MPIfG Working Paper 00/3, March 2000

Institutions in Comparative Policy Research

by Fritz W. Scharpf 

Prof. Fritz W. Scharpf is Director at the Max Planck Institute for the Study of Societies, Cologne.

1Introduction

Varieties of the "new institutionalism" hold a promise of theoretical integration across several sub-disciplines of the social sciences and a wide range of research fields, including comparative politics, the study of European integration, comparative political economy, comparative industrial relations, or comparative industrial governance (Hall & Taylor 1996; Jupille & Caporaso 1999; Thelen 1999). There are thus good reasons to explore the usefulness of institutional explanations in comparative policy research as well. But in doing so, we need to be aware of the special conditions that complicate their application in this particular field. To begin with, it seems useful to specify the ways in which institutionalist and policy perspectives may intersect.

The policy perspective focuses on two different questions which I described as being "problem-oriented" and "interaction-oriented" (Scharpf 1997, 10-12). Under the first perspective, policy research will analyze the nature and the causes of (societal) problems that (public) policy is expected to resolve as well as the (empirical or potential) effectiveness of policy responses to these problems. Quite obviously, much of the substantive knowledge required in problem-oriented policy analyses will not come from political science but from other disciplines, such as macro-economics, labor-market economics, public health, biology or climatology. Within the perspective of interaction-oriented policy research, by contrast, political-science knowledge is likely to dominate analyses of the interactions among policy makers, and of the conditions which favor or impede their ability to adopt and implement those policy responses which problem-oriented analyses have identified as being potentially effective. Similarly, the study of institutions also includes two distinct perspectives, one focusing on the consequences that institutions may have for actors and actions within their domains, the other one focusing on the genesis of and transformation of institutional arrangements themselves. If we combine these perspectives in a fourfold table, it is clear that interesting research questions may in fact be located in all four cells.

In the first cell, one would locate "functionalist" attempts to explain the existence of specific institutions by their ability to solve certain societal or economic problems - a perspective which, for instance, informs the institutionalism of transaction-cost economics (Williamson 1985). In the second cell, the perspective is reversed, asking how the existence of given institutions contributes to the emergence or avoidance of certain societal or economic problems - a perspective adopted by "structuralist" studies which, for instance, are attributing differences in the rate of inflation to the existence or absence of centralized wage-setting institutions (Calmfors & Driffil 1988) or of independent central banks (Cukierman 1992). In the third cell, research could focus on how institutional change may be explained as the outcome of strategic interactions among purposeful and resourceful actors - a perspective which, for instance, has been highly productive in explaining major institutional changes in the European Union as the outcome of strategic interactions among national governments (Moravcsik 1998).

The fourth cell, finally, identifies the perspective on institutions that is characteristic of interaction-oriented policy research. It treats institutions as one set of factors affecting the interactions among policy actors, and hence the greater or lesser capacity of policy-making systems to adopt and implement effective responses to policy problems. This is the perspective adopted in the present article. It differs from the "structuralism" in cell 2 by its actor-centered character. Actors and their interacting choices, rather than institutions, are assumed to be the proximate causes of policy responses whereas institutional conditions, to the extent that they are able to influence actor choices, are conceptualized as remote causes (Scharpf 1997). But since actors differ in their orientations and capabilities, and since we also need to take account of the problem perspective, the search for, or the use of, institutional explanations faces characteristic difficulties in the context of in interaction-oriented policy research. These difficulties will be discussed in the following section before I will then turn to the uses of institutional explanations in comparative policy studies.[1]

2The Elusive Quest for Generality  

To a greater degree than is otherwise true in the social sciences, policy research aspires to pragmatic usefulness in the sense that it should be able to provide information which (if heeded by policy makers - which is another question altogether) could contribute to the design of effective and feasible policy responses to given societal problems. At the practical end of a continuum, this calls for in-depth analyses of specific policy problems and interaction constellations that may best be done by consultants or by the in-house staffs of ministerial departments and other policy-making organizations, rather than by theory-oriented academic research. At the same time, however, such applied work would greatly benefit from being able to draw on empirically validated theoretical propositions specifying general causal mechanisms affecting the feasibility and effectiveness of policy options

Given the multi-dimensionality and variety of real-world policy problems, however, any general theoretical proposition can at best cover partial aspects that need to be integrated with other partial theories in the development of effective policy designs - just as the solution of any complex engineering problem will draw on a wide range of distinct natural-science theories. But in policy research even the search for partial theories is affected by the real complexity of its subject matter. While the natural sciences can often rely on experimental designs in order to isolate the causal effect of a single factor, this is not usually possible in the social sciences. Here the comparative empirical study of real-world phenomena is generally our only way to discover causal relationships. But regardless of whether comparison is inter-temporal, cross-sectional, or cross-national (which will be my focus here): If institutional conditions are thought to matter, they are in themselves very complex "factors" with a high degree of variability across time and space. Moreover, if the dependent variable is to be policy responses, interaction-oriented policy research must also consider at least two additional sets of factors that are likely to have causal influence - the characteristics of the policy problems faced, and the characteristics of the policy actors involved. These conditions constrain the design of theory oriented and empirical policy research (Scharpf 1997, chapter 3).

The standard way of dealing with complex factor constellations in empirical research is through multivariate statistical analyses that seek to identify the causal effect of specific variables while trying to control for the influence of other factors. Since internationally comparative policy studies are inevitably plagued with the "small-n problem" of too many variables and too few cases, it has become common practice to multiply the number of available observations by relying on cross-country pooled time-series data (Beck & Katz 1995). Their usefulness is limited, however, by the fact that some of the factors that influence outcomes may be both country-specific and relatively stable over time, so that the multiplication of observations does not increase the available information to nearly the same degree. The same is true if exogenous shocks (like the oil-price crises of 1973-1975 and of 1979-81) affect all countries at the same time. If these "fixed effects" are then accounted for by the introduction of country and year dummies in the regression equations, what is left is statistical information about relationships among variables that are cleaned of all influences that are specific for a given country or a given time period.

In the field of comparative political economy, there is a growing and methodologically sophisticated literature relying on these remedies for the small-n-disease. They seem most useful for the identification of stable ceteris-paribus relationships of an essentially "structural" character. Examples are studies of the relative influence of economic growth, trade openness, capital mobility and other "economic" factors on the one hand, and of the political and organizational strength of left-of-center political parties, of labor unions, or corporatist institutional arrangements and other "institutional" factors on the other hand, on levels of taxation and social spending (Rodrick 1997; Garrett 1998; Garrett & Mitchell 1999; Hicks 1999). Similar methods are used to assess the effects of wage dispersion on private service employment (Iversen & Wren 1998), of central-bank independence on inflation (Cukierman 1992 ), or of more or less centralized wage-setting institutions on inflation and unemployment (Calmfors & Driffil 1988; Soskice & Iversen 1999; Iversen 1999). From the perspective of developing politico-economic theory, these are very useful studies, even though their reach is constrained either by the availability of quantitative time-series data that are cross-nationally standardized, and by the need to reduce complex qualitative factors to quasi-quantitative indicators or to dummy variables.

From a policy perspective, however, the information that is screened out through the use of country and year dummies in pooled cross-section time-series regressions may be more important than the statistical regularities that are discovered. Exogenous challenges may change radically over time, and even if all countries were confronted with the same challenge at a given time (say, the dramatic rise of real interest rates in the international capital markets of the early 1980s), countries with different economic and social structures, and with different policy legacies may differ greatly in their vulnerability and hence in problems that their policy systems must deal with. Moreover, even where these "fixed effects" play no role, the information used in multivariate regressions is generally too "thin", and the probabilistic effects identified are too uncertain to provide much guidance for policy choices in specific historical constellations.

