The development of the welfare state has been a productive field for studying American political development.Footnote ¹ As a “mature” field of inquiry, the comparative welfare state literature has also been a testing ground for theories of policymaking, democratic governance, or state-society relations, and it has been an area where political scientists and historians could benefit from one another.Footnote ² To the extent that the study of (welfare-) policymaking is internationally comparative, it is drawn toward country-level generalizations about institutional configurations. The ambition of current comparative politics scholarship is aimed at theorizing the development, determinants, and effects of institutional models: distinct worlds of welfare,Footnote ³varieties of capitalism,Footnote ⁴ or patterns of democracy.Footnote ⁵ However, how do such models relate to the complex dynamics of policymaking within each country, and what is the empirical basis of these generalizations about institutional regimes?

One important source of evidence is quantitative macro-comparative analysis, testing hypotheses about associations among aggregate attributes of national social protection systems, political institutions, and the social and economic contexts in which they operate.Footnote ⁶ An equally important empirical tradition in the comparative literature is case study research. This qualitative work is often comparative, just as the quantitative studies are, but its primary goal is not the broad range of its empirical findings, but its ability to observe national institutional structures or patterns of political mobilization in conjunction with the political dynamics of social policy development. “Public policy is the most promising way of watching states in action,”Footnote ⁷ and thus the focus of qualitative case studies is often on instances of policymaking, rather than institutional or socioeconomic structures. Instances of policymaking bring to the fore the actors involved in reform decisions, the preferences they articulate, and the political influence they use in order to reach their goals. Statements about national institutional fmodels implicitly assume that policymaking can be found to be distinctive for each country (or welfare regime), and that political processes create the institutional regimes with their characteristics. Comparative case study research, in other words, examines national models “at work” by moving to a lower level of observation. In this it resembles policy history, the historical analysis of contemporary policy problems.Footnote ⁸ But while scholars of policy history are interested in the development of public policies per se, comparativists often aim at generalizations broader than findings about public policymaking,Footnote ⁹ and historiographic accounts of different countries’ institutional development are important primarily as sources of evidence for claims about macro-institutional structures. In this endeavor, public policies and the welfare state serve as a “convenient window into some larger relation of power.”Footnote ¹⁰

This article addresses some of the methodological questions associated with the standard practice of studying events in the history of national social protection systems and drawing inferences about national models of the welfare state. First, I will highlight some of the “occupational hazards” that are associated with this standard practice. Then I will review the three most important ways in which scholars working in this vein (implicitly or explicitly) connect their empirical focus on political events with the generalized claims about institutional models that they seek to validate. Specifically, I will first discuss the notions of national welfare models as the cumulative result of narrower public policy choices. Instances of social policy reform, according to this view, are data-points; examining them yields pars-pro-toto images of the political logic of the welfare state. Policy history, the second way of relating welfare policies to welfare states, locates specific instances of political choice on a larger policy trajectory. The events become intelligible in the context of a long-term development that is distinctive of the country under review. Policy typologies form the third approach to the study of specific policy choices and the country-level generalizations they may or may not engender. Typologies allow for within-country comparisons, and by this determine the shared—and possibly systemic—features of a wide range of instances of political choice.

The article introduces these three ways of connecting the study of specific social policy reforms to the broader themes of comparative welfare state research. I ask how valid statements about national welfare systems can be made on this basis and explain the rationale of the answer given by each of the three approaches. I close with a plea for policy typologies and point to that comparative tool’s potential contribution in establishing convincing connections between events and macro-institutional models. One of my goals is to describe ways of bridging the gap between the rich literatures on public policy and policy history and the comparative welfare state research with its generalizing ambitions.

