This Forum was first published in the ECSA Review, Spring 1999 (XII: 2), pp. 2-9.

Four ECSA members reflect on institutionalism in European integration studies. A collective bibliography appears at the end of this Forum.

Vivien A. Schmidt

THIS ECSA REVIEW FORUM was inspired by Robert Bates’ polemic on area studies, and in particular his argument that area studies is largely irrelevant where it does not aspire to "science," by which he means rational choice institutionalism (Bates l997). The object of the Forum is not so much to respond directly to Bates’ arguments as to consider the question of which approach or approaches are most appropriate to the study of European politics.

In this forum, we have three very different pieces. The first, by George Tsebelis, argues that institutional analysis in the rational choice tradition can provide insights for the explanation of European integration that go beyond intergovernmentalist or neofunctionalist theories. The second, by Thomas Risse, maintains that in explaining the European Union, rather than following rational choice institutionalism, whether methodologically "hard" or "soft," one should consider the use not just of historical institutionalism, the other mainstream methodology, but also of sociological institutionalism. The third, by Fritz W. Scharpf, takes a position somewhere between the other two pieces, by pointing to the dangers of overgeneralization from empirical cases in the pursuit of theories of the rational choice variety and, by implication, by leaving an opening for both historical and sociological institutionalism where rational choice institutionalism does not or can not sufficiently explain.

This introductory essay, rather than coming down on one or another side of the methodological discussion, seeks to put the debate in perspective, by considering the epistemological differences in the three kinds of "institutionalisms" evoked in the Forum pieces. These differences involve such things as the underlying assumptions about the appropriate objects of explanation, the goals of explanation, the kinds of theories or generalizations necessary to explanation, the kinds of "facts" to explain, and the standards of evaluation for explanation. The overall goal of this essay is to argue for tolerance for all three varieties of institutionalism, used where appropriate, much in keeping with the plea by Ian Lustig in the rejoinder to the Bates piece in PS (l997).

There is no room here to go into a detailed discussion of the three institutionalisms or of the controversy (for this, see Hall and Taylor 1994; Pierson 1996; and Thelen l998). All that is possible in this short space is to point to the differences among approaches in an effort to show that these are indeed very different enterprises, each of which provides different insights into political reality.

Broadly speaking, the three approaches have very different objects and goals of explanation: Rational choice institutionalism focuses on intentional, interest-motivated action and seeks to make universal generalizations or predictions about what rational actors will do within a given set of institutions, seen as structures of incentives. Historical institutionalism concentrates instead on the origins and development of the institutions themselves, seen as institutional structures and processes, which are explained by the (often unintended) outcomes of purposeful choices and historically unique initial conditions. Sociological institutionalism concerns itself with culturally framed actions, ideas, and identities that follow from culturally-specific rules and norms. These may or may not be "rational" in the stricter rational choice sense or predictable by way of universal generalizations, although they may be "expectable" within a given cultural context. The three approaches also have very different standards of evaluation, with the sociological approach referring to the "logic of appropriateness," the historical approach following the "logic of path-dependence," and the rationalist approach invoking the "logic of interest."

Naturally, there is overlap among the three approaches, given that interest is generally given expression through culture and can be traced throughout history, that history shapes interest and culture, and that culture gives both history and interest meaning. It stands to reason (whether understood in rationalist, historical, or sociological terms), therefore, that practitioners of any one of the methodologies would claim to account for the objects of explanation of the others, whether because interest can be seen as "nested" or "embedded" in historically evolving institutions or culturally derived norms, because history can be described as the product of the interest-motivated and norm-driven actions of individuals, or because culture can be interpreted as the expression of individuals’ cumulative interests or seen as embodied in the community’s evolving institutional structures and processes. But whatever the subsumability of one institutionalism by another, the three nevertheless remain methodologically distinct enterprises, each with a different set of insights into political reality, each with different limitations.