Turning instead to historical studies of single cases, which may do justice to the complexity of interacting factors, cannot be the answer, since these will not allow lessons to be derived for other cases, let alone cumulative theory development.[2] For some purposes, it may be useful to combine deductive theoretical work and statistical analyses of some empirical indicators with case studies that explore the more complex specific antecedent conditions (Coppedge 1999). An example is Iversen's (1999) theoretical and econometric analysis of the joint effects of (accommodating or non-accommodating) monetary policy regimes and (more or less centralized) wage-setting institutions on inflation and unemployment, backed up by more detailed historical explorations of developments in a few European countries. Here, statistical analyses are used to identify probabilistic regularities, while the influence of various (and changing) institutions on policy choices is presented in the narrative mode. For Iversen's purposes, that seems a perfectly valid solution.[3]

But what if we are primarily, or at least equally, interested in empirically supported generalizations about the influence of a greater variety of policy-making institutions on the capacity for effective policy responses in a wider range of policy areas? From what I have said, it follows that comparative case studies are likely to be better suited to this task than multivariate regressions based on quantifiable data and indicators. But it also follows that in order to arrive at potentially generalizable conclusions, we must then find ways to cope with the excessive variety and complexity of causal constellations. This can often be achieved by focusing the comparison on a subset of cases in which it is possible to hold other contingent conditions sufficiently constant to allow the influence of institutional variations to be identified with some confidence. But in order to do so, we must have a theoretical understanding of the contingencies that we are trying to hold constant.

3Contingencies

The effect of institutional conditions on the effectiveness of policy choices is contingent on two broad sets of non-institutional factors - the nature of the problems or challenges that policy is supposed to meet and the normative and cognitive orientations of policy actors involved. Both are generally ignored in "structuralist" theories asserting the unconditional superiority or inferiority of certain policy-making institutions, such as "neo-corporatism" or the "Westminster" model.

Thus the multiple-veto characteristics of German political institutions may indeed impede policy changes (Tsebelis 1995) - but compared to the stop-go policies facilitated by the single-actor British political system, this condition favored successful German economic policy from the 1950s to the end of the 1980s. In the 1990s, however, when new economic challenges would have required major policy changes, the same institutional conditions are considered causes of German policy failures (Manow & Seils 2000). Conversely, institutional conditions and the challenges arising from the external policy environment did not change between 1978 and 1979 in Britain, or between 1983 and 1984 in New Zealand. Nevertheless, radical policy changes were brought about by new governments whose cognitive and normative orientations differed from those of their predecessors (Rhodes 2000; Schwartz 2000). Both of these contingencies can be specified more precisely.

3.1Policy Challengens

In the political economy of advanced welfare states, policy challenges are themselves a complex concept that is best defined by the interaction among three sets of factors - changes in the policy environment impacting on more or less vulnerable socio-economic structures and on more or less vulnerable policy legacies.

For examples of recent changes in the policy environment, one may think of the two oil price crises, of the fall of the Berlin Wall, of the completion of the European internal market and of the European Monetary Union, but also of demographic changes resulting in the "graying" of the population, or of changes in family structures increasing the labor-force participation of women. Such changes will often affect several or many countries at the same time. But whether and how these will lead to policy problems is likely to be conditioned by the characteristics of domestic socio-economic structures and policy legacies.

The importance of differences in socio-economic structures is obvious: The second oil-price crisis constituted a different type of policy challenge for oil-exporting Norway than it did for oil-importing welfare states. Similarly, the Danish economy, which is dominated by small, family-owned enterprises, was less affected by recent increases in capital mobility than was true of Sweden, where large, multinational enterprises play a major role (Benner & Vad 2000). The second set of mediating factors is described by the concept of policy legacies (Skocpol & Weir 1985) which refers to existing policies and the practices, and expectations based on them. Since their theoretical importance is less obvious, a clarification seems useful.

Even though policy legacies are the product of past political choices, they are not necessarily at the disposition of present policy makers. For one thing, not everything can be changed at the same time in modern, highly differentiated policy systems. In any one policy area, therefore, the body of existing policy in other areas must mostly be considered an invariant environment of present policy choices. But even where changes are considered, they are likely to be impeded by policy inertia. In multi-actor policy systems with high consensus requirements, innovators will be at a competitive disadvantage in interactions with the beneficiaries and defenders of the status quo (Scharpf 1988). But even if formal (majoritarian) decision rules do not have a conservative bias, the status quo is favored by the fact that a proposed innovation must not merely appear superior in the abstract, but that the expected gains must be big enough to also cover the costs of transition - and that these are likely to increase with the extent to which the new policy would depart from status-quo solutions. In extreme cases, the status quo may be protected by "lock-in effects" (David 1985; Arthur 1989; 1990), and even when policy change is feasible, it is likely to be "path dependent" (Pierson 1996; 1997; Thelen 1999) in the sense that only certain goals can be reached from a given starting position.

As a consequence, similar changes in the policy environment may constitute problems differing in nature or in severity, depending on the accidental goodness-of-fit between these changes and existing national policy legacies. Thus, the policy problem generated by the first oil-price crisis was massive job losses in countries like Germany, where the monetary regime was non-accommodating, while rampant inflation became the dominant problem in Britain and other countries with an accommodating monetary legacy (Scharpf 1991). Similarly, the "graying of the population" constitutes a massive financing problem in welfare states whose earnings-related pensions are financed by the "pay-as-you-go" contributions of a shrinking active population, whereas "funded" pension systems are much less threatened. Under conditions of increasing international competition and capital mobility, moreover, the characteristic differences in the policy legacies of Scandinavian, Continental, and Anglo-Saxon welfare states (Esping-Andersen 1990; 1999), of national production systems (Soskice 1999; Estevez-Abe et al. 1999) and of industrial-relations systems (Crouch 1993; Golden et al. 1999) have come to matter very much for the problem load that must be faced by national policy systems (Scharpf 2000a; Scharpf & Schmidt 2000).

3.2Actor Orientations

In combination, external changes impacting on given socio-economic structures and policy legacies will create time and country-specific patterns of vulnerabilities which, in democratic polities, are transformed into challenges to which policy actors may have to respond. That these responses are influenced by the institutional setting is generally assumed, but the dominant strands of current institutionalist theorizing, "rational-choice institutionalism" and "sociological institutionalism" (Hall & Taylor 1996; Thelen 1999), [4] differ in their conceptualization of these influences, even though both make quite strong claims about the value of institutions as predictors of what actors will in fact do.

In sociological institutionalism, institutions are defined very broadly, so as to include not only externally imposed and sanctioned rules, but also unquestioned routines and standard operating procedures and, more importantly, socially constructed and culturally taken-for-granted world views and shared normative notions of "appropriateness" (Berger & Luckmann 1966; March & Olsen 1989; DiMaggio & Powell 1991; Zucker 1991). In that view, therefore, institutions will define not only what actors can do, but also their perceptions and preferences - and thus what they will want to do.

In rational-choice institutionalism, by contrast, institutional rules are understood as external constraints and incentives structuring the purposeful choices of self-interested rational actors (Shepsle 1989; North 1990). But since actors are also assumed to have standardized and stable preferences defined by their personal or organizational self-interest, and since they are assumed to be rational in the sense that their perceptions can be taken to be correct representations of the objective situation, and that their cognitive capabilities are sufficient to identify the consequences of available options for their self-interest, the knowledge of institutions will also allow predictions and explanations of what actors will in fact do. As George Tsebelis (1999a, 4) has put it: "Since institutions determine the choices of actors, the sequence of moves, as well as the information they control, different institutional structures will produce different strategies of the actors, and different outcomes of their interactions."

In the light of empirical policy research, however, both of these theoretical positions appear much too deterministic.[5] Above, I referred to the examples of Britain in 1978/79 and New Zealand in 1983/84, but the point can be stated more generally: Even if external and institutional conditions remain constant, policies can change if the cognitive and normative orientations of policy makers change - as they did when Keynesianism was replaced by monetarist and neo-liberal economic paradigms (Hall 1992; 1993). In our own framework of "actor centered institutionalism" (Mayntz & Scharpf 1995; Scharpf 1997), we therefore find it necessary to treat actor orientations (i.e., their preferences and perceptions) as a theoretically distinct category - influenced, but not determined by the institutional framework within which interactions occur.