QUALITATIVE COMPARATIVE CASE STUDIES: RATIONALE AND POTENTIAL HAZARDS

The study of public policymaking has important insights to offer for internationally comparative institutionalism. Public policy research is important as an evidentiary basis for theories about national institutional models. Such models essentially depict distinct ways of organizing power in a political system,Footnote ¹¹ of dealing with similar problems via different statutory tools,Footnote ¹² or, last but not least, of organizing social protection. This decision to study welfare states as macro-political phenomena to some extent reflects conscious meta-theoretical choices made by comparativists, which are “shaped in large part by the enduring reality of countries, nations, and other large (and imposing) political entities.”Footnote ¹³ In that sense, comparativists respond to intuitively plausible questions about intuitively significant political units—questions that policymakers and the general public may legitimately ask, and to that social scientists may legitimately feel obliged to answer. In addition to this practical and political motivation, broad agreement on the macro-level as the appropriate analytic level also has theoretical and methodological reasons. Explaining the persistence of cross-national differences has been “a shared theoretical project aimed at the middle range”Footnote ¹⁴ that institutionalist scholars had started to carve out for themselves beginning in the 1970s. In the more recent forms of institutional analysis, moreover, explanatory ambitions are drawn toward the macro-polity for a more specific theoretical reason: to the extent that explanations are statements about regimes—that is, configurations of institutions that are “closely linked and jointly constitute a virtually inseparable unity”Footnote ¹⁵—empirical research must focus on these macro-political interconnections and cannot regard their constituent parts (such as the different policy fields or specific events) in isolation. Comparisons of national institutional settings are thus a terrain rightly claimed by institutionalist scholars to answer questions of political significance, gain a comparative sense of proportion in their assessments about policy developments, and advance a shared project of mid-range theory-building.

While theoretical, political, and ontological commitments center on macro-social phenomena,Footnote ¹⁶ efforts at empirically validating such claims work on lower levels. The observational units in qualitative research on regimes are not primarily the institutions of the macro-polity, but institutions and political dynamics in specific policy fields, as well as crucial events, that is, decision making and reform debates at significant turning points in a country’s (social-) policy history. In substantive terms, these specific instances relate to rather narrow, often outright technical issues, such as modes of calculating public pensions or criteria for access to public health insurance systems. Understanding the narrow technical issues is not an end in itself, though. Researchers study them because they reflect central organizing principles of the national models. They convey important insights about the rationale of the institutional regime and show its “political system in action.” Policymaking, therefore, is the stuff of which comparative welfare state research is made, but much of that research is not primarily interested in policymaking per se. The goal remains how to explain national models, and often even larger units, such as clusters of countries that form distinctive welfare regimes or production models.

Deriving rather wide-ranging generalizations about macro-political regimes from rather narrow and issue-specific empirical observations is intuitively plausible for two reasons: On the one hand, events and narrow issues are the only observational units available for certain types of information. Instances of specific policy change are the occasions that bring to the fore the arguments raised, the compromises reached, the coalitions forged, or the new solutions worked out. To learn something about the functioning of an institutional model, one must therefore observe events. On the other hand, specific issues and the settlements reached about rather technical details in their sum constitute the country’s institutional system. Modes of calculating public pension levels or the criteria for access to public health insurance systems are relevant observational categories because they are the building blocks of social insurance systems and hence constituent parts of the macro-polities.Footnote ¹⁷ It seems both necessary and justified to resort to lower-level observational units for empirically validating claims about institutional generalizations. Strong intuition and empirical necessity may not be enough, however, and it might be worthwhile to make the implicit connection between observations about events and broad institutional claims more explicit than scholars usually do. “Unconscious conceptualizing” has certain risks here for three reasons:

First, the dominance of national models and welfare regimes in theories of comparative welfare state development often makes it difficult to make sense of empirical findings that deviate from the model. Institutional regimes provide no language, few concepts, and no sense of proportion for evaluating such observations. While one can always conceive of regime generalizations as ideal types and empirically examine divergence of empirical observations from ideal types, it remains unclear how much theoretical understanding is gained from this effort. Within the boundaries of Esping-Andersen’s Conservative welfare regime, there is no convincing way of systematically thinking about the politics of universalistic family policy, attempts at tearing down organizational boundaries in the fragmented social insurance system, or equalizing benefits for blue- and white-collar employees. All these policies were enacted in the prototype of the Conservative welfare regime, West Germany’s welfare state. And indeed many of the comparative case studies of the welfare state in Germany long downplayed the significance of these tendencies. There is a real danger, therefore, that the strong hypotheses about national models blind us to variability within the regimes, so that descriptively comparative research will reify rather than empirically validate homogeneous organizing principles within the regimes. It then becomes difficult to diagnose processes of slow erosion,Footnote ¹⁸ or to diagnose institutional change as sequences of small and unspectacular decisions.Footnote ¹⁹ Observations are drawn toward the national model, the dominant interpretation, which serves as a focal point and directs empirical attention.