Rational choice institutionalism works best at identifying the interests and motivations behind rational actors’ behavior within given institutional settings. The deductive nature of its approach to explanation means that it not only is tremendously helpful at capturing the range of reasons actors would normally have for any given action within a given institutional incentive structure and at predicting likely outcomes, but also at bringing out anomalies or actions that are unexpected given the general theory (Scharpf l997). However, for the most part it cannot explain these anomalies if they depart radically from interest-motivated action (and therefore might better be explained in sociological or historical institutional terms). Moreover, because of its very deductiveness, along with a theoretical generality that often starts from universal claims about rationality, rational choice institutionalism has difficulty explaining any one individual’s reason for action within a given context or any particular set of real political events (see the critique by Green and Shapiro l994)—although one could argue that the recent attempt to ‘contextualize’ analyses through ‘analytic narratives’ represents a corrective to this problem (Bates et al., 1998; Levi, n/a). Generally, however, individuals qua individuals are not present here. Explanation is also static, focused on equilibrium conditions, and therefore has difficulty accounting for change over time. Finally, rational choice institutionalism’s emphasis on the self-interested nature of human motivation, especially where it is assumed to be economic self-interest, is value-laden, and can appear economically deterministic.

Historical institutionalism, by contrast, works best at delineating the origins and development of institutional structures and processes over time. It tends to emphasize sequences in development, timing of events, and phases of political change. Interests, moreover, rather than being universally defined, are contextual (Zysman l994; Thelen l998). Historical institutionalism tends to be less universalistic in its generalizations and more "mid-range" in its theory-building by focusing on changes in a limited number of countries unified in space and/or time or on a specific kind of phenomenon that occurs in or affects a range of countries at one time or across time (Thelen l998). But although more particular in its generalizations, the "new" historical institutionalism rarely stays at the level of the "mere story-telling" of which it is sometimes accused by rational choice institutionalists. Noticeably absent is the focus on "great men" or "great moments" characteristic of more traditional historical approaches to political history. In fact, the macro-historical approach prevalent in most accounts tends to emphasize structures and processes much more than the events out of which they are constructed, let alone the individuals whose actions and interests spurred those events. Here, too, then, there are no individual actors as such. But rationality in the strict rational choice sense is present only insofar as the institutions are the intended consequences of actors’ choices, which is often not the case, given the unintended consequences of intentional action and the unpredictability of intervening events. Thus, the "micro-foundational logic," as rationalists put it, is generally missing from this macro-historical work. Rather than appearing economically deterministic, it can appear historically deterministic or even mechanistic, where it focuses exclusively on continuities and path-dependency. The "critical junctures" literature is a corrective to this problem, since it looks at the ‘configurative’ moments; but in these moments it too looks to groups or classes of actors and developing processes rather than any one set of individual actors or events.

Sociological institutionalism, finally, works best at delineating the shared understandings and norms that frame action, shape identities, influence interests, and affect what are perceived as problems and what are conceived as solutions. Rather than being too general, it is sometimes accused of being too specific, and the ‘cultural knowledge’ it provides is useful mainly as preliminary to rational choice universalization. But when the objects of sociological institutionalism are subsumed under rational choice explanation, often the very essence of sociological institutionalism, the norms, rules, and reasons which are culturally unique or anomalous because they do not fit generally expected interest-motivations, get lost. Because sociological institutional explanations are arrived at inductively rather than deductively, they can lend insight into individuals’ reasons for action in ways that rational choice institutionalism cannot, whether or not they depart from the norm. Moreover, because of their ability to account contextually for individuals’ reasons for action, they are better able to explain the events out of which historical institutional explanations are constructed. However, because sociological institutionalism makes no universalistic claims about rationality and is generally focused on explanation within rather than across cultures, it risks an implicit relativism which leads some to question whether sociological institutionalism allows for any cross-national generalizations at all. In fact, generalizations are possible here too, by invoking similarities as well as differences in cultural norms and identities, much in the way of historical institutionalism with country-specific institutional structures and processes. The resulting explanation, however, involves a lower level of generality and less parsimonious, "thicker description" than in historical institutionalism, let alone rational choice institutionalism. Finally, rather than appearing either economically or historically deterministic, sociological institutionalism can appear culturally deterministic where it emphasizes the cultural routines and rituals to the exclusion of individual action which breaks out of the cultural norm, i.e., rule-creating action as opposed to rule-following action. Moreover, like the rational choice approach, it too can be too static or equilibrium-focused, and unable to account for change over time—although where it takes an historical perspective, it can also show how norms are institutionalized (see Katzenstein l996).