In our view, actor preferences[6] have at least two dimensions, individual and organizational self-interest on the one hand, and (internalized) normative obligations and aspirations on the other hand.[7] In the organization-theoretic literature, this corresponds to the distinction between "system maintenance" and "goal attainment" (Etzioni 1964, 16-19), and there it is generally assumed that in cases of conflict the former takes precedence over the latter.[8] In any case, the "goals" of corporate and collective actors are strongly influenced by the institutional rules to which they owe their existence and by institutional and cultural norms which define the criteria of their success or failure.[9] For that reason, they will vary greatly between different types of actors - political parties, government ministries, unions, central banks, etc. -, and in time and place. By contrast, the "maintenance" or survival interests in assuring adequate organizational resources, defending organizational autonomy, and (where institutionally relevant) achieving competitive success, are likely to be more uniform and constant - which allows fairly general and reliable predictions of organizational responses to institutional incentives (and hence useful suggestions for institutional design).[10]

Like actor preferences, cognitive orientations are also shaped by institutional norms and incentives defining the role-specific content of "conventional wisdom" and "selective perceptions" (Dearborn and Simon 1958). Central banks will pay more attention than union leaders to early indicators of inflation. Moreover, certain policy areas may be dominated by the specific paradigms (Hall 1993) of particular "advocacy coalitions" (Sabatier 1987; 1999) or "epistemic communities" (Haas 1992). Beyond that, the capacity for effective policy responses is affected not only by the quantity and quality, but also by the diversity of policy-relevant information and analysis provided by an institutionalized information infrastructure. Thus, in the heyday of Austro-Keynesianism in the 1970s, policy coordination in Austria was greatly facilitated by the fact that the government, the political parties and the social partners relied on the analyses provided by a single economic research institute. In Germany, by contrast, unions and employers maintain separate research institutes, the federal government and the Länder support altogether six such institutes; the federal labor administration as well as the Bundesbank maintain large in-house research capacities, the independent Council of Economic Advisors relies on its own research staff, and the big commercial banks also have their own macro-economic research departments. All of these provide not only data, explanations, and policy recommendations for their respective sponsors, but they also publish their often inconsistent findings and contradictory interpretations - which are then debated in the financial press and in the economic and financial pages of the general press, and which allow very different policy proposals to claim "scientific" support.

The downside of the monopoly model are, of course, the risks of "groupthink" (Janis 1972) - that is, of the failure to pay attention to observations, interpretations and recommendations that do not conform to the dominant world view. This was arguably the case in Britain in the early 1970s, when policy makers in the Treasury continued to rely on the Keynesian recommendations derived from the single macro-economic simulation model even though the economy had ceased to respond as predicted (Hall 1992). But when the analyses of institutionalized information monopolists do fit the problem (and if they are believed) they will facilitate effective problem-solving in single-actor systems and effective coordination in multi-actor systems. The pluralistic model, by contrast, will provide protection against the institutionalization of error. But the cacophony of voices may render coherent policy choices difficult even in single-actor systems, and is likely to work against cooperative or at least coordinated problem solving in multi-actor systems.

4Dimensions of Institutional Effects

We thus have three major sets of variables that will affect the capacity of a system to come up with effective policy responses: the nature of the policy problem, the orientations of policy actors, and the characteristics of the institutional setting. Since each of these will vary greatly over time and space, comparative policy research will be faced with a dilemma: If we try to cover a variety of significant policy choices in a larger set of countries (say, the usual OECD-18 set), the number and diversity of distinct factor constellations will overwhelm available methods of multivariate statistical analysis as well as Ragin's (1987) Boolean "qualitative comparative analysis". And if we try to instead select our cases by the logic of "most similar systems" or "most different systems" designs (Przeworski & Teune 1970) in order to identify the influence of a single set of factors, we are likely to end up with very few cases at best. Thus, focusing on Britain in 1978/79 or on New Zealand in 1983/84 does in fact allow us to identify the influence of actor orientations while problems and institutions are held constant. Similarly, in my "Crisis and Choice in European Social Democracy" (Scharpf 1991), I selected four countries (Austria, Britain, Germany and Sweden) which during the 1970s were faced with the same problems of "stagflation", and which at the time had governments with strong preferences for the maintenance of full employment and with cognitive orientations shaped by the Keynesian paradigm. Within this design it was then relatively easy to relate differences in policy choices to differences of institutional structures (in particular, central-bank independence and concentrated or fragmented collective bargaining institutions).

4.1The Need for Institutionalist Working Hypotheses

In comparative policy research, however, we want to study a larger population of countries and policy choices, or seek to explain cases selected for their topical interest, rather than by the logic of a variance-controlling research designs. Still, for the reasons discussed above, a purely empirical approach would not serve our purposes. Instead, we might rely on theoretically grounded working hypotheses, rather than empirical research, for some causal influences in order to simplify and guide our empirical search for other explanatory factors. Preferably, these should be hypotheses with high predictive power - meaning that they should allow us to form precise expectations about significant causes and effects on the basis of relatively limited and easily obtainable information (Scharpf 1997, chapter 1).

Such hypotheses are clearly not available for all factors involved in our complex constellations. We have no general theory that would allow us to predict the occurrence of policy problems or the characteristics of effective solutions. These need to be identified by case-specific problem and policy analyses. Similarly, we have no general theory predicting the type of institutional settings within which policy interactions take place. They also need to be identified empirically in the specific cases at hand. But once we have obtained both of these sets of empirical information, we can use them to formulate theoretically grounded expectations of the policy responses which we should expect under the circumstances.

Substantive problem and policy analyses will generate policy recommendations of a prescriptive character which, however, can be converted into empirically relevant predictions if they are coupled with the assumption that the perceptions of all policy actors are identical with those of the policy analyst, and that the preferences of all policy actors are exclusively defined by a normative commitment to maximizing the public interest as defined by the policy recommendations of the analyst. If these (unrealistic) assumptions were true, institutions would only play a very limited role. But since we have reason to think that institutions matter, we also need hypotheses that predict their influence on the effectiveness of policy responses.

This influence is most obvious where formal institutional rules impose direct constraints on the repertoire of permissible policy choices. But we expect that within this repertoire the actual choice among permissible options is also affected by institutional rules prescribing actor constellations and modes of interaction, and by institutional incentives and disincentives affecting actor preferences. For formulating such predictions, rational-choice institutionalism has a clear advantage over the less information-efficient sociological variety. Its predictive power depends precisely on its narrow focus on the incentive-effect of institutional rules on organizational self-interest and on the assumption of complete information and adequate information processing capacity. Thus, if we know how the organizational self-interest of specific policy actors would be affected the available responses to given policy challenges, and if we know the actor constellation and the applicable rules of the game, we can then apply a range of analytical (essentially game-theoretic) tools to predict the policy outcome that interactions between rational and self-interested policy actors would produce in the specific case.

From what was said above, it follos of course that both of these predictions are based on unrealistic assumptions and thus will often be false. But that does not mean that they would be useless.[11] I will return to this point in the concluding section, and now turn to a brief overview of the institutional hypotheses that are likely to play a major role in comparative policy studies.

4.2Constraints on Permissible Policy Options

Countries differ in the range of institutionally permissible policy options. An example is the power of governments to determine wages and working time - an option which was routinely exercised by Belgian governments in the 1980s and 1990s, and which is available in most other countries as well, but which is ruled out in Germany by the constitutional guarantee of collective bargaining. Similarly, the judicial interpretation of the German constitution imposes narrow constraints on the permissible extent of property taxation, and it is claimed to rule out a Scandinavian-type "dual income tax" which would tax capital incomes more lightly than incomes from other sources. Admittedly, the German case represents an extreme example of legalism-cum-judicial-activism, but similar legal fetters - for instance on the public debt - can be found in other constitutional democracies as well.

More important for most countries is the increasing tightness of international legal constraints. Thus, the blatant protectionism which shielded import-substituting industrialization in Australia and New Zealand well into the 1980s (Schwarz 2000) would now be liable to massive legal sanctions by the World Trade Organization (WTO). Similarly, the tight control of capital transfers, and the highly discriminatory regulation of credit markets which facilitated the success of macro-economic full-employment strategies in Sweden until the mid 1980s (Benner & Vad 2000), would now be ruled out by EU directives liberalizing capital markets and financial services. More generally, the rules of "negative integration" and in particular, European competition law have become a major constraint on all economic-policy options that could be construed as inhibiting or distorting free competition in the markets of EU member states (Scharpf 1999a). And countries that have joined the European Monetary Union not only had to transfer their monetary competencies to the (totally independent) European Central Bank, but they also had to accept legally binding constraints on their fiscal-policy options. As a consequence, the option of achieving full employment through Keynesian strategies of demand reflation is no longer available at the national level among EMU member states.

