Early Approaches to Explain the Emergence and Expansion of Welfare States

The early approaches to explain the spectacular expansion of Western welfare states in the 1960s and 1970s were mostly functionalist; they assumed that coping with the consequences of the war and the progression of industrialization created social policy needs and demands as well as economic growth—and hence the ability to meet these demands (e.g., Titmus, 1958; Peacock & Wiseman, 1961; Wilensky, 1975; and the canonical Flora & Alber, 1981). According to this account, the rapidly expanding old-age, accident, widow, illness, and disability social security schemes were primarily a functional response to expanding risks in the context of the waning family unit and community solidarity, caused by industrialization. These explanations of the sociopolitical process remained largely blind to the differences across countries, political actors, and power relations.

The party-centered theories of the 1970s should be regarded mainly as political science’s first genuinely political response to this early functionalism. Two approaches are particularly important: partisan theory and power resource theory. Partisan theory argued that the ideological orientation of parties and governments shapes the generosity of social security programs. Both early and later contributions along this line of reasoning addressed primarily the explanation of public expenditures in different areas of social policy (Castles & McKinlay, 1979 and Hewitt, 1977 are examples of early contributions; more recent contributions include Schmidt, 1997; Castles, 2009; and Zohlnhöfer et al., 2013).

Within partisan theory, what constitutes the basis for the ideological orientation of parties remains controversial. The power resource approach had from the start a narrower theoretical focus (e.g., Stephens, 1979; Korpi, 1983; and Esping-Andersen, 1985 as examples of early contributions and Korpi & Palme, 2003; Allan & Scruggs, 2004; Bradley et al., 2003; Huber & Stephens, 2001 as later ones). This approach conceptualizes social policy as the actual power resource in the class conflict between labor and capital. This means that social benefits are the goal of labor mobilization, because they reduce workers’ dependency on the commodification of their labor power (what Esping-Andersen, 1990 called “decommodification”). On the other hand, it means that social policy achievements are themselves resources for the mobilization and politicization of the “democratic class struggle,” that is, the conflict in the parliamentary arena between social-democratic labor parties and the parties representing the interests of capital. The radically new—and provocative—element of power resource theory was its obvious root in Marxist class analysis, combined, however, with a steadfast faith in and commitment to nonrevolutionary parliamentarian-democratic distributional politics. It brought the powerful proposition that the welfare state could be an effective instrument to overcome capitalism democratically.

Despite these differences in their theoretical foundations, partisan theory and power resource theory agree on a key hypothesis: the stronger the position of social-democratic parties in parliament and the government, the more generous the design of social policy. In the case of the power resource theory, this linear hypothesis was complemented by contributions that pointed out the importance of specific distributional coalitions. Esping-Andersen (1985) emphasized, for instance, the alliance between social-democratic and agrarian interests for the creation of the Nordic welfare state model, and Van Kersbergen (1995) added to this approach a theory on the influence of Christian Democracy regarding the structure of continental European welfare states. These contributions provided a nuanced picture not only of the size but also of the structure of these welfare states: universalistic-egalitarian in the North, stratified in Europe’s center.⁵

Besides the functionalist and party-centered approaches to explaining the expansion of social policy schemes, a third important line of research in this early phase of social-policy research is institutionalist theories. Early institution-based contributions argued that similar structural trends may generate different social-policy responses depending on the degree of institutional decentralization and the autonomy of the state (Heclo, 1974; Orloff & Skocpol, 1984; Skocpol, 1992). In reaction to power resource theory, scholars devised arguments that conceptualized the effects of power configurations to be conditional on institutional veto points (Immergut, 1992; Huber, Ragin, & Stephens, 1993; Bonoli, 2000). These arguments did not refute the basic hypotheses of party-centered theories, but added another variable: veto points open up the system, divide power, and require compromise. Thereby, they moderate the influence of partisan governmental power.

Welfare State Typologies: Two Versions of Regime Theory

The early theories of the emergence and formation of the modern welfare states were followed—consequentially—by literature that categorized the outcome of this formation. The resulting typologies, developed in the early 1990s and the early 2000s, became immensely important for the field of social policy, especially in the form of regime theories. Regime theories are equilibrium theories, because they identify an interdependent and self-stabilizing network of institutions. They are heuristic instruments that postulate a number of ideal-typical configurations, to which real-world social policies can be compared. In welfare state research, two regime theories are particularly important: the “worlds of welfare capitalism” (Esping-Andersen, 1990) and the “varieties of capitalism” (Hall & Soskice, 2001).

