.0779)

Domestic and international pressures

Public sector debt

−0.0055 (0.0086)

−0.0052 (0.0088)

−0.0074* (0.0029)

−0.0083* (0.0027)

Direct foreign investment flows

−0.0226 (0.1081)

0.0494* (0.0231)

— 0.0170

International capital market flows

−0.0215 (0.0688)

0.0170 (0.0181)

General model

Left government

−0.0008 (0.0028)

−0.0009 (0.0028)

0.0005 (0.0009)

0.0006 (0.0009)

Christian democratic government

−0.0068 (0.0054)

−0.0065 (0.0053)

−0.0012 (0.0012)

−0.0009 (0.0012)

Elderly population

0.2154a (0.1518)

0.2125a (0.1511)

−0.0284 (0.0319)

0.0008 (0.0319)

Unemployment rate

0.3330* (0.0541)

0.3326* (0.0532)

0.0571* (0.0108)

0.0504* (0.0107)

Inflation rate

0.0518* (0.0314)

0.0492* (0.0304)

0.0030 (0.0062)

0.0035 (0.0064)

Level of affluence

0.3995* (0.1135)

0.3912* (0.1139)

0.1396* (0.0278)

0.1437* (0.0276)

Economic growth rate

−0.1078* (0.0290)

−0.1057* (0.0287)

−0.0144* (0.0042)

−0.0152* (0.0042)

Trade openness

−0.0558* (0.0138)

−0.0520* (0.0130)

−0.0044* (0.0010)

−0.0034* (0.0015)

Intercept

6.1500

6.2043

−0.5301

−0.8575

 Buse R2

0.8120

0.7960

0.8530

0.8439

Note: Each model is estimated with 1980–95 data by Ordinary Least Squares; equations are first‐order autoregressive. The table reports OLS unstandardized regression coefficients and panel correct standard errors. All models include nation‐specific dichotomous variables (if t > 1.00).

(*) Significant at 0.05.

(a) Significant at 0.10 level.

of social welfare provision. As to the direct roles of political institutions in shaping 1980 to 1995 changes in income maintenance and social service outlays, findings underscore the importance of all three dimensions of national institutions. This is particularly true for social corporatism and (p.230) decentralization which have, net of the welfare effects of other forces, substantively large and significant impacts in both areas of welfare. Electoral inclusiveness has significant direct effects on variations in cash income maintenance but not on social services.

The second panel of the table reports the direct social welfare effects of public sector debt, direct foreign investment flows, and borrowing on international capital markets; these results largely mirror those for total social welfare effort. The two dimensions of capital mobility are largely unrelated to social welfare provision; in the case of foreign direct investment, international capital mobility is positively related to social services. With regard to this latter finding, the relationship between increases in social services (some of which are directly oriented to assisting families manage the combined demands of work and family life) and rises in international capital markets is consistent with the view that policy makers in (some) welfare states have increasingly emphasized human capital development and associated social policies in an era of internationalization of markets (e.g. Garrett 1998a, 1998b; Esping‐Andersen 1996a). With respect to the direct effects of domestic fiscal stress, findings suggest that the impacts of fiscal imbalance on social welfare provision come primarily through downward pressures on social services. In fact, this finding is consistent with the conventional notion that more ‘discretionary’ social service budgets are relatively easier to cut than basic social insurance budgets whose relative importance to economic well‐being, longer‐lived status, and legal foundations pose relatively greater political difficulties for retrenchment‐minded incumbent governments. However, as I show below, in institutional contexts that are conducive to welfare retrenchment, rises in debt have negative effects on cash transfers.

Finally, the last panel of Table 7.6 reports evidence to suggest that the elderly population; unemployment, inflation, and economic growth rates; and the level of affluence play important roles in shaping contemporary income maintenance and social service spending. Trade openness (insignificant in general models of total social welfare effort) has small negative impacts on both income maintenance and social service outlays. Contrary to findings on the direct effects of international capital mobility, this finding is consistent with contemporary proponents of the internationalization explanation for welfare state retrenchment. However, in the absence of analysis of more detailed measures of trade flows, competitiveness indicators, and related dimensions of markets for goods and services, it is difficult to draw strong conclusions about the welfare policy roles of trade.