4.3Actor Constellations and Modes of Interaction

In addition to imposing substantive prohibitions, institutional rules also define the constellations of actors that may participate in the adoption and implementation of policy responses as well as their permissible modes of interaction - which could be classified as "mutual adjustment", "negotiated agreement", "voting" or "hierarchical direction" (Scharpf 1997). Together, these rules determine the most basic and policy-relevant characteristic of the institutional setting, namely the number of formal "veto positions" (Tsebelis 1995). In extreme simplification, one may thus speak of "single-actor constellations " in which all relevant policy choices are potentially determined by the preferences and perceptions prevailing in a unified action center, or of "multiple-actor constellations" in which effective policy depends on the choices of several independent actors that may be acting from separate and potentially conflicting preferences and perceptions.[12] Here, policy can be blocked at multiple veto positions, and effective action will depend on negotiated agreement. The theoretical expectation is that, everything else being equal, the adoption of policy changes will be more difficult in multiple-actor than in single-actor constellations.

In practice, most policy choices result from multi-actor interactions. But some countries, whose political institutions approximate the ideal "Westminster model", have the option of treating any major policy problem in a single-actor constellation (Wilson 1994). In other words, hierarchical direction is an institutionally available mode of interaction[13]. This ideal type was approximated in Britain under Thatcher, and in New Zealand until the 1993 reform of the electoral system, but despite the existence of coalition governments, France also comes close to the single-actor pattern (Schmidt l996, chapters 1 and 2). Germany, by contrast, is characterized by a multi-actor constellation with coalition governments and a bicameral parliament at the center, autonomous subnational governments, an independent central bank, an independent constitutional court, and with constitutional guarantees of autonomous collective bargaining by unions and employers' associations, and an even more extreme example of a multiple-veto system is the Swiss combination of bicameralism, all-party coalitions, and the referendum.

4.4Institutions as Structures of Incentives

Institutional rules will affect policy responses not only by restricting options and by constituting actor constellations and regulating their modes of interaction but also by structuring the incentives of the participating actors. In rational-choice institutionalism, these incentives are defined by reference to the self-interest of the corporate and collective actors involved in the policy process - e.g., governments, political parties, central banks, labor unions, their sub-units or the individuals acting for them.[14] In single-actor systems, the incentives that have the most direct effect on policy choices are constituted by the mechanisms of political accountability. In multi-actor systems, accountability is weakened, and policy outcomes are more affected by incentives favoring cooperation or conflict among the veto-actors.

4.4.1Accountability Incentives

Accountability is most clearly institutionalized in Westminster-type political systems in which all policy competencies are concentrated in a central government whose choices are controlled by the winner in periodic two-party electoral competition. Under the restrictive assumption of a one-dimensional and single-peaked distribution of voter preferences, this model creates incentives for self-interested governments to adopt policies corresponding to the preferences of the median voter (Downs 1957), and under even more restrictive assumptions of complete information all around, these policies would maximize aggregate welfare (Mueller 1989). It should not be forgotten, however, that under the same complete-information assumptions, and in the absence of transaction costs, the Coase-Theorem predicts that welfare-maximizing outcomes would also be achieved in consociational democracies through negotiations among parties and associations representing the interests of affected groups (Coase 1960; Scharpf 1997, chapters 6 and 8). Under idealized conditions, in other words, democratic accountability would assure effective policy responses in multi-actor as well as in single-actor constellations.

In the real world, however, policy information is incomplete and contested, and inter-group negotiations are impeded by high transaction costs. Under these conditions, it is plausible that voters as well as the members of organized interest groups will oppose policy changes whose immediate impact on their status-quo interest position is negative. Then the institutions of democratic accountability create incentives favoring policies maximizing short-term benefits and avoiding short-term costs for voters and interest groups - which may prevent the adoption of effective policy responses to manifest problems. But the extent to which this is likely varies with the specifics of institutional arrangements.

For one thing, an increasing number of countries have exempted certain policy functions from the control of democratically accountable governments and legislative majorities - by providing constitutional courts, independent central banks or cartel offices with a greater, though surely not unlimited,[15] freedom to pursue unpopular policies. To an even greater (and from the perspective of democratic accountability much more problematic) degree, this is also true of the European Central Bank, of the European Commission, and of the European Court of Justice in its role as interpreter and guardian of Treaty obligations - under institutional rules where its interpretation of the Treaties could only be reversed by unanimous agreement among member governments that needs to be ratified by the parliaments of all member states (Scharpf 1999a). On the other hand, the negative incentive-effect of democratic accountability increases with the proximity of elections and their expected salience for the parties in power. Thus, the fact the legislative term was shortened to three years in 1970 is said to have reduced the capacity of Swedish governments to govern from a long-term perspective (Immergut 1999), and in France shorter intervals between parliamentary and presidential elections have been shown to affect the willingness of governments to maintain potentially unpopular initiatives (Bonoli 1997). In Germany, the government is even less insulated from short-term electoral pressures since the four-year parliamentary term is punctuated by altogether 16 Land elections that affect the partisan composition of the Bundesrat. By contrast, the realization of Margaret Thatcher's radical reforms benefited from the five-year interval between general elections, and even more from the fact that she was facing a divided opposition which, under first-past-the-post election rules, practically ruled out electoral defeat.

It is clear, then, that national institutions differ greatly in the extent to which they expose governments to incentives favoring policies responding to short-term voter reactions, rather than strategies which, though initially unpopular, may demonstrate their effectiveness and convince voters in the longer term. But it should be kept in mind that, first, the organizational self-interest that is affected by these incentives does not circumscribe the full range of action-relevant preferences. Political parties and governments may be committed to normative goals and role obligations that will override electoral concerns - as was true of the Lubbers coalition government in the Netherlands which stuck to its unpopular retrenchment program in the face of practically certain electoral defeat in 1994 (Visser and Hemerijck 1997). Even more important, governments are not necessarily helpless when faced with negative voter reactions. They may be able to engage opposition parties, organized interests, the media and the general public in policy-oriented discourses that explain and justify the course of action that is proposed and the sacrifices associated with it. Again, however, their capacity to do so is influenced by institutional conditions - it is greater in single-actor polities, where the government can speak with one voice, than it is in multiple-veto constellations where policy choices will often have the character of compromises among conflicting goals that are difficult to justify from the positions of any one of the parties involved in their adoption (Schmidt 2000).

4.4.2Incentives to Collaborate in Multi-Actor Systems

Depending on whether unilateral policy action is, or is not, a realizable option, it is useful to distinguish between two variants of multi-actor systems: In the first case, which one may call "policy fragmentation", some of the instruments of public policy that are required for the adoption of a successful policy response to a given problem are not under the control of a single actor but are exercised by independent agents - a formally independent constitutional court, a de-facto independent central bank or cartel office, autonomous subnational governments, and in Europe, the EU Commission and the European Court of Justice. The effect of policy fragmentation is, first, to limit the range of policy options which a single actor may unilaterally adopt and implement and, second, to raise the possibility that mutually inconsistent unilateral policy choices will interfere with each other; and even if some coordination is achieved by actors responding to others' choices in the mode of "mutual adjustment", the outcome may still be suboptimal.[16] If that is to be avoided, coordination through negotiated agreement[17] among independent actors may be required.

In the second case, which I have called "joint decision making" (Scharpf 1988), none of the parties is allowed (or physically able) to act unilaterally. Here, effective policy responses can only be adopted and implemented with the agreement of several independent actors. That may be true even in single-party governments if ministers are representatives of intra-party factions that cannot be ignored by the prime minister, or if ministries enjoy a large degree of constitutional autonomy so that policy conflicts among them must be resolved through inter-ministerial negotiations or through the cabinet as a collegiate body (Mayntz & Scharpf 1975). In coalition governments, the agreement of separate political parties, each with its own policy goals and organizational self interest, is generally necessary for important policy choices (Budge & Laver 1993), and in bicameral legislatures the second chamber may have a veto that is exercised by partisan majorities differing from those in the first chamber.