Esping-Andersen’s (1990) conceptualization of three “worlds” of social policy, which differ according to their politico-ideological origins and the resulting institutional peculiarities, can justifiably be called the single most important and most cited contribution of social policy research. Based on the power resource theory, Esping-Andersen argues that the Western welfare states differ systematically along two dimensions: the workers’ degree of independence from the labor market as a source of subsistence (“decommodification”) and the degree of social inequality that social policy either creates or mitigates (“stratification”). The social democratic welfare states of Northern Europe feature a generous, universalistic-egalitarian profile of social security schemes, while the liberal, Anglo-Saxon welfare states practice egalitarian, but only means-tested, poverty avoidance. And the continental, Christian democratic welfare states established generous, but strongly stratifying, social insurance systems. Each of the institutional configurations generates, according to Esping-Andersen, a specific political support coalition that stabilizes the configuration in place.

The theory of the “three worlds” was used in countless contributions, was criticized, and was further developed.⁶ Only two objections turned out to be theoretically important and consequential. First, scholars showed that the Southern European welfare states, characterized by low benefit levels and clientelistic practices, need to be distinguished analytically from the other types (Ferrera, 1996; Bonoli, 1997; Manow, 2013). Second, the feminist critique of Esping-Andersen pointed out the necessity to expand the perspective to other policies; authors like Ann Shola Orloff (1993) and Jane Lewis (1993) argued that Esping-Andersen’s focus on wage replacement and its decommodifying effects fails to address the life chances of women. Defamilization is—especially for women—at least as important as decommodification, because it accounts for gender-specific dependencies. Subsequently, scholars paid more attention to family policies, and Esping-Andersen (1999) himself acknowledged the critique by extending his typology in a follow-up study. Today, several studies emphasize the tendency toward hybrid forms of welfare states (e.g., Palier, 2010) that could increasingly blur the differences between the three worlds. These adjustment processes occur, however, in a severely institutionally constrained way, which stabilizes regime differences (Beramendi et al., 2015).

At the beginning of the 2000s, another regime theory emerged that attracted extensive attention in welfare state research—the “varieties of capitalism” (Hall & Soskice, 2001)—because it offered a different interpretation of the creation and role of social policies. The basic assumptions of this approach fundamentally contradict power resource theory: the key driver of social-policy development, according to this account, is not class conflict but the cross-class interests of workers and employers in the most productive firms and sectors of the economy (Mares, 2003). Producer group politics are thus this approach’s linchpin. Social policy is not seen as a decommodifying, redistributive instrument but as an institutional incentive structure for workers (Estevez-Abe et al., 2001). In the coordinated market economies of Northern and Continental Europe, the key industries require workers with firm- and industry-specific skills. But for individuals, it is rational to invest in these “specific skills” only if they are protected by generous state, sectoral, or firm-level social insurance schemes or by employment protection. In the liberal market economies of the Anglo-Saxon world, by contrast, producers need flexibility and workers with general skills. For this reason, social insurance is less important in these countries. Varieties of capitalism conceptualized social policies purely in insurance terms (as opposed to redistribution), and thereby sparked a strong and growing literature on insurance motives as drivers of social policy preferences, which became relevant and visible beyond the actual varieties of capitalism theory (Iversen & Soskice, 2001; Rehm, 2011; Rehm et al., 2012). Also in contrast to power resource theory, which underlies Esping-Andersen’s regime theory, the theory of liberal versus coordinated market economies pursues an individualistic, rationalist, and largely functionalist approach. Social policy is an instrument of successful economic policy and not one of distributional compensation; varieties of capitalism focuses on efficiency instead of distributional policy, and on insurance instead of redistribution.

The debate between the two approaches is not yet resolved and takes place mostly in the field of historical, qualitative research, because their main difference lies in the role of capital for the historical introduction of major social insurance systems (cf. Mares, 2003 or Swenson, 2001 vs. Korpi, 2006 and Paster, 2011). It is difficult to deny that varieties of capitalism can—at best—explain only parts of social policy (such as the most important social insurance programs). Large segments of the welfare state (e.g., social assistance) can hardly be explained without reference to distributive power struggles.