Further evidence on the effects of international capital mobility and fiscal stress, and findings about the paper's central proposition—that domestic and international pressures are mediated by national political institutions—are reported in Tables 7.6A and 7.6B. As the first table reveals, the findings reported above on the central role of political institutions in shaping social (p.231)

Table 7.6A. The Mediation of Domestic and International Pressures on Cash Income Maintenance by Political Institutions

Public sector debt

Direct foreign investment

International capital markets

Political institutions

Social corporatism*domestic or international factor of column

0.0218* (0.0089)

0.3549* (0.1181)

0.2594* (0.0741)

Electoral inclusiveness*domestic or international factor

−0.0045 (0.0066)

0.0364 (0.0674)

0.2318* (0.0701)

Decentralization*domestic or international factor

−0.0144a (0.0109)

−0.0846 (0.1307)

−0.2814* (0.0798)

Welfare state programme structure

Universalism*domestic or international factor

0.0079* (0.0024)

0.1173* (0.0334)

0.0420* (0.0241)

Conservatism*domestic or international factor

−0.0044a (0.0024)

0.0142 (0.0238)

0.0100 (0.0196)

Liberalism*domestic or international factor

−0.0038* (0.0021)

−0.0493* (0.0271)

−0.0406* (0.0117)

Note: Interactions are estimated in the focal equations of Tables 7.6. Estimation is by OLS with AR1 autoregressive parameters and panel correct standard errors. Direct effects of Table 7.6 factors remain universally unchanged in terms of significance levels and substantive effects and, hence, full table results are not reported. Tables are available from the author.

(*) Significant at 0.05 level.

(a) Significant at the 0.10 level.

See Appendix Table 7.3 for full results of interaction analysis.

Table 7.6B. The Mediation of Domestic and International Pressures on Social Welfare Services by Political Institutions

Public sector debt

Direct foreign investment

International capital markets

Political institutions

 Social corporatism*domestic or international factor of column

0.0015 (0.0027)

0.1096* (0.0317)

0.0117 (0.0211)

 Electoral inclusiveness*domestic or international factor

−0.0010 (0.0012)

−0.0013 (0.0400)

0.290* (0.0165)

 Decentralization*domestic or international factor

−0.0037* (0.0020)

−0.0244 (0.0251)

−0.0213 (0.0220)

Welfare state programme structure

 Universalism*domestic or international factor

0.0009a (0.0006)

0.0332* (0.0123)

0.0105a (0.0070)

 Conservatism*domestic or international factor

−0.0008 (0.0006)

−0.0052 (0.0056)

0.0058 (0.0060)

 Liberalism*domestic or international factor

−0.0008* (0.0006)

−0.0252* (0.0084)

0.0058 (0.0060)

See note to Table 7.6A.

(p.232) welfare responsiveness to domestic and international pressures are reproduced in the case of cash income maintenance programmes. Thirteen of the eighteen interactions between domestic and international forces, on the one hand, and political institutions, on the other, are moderately significant; eleven are significant at conventional levels and beyond. In fact, the role of institutions in mediating the impact of domestic fiscal stress is more pronounced in the case of income maintenance than in the instance of total social protection. Fiscal stress is negatively related to income maintenance at low levels of social corporatism and universalism and at high levels of decentralization of policy making authority and liberalism (information for derivations from interactions is reported in Appendix Table 7.3).

Turning to the findings on the role of political institutions in mediating domestic and international pressures on social services, Table 7.6B suggests that political institutions, while not irrelevant, are less systematically import‐ant to determining the magnitude of social welfare impacts of fiscal stress and international capital mobility. Only eight of the possible eighteen relationships are statistically significant. However, in the case of direct foreign investment for instance, the findings indicate that welfare states nested in the context of social corporatist interest representational systems and universalistic programme structures will experience fewer cuts in social services than liberal welfare states. Indeed, using the mathematics of interactions, one can derive that an increase in the level of foreign direct investment of 1 per cent of GDP is associated with a decline in social services equal to 0.23 per cent of GDP in strongly liberal welfare states (i.e. those scoring 12 on the liberalism scale: Canada and the United States). Although limited in number, additional institutionally mediated effects of domestic and international forces exist for social services.