The second reason why making the connection between events and macro-institutional structures explicit might be useful is in the realm of explanation. The price of generalization is typically a set of rather invasive ceteris-paribus assumptions, which allow researchers to focus on only certain causal factors and push others to the sidelines. For instance, left-power-resource accounts of welfare state development implicitly assume that the preferences of workers are essentially the same across diverse countries and historical settings.Footnote ²⁰ Only on that basis are they able to draw general conclusions from their analyses of the organizational and electoral strength of the working class. On the other hand, careful analyses of the preferences of the crucial actors in the polity implicitly assume that these actors are influential more or less regardless of when and in what institutional contexts reform decisions are being made.Footnote ²¹ If they drop that assumption, there are too many variables for them to arrive at a generalized statement about a national policy trajectory. Historical accounts of public policymaking demonstrate that such restrictions in the set of causally relevant variables are unwarranted, but the small number of cases with which qualitative comparisons on the macro-level operate often makes it difficult to work with more complex explanatory settings.

The third reason why inferences from observed political dynamics to macro-institutional regimes are less straightforward than intuitively plausible points to ontological troubles: moving to the more specific public policy choices is justified and required, as argued above, because principles of social security and evolving features of welfare regimes by necessity occur in the form of concrete policies and specific reform decisions in particular policy fields. But then welfare state development may not have their most significant causes on the macro-level. There are very few political decisions directly about the big themes of comparative welfare state research, such as decommodification, universalism, or equality. The prevalent form of social policy choices instead is about tangible things like the conditions under which the self-employed should gain access to the public health insurance system, the differential pension entitlements of the widows of blue- and white-collar workers, or details in the formula for calculating pension benefits. As a result, preferences about welfare policies or political struggles have important pragmatic and situation-specific dimensions. These can be the regulatory implications of social policy that Peter Swenson analyzes,Footnote ²² or they can be straightforward financial advantages, but there are also implications for less tangible aspects such as status recognition or consequences for who controls the present and future design and implementation of important welfare programs. Such causally significant aspects are easily left behind when “climbing up the ladder of abstraction” and theorizing institutional models.Footnote ²³

Therefore, the broad macro-political outcomes—the institutional structures and patterns of stratification that we encounter as welfare regimes—are insightful as overall characterizations of welfare states and important in cross-national comparisons. However, in the politics of welfare reform, such macro-political outcomes reside primarily in the realm of rhetoric, which is also where to locate the programmatic commitments held by social reformers. Overarching ideological commitments do not alone determine policymaking, and more often than not they play a minor role. Organizations or individuals who generally uphold the banner of social solidarity often favor market-oriented policies in specific choices, and advocates of individual responsibility and social differentiation often opt for broadly redistributive schemes. The decisions of social reformers in actual instances of social reform then often seem anomalous from the perspective of macro-political variables. Such reforms can be explained, but only on a level of analysis below the country-generalization and without interpreting observations only through the lens of national institutional models. This is the ontological difficulty: when studying specific policies, one does not find mere miniature versions of broad redistributive principles or regime characteristics. There is no “self-similarity” across analytic levels in the politics of social reform. Different levels of analysis reveal different pictures of the reform dynamic instead of only pictures at different scales.