Explanations of the Retrenchment and the Restructuring of Welfare States

The three decades after World War II saw an extraordinarily high level of economic growth, and this growth fostered the expansion of social policy. Since the 1980s, the structural context of social policy has changed dramatically: slowing economic growth and deindustrialization, consequently increasing structural unemployment, demographic aging, and declining birth rates are merely the most important trends that strain social policy and welfare states. In the 1990s, the question emerged whether social policy could even still be funded or whether the globalization-related strengthening of capital interests would radically erode social policy programs (Kuhnle, 2000).

In this environment, Paul Pierson’s (1996, 2001a) “New Politics of the Welfare State” generated enormous resonance in political science. Pierson argued that social policy arrived in an “era of permanent austerity.” This austerity would not lead to a simple inversion of the power resources theory (i.e., to a strengthening of capital interests against labor interests and hence to the dismantling of social policy programs). Rather, the “mature welfare state” is, according to Pierson, institutionally and politically stabilized: the established social policies lead to a quasi-automatic growth of the welfare state, because the demographic and structural transformations make expenditures grow even in the absence of political reforms (Pierson, 2001b). In addition, Pierson made the case that benefit cuts are politically unlikely, because the welfare state creates its own support through power asymmetries, path dependencies, and policy feedbacks. Because larger and larger parts of the population have earned claims on the welfare state through paid contributions (and have become “stakeholders”), its support has grown far beyond the confines of the social democratic electorate. No political party would dare to cut benefits, lest it be punished in the next election.

The evidence for this stability thesis is empirically disputed (Green-Pedersen, 2004; Starke, 2006). In particular, the proponents of power resource theory disagreed with Pierson and criticized his choice of dependent variable. While Pierson shows the stability mainly using expenditure data, they provided analyses that demonstrated the restraining effect of right-wing, conservative governments—even in recent years—on actual benefit levels (Korpi & Palme, 2003; Allan & Scruggs, 2004; Pontusson & Weisstanner, 2018) and public employment (Clayton & Pontusson, 1998). More recent contributions have suggested a useful distinction between policy areas that can explain the institutional conditions under which partisan politics remain effective or, respectively, undercut. Jensen (2012) and Zohlnhöfer et al. (2013) argue that social insurance schemes that target life-course risks and reach large parts of the population—especially the old-age provision—are indeed stabilized by institutional and electoral mechanisms, as Pierson asserts. In these policy fields, partisan differences in policymaking shrink. The fields that entail more redistribution between social classes (because risks are concentrated and benefits accrue mainly to the less privileged, especially unemployment benefits) continue to exhibit ideological partisan differences.

Pierson’s New Politics thesis may hold empirically only in the presence of certain conditions. His contribution to social policy research, however, extends much further. The most important consequence of Pierson’s argument for research on the transformation of the welfare state was steering theoretical attention away from blunt benefit levels towards the institutional design of social policies and its consequences for political dynamics. The key insight of social policy research in the last two decades was essentially a very old one: as Lowi (1972) already noted, policy shapes politics. The political conflict over reforming a policy—against the background of the existing policy—is not the same as the one over introducing a new policy.

From this insight, two strands of literature emerged that explain the recent transformation of social policy from a new, more institutionalist perspective. One growing body of literature concerns “endogenous” distributional effects. It examines how a policy’s distributional effects change over time due to an accumulation of marginal policy reforms (Streeck & Thelen, 2005) or even in the absence of any reform (Hacker, 2004). The underlying idea is that social policy benefits may be dismantled, for instance, by not adapting the policy to a changing situation (what Hacker, 2004 calls “drift”). The Literature on “dualization” offers another example: when fewer and fewer workers fulfill a social policy’s eligibility criteria (e.g., as a consequence of structurally growing atypical or marginal employment), the welfare state is selectively retrenched even if the benefit levels for the (shrinking) core workforce remain stable (Palier & Thelen, 2010; Emmenegger et al., 2012).

Besides these contributions, another strand of literature emerged that theorizes and examines the increasing multidimensionality of welfare politics. Precisely because the institutional distributional effects of social policy are so different, existing welfare states “create” differing stakeholder groups that may cross-cut traditional class conflicts. Giuliano Bonoli (2005) argues, for instance, that the interaction of industrial welfare states with post-industrial societal changes generates “new social risks,” whose opponents and supporters are not adequately captured by the simple labor-capital antagonism. Others contend, in a similar vein, that the left (social democracy and unions) needs to be reconceptualized with regard to the diverging sociopolitical interests of labor market insiders and outsiders (Rueda, 2005; Häusermann, 2010).