Given that no policy system is truly single-actor, the difficulties of effective action depend, of course, on the substantive convergence or divergence of policy preferences among the veto players involved (Tsebelis 1995; 1999b). In contrast to mutual adjustment, the Coase-Theorem (Coase 1960) assures us that in the absence of transaction costs negotiations are able to achieve efficient outcomes in all constellations where potential gains from cooperation exist. But since transaction costs are never absent, the probability of agreement depends very much on issue-independent incentives favoring cooperation or conflict that are inherent in different institutional settings.

General incentives for cooperation should be strongest within single-party and unitary governments of the Westminster model. Even if intra-party factions may compete with each other for policy influence and career opportunities, and even if individual ministerial departments may enjoy considerable autonomy, all office holders of the governing party must win or lose elections together, and thus should have strong incentives to cooperate on successful policies. Similar incentives also exist among the parties in a coalition government that should have a common interest in the success of their government. But since each of them must ultimately face its own clienteles and voters separately, there are also incentives to sharpen one's own profile through tough bargaining strategies and even open conflicts within the coalition - which may then delay, distort, or even block effective solutions. Blockage is more likely in the "end-game" phase of an election period than in the "honeymoon" of a new coalition, and it is generally more likely in multi-party systems with an expectation of rapidly changing governing coalitions - so that unilateral concessions in the interest of longer-term cooperation are unlikely to have high future payoffs. This was true of multi-party coalition governments in the French IVth Republic, and of Italian governments before the early 1990s, whose capacity for effective policy responses was generally very low. By contrast, the Danish multi-party system seems to have evolved into a pattern of relatively stable minority governments that are free to seek the support of variable ad-hoc majorities for their policy initiatives, and that are thus able to achieve a remarkably high degree of problem-solving capacity ( Benner & Vad 2000).[18] Thus, its ad-hoc coalitions may be formed exclusively on the basis of convergent substantive policy interests, whereas longer-term government coalitions are usually based on agreements that rule out voting with the opposition against another coalition party - with the consequence that effective policy responses may be blocked within the coalition even though they would have majority support in the full parliament.

At the opposite extreme are constellations of "divided government" where institutional incentives favor conflict, rather than cooperation. This is true in bicameral legislatures, when partisan majorities in both houses are opposed to each other (Laver & Shepsle 1991; Krehbiel 1996). Somewhat similar conditions may obtain in all constitutional democracies when effective policy solutions require constitutional amendments[19] that can only be passed by super-majorities including parties in opposition to the government of the day. Under these conditions, even compromises that would advance the substantive policy goals of the opposition may be blocked. Since governments and opposition parties are engaged in zero-sum competition for the votes of the same electorate, the opposition has not only a substantive interest in promoting its own policy goals through favorable compromises, but also a competitive interest[20] in defeating government initiatives in order to undermine its reputation for competent and successful political action. Thus conditions of divided government are particularly unfavorable for effective but unpopular policy responses with a high party-political salience (Laver & Shepsle 1991; Alt & Lowry 1994).[21]

In between these extremes are located attempts to achieve effective policy responses through "joint decision making" between governments and independent public agencies or social-partnership organizations, or among different national or subnational governments with autonomous jurisdictions and separate constituencies (as is true of negotiations in the EU Councils of Ministers). Here, agreement is not facilitated by a generally shared interest in the success of a joint venture, but neither is it impeded by disincentives arising from political competition. As long as each party is unchallenged in its own institutional domain, self-interested cooperation on substantive solutions has a chance under the "Coase Theorem" whenever the policy goals pursued by all parties are better achieved through concerted, rather than through separate action. In fact, much public-sector problem-solving does occur in the form of agreements among local or regional or national governments, and the important achievements of European integration are best explained in terms of intergovernmental agreements as well (Moravcsik 1998).

Going beyond the Coase Theorem (which presupposes that agreements can only be achieved if all parties involved will benefit from it), long-standing patterns of successful substantive collaboration may create additional generalized incentives to maintain cooperation even if one or the other party must accept short-term sacrifices to its own institutional or policy interests. Such incentives are said to explain the long-term success of social-partnership institutions in Austria (Marin 1987; Heinisch 1999). Nevertheless, even institutionalized "corporatist" cooperation cannot be taken for granted. The breakdown of "Concerted Action" in Germany in the early 1970s (Scharpf 1991, chapter 7), the complete, but temporary, suspension of coordinated wage setting in the Netherlands during the 1970s (Visser & Hemerijck 1997), and the erosion of centralized wage setting in Sweden in the 1980s (Compston 1995; Pestoff 1995) demonstrate that generalized incentives are unlikely to maintain cooperation in the face of sustained conflicts over either policy goals or institutional self-interest among the parties involved.

5Strategies for Comparative Policy Research

This overview of hypotheses regarding institutional incentives and their policy-relevant effects on organizational self-interest of corporate and collective actors could be extended from the arenas of parliamentary and intergovernmental interaction to other arenas in which important policy choices are determined - e.g., interactions in monetary policy, industrial relations, or European regulation. Moreover, the hypotheses themselves could and should be formulated with greater differentiation and precision regarding the assumed organizational self-interest of different types of actors and the likely incentive effects of different types of institutional arrangements. Given the deductive methods of rational-choice institutionalism, the formulation of such hypotheses would not pose serious difficulties - and given a suitable data base, some of these hypotheses may also be corroborated by statistical analyses (see e.g., Bawn 1999; Tsebelis 1999b).

As I emphasized above, however, in the specific case the hypotheses derived from rational-choice institutionalism will often turn out false. So what is the role that they could and should play in comparative policy research that is less interested in statistical probabilities than in explaining the outcomes of individual cases? The answer is that having such hypotheses is as useful for comparative case studies as having a regression line in a scattergram is for the exploration of quantitative relationships: It focuses attention on the "residuals" - i.e., on cases where the prediction is not confirmed.

In fact, from what I said about the relationship between problem-oriented and interaction-oriented policy research, it follows that we are able to work with two different sets of theory-based hypotheses: On the one hand, problem-oriented policy analyses predict the choices we should expect if all policy actors had complete information and were exclusively motivated to realize public-interest maximizing outcomes. On the other hand, rational-choice institutionalist analyses predict the choices we should expect if all policy actors had complete information and were exclusively motivated to realize self-interest maximizing outcomes. If these two predictions agree, we could conclude that institutional incentives are favoring the adoption of effective policy responses to the problem in question. If they disagree, such responses are made more difficult by institutional obstacles. But since (multi-purpose) political institutions can rarely be optimized with a view to specific policy problems, [22] this abstract evaluation of institutional incentive-effects is not our main concern.

However, empirical policy research benefits very much from being able to compare the observed outcome of policy interactions to these two theory-based predictions, both of which rely on highly stylized assumptions about actor orientations. If the outcome should agree with either of these, we could content ourselves with having obtained a satisfactory explanation. But when that is not the case (i.e., most of the time), we now have at least a clear idea of where to search for explanations. Since both the nature of the problem and the institutional incentives have already been accounted for, the explanation must be found in actor orientations that differ from the stylized assumptions underlying our predictions. Thus we now must seek empirical information on the actual preferences and perceptions of the policy actors involved.

More specifically, we should seek for cognitive orientations that deviate from the information and cause-and-effect hypotheses of the substantive problem and policy analyses which we consider to be true. At the same time, we should also seek for actor preferences that differ from either the perfectly public-interested or the perfectly self-interested motivations assumed in our hypotheses. In doing so, it is often very useful - in the spirit of Lindenberg's (1990) "method of decreasing abstraction" - to begin by focusing on the institutionalized "norms of appropriateness" that are emphasized by sociological institutionalism (March & Olsen 1989), and to move on to searching for more idiosyncratic normative orientations and identity concepts only when the more stylized institutional hypotheses fail to explain choices. In any case, however, the discrepancy between theoretical predictions and observed policy choices will guide and greatly simplify the empirical search for those preferences and perceptions that actually can explain the failure to adopt effective policy responses.