5. Conclusions

The preceding analysis has presented theory and evidence that highlight the importance of national political institutions in shaping social policy responses to domestic fiscal stress and internationalization. I have emphasized the ways in which institutions provide (restrict) opportunities for collective action, influence the political power of social groups and classes, and promote (impede) the development and maintenance of value orientations that support social welfare states. To summarize the central findings about rising fiscal imbalance, internationalization, and institutions, the welfare state effects of notable domestic and international pressures vary systematically across institutional contexts: where institutions of collective interest representation—social corporatism and inclusive electoral institutions—are strong, where (p.233) policy making authority is centralized, and where the welfare state is based on the principle of universalism, the effects of fiscal stress and international capital mobility are absent, or they are positive in the sense that they suggest economic and political interests opposed to neoliberal reforms, or adversely affected by domestic and international structural changes, have been successful in defending the welfare state. Where corporatism and electoral inclusiveness are weak, where the dispersion of authority is substantial, and where liberal principles structure the welfare state, rises in public sector debt and international capital mobility are associated with downward pressures on social welfare provision. Indeed, countries such as Canada and the United States, and to a lessor extent Australia and the United Kingdom, seem particularly likely to have experienced a broad array of domestic neoliberal policy reforms and budget cuts with rises in debt and capital mobility.

On a more cautionary note, I should emphasize that the findings of the present study suggest a number of possibilities with regard to further welfare state retrenchment and the future impacts of demographic shifts, domestic economic stagnation, fiscal imbalances, and internationalization in domestic policy determination. First, my findings indicate that if social corporatism is further weakened (e.g. widespread declines in corporatist interest intermediation comparable to 1980s and 1990s changes in Sweden), a viable institutional mechanism for labour to resist domestic and international retrenchment pressures will be significantly diminished. Second, to the extent that authority for national policy formulation is further devolved to subnational jurisdictions (e.g. the contemporary USA and elsewhere) or to the extent power is ceded to supranational federations, most notably the European Union, the findings presented here suggest that the development of the resultant systems of decentralized authority may undercut the maintenance or expansion of systems of ample social protection (also see Huber and Stephens 1992). Finally, to the extent that governments continue to implement means‐testing and related neoliberal principles (see Swank 2000, forthcoming, for a survey of this development in social democratic, Christian democratic, and liberal welfare states), the theory and results presented above suggest that the chances for further retrenchment in the face of domestic and international pressures increase in the long term. In each case, more significant cuts in the welfare state are likely as institutional mechanisms that support the welfare state are eroded. However, as noted in the introduction, the pace of these changes and their scope seem contingent on the conscious policy choices of elected governments and the institutional contexts in which they operate.

(p.234)

Appendix Table 7.1. Principal Variables

TOTAL SOCIAL WELFARE

Total government expenditure for social welfare programmes as a percentage of GDP.

CASH INCOME MAINTENANCE

Total cash benefits for families and sick pay, the elderly, disabled, and unemployed as a percentage of GDP.

SOCIAL SERVICES

Total public social services to families, the elderly, and disabled as a percentage of GDP.

DIRECT INVESTMENT

Average (lags 1, 2, and 3 years) of inflows and outflows of direct investment as a percentage of GDP.

CAPITAL MARKETS

Average (lags 1, 2, and 3 years) of borrowing on international capital markets (e.g. bonds, equities, bank borrowing) as a percentage of GDP.

SOCIAL CORPORATISM

Weighted standard score index of union density, confederal power, and level of wage bargaining. Union density is measured as union membership (excluding retired, self‐employed, and unemployed) as a percentage of wage and salary workers; confederal power is an unweighted standard score index of power of appointment, veto over wage agreements, veto over strikes, control of strike funds, and involvement in wage setting of largest union confederation; and level of wage bargaining is a 1–4 scale of centralization of wage bargaining in the economy.