In this context, one can also note the emerging literature on migration and social policy, which points out that the generosity of social policy does not necessarily include the integration of migrants into social security systems (Sainsbury, 2012; Eugster, 2013). The explicit exclusion of immigrants to maintain the social-policy privileges of a nationally defined population is called welfare chauvinism. Hence, welfare chauvinism can also split the interests of labor.

The politically relevant groups in all of these contributions (new social risk bearers, labor market outsiders, migrants) are not—or are only insufficiently—taken into account by the existing welfare states because they were tailored historically and politically to the needs of the core industrial workforce. The interests of these groups do not necessarily converge with those of this core workforce and defy the simple left-right categorization that underlies partisan or power resource theory. These more recent approaches require therefore a more complex, multidimensional understanding of sociopolitical distributional conflicts. This multidimensionality of cleavages accounts for flexible and changing coalition patterns, which cannot be captured by the traditional partisan approaches (Bonoli & Natali, 2012; Häusermann, 2012; Beramendi et al., 2015).

Recent Debates and Approaches

Welfare state research is developing at least as dynamically as welfare states themselves. For that reason, it is useful to point out three current debates and trends in social policy research that have not yet been fully developed as distinct theoretical approaches, but that complement or challenge traditional approaches.

First, welfare state research has turned its attention more strongly to political economy approaches. A rise in micro-level centered research questions and explanations is the most visible consequence of this trend. It implies a stronger emphasis on individual-level research questions and designs. The research on the (institutional) determinants of people’s social policy attitudes and preferences is in fact not new (cf. Svallfors, 1997; Larsen, 2008). More recent studies, however, focus more on policy processes and “demand-driven” policy change, and less on attitudes themselves (e.g., Brooks & Manza, 2007; Gingrich & Ansell, 2012; Rehm, 2011, 2016; Rehm et al., 2012; Häusermann et al., 2015, 2016). This trend can be understood as the effort to provide a micro-foundation for the meso and macro theories of social policy development. This aim is all the more important as many traditional theories, like power resource and partisan theory, rest upon the aggregation of assumed, rather than empirically observed, interests.

This point leads to the second more recent and unresolved debate: the question of the explanatory power of classic party-political approaches of social policy (for an overview, see Häusermann et al., 2013). Partisan theory is one of social policy research’s most established approaches. Whether social democratic parties still implement more generous social policy reforms than conservative parties is disputed. But generally, we have seen the explanatory power of the “partisan color” of the government decline since the 1980s (Huber & Stephens, 2001; Zohlnhöfer et al., 2013). What most studies leave unanswered are the reasons for this waning explanatory power. In most cases, they put forward theoretical explanations like globalization constraints, financial pressure, or parties’ ideological convergence, thereby neglecting to a large extent the fact that parties and party systems have changed fundamentally in the last few decades. The emergence of right-wing populist parties, which mobilize mostly workers (Rydgren, 2012), and the transformation of social democracy into culturally liberal parties of the middle class (Kitschelt, 1994; Beramendi et al., 2015) have fundamentally altered the electoral base of traditionally “left” and “right” parties. This transformation has not only created a new dynamic of party competition (Kitschelt, 2001), but has also led to changes in parties’ social policy positions (Häusermann, 2010; Gingrich & Häusermann, 2015).

Finally, a substantive debate on social policy is becoming increasingly important. This debate—beyond the retrenchment or restructuring of existing social policy—is about the goals of social policy in a post-industrial society, because striving to protect income increasingly disregards the reality of eroding labor market security and family structures (Esping-Andersen, 2002). Recent contributions argue that social policy may (need to) transform into “social investment policy,” enabling people ex ante to find gainful employment instead of alleviating poverty ex post through transfers and compensating income loss. Key elements of such enabling and activating policies are investments in (early childhood, regular, and vocational) education, child care services for working parents, and activating labor market measures and training for job seekers (Morel et al., 2012; Hemerijck, 2013; Hemerijck, 2017). This debate on compensating versus activating social policy raises the fundamental question of the research subject and the central “dependent variables” of social-policy research, which can (still) only insufficiently be measured with the field’s usual data sources. The theory building on the politics of social investment policy is only in its infancy (Garritzmann et al., 2017).