To give an example: After the Bundesbank had switched to monetarism in the mid 1970s, all countries that had pegged their currencies to the German mark (Austria, Denmark, the Netherlands, and Belgium) were confronted with the same hard-currency environment which, in the absence of union wage restraint, would produce high unemployment. When faced with this challenge, rational-institutionalism would tell us, the centralization or fragmentation of wage-setting institutions should not matter: Large or small unions alike should find it in their organizational self-interest to save the jobs of their members through wage restraint. In Germany and Austria, this is what happened within a year or two, whereas collective bargaining in Denmark, Belgium and the Netherlands continued to generate wage inflation even under conditions of steeply rising unemployment. Since institutional incentives cannot explain these differences, the comparison would then have to focus on the influence of cognitive misperceptions and normative (ideological) orientations that prevented the unions in some countries, but not in others, from fully appreciating the impact of a change in (German) monetary policy on their own employment situation (Scharpf 2000b).

The example has several interesting implications for research strategies: First, what counts as a "case" in comparative case studies depends on what is to be explained. In comparative policy research, that is trying to explain differences in policy response to a given challenge, the case is defined by the set of interactions connecting the challenge to the response. Since the historical processes we are studying contain a considerable variety of (often nested) policy challenges, we have considerable freedom in selecting sets of cases which, at minimum, are similar with regard to the policy challenge that was being faced. In the example, we could have focused on the full set of countries that were members in the European currency "snake" in 1973/74, asking why some of them responded to the German switch to monetarism by quitting the snake, whereas others maintained their commitment to a hard-currency position. Among the subset of soft-currency countries, we could then compare responses to the challenge of escalating rates of inflation; among the hard-currency countries, in turn, comparison would focus on responses to rising unemployment, and so on. Among these sets defined by similar challenges, moreover, it should usually be possible to identify subsets of cases that have additional causal factors in common - which would then permit a more precise focus on a narrower range of explanatory variables.

Combined with the use of rational-choice working hypotheses, then, structured comparisons within varying subsets of cases seem to be our best hope for building a body of generalizable knowledge about the causal relations between types of policy challenges, types of institutional structures, and actor orientations. Moreover, by switching between overlapping subsets of cases defined by either common challenges, or common actor orientations or common institutions, we should be able to increase our confidence in the explanations discovered in each of these dimensions. These explanations will surely not amount to a general and comprehensive theory, but they will go beyond the ad-hoc explanations that are otherwise characteristic of historical case studies (Scharpf 2000b).

References

Alt, James E. & Lowry, Robert C. (1994). Divided government, fiscal institutions, and budget deficits: Evidence from the states. American Political Science Review, 88, 811 - 828.

Arthur, W. Brian. (1989). Competing technologies, increasing returns, and lock-in by historical events. Economic Journal, 99, 116 - 131.

Arthur, W. Brian. (1990). Positive feedback in the economy. Scientific American, February 1990, 80 - 85.

Bawn, Kathleen (1999). Money and majorities in the Federal Republic of Germany: Evidence for a veto players model of government spending. American Journal of Political Science, 43, 707-736.

Bates, Robert H., Greif, Avner, Levi, Margaret, Rosenthal, Jean-Laurent, & Weingast, Barry R. (1998). Analytic narratives. Princeton, NJ: Princeton University Press.

Beck, Neal, & Katz, Jonathan. (1995). What to do (and not to do) with time-series cross-sectional data in comparative politics. American Political Science Review, 89, 634 - 647.

Benner, Mats, & Vad, Torben B. (2000). Sweden and Denmark: Defending the welfare state. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open econ-omy. Volume II: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

Berger, Peter L., & Luckmann, Thomas. (1966). The social construction of reality. Garden City: Dounbleday.

Blom-Hansen, Jens. (1999). Avoiding the "joint-decision trap": Lessons from intergovernmental relations in Scandinavia. European Journal of Political Research, 35, 35 - 67.

Bonoli, Giuliano. (1997). Pension politics in France: Patterns of co-operation and conflict in two recent reforms. West European Politics, 20, 111 - 124.

Budge, Ian, & Laver, Michael. (1993). The policy basis of government coalitions: A comparative investigation. British Journal of Political Science, 23, 499 - 519.

Calmfors, Lars, & Driffil, John. (1988). Centralization of wage bargaining. Economic Policy, 6, 14 - 61.

Coase, Ronald H. (1960). The problem of social cost. Journal of Law and Economics, 3, 1 - 44.

Compston, Hugh. (1995). Union participation in economic policy making in Scandinavia, 1970 - 1993. West European Politics, 18 (1), 98 - 115.

Coppedge, Michael. (1999). Thickening thin concepts and theories: Combining large N and small in comparative politics. Comparative Politics, 31, 465 - 476.

Crouch, Colin. (1993). Industrial relations and European state traditions. Oxford: Clarendon Press.

Cukierman, Alex. (1992). Central bank strategy, credibility, and independence. Cambridge, MA: MIT Press.

David, Paul A. (1985). Clio and the economics of QWERTY. American Economic History, 75, 332 - 337.

Dearborn, DeWitt C., & Simon, Herbert A. (1958). Selective perception: A note on the departmental identification of executives. Sociometry, 21, 140 - 144.

DiMaggio, Paul, & Powell, Walter. (1991). Introduction. In Walter Powell & Paul DiMaggio (Eds.), The new institutionalism in organizational analysis (pp. 1 - 40). Chicago: University of Chicago Press.

Downs, Anthony. (1957). An economic theory of democracy. New York: Harper & Row.

Downs, Anthony. (1967). Inside bureaucracy. Boston: Little Brown.

Esping-Andersen, Gøsta. (1990). The three worlds of welfare capitalism. Princeton, NJ: Princeton University Press.

Esping-Andersen, Gøsta. (1999). Social foundations of postindustrial economies. Oxford: Oxford University Press.

Estevez-Abe, Margarita, Iversen, Torben, & Soskice, David. (1999, September 2 - 5). Social protection and the formation of skills: A reinterpretation of the welfare state. Paper prepared for presentation at 95th American Political Science Association Meeting, Atlanta.

Etzioni, Amitai. (1964). Modern organizations. Englewood Cliffs, NJ: Prentice-Hall.

Garrett, Geoffrey. (1998). Global markets and national politics. International Organization, 52, 787 - 824.

Garrett, Geoffrey, & Mitchell, Deborah. (1999). Globalization and the welfare state. Ms. Department of Political Science. Yale University.

Golden, Miriam, Wallerstein, Michael, & Lange, Peter. (1999). Postwar trade-union organization and industrial relations in twelve countries. In Herbert Kitschelt, Peter Lange, Gary Marks, & John D. Stephens (Eds.), Continuity and change in contemporary capitalism (pp. 194 - 230). Cambridge, UK: Cambridge University Press.

Green-Pedersen, Christoffer. (1999). Welfare-state retrenchment in Denmark and the Netherlands 1982 - 1998. The role of party competition and party consensus. Ms. Department of Political Science. University of Aarhus, Denmark.

Haas, Peter M. (1992). Introduction: Epistemic communities and international policy coordination. International Organization, 46 (Special Issue), 1 - 35.

Hall, Peter A. (1986). Governing the economy: The politics of state intervention in Britain and France. Cambridge, UK: Polity Press.

Hall, Peter A. (1992). The movement from Keynesianism to monetarism: Institutional analysis and British economic policy in the 1970s. In Sven Steinmo, Kathleen Thelen, & Frank Longstreth (Eds.), Structuring politics. Historical institutionalism in comparative analysis (pp. 90 - 113). Cambridge, UK: Cambridge University Press.

Hall, Peter A. (1993). Policy paradigms, social learning, and the state. The case of economic policy making in Britain. Comparative Politics, 25, 275 - 296.

Hall, Peter A., & Taylor, Rosemary C. R. (1996). Political science and the three new institutionalisms. Political Studies, 44, 936 - 957.