Each of the three institutionalisms, thus, offers a different perspective on political reality, each with different objects, goals, and standards of explanation, each with different advantages and disadvantages. So which institutionalism is most appropriate to the study of Europe? In the three selections that follow, we find different responses to this question. This essay suggests, however, that because each offers different insights into reality, Europeanists should use whichever of the three institutionalisms (among other methods) is appropriate to elucidating the problem at hand. And to gain a full sense of reality, they could try to combine all three. A discussion of how to accomplish this, though, must be left for another time.

Vivien A. Schmidt is professor of international relations at Boston University, an elected member of the ECSA Executive Committee, and the author of many publications on European political economy and public policy.
 

George Tsebelis

IT IS QUITE FREQUENT for empirical analyses to start by describing two theoretical approaches to the study of the European Union, intergovernmentalism and neofunctionalism, and then to proceed by stating that none of them can capture the reality of the phenomenon studied (see Cameron (1992) and Lange (1992) as examples). I want to discuss a third theoretical approach, institutional analysis, and suggest that if the other two do not describe well what we see before us, we may want to turn to institutional analysis to understand micro and macro phenomena of integration. In this note I will present the logic of institutional analysis, explain its differences from other approaches, and present some important phenomena in the history of the EU that can readily be understood on the basis of such an approach.

First, at the risk of oversimplifying a little, let me outline the important differences between the three approaches. For intergovernmentalists, governments are the ultimate decisionmakers in the EU, they define the process of integration and set its limits; as a consequence, their study of the EU focuses on the defining moments of integration, the signature of treaties (Moravcsik forthcoming). For neofunctionalists, governments are not the sole important actors in the EU. The focus shifts to the process "whereby political actors in several distinct national settings are persuaded to shift their loyalties, expectations, and political activities towards a new and larger center, whose institutions possess or demand jurisdiction over the pre-existing national states." The quote comes from Haas (1961: 366) and if the reader thinks it is too wide and imprecise she will not find any objection here. Several attempts have been made to apply neofunctionalist principles as well as elaborate on them (Burley and Mattli 1993, Stone Sweet and Sandholtz 1997).

For institutional approaches, the main focus is on the institutional structure of the EU. Why do institutions take such a preeminent role as opposed to say ideas, identities, processes, national interests, spillover effects, or other concepts that could or have been the starting point of the analysis? Let us start from the simplest possible understanding of human interaction. In such an understanding, there are three necessary concepts: the players (individual or collective) involved in the interaction, their strategies (which jointly determine the outcome), and the payoffs that they receive at the end of their interaction. In fact, in game theory these three concepts are sufficient for the description of any game.

If we look closer at the concept of "strategies" we will see that it depends on the sequence of moves that define the game, on the set of choices and information that each player has at the moment that is called upon to move. These parameters are determined by the institutional structure of the situation. Formal institutions specify that legislation starts with the introduction of a draft of a directive or a regulation by the Commission to the Parliament, and ends by the approval by the Council. Formal institutions specify what is permitted and what is not: for example, the treaties specify that environmental issues are today (but not in the sixties) within the jurisdiction of the EU. Formal institutions also specify what the set of choices is for each of the actors: for example if the EP wants to move a paragraph from one point of a bill to another, it has to introduce two amendments, one deleting the original text, and the second reintroducing it in the new position.

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. Consequently, one can study institutions in order to see how they are systematically associated with specific outcomes. The study can be theoretical, (on the basis of a formal or informal model) or empirical (association of observed frequencies or patterns of interaction among different actors). Studies like Cederman and Schneider (1994), Crombez (1996,1997), Garrett (1992, 1995), Garrett and Tsebelis (1996), Moser (1996), Streunenberg (1994), Steunenberg et. al. (1996), Tsebelis (1994) fall in the first category. Studies like Kreppel and Tsebelis (forthcoming) and Tsebelis and Kalandrakis (forthcoming) in the second.

Studying EU institutions in order to understand the results they produce is very similar to what other branches of comparative politics do. For example, studying the effect of the electoral system on the party system of a country (Duverger’s law) or of the influence of committees on legislative outcomes is part of the same conceptual approach, institutional analysis. Once the importance of institutions as a starting point is established, there are two different ways that one can proceed in their study: use them either as independent variables (study their effects) or as dependent variables (study their origins). The first approach, of using institutions in order to explain patterns of outcomes is logically prior to, and easier than explaining why certain institutions were adopted. So, I will start with institutions as independent variables.