INCLUSIVE ELECTORAL INSTITUTIONS

Weighted standard score index of degree of proportional representation (0, 1, 2 scale) and effective number of legislative parties (computed using the Laasko and Taagepera index as presented in Lijphart 1984: 120).

DECENTRALIZATION

Weighted standard score index of federalism (0, 1, 2 scale) and of authority bicameralism (0, 1, 2 scale).

PRESIDENTIALISM

Dichotomous variable for presidential (1) and parliamentary systems (0.00).

OLD

The percentage of the population 65 years of age or older.

UNEMPLOYMENT

The percentage of the civilian labour force unemployed.

INFLATION

Year‐to‐year percentage changes in the Consumer Price Index.

GROWTH

Percentage change in real GDP.

AFFLUENCE

Per capita GDP in constant (1985) international prices (1000s of dollars)

LEFT

Annual percentage of left party cabinet portfolios.

CHRISTIAN DEMOCRAT

Annual percentage of Christian democratic party cabinet portfolios.

TRADE

Real imports plus real exports as a percentage of real GDP.

(p.235)

Appendix Table 7.2. Data Sources

International variables

Total inflows and outflows of direct investment, portfolio investment, and bank lending in millions of current US dollars. Source: IMF, Balance of Payments Statistics (Washington, DC, selected years).

Total borrowing on international capital markets. Source: OECD, International Capital Market Statistics (Paris: OECD, 1996).

Exports and imports of goods and services in millions (billions for Italy and Japan) of national currency units. Source: OECD, National Accounts of OECD Member Countries (Paris: OECD, various years).

Data for computation of variables measuring the social welfare state

Total social welfare outlays. Source: for 1991–3, and for some countries for 1980–93, OECD, Social Expenditure Statistics of OECD Member Countries, Labour Market and Social Policy Occasional Papers, No. 17 (Paris: OECD, 1996). For 1961–91, OECD, New Directions in Social Policy in OECD Countries (Paris: OECD, 1994); for 1994–5, Work File, OECD Social Expenditure Data Base, 1998.

Cash income maintenance and social services. Work File of the OECD Social Expenditure Data Base, 1998.

Political data

Left party cabinet portfolios as a percentage of all cabinet portfolios. Source (for portfolios): Eric Browne and John Dreijmanis, Government Coalitions in Western Democracies (Longman, 1982); Keesings Contemporary Archives (selected years). Sources (for classification): (1) Francis Castles and Peter Mair, ‘Left–Right Political Scales: Some ÒExpertÓ Judgments’, European Journal of Political Research, 12 (1984), 73–88. (2) Political Handbook of the World (New York: Simon and Schuster, selected years). (3) Country specific sources.

Political institutions: Union membership. Source: Jelle Visser, ‘Trade Union Membership Database’, Typescript, Sociology of Organizations Research Unit, Department of Sociology, University of Amsterdam, Mar. 1992; ‘Unionization Trends Revisited’, Centre for Research of European Societies and Industrial Relations (CESAR), Research Paper 1996/2, Feb. 1996. Data on elements of confederal power, level of wage bargaining, and related union measures are from, Miriam Golden, Michael Wallerstein, and Peter Lange for access to the database, ‘Union Centralization among Advanced Industrial Societies’. (Generally, the Visser data extend to 1993 and the Golden–Wallerstein–Lange data extend to 1992. To estimate 1993 and 1994, necessary for previous analyses, I used country‐specific sources as well as linear extrapola‐tion of data series); Union confederation power in the policy process. Source: Table A.1 in Boreham and Compston (1992); data on degree of proportional representation, federalism, bicameralism, and use of referendums are from Huber, Ragin, and Stephens (1993); data on legislative seats are from Mackie and Rose, The International Almanac of Electoral History and annual ‘political data handbooks’in European Journal of Political Research.

Welfare State Institutions: Ranking of nations of socialism, conservatism, and liberalism, as defined above, are from Esping‐Andersen (1990).