Heinisch, Reinhard. (1999). Coping with the single market: Corporatist response strategies in Germany and Austria. Ms. University of Pittsburgh/Johnstown.

Hicks, Alexander. (1999). Social democracy and welfare capitalism: A century of income security politics. Ithaca, NY: Cornell University Press.

Immergut, Ellen M. (1999, September 9 - 12). Political institutions and the Swedish welfare state. Paper presented at the International Sociological Association RC 19 meeting, Societal Changes and Social Policy Responses, Prague. Electronic version: http://www.cas.cz/zahr_odbor/fp/Immer1.doc.

Iversen, Torben. (1999). Contested economic institutions. The politics of macroeconomics and wage bargaining in advanced democracies. Cambridge, UK: Cambridge University Press.

Iversen, Torben, & Wren, Anne. (1998). Equality, employment, and budgetary restraint: The trilemma of the service economy. World Politics, 50, 507 - 546.

Janis, Irving L. (1972). Victims of Groupthink. Boston, MASS: Houghton Mifflin.

Jupille, Joseph, & Caporaso, James A. (1999). Institutionalism and the European Union: Beyond international relations and comparative politics. Annual Review of Political Science, 2, 429 - 444.

Keohane, Robert O. (1984). After hegemony. Cooperation and discord in the world political economy. Princeton, NJ: Princeton University Press.

Krehbiel, Keith. (1996). Institutional and partisan sources of gridlock. A theory of divided and unified government. Journal of Theoretical Politics, 8, 7 - 40.

Laver, Michael, & Shepsle, Kenneth A. (1991). Divided government: America is not "exceptional". Governance, 4, 250 - 269.

Levy, Jonah D. (1999). Tocqueville's revenge. State, society, and economy in contemporary France. Cambridge, MA: Harvard University Press.

Lijphart, Arend. (1984). Democracies. Patterns of majoritarian and consensus government in twenty-one countries. New Haven: Yale University Press.

Lindenberg, Siegwart. (1989). Choice and culture: The behavioral basis of cultural impact on transactions. In Hans Haferkamp (Ed.), Social structure and culture (pp. 175 - 200). Berlin: de Gruyter.

Lindenberg, Siegwart. (1990). Homo socio-oeconomicus: The emergence of a general model of man in the social sciences. Journal of Institutional and Theoretical Economics, 146, 727 - 748.

Manow, Philip. (1999). Liberalism's and corporatism's uneasy compromise in postwar Germany. Ms. Cologne: Max Planck Institute for the Study of Societies.

Manow, Philip, & Seils, Eric. (2000). Adjusting badly: The German welfare state, structural change, and the open economy. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open economy. Volume II: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

March, James G., & Olsen, Johan P. (1989). Rediscovering institutions. New York: Free Press.

Marin, Bernd. (1987). Neo-corporatism and the "consociational" heritage. From consociationalism to technocorporatism: The Austrian case as a model-generator? In Ilja Scholten (Ed.), Political stability and neo-corporatism. Corporatist integration and societal cleavages in western Europe (pp. 39 - 69). London: Sage.

Mayntz, Renate, & Fritz W. Scharpf. (1975). Policy making in the German federal bureaucracy. Amsterdam: Elsevier.

Mayntz, Renate, & Fritz W. Scharpf. (1995). Der Ansatz des akteurzentrierten Institutionalismus. In Renate Mayntz & Fritz W. Scharpf (Eds.), Steuerung und Selbstorganisation in staatsnahen Sektoren (pp. 39 - 72). Frankfurt a. M.: Campus.

Moravcsik, Andrew. (1994). Why the European community strengthens the state: Domestic politics and international cooperation. (Working Paper Series No.52). Cambridge, MA: Center for European Studies, Harvard University.

Moravcsik, Andrew. (1998). The choice for Europe. Social purpose and state power from Messina to Maastricht. Ithaca, NY: Cornell University Press.

Mueller, Dennis C. (1989). Public choice II. Cambridge, UK: Cambridge University Press.

North, Douglas C. (1990). Institutions, institutional change, and economic performance. Cambridge, UK: Cambridge University Press.

Pestoff, Victor A. (1995) Towards a new Swedish model of collective bargaining and politics. In Colin Crouch & Franz Traxler (Eds.), Organized industrial relations in Europe. What future? (pp. 151-182). Aldershot: Avebury.

Pierson, Paul. (1996). The path to European political integration: A historical institutionalist analysis. Comparative Political Studies, 29, 123 - 163.

Pierson, Paul. (1997). Path dependency, increasing returns, and the study of politics. (Working Paper Series No. 7.7). Cambridge, MA: Center for European Studies, Harvard University.

Przeworski, Adam, & Teune, Henry. (1970). The logic of comparative social inquiry. New York: John Wiley.

Ragin, Charles C. (1987). The comparative method. Moving beyond qualitative and quantitative strategies. Berkeley: University of California Press.

Rhodes, Martin. (2000). Restructuring the British welfare state: Between domestic constraints and global imperatives. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open economy. Volume II: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

Rodrick, Dani. (1997). Has economic integration gone too far? Washington, DC: Institute for International Economics.

Sabatier, Paul A. (1987). Knowledge, policy-oriented learning, and policy change: An advocacy coalition framework. Knowledge: Creation, Diffusion, Utilization, 8 (4), 649 - 692.

Sabatier, Paul A. (Ed.). (1999). Theories of the policy process. Boulder: Westview.

Scharpf, Fritz W. (1987). A game-theoretical interpretation of inflation and unemployment in western Europe. Journal of Public Policy, 7, 227 - 257 (reprinted in Scharpf 1997, Appendix).

Scharpf, Fritz W. (1988). The joint decision trap: Lessons from German federalism and European integration. Public Administration, 66, 239 - 278.

Scharpf, Fritz W. (1991). Crisis and choice in European social democracy. Ithaca, NY: Cornell University Press.

Scharpf, Fritz W. (1997). Games real actors play. Actor-centered institutionalism in policy research. Boulder: Westview.

Scharpf, Fritz W. (1999a). Governing in Europe. Effective and democratic? Oxford, UK: Oxford University Press.

Scharpf, Fritz W. (1999b). Approaches to the study of European politics. ECSA Review, 12 (2), 7 - 9.

Scharpf, Fritz W. (2000a). The viability of advanced welfare states in the international economy: Vulnerabilities and options. Journal of European Public Policy, 7, 190 - 228.

Scharpf, Fritz W. (2000b). Economic changes, vulnerabilities, and institutional capabilities. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open economy. Volume I: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

Scharpf, Fritz W., & Schmidt, Vivien A. (Eds.). (2000). Welfare and work in the open economy. Volume I & II: Diverse responses to common challenges. Oxford, UK: Oxford University Press, forthcoming.

Schmidt, Vivien A. (1996). From state to market? The transformation of French business and government. Cambridge, UK: Cambridge University Press.

Schmidt, Vivien A. (2000). Discourse and values. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open economy. Volume I: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

Schwartz, Herman. (2000). Internationalization and two liberal welfare states: Australia and New Zealand. Forthcoming in Fritz W. Scharpf & Vivien A. Schmidt (Eds.), Welfare and work in the open economy. Volume II: Diverse responses to common challenges. Oxford, UK: Oxford University Press.

Shepsle, Kenneth A. (1989). Studying institutions: Some lessons from the rational choice approach. Theoretical Politics, 1, 131 - 147.

Skocpol, Theda, & Weir, Margaret. (1985). State structures and the possibilities for "Keynesian" responses to the great depression in Sweden, Britain, and the United States. In Peter B. Evans, Dietrich Rueschemeyer, & Theda Skocpol (Eds.), Bringing the state back in (pp. 107 - 163). Cambridge, UK: Cambridge University Press.

Soskice, David. (1999). Divergent production regimes: Coordinated and uncoordinated market economies in the 1980s and 1990s. In Herbert Kitschelt, Peter Lange, Gary Marks, & John D. Stephens (Eds.), Continuity and change in contemporary capitalism (pp. 101 - 134). Cambridge, UK: Cambridge University Press.