The modus operandi of the study of the effects of institutions in rational choice theory is equilibrium analysis: actors are assumed to be rational, and trying to maximize the achievement of their goals (or their payoffs), subject to institutional constraints, and to what other actors are doing. If all actors are maximizing, the outcome is the best that each one of them can do, given the choice of the others, which is the essence of equilibrium analysis: no actor can unilaterally improve his situation (because, since he is rational he has selected the best choice given the choices of others). How can we place these theoretical ideas in the analysis of the EU?

There are a series of institutions that have played an important role in the evolution of the EU. Let us take the European Court of Justice for example. The ECJ established with a series of decisions the supremacy of European law over national law, the direct effect of European legislation, the principle of mutual recognition, which established the common market in practice. All these decisions were made in the late sixties and in the seventies. However, one would be hard pressed to find decisions of equal importance in the late 1980s or in the 1990s. Why?

There is a very simple explanation in institutional terms: the Luxembourg Compromise had forced unanimity in European decisionmaking before the Single European Act. At that time (1987) new institutions were introduced. These institutions prescribed two things: first, that decisions in the Council would be made by qualified majority, second that the proposals would originate with pro-European actors (the Commission and the Parliament). These new rules made it possible for European institutions to make decisions on important issues like the establishment of the single market, through the introduction of hundreds of directives and regulations.

Let us now focus on the "democratic deficit." While different analysts mean different things by the term, one common concern is the role that the EP plays in decisionmaking in the EU. What is this role? Again, institutional analysis can identify a series of parameters that increased the role of the EP over time. The first would be the isoglucose case in which the ECJ ruled that the Council would have to wait before making a decision until the EP gave its consultative opinion. The Court used the fact that the EP was directly elected as the foundation of its reasoning. What difference does it make in the influence of the EP if it is directly elected, or if the Council has to wait for its consultative opinion? A series of bargaining models have demonstrated the importance of delaying in shifting outcomes towards the ideal point of the actor that can introduce these delays (Tsebelis and Money (1997)). As a consequence, the isoglucose procedure marks a significant shift in the role of the EP, despite the fact that the consultation procedure was not officially modified. But there are more significant changes: the Single European Act endowed the EP with the power of a conditional agenda setter (Tsebelis 1994), Maastricht introduced codecision I which gave the parliament veto power (while taking away conditional agenda setting) (Tsebelis (1997)), and Amsterdam turned the EP into a co-equal legislator with the Council through codecision II. These institutional changes have as a result the overall increase of the power of the EP.

How about the role of the Commission? While the Commission introduces legislation in all legislative procedures, but the reason that it has been very powerful is that in the conciliation and cooperation procedures the Council decides on a Commission proposal and can accept it by qualified majority and modify it by unanimity. This asymmetry between accepting and modifying the Commission proposal was the source of Commission power (see Tsebelis and Kreppel 1998). However, in the codecision procedure (both versions I and II) the Council and the Parliament can agree between then in the last rounds of the procedure and overrule Commission objections. This modification has the result of reducing the importance of the Commission in legislation (Garrett and Tsebelis 1996).

So, changes in institutional structures had as results shifts of power among the different actors, as well as shifts in the visibility of these actors. These questions that arise naturally inside an institutional analysis framework are existent neither in intergovernmentalism nor in neofunctionalism.

Understanding how institutions operate is the first step to performing a successful institutional change. The EU has undergone institutional changes very frequently (three significant changes the last twelve years). These changes were performed with specific targets in mind, and presumably previous performance was measured against the targets before the next change was undertaken. For example, the role of the Commission and the role of the Parliament were part of the institutional debates, and each constitutional revision of the EU had as target the increase of the power of the Parliament and the decrease of the power of the Commission. There are two different ways of studying institutional change: one is the replication of the principles of the analysis above (rational choice institutionalism) and the other is historical institutionalism. Space constraints will restrict me to what I consider the major difference between the two. Interested readers can read Tsebelis (1990) and Pierson’s analysis (1996) to have a more extensive and precise understanding of the differences.

In my view the fundamental difference between rational choice and historical institutionalism is that rational choice performs equilibrium analysis, while historical institutionalism does not