The conceptual implications of this are substantial: the assumption that properties of the institutional regime or the broad programmatic goals of political camps will always “trickle down” into specific policy choices does not hold.Footnote ²⁴ The assumption that the preferences revealed in specific reform decisions will always “percolate up” and have some congruent manifestation in broader institutional patterns is similarly problematic. Comparative welfare state researchers’ interest in macro-institutional models is reasonable and legitimate, as argued above, but there is no way around studying specific political processes and examining power as it is deployed (say, in rulings on who gets access to welfare benefits) and preferences in their immediate, concrete, and binding expression (say, in the context of parliamentary hearings). These specific events have patterns of support and opposition tied to expectations about specific, tangible effects of policies, and there is much to be gained from making explicit precisely how they relate to the macro-institutional regimes.

This calls for conceptual ideas about the relationship between institutional structure and political process, and it calls for analytic procedures for establishing that relationship empirically. Such claims and methodologies exist, but they often remain implicit. The next sections therefore explicate three ways in which scholars imagine and implicitly theorize the relationship between the specific events they study and the generalizations that interest them in comparative perspective.

EVENTS AS “DATA POINTS” AND INSTITUTIONAL MODELS AS THEIR CUMULATIVE RESULTS

The most straightforward notion of the relationship between political decision and institutional generalization views the latter as cumulative result. In this view, the regime features of different welfare states are the result of institution-building, that is, of all the welfare reforms in a country’s social policy history. These reforms can be major turning points in the country’s welfare history, but as Jacob Hacker has shown, they can also be relatively small and seemingly technical policy changes that over time develop large effects on the functioning of existing institutions.Footnote ²⁵ Here the relationship between the specific policies and the broad generalizations about institutional systems is one of representativeness: the political dynamic of each instance can be seen as typical of the politics of welfare state development in a country, so that studying it will yield broader insights. Peter Katzenstein’s seminal interpretation of the German political economy is perhaps the best example of such an argument: studying the institutions, policy processes, and goals in a range of specific policy fields, he finds similarity in policy processes across these distinct policy sectors—policymaking under conditions of “semisovereignty,” as he calls it—and he concludes that “public policies [are] shaped primarily by the way a county organizes political power . . . , [rather than] the functional imperatives of particular problems.”Footnote ²⁶

Many studies in Katzenstein’s footsteps have set out to examine national institutional models or “policy styles” by conducting in-depth case studies of particular policy sectors or reform processes. Scholars have examined pension policymaking in Germany and have drawn conclusions about the transformation of the German welfare state,Footnote ²⁷ for instance, or they have examined industrial relations in the German metalworking industry and drawn conclusions about the German model of industrial relations.Footnote ²⁸ The underlying assumption of this work is that the process of reform activity is relatively coherent and distinctive in each country and (more) dissimilar across countries, so that by studying the political system “at work” in one sector one will learn something about this coherent approach and ultimately about the institutional model. Historical institutionalist research in this vein is “interested in how policies can help explain institutions, rather than understanding the histories of the policies as an end in itself.”Footnote ²⁹ Thus, for instance, when Ellen Immergut studies health reforms in France, Switzerland, and Sweden, she is not so much interested in the result of these reforms as in the effect of veto players on the ability to enact health reforms at all.Footnote ³⁰ Katzenstein is interested in Germany’s consensual policy process, not in details of policy outcomes in the different sectors. The underlying assumption here is that the issue-specific reform processes can be used as a window for learning more general things about the polity and the policy style. Katzenstein himself has noted the character of his main finding as a political process outcome (rather than a substantive policy outcome). He depicted Germany’s sovereign state as one that favors incremetalism over large-scale policy innovation, but he emphasized that “it is problematic to move from this characterization of West Germany to an evaluation of the ‘success’ or ‘failure’ of different policies.”Footnote ³¹

In the politics of welfare state development, coherence and distinctiveness are difficult enough to make out in political processes. When it comes to generalizations about the results of political processes—the substantive contents and direction of social reform—the sheer variety of substantive policy options across time and policy field sheds doubts on the possibility to make out cumulative patterns. In fact, there is much to be said for the reverse expectation, “that the differences between countries in a given policy area are often not very pronounced, and that there is considerable variance within countries from policy field to policy field.”Footnote ³² Thus, too often the coherent policy style and institutional model enter the analyses without explicit empirical evidence for greater similarity within countries than across countries. If there are doubts about coherence, however, the cumulation-template for choosing specific instances in order to learn something about the polity rests on unstable foundations, and Evelyne Huber and John Stephens are right when they say that “a simple cumulation of the causes of short-term change will result in a misleading picture of the causes of long-term change.”Footnote ³³