Socio‐economic Data

Consumer price index. Source: IMF, International Financial Statistics (Washington, DC: IMF, various years).

Percentage of the civilian labour force unemployed, population 65 and older, population 15 and under; total population; and wage and salary workers. Source: OECD, Labor Force Statistics (Paris: OECD, various years).

GDP deflator, consumer price index, and Gross Domestic Product (national currency units and current US dollars). Source: OECD, National Accounts (Paris: OECD, various years).

Real Per Capital GDP in constant (1985) international prices. Source: The Penn World Table (Mark 5.6). National Bureau of Economic Research (http://www.nber.org).

Public Sector Gross Debt. OECD, Economic Outlook: Data Appendix (Paris: OECD, various numbers).

(p.236)

Appendix Table 7.3. Estimates for Components of Interactions in Tables 7.5A, 7.6A, and 7.6B

Social corporatism

Electoral inclusiveness

Decentralization of authority

Universalism

Conservatism

Liberalism

Total social welfare

Public debt

−0.0225*

−0.0145

−0.0225*

−0.0451*

−0.0086

−0.0260*

Institution (column)

1.4092*

1.3987*

−3.2774*

0.4824A

0.6813*

−1.1965*

Debt*institution

0.0146A

−0.0089

−0.0173A

0.0055*

−0.0026

0.0024

Direct investment

0.0687

0.0094

0.0281

−0.9299

−0.0064

0.6380*

Institution

0.1537

0.4698A

−2.3163*

0.3540A

0.2703

−0.1520

Direct investment* institution

0.4222*

0.1887*

−0.3666*

0.2173*

0.0317

−0.0964*

International borrowing

−0.1115A

−0.0302

−0.1030A

−0.3517*

−0.1273A

0.1682A

Institution

0.4033

0.3005

−3.1053*

0.5917*

0.1018

−0.1098

International borrowing*institution

0.2236*

0.3275*

−0.3647*

0.0561*

0.0689*

−0.0342*

Cash transfers

Public debt

−0.0124A

−0.0027

−0.0101

−0.0421*

0.0147

0.0118

Institution (column)

−0.4215

0.9027*

−3.3222*

−0.6833*

0.4060*

0.3740*

Debt*institution

0.0218*

−0.0045

−0.0144A

0.0079*

−0.0044A

−0.0038*

Direct investment

−0.0520

−0.0530

−0.0403

−0.6120*

−0.0563

0.2137

Institution

0.0974

0.0542*

−3.9168*

−0.2624A

0.1165

0.2694*

Direct investment* institution

0.3449*

0.0364

−0.0846

0.1173*

0.0142

−0.0493*

International borrowing

−0.1844*

−0.0047

−0.1351*

−0.2773*

−0.0067

0.1545*

Institution

−0.0966

−0.0556

−3.7638*

−0.2225

0.1191A

0.4000*

International borrowing*institution

0.2594*

0.2318*

−0.2814*

0.0420*

0.0100

−0.0406*

Social services

Public debt

−0.0078*

−0.0070*

−0.0085*

−0.0113*

−0.0048

−0.0045

Institution (column)

0.2088

0.0985

−0.2736*

0.0821

0.0069

−0.0006

Debt*institution

0.0015

−0.0010

−0.0037*

0.0009A

−0.0008

−0.0008

Direct investment

0.0387*

0.0503*

0.0440*

−0.1327*

0.0670

0.1226*

Institution

0.0810

0.0289

−0.4469*

0.1320*

−0.0226

−0.0212

Direct investment* institution

0.1096*

−0.0013

−0.0244

0.0332*

−0.0052

−0.0252*

International borrowing

0.0100

0.0113

0.0087

−0.0549A

0.0048

0.0244

Institution

0.2528*

−0.0478

−0.4674*

0.1726*

−0.0529*

−0.0692

International borrowing* institution

0.0117

0.0290*

−0.0217

0.0105A

0.0058

0.0058

(p.237)