Soskice, David, & Iversen, Torben. (2000). The non-neutrality of monetary policy with large wage or price setters. Forthcoming in Quarterly Journal of Economics 15 (1).

Steinmo, Sven, & Thelen, Kathleen. (1992). Historical institutionalism in comparative politics. In Sven Steinmo, Kathleen Thelen, & Frank Longstreth (Eds.), Structuring politics: Historical institutionalism in comparative analysis (pp. 1 - 32). Cambridge, UK: Cambridge University Press.

Thelen, Kathleen. (1999). Historical institutionalism in comparative politics. Annual Review of Political Science, 2, 369 - 404.

Tsebelis, George. (1995). Decision-making in political systems: Veto players in presidentialism, parliamentarism, multicameralism, and multipartyism. British Journal of Political Science, 25, 289 - 326.

Tsebelis, George. (1999a). Approaches to the study of European politics. ECSA Review, 12 (2), 4 - 6.

Tsebelis, George. (1999b). Veto players and law production in parliamentary democracies: An empirical analysis. American Political Science Review, 93, 591 - 608.

Visser, Jelle, & Hemerijck, Anton. (1997). "A Dutch miracle". Job growth, welfare reform and corporatism in the Netherlands. Amsterdam: Amsterdam University Press.

Waltz, Kenneth N. (1979). Theory of international politics. New York: McGraw-Hill.

Weaver, R. Kent. (1987). The politics of blame avoidance. Journal of Public Policy, 6, 371 - 398.

Williamson, Oliver E. (1985). The economic institutions of capitalism. Firms, markets, relational contracting. New York: Free Press.

Wilson, Graham K. (1994). The Westminster model in comparative politics. In Ian Budge & David MacKay (Eds.), Developing democracy: Comparative research in honour of J. F. P. Blondel. London: Sage.

Zucker, Lynn G. (1991). The role of institutionalization in cultural persistence. In Walter Powell, & Paul DiMaggio (Eds.), The new institutionalism in organizational analysis (pp. 83 - 107). Chicago: University of Chicago Press.

Endnotes

1   Most of my examples will be taken from the fields which I know best - cross-national comparative studies of welfare-state and employment policy responses to changes in the international economic environment (Scharpf 1991; Scharpf and Schmidt 2000).

2   Worse yet, if such case studies are written with the self-conscious intent of producing inductive generalizations, they are likely to contribute to the endless cycle of over-generalized propositions and false falsifications that plagues not only the study of European politics (Scharpf 1999b).

3   Another interesting combination of methods is used by Hicks (1999) who relies on multivariate regressions where quantitative time-series data are available, but resorts to Ragin's (1987) "qualitative comparative analysis" where only qualitative data can be obtained.

4   "Historical institutionalism", the third strand discussed in these overviews, is primarily interested in the path-dependent evolution of institutional forms (Steinmo and Thelen 1992; Pierson 1997), but has no distinct theory of its own about the way institutions influence actors' choices. Some authors lean more toward rational-choice assumptions, others toward social-constructivist interpretations.

5   I should add that my presentation of the differences between the two is also overdrawn. For a more integrative discussion, see Thelen (1999). From the rational-choice side, a useful theoretical bridge is built by Lindenberg's (1989) concept of socio-culturally defined "production functions" through which (or the social roles in which) the abstract individual self-interest in physical well-being and social recognition must be concretely pursued. For the organizational self-interest of corporate actors, this concept seems even more pertinent.

6   Preferences, in our view must be understood as relatively stable ("pre-strategic") criteria for the evaluation of outcomes, rather than as tentative commitments to a particular strategy (which may of course be changed quite easily by "arguments" providing new information). If this distinction is kept in mind, much fruitless debate over the need to "endogenize preferences" in analyses of policy interactions could be avoided.

7   A third dimension is defined by identity concepts that may be more specific than institutionalized norms and role obligations (Scharpf 1997, chapter 3).

8   The point is nicely made in the dictum, ascribed to Lyndon Johnson, that "you got to be re-elected to be a statesman".

9   See the reference to "production functions" in note 5, above.

10   It is puzzling that in modeling the effect of institutional rules on policy choices, Tsebelis (1995; 1999a) considers only the "goal" dimension (expressed by the ideological distance between veto players) while ignoring the "maintenance" dimension (e.g., the effect of competitive incentives on the organizational self-interest of corporate actors and hence on their willingness to reach agreement).

11   In this regard, the approach discussed here differs from the "analytic-narratives" approach (Bates et al.1998) which expects rational-choice analyses to provide complete explanations of historically unique cases.

12   The same kind of descriptors may be used to characterize wage-setting systems. Here, effective decisions may, again, be vertically centralized and horizontally concentrated in the arena of peak-level negotiations among monopoly associations of capital and labor; located at the level national industrial sectors and branches; or decentralized to regional negotiations, to firm-level or shop floor negotiations, or even to individual employment contracts.

13   The center would be hopelessly overloaded if choices were in fact made there in many cases (Downs 1967). Nevertheless, the fact that any decision could be centralized is likely to affect the choices made "in the shadow of hierarchy" at the level of departments or agencies (Scharpf 1997, chapter 9).

14   I will not here go into intricacies of when it is useful to ascribe actor qualities to corporate and collective actors, and when it is necessary to focus on sub-units thereof or even on individuals (Scharpf 1997, chapter 3), except to note that in general the incentives for individuals acting in the name of corporate and collective actors are closely linked to their success in terms of organizational self-interest.

15   The celebrated independence of the German Bundesbank could have been abolished by ordinary legislation. That it was maintained even in major conflicts with governments of the day, from Konrad Adenauer to Helmut Kohl, was due to broad public support for its institutional independence.

16   More precisely, the efficiency of the outcome depends entirely on the underlying constellations of actor interests. If they correspond to a "Battle of the Sexes" - which is the constellation that Carey (this volume) is primarily considering - mutual adjustment may be highly efficient. A real-world example was the adjustment of German unions after 1975 to the monetary policy announced annually by the Bundesbank. But if the underlying constellation should resemble a Chicken Game or the Prisoner's Dilemma, mutual adjustment would produce highly undesirable policy outcomes. A real-world example was the inflationary wage competition among British unions in the 1970s (Scharpf 1987).

17   Negotiated coordination may take two forms: "Negative coordination" merely avoids policies that would have negative external effects on other policy areas, whereas ""positive coordination" attempts to maximize the gains from cooperation (Scharpf 1997, chapter 6).

18   Blom-Hansen (1999) points to a second mechanism facilitating effective problem solving in Scandinavian countries, namely the government's choice among alternative arenas of decision making - parliamentary, intergovernmental, and corporatist.

19   Such conditions are most likely to occur in countries where an "activist" constitutional court is deeply involved in substantive policy choices - which is true of courts in Germany and Denmark, and increasingly in Italy and France. In that sense, constitutional courts may be able to create the very political deadlocks which George Tsebelis (1999a) sees as their major source of power.

20   Competitive incentives may be softened by expectations of an imminent return to government. Thus, in Sweden in the early 1990s, opposition social democrats collaborated with the conservative government in designing cuts in welfare-state expenditures they considered necessary. For the Netherlands in the 1980s, it is also suggested that the social democrats did not attack on the unpopular reforms of the christian-liberal government because they hoped to join a future coalition with the christian democrats (Green-Pedersen 1999).

21   There is a parallel in International Relations, where the "Realist" school assumes a dominance of competitive incentives that force states to pursue "relative", rather than "absolute gains" (Waltz 1979) whereas "Liberal" theorists assert the theoretical possibility and practical importance of international agreements capturing "absolute" gains from cooperation (Keohane 1984).

22   The exception are institutions that are explicitly designed to serve a single purpose - as may be true of the European Central Bank which is supposed to be exclusively committed to price stability.

Copyright © 2000 Fritz W. Scharpf

No part of this publication may be reproduced or transmitted without permission in writing from the author.
Jegliche Vervielfältigung und Verbreitung, auch auszugsweise, bedarf der Zustimmung des Autors.

MPI für Gesellschaftsforschung, Paulstr. 3, 50676 Köln, Germany