Huber and Stephens deal with the problems associated with the notion of cumulatively effective causes of short-term events in a specific way. In their study, generalized statements about the causes of welfare state development are derived from statistical analyses of macro-variables. The general model thus obtained is then applied to historical and comparative case studies. Huber and Stephens thus maintain the notion of case studies as describing roughly representative “data points” that can be used to illustrate the causal model at work and lend further support to its propositions by being able to “establish causality through tracing the links between events and actors’ behavior in the historical narrative” (32). In this conception, specific events do cumulatively make up the welfare state, but these events are not a good empirical basis for establishing that claim. Instead, the causal argument hinges on the statistical analysis, and “narrative and analytical comparisons are oriented by . . . [the] theoretical framework, and the results of . . . [the] quantitative analysis” (115). This way of linking the study of events with the claims about institutional models is typical of comparative work that combines qualitative and quantitative methods. It is problematic because it relies on far-reaching (and not always warranted) assumptions about the causes of supposed coherence within national models. These assumptions are postulates about the structural conditions that push all public policy decisions in a similar direction in reform after reform, so that a distinctive pattern obtains in each country. Qualitative case studies do not do much analytic work here at all. They primarily serve illustrative purposes, lending additional plausibility to arguments that essentially rest on broad comparisons of macro-level attributes of national models across counties and time periods. Generalization, then, is relatively unproblematic, simply because the lower-level units of analysis play no essential role in the argument.

So there is a choice in the conception of specific events as representative instances: either a grid of macro-level comparisons establishes a general explanatory claim or the case studies merely illustrate that claim. In that case, historical narrative plays virtually no analytic role. Or case studies of events and policy fields are taken seriously as explanatory tools; in that case, their status as “proof” for propositions about the macro-institutional model remains ultimately uncertain. Many of these interpretations of macro-institutional characteristics are enormously convincing and successful (Katzenstein’s is the first example), but this success does not rest on a transparent analytical procedure linking specific events to institutional generalizations.Footnote ³⁴

FORMATIVE MOMENTS IN POLICY HISTORY

The second way of relating instances of decision making to statements about institutional models studies specific events not because they are typical, but because they are constitutive. This answers the generalization question by pointing to the sheer importance of the events studied. These are often the formative moments or the crucial turning points in a country’s (social policy) history, where broader significance is not a question. Studying these crucial milestones in the development of a country’s welfare state is important in and of itself—after all, as path-dependence arguments emphasize, they leave their mark on all subsequent social policy legislation. Large parts of the case-study literature in qualitative welfare state research implicitly endorse this theory of the connection between reform processes and institutional structures. Confidence about the general significance of the findings about the New Deal, the Bismarckian social insurance legislation, or the postwar reconstruction of European social security systems during the 1950s is based on a notion of critical junctures, according to which formative events at crucial turning points will have lasting impact on a country’s subsequent history.Footnote ³⁵ Arguments that use this template do not claim that the reform episodes studied are in any sense typical or representative of welfare policy in that country.Footnote ³⁶ They do claim, however, that their case studies identify some of the crucial causal dynamics that will persist beyond the scope of their inquiries. Persistence of the policy pattern and the institutional form can result from causal dynamics that remain salient beyond events examined directly in the case studies.Footnote ³⁷ Persistence can also result from the self-reinforcing processes and feedback effects that analysts of path dependence have examined.Footnote ³⁸ Case studies of formative moments or crucial turning points therefore rest on an implicit theory of critical junctures, which locates the study of formative events in a broader time frame. Very often this has remained implicit, which in turn has called into question the generalizing causal claims about formative events. Critics of formative-events reasoning have called for explicit analyses of the broader temporal contexts in which the crucial turning points are situated.Footnote ³⁹