Appendix Table 7.4. The Social Welfare State, 1980–95

1980

1985

1990

1995

All countries

Total social welfare

19.9

21.4

22.0

24.0

Cash transfers

13.1

14.4

14.4

16.0

Social services

6.8

7.0

7.5

8.0

Health care

5.5

5.8

6.0

6.3

Non‐health services

1.2

1.2

1.5

1.7

Social wage

38.7

44.0

43.6

43.6

Public health share (% total)

76.8

76.8

76.9

75.8

Universal welfare states

Total social welfare

24.1

24.9

27.3

29.3

Cash transfers

15.2

16.0

17.5

19.4

Social services

8.9

8.9

9.8

9.9

Social wage

54.2

65.3

66.1

65.4

Public health share

84.0

83.6

82.6

80.8

Conservative welfare states

Total social welfare

22.6

24.4

23.7

26.1

Cash transfers

16.2

17.8

16.8

18.5

Social services

6.4

6.6

6.9

7.6

Social wage

32.4

35.4

35.9

35.7

Public health share

78.0

77.9

78.3

77.8

Liberal welfare states

Total social welfare

13.1

13.6

14.9

16.8

Cash transfers

8.0

8.4

9.0

10.3

Social services

5.0

5.2

5.9

6.5

Social wage

29.5

29.9

28.7

29.6

Public health share

68.3

68.1

68.7

69.0

Universal, conservative, and liberal as defined in text

Countries: universal—Denmark, Finland, Netherlands, Norway,

Sweden; conservative—Austria, Belgium, France, Germany, Italy; liberal—Australia, Canada, Japan, United Kingdom, United States.

Total social welfare

Total social welfare expenditures (see Appendix) as a percentage of GDP.

Cash transfers

Cash old‐age and disability, unemployment assistance, sickness, family support, low‐income, and miscellaneous benefits as percentage of GDP.

Social services

Government health care outlays and non‐health services (elderly, disabled, families) as percentage of GDP.

Social wage

Gross income replacement rate for first‐year of benefits for a worker at the average production worker's income.

Public health share

Government health care as percentage of total health care spending.

Sources: The OECD Social Expenditure Data Base 1998, Work File (Preliminary data); OECD. Health Data 98. CD ROM Version (Paris: OECD); OECD Database on Unemployment Benefit Entitlements and Replacement Rates (Paris: OECD, forthcoming).

Notes:

The original draft of this paper was presented at the ‘Workshop on the New Politics of the Welfare State’, Center for European Studies, Harvard University, Cambridge, Mass., 30 October–1 November 1998. I would like to thank the German Marshall Fund of the United States and the Marquette University Committee on Research for generous financial support; Marco Doudeijns and Willem Adema of the Social Policy Division, Organization for Economic Cooperation and Development, for unpublished data; Keith Banting, Francis Castles, Markus Crepaz, John Freeman, Geoffrey Garrett, Miriam Golden, Alex Hicks, Cathie Jo Martin, Paul Pierson, Jonas Pontusson, Dennis Quinn, Michael Shalev, and John Stephens for helpful comments on work incorporated in this paper; the other authors of this volume for comments on the original draft; and Dengming Chen, Craig Goodman, William Muck, and William Nichols for exceptional research assistance.

(1) One might also cite the (indexing based) expenditure impacts of inflation, rising wage inequality, and changes in family and labour market structure. However, in terms of the size of pools of claimants and potential impacts on the welfare state, the potential consequences of demographic and employment trends stand out.

(2) The causes of the internationalization of finance have been analysed extensively. See Carnoy et al. 1993; McKenzie and Lee 1991; Williams 1993; and IMF 1991, for alternative perspectives on technical and economic determinants. In addition, several recent studies have highlighted the importance of the formal deregulation of international and national financial markets and, in turn, the role of political interests, actors, and institutions in shaping this process (e.g. Helleiner 1994; Sobel 1994; Quinn and Inclan 1997). For a synoptic survey of the contemporary literature on these issues, see Cohen 1996.

(3) The fifteen nations included here are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, Norway, Sweden, the UK, and the USA. Ireland, New Zealand, and Switzerland as well as small developed democracies (e.g. Iceland, Luxembourg) are excluded in much of the subsequent analysis because of the unavailability of data on one or more key variables discussed below.

(4) Moreover, Kurzer (1993), Moses (1994), and others have argued that capital's exit option weakens social democratic parties, unions, and neocorporatist institutions, important sources of the development and defence of the welfare state (e.g. Stephens 1979; Hicks and Swank 1992). With respect to revenues, Steinmo (1993: esp. ch. 6, 1994) has argued that redistributive taxation, always a difficult political and economic objective, is made nearly impossible by the ability of capital to move across national borders. Huber and Stephens (1998) have argued that international capital mobility has undercut the ability of governments in large social democratic welfare states to maintain low unemployment and hence the affordability of the welfare state.

(5) Cameron (1978), Stephens (1979), Ruggie (1982), and Katzenstein (1985b) have argued that the income maintenance programmes of small open political economies, and complementary regulatory, labour market, and related policies, are central components of national strategies for adaptation to world markets (and see Garrett 1998a, 1998b; Pauly 1995; and Rodrik 1997 for contemporary extensions).

(6) My theoretical argument draws on important work by Esping‐Andersen (1990, 1996a), Crepaz and collaborators (Birchfield and Crepaz 1998; Crepaz and Birchfield, no date), Garrett and Lange (1991, 1995), Pierson (1994, 1996), and Rothstein (1998b). A more complete statement, esp. with regard to the institutional mediation of internationalization, appears in Swank (1998b, forthcoming).

(7) Rhodes's analysis highlights the importance of negotiated social pacts not only in heretofore relatively corporatist polities (e.g. the Netherlands), but in nations where traditional corporatist institutions and practices have been less developed (e.g. Italy).

(8) The logic of institutional and partisan veto points, or what Birchfield and Crepaz (1998) label competitive and collective veto points, is fully explored by Tsebelis (1995). In that work, Tsebelis argues that the number of institutional or partisan veto players (the latter being, for instance, parties in lower parliamentary chambers) will be related to slower policy change, subject to the degree of divergence among veto players and the cohesion of the constituent units of veto players.

(9) An important exception is the union‐administered (Ghent) system of unemployment insurance present in several universalistic welfare states.

(10) Svallfors (1995) presents detailed panel data on attitudes towards social welfare policies in Sweden. Consistent with the theoretical and formal analyses of Moene and Wallerstein and the arguments and evidence of Rothstein (1998b), Svallfors documents the continuation of high levels of support for much of the Swedish welfare state with one exception: a clear erosion of support for means‐tested programmes, most notably social assistance (cf. Pierson 1994).

(11) Two points are in order. First, although aggregate welfare benefit spending has been criticized as a blunt measure of social policy (e.g. Esping‐Andersen 1990), it is important to note that aggregate benefit spending is highly correlated with more theoretically and substantively important outcomes such as income redistribution (see Korpi and Palme 1998 for the most recent evidence). Indeed, with proper need, income, and business cycle controls, such measures provide useful (albeit very imprecise) summary indicators of benefit generosity and eligibility standards. Second, spending measures in part reflect concrete policy changes that have been made years before; this is particularly true in the case of pensions. (I thank John Myles for drawing this to my attention.) However, many concrete policy changes affect aggregate benefit outlays in the short term (e.g. changes in indexing, waiting days, some benefit and eligibility criteria, and so forth). Many of these reforms, and their determinants, are captured by models such as those used below. Generally, one should confirm results of benefit outlay analysis with models of more concrete policy change. I have done that in complementary work and results reported below are universally confirmed. (For instance, in Swank, forthcoming, an identical pattern of institutionally mediated effects of internationalization are reported for analysis of the ‘social wage’, or income replacement rates for the average unemployed production worker.)

(12) In complementary work on the effects of internationalization on tax policy and on the welfare state during the 1965 to 1993 era (Swank 1998a, 2000), I show that these categorical measures of actual capital flows are significantly associated with total capital flows and measures of capital and financial market liberalization. Results reported below for foreign direct investment and capital market borrowing do not change with alternative measures of international capital mobility.