This criticism emphasizes the temporal relationship between context, political process, and the institutional structures and seeks to avoid the flaws of “snapshot-views” of crucial events in the history of a policy field. The findings of such short-term analyses are biased, the argument goes, because they focus only on the immediate sources of policy change, the catalysts, and tend to leave out the slow-moving, contextual factors that may be important causes of an event.Footnote ⁴⁰ Moreover, because of the short time frame, “snapshot conceptualizations” do not allow for the acknowledgment of the significance of particular actors’ agenda-setting power in limiting the range of feasible policy options.Footnote ⁴¹ They are “unable to assess whether or not the entire policy terrain . . . is tilted in favor or against” the interests of certain actors.Footnote ⁴² Furthermore, with too strong a focus on the enactment of a particular policy, and without enough attention paid to that policy’s more distant lineages and its development after enactment, “snapshot-arguments” are in danger of becoming what Paul Pierson has called “actor centered functionalist” in that they explain any institutional pattern simply by working backward from the established institutions to the presumed interests at work in their formation.Footnote ⁴³

To some extent, the critique of “snapshot-images” is overdrawn because it disregards the implicit notions of critical junctures with which case studies of formative events have always positioned their cases, as described above. Some of the criticism is well taken, however, as it calls attention to the significance of less important political decisions about seemingly small technical details, which develop important effects over time. Accounts of policy development combine detailed analyses of the dramatic choice points in a country’s social policy history with careful investigation of the antecedents and consequences of the reform processes and explanations of less visible dynamics of implementation, routine legislation, and incremental policy change. Jacob Hacker’s Divided Welfare State and Kathleen Thelen’s How Institutions Evolve are superb examples of this genre. They present coherent overall characterizations of their institutional macro-histories, the development of the public/private welfare mix in the United States, and the development of the vocational education system (primarily) in Germany. Jacob Hacker describes the difficulty of this mode of analysis: “For the analyst of a regime, the challenge is to tease out patterns within the whole and try to identify their sources and their effects, while at the same time remaining attentive to the complex individual variations over time and across constituent elements.”Footnote ⁴⁴ As a result, empirical attention is not exclusively focused on the formative events in the policy histories, and the narrative also covers seemingly less important events.

“Cases” in this genre refer to a policy trajectory in the general development of the country’s social protection system or—more frequently—the overall evolution of a particular policy field in a country. In a sense, they present a “processual policy outcome” and then explain the causal mechanisms underlying this overall development.Footnote ⁴⁵ In these explanations, policy feedback and co-evolution (complementary developments in adjacent policy fields) play important roles. Together with the long time frame of analysis, this attention paid to interrelationships across time and policy fields puts policy-development approaches in a good position to connect statements about political dynamics with those about institutional models; they enable the researcher to understand “how institutional complementarities are forged historically because they alone provide the necessary causal links between the specific episodes—across different periods and often involving different political actors and coalitions—in which successive rounds of institutional innovation occur. Such feedback effects establish the raw materials (constellation of players, definitions of interests) with which policy entrepreneurs, as well as defenders and challengers of existing institutions have to work.”Footnote ⁴⁶

Policy-development analysis thus avoids examining individual reform episodes in isolation. It avoids comparing individual reform decisions in a policy history to contrast the political dynamics and contextual factors involved. Comparisons play a role only across countries (in How Institutions Evolve) or policy fields (In The Divided Welfare State), but not among reform initiatives, and rarely within countries or policy fields.

Arguments about policy trajectories have a strong theoretical commitment to finding more or less coherent policy trajectories in national institutional regimes. That is the generalizing ambition in this genre. Policy legacies, that is, the constraints on innovation imposed by an institutional setting, learned problem-solving behaviors, or shared normative beliefs all ensure that a nation’s policy history is not random. Yet it would be useful to go beyond these theoretical positions and develop more explicit analytical procedures for empirically validating judgments about country-typical trajectories of policymaking. There is a certain danger that those remain fairly close to conventional wisdom about the political history of a country. This risk is partly the result of the fact that the moving picture is supposed to be more than a mere random walk of decision making in all kinds of directions and explicitly seeks to present a narrative of some coherence.Footnote ⁴⁷ Strong notions of unified trajectories are risky because they may direct analytic attention only in certain directions, while divergent tendencies might remain below the threshold of selective attention. Jacob Hacker is aware of this, and he introduces internationally comparative perspectives and counterfactual thought experiments to validate his claims.Footnote ⁴⁸ In Thelen’s How Institutions Evolve, the explicitly comparative framework presents alternative scenarios and acts as a precaution against “overly inevitable” conceptions of policy development. These comparative perspectives (across country or across policy field) are important safeguards against parochial views on policy development, and they bring into sharper relief the distinctive features of particular developments.

POLICY TYPOLOGIES

Policy typologies take another approach to reconciling the detailed study of political dynamics with the ambition to say something about a broader institutional pattern. Attention focuses on individual instances of social reform within one country, but instead of trying to identify the common thread throughout the history of a country’s major (and minor) welfare reforms, policy typologies are designed to examine similarities and differences among these moments of political choice in a comparative setting. An early argument in favor of policy typologies has been put forward by Theodore Lowi, who introduced them as tools for comparing policymaking in a systematic way, and whose critique of qualitative (single) case studies echoes many of the concerns outlined above. While acknowledging the richness of qualitative case studies in empirical detail, Lowi is bothered by their seeming inability to “break through the weak and designative vocabulary of public law”Footnote ⁴⁹ and move the theoretical claims that can be derived from case studies beyond their (then) current status as “self-validating standpoints”: “At the end of an empirical study, neither approach [in the realm of qualitative case studies] affords a means for cumulating the data and findings in coherent and logical abstractions with other findings. They merely provide the basis for repeating the assumptions of the beginning. . . . One debilitating handicap of all case studies [is] the problem of uniqueness. . . . How are we to know the extent to which the patterns the authors discover apply to all cases? How are we to know at least toward what class of cases we can generalize?”Footnote ⁵⁰

Lowi expresses the same concerns that Jacob Hacker points out when he sees a danger of case studies merely listing “one damn thing after another—study piled upon study, fact upon fact, without adequate integration, explanation, or advancement.”Footnote ⁵¹ Yet Lowi’s solution is a different one. He proposes policy typologies to identify the important dimensions on which policies differ and work through the political process implications of different types of policies. Such typologies are general interpretative schemes for distinguishing issue areas, because “each arena [i.e., type of policy] tends to develop its own characteristic political structure, political process, elites and group relations,”Footnote ⁵² and as a result “a distinct set of moral and political-process consequences is associated with each kind of governmental commitment or political intervention.”Footnote ⁵³ This is Lowi’s version of the famous dictum that policies shape politics.

Despite similar concerns about the generalizing potential of case studies, Lowi’s approach differs from that put forward by the advocates of policy-development studies described in the previous section. According to their view, policies shape politics by way of the linkages across time that exist among policymaking episodes, where ignoring these feedback effects would miss important aspects of public policy. When Paul Pierson quotes Schattschneider with the dictum that policies shape politics, he refers to these intertemporal connections.Footnote ⁵⁴ In Lowi’s approach, by contrast, policies determine politics not over time, but immediately, by way of their substantively different reform options, which produce systematic expectations about the actors most actively involved in reform episodes, the kind of salient demands that the different actors raise,Footnote ⁵⁵ and, given a certain distribution of political influence, the kind of reform measures likely to be adopted. As policies select the participants in policy processes,Footnote ⁵⁶ they reveal much about the policymaking dynamic in social policy reforms, so that a good policy classification can supply a first step toward both explanation and generalization.