(13) It is important to note that available data on total social welfare spending across the developed democracies for a long time span contain a number of breaks in the data series. To correctly test the paper's central propositions for the longer time‐series, I control for eight significant series breaks through the use of dummy variables (i.e. 0 for country years before and outside of the country series break, 1.00 for country years following the break). For the 1980 to 1995 period, the OECD database utilized here provides a near universally consistent set of time‐series cross‐section data. I select the late 1970s as a threshold point for two sets of reasons. First, the late 1970s roughly demarcates the acceleration of capital mobility, domestic fiscal stress, the ascendance of neoliberal macroeconomic orthodoxy, and the weakening of Keynesian welfare state policies and political alliances (e.g. contributions to Crafts and Toniolo 1996; Scharpf 1991). Second, data for comprehensive measures of fiscal stress and for disaggregated social welfare effort are only available from the late 1970s on.

(14) For basic models reported in the paper, I used ten‐year averages of the percentage of cabinet portfolios held by parties of the left and Christian democracy. In supplementary analyses, I explored various (re)specifications of these variables (e.g. short‐term and long‐term lags, cumulative years in office, and so forth). Generally, these alternative specifications are consistent with results discussed below for the focal measures. I provide an overview of partisan effects below.

(15) For instance, the correlation between the focal measure of social corporatism and Boreham and Compston's (1992) measure of the incorporation of labour in the national policy process is 0.759 (using 1970–1986 time‐series data from thirteen nations). As to union concentration, I exclude for the present analysis exploration of independent effects of concentration, or the degree to which union members are concentrated in one or few national peak associations.

(16) In companion work (e.g. Swank, 1998b, forthcoming), I have also included the use of referendums as an institutional dimension. However, because Switzerland is excluded in the present work on the basis of data unavailability, I exclude referendums here. Without Switzerland, there is very little variation in the use of referendums across the developed democracies as measured by Lijphart (1984) or Huber, Ragin, and Stephens (1993). Moreover, while the use of referendums ‘loads’ on the decentralization dimension with federalism and bicameralism, its inclusion or exclusion makes no difference to results reported below.

(17) I also examined different features of fiscal decentralization (i.e. the central government's share of total revenue) that have been highlighted in the literature (see Castles 1998). However, measures of fiscal centralization do not ‘load’ on any of the focal institutional dimensions nor do they have direct or indirect effects in empirical models discussed below.

(18) It may be helpful to discuss the interpretation of interactions. Briefly, the interaction of, let us say, X1 and X2, when the explanandum is Y, will tell us whether the effect of X2 on Y varies with levels of X1 (or vice versa). The significance test for the interaction simply tells us whether differences in the effect of X2 at different levels of X1 are significantly different from zero. The interaction term itself, when multiplied by a value of X1 and added to the coefficient of X2, becomes the slope for the effect of X2 at that level of X1. Moreover, standard errors necessary for testing the significance of the effects of X2 at some level of X1 are easily derived (see Friedrich 1982).

(19) In complementary analysis of 1965–93 impacts of a variety of dimensions of international capital mobility, I find that capital mobility interacts with high short‐term deficits. Specifically, when deficits exceed roughly 10% of GDP, international capital mobility begins to exert downward pressures on national welfare budgets. Coupled with results presented here, these findings suggest that policy makers in developed democracies enjoy some latitude in borrowing even in the context of open capital markfiets. It is only when debt and deficits exceed a relatively high threshold (or emerge in an institutional context conducive to retrenchment), that international capital market discipline creates significant downward pressure on the welfare state. In Swank (forthcoming), I present these quantitative findings and a variety of supportive case study evidence.

(20) This section draws from my work on social democratic welfare states in global markets (see Swank 2000, forthcoming, esp. ch. 4).

(21) In addition to these changes in social welfare programmes, 1990 marked ‘the tax reform of the century’ in which marginal rates on individuals and capital were reduced substantially (along with the removal of a variety of exemptions, allowances, and deductions which benefited upper tier earners and corporations).