Few studies in the comparative welfare state literature explicitly use the language of policy typologies. In terms of the analytic procedures it employs, however, Isabela Mares’s study about employer preferences in French and German social policy decisions can be “co-opted” into this mode of analysis and in fact represents a very good example: the empirical centerpiece of the study consists of testing a set of theoretical propositions “in a large number of cases, spanning nearly a century of social policy development.”Footnote ⁵⁷ Although “Germany and France have been selected as country cases in an effort to maximize variation along crucial independent variables” (ibid.), the crucial comparative units are instances of major social reform (such as the introduction of accident insurance, the introduction of unemployment insurance, attempts at unifying social insurance immediately after WWII, early retirement policies in the 1980s). Reform options on the agenda in each of these episodes can be described in terms of two crucial causally significant dimensions,Footnote ⁵⁸ and accordingly the constellation of (employer-) preferences that each reform initiative invokes, and the pattern of opposition and support will be a different one in each episode. This kind of typological approach allows for comparison among instances of public policymaking “below” the country-level of generalization. Besides contrasting policymaking episodes in different contexts (at different times and in different policy fields, but within one country), the goal of such comparisons is to identify the causal structure of different types of decisions about the welfare state. Whether or not Peter Katzenstein is right in saying that the logic of the country is more important in shaping public policies than the functional imperatives of different policy fields can then be decided as a matter of empirical observation, rather than as a conceptual supposition.Footnote ⁵⁹ That is a big advantage.

Statements about the macro-polity thus become empirical outcomes. In policy history studies, by contrast, there is the risk that they enter the analysis by way of assumption. Locating each case of welfare reform on the policy trajectory of a country would have to take for granted some minimal degree of coherence, as well as some initial ideas about characteristic features of this national policy trajectory, because the historical work itself provides few criteria for adjudicating between model-conformist observations and path departures. Typologies, by contrast, are a powerful tool for comparative analysis. Alexander George and George Bennett explicitly introduce them as analytic tools for analyses below the country-level of generalization in what they call “within-case-analyses.” Policy typologies also provide a conceptualization for different permutations of one phenomenon.Footnote ⁶⁰ They provide a “conceptual map” of the policy options on the reform agendas, one that indicates both the common features and the variations among policy options. Following Lowi’s suggestion to derive political process consequences from these differences, it becomes possible to identify crucial interests in the different reform options, as Mares did, and to develop middle-range and contingent generalizations about the process and likely result of social reforms. Typologies, George and Bennett say, can be used for both generalizing and differentiating purposes.Footnote ⁶¹ They generalize by emphasizing the systematic connections among types and depict them as part of a system of (interconnected) phenomena,Footnote ⁶² and they differentiate by clearly defining the variables and their values that are used to construct the type concept.Footnote ⁶³ Even if empirically it turns out not to be possible to identify an overarching national model of politics across all times and all policy fields, it may be possible to arrive at a generalization about the causal mechanism that generates public policy decisions. The operation of this mechanism will be contingent on the type of policy decision, but the mechanism might be able to travel across case and context.

It is necessary to emphasize the distinction between policy typologies and regime typologies. The latter are popular among comparativists. Many scholars value their heuristic functions as useful tools for comparing countries and a “convenience to talk about an interaction effect.”Footnote ⁶⁴ Regime typologies lay out a map of the causal structure explaining the institutional regimes described by the types.Footnote ⁶⁵ Remaining thoroughly on the macro-level, they do not gain much leverage over the question of how to link institutional regimes with case studies of policymaking, however. Policy typologies is a tool borrowed from policy analysis. They deliberately juxtapose policymaking episodes and consider them analytically (though not observationally) distinct for the purpose of comparative analysis. This helps solve some of the problems that have been identified as hazards of “snapshot views” of policymaking episodes (such as “defining away” structural context conditions, disregarding the role of agenda-setting power, or overall gradual shifts of the balance of power among the participants of reform controversies). Rather than defining away the structural and contextual conditions of decision episodes and their numerous interlinkages, explicitly comparing these instances establishes controls for contextual influences. The procedure thus holds the promise of contributing to a genuinely political theory of welfare state development, one that recognizes the role of agency and political dynamics in the making of the welfare state and seeks to separate it analytically (though not observationally) from all the structural constraints and contextual factors that also influence social reform.Footnote ⁶⁶ Case studies of policymaking are necessary for understanding the development of welfare states, as argued above, and policy typologies are a very potent way of thinking through the case studies’ potential for generalization about national institutional systems and about the political dynamics that produce and reproduce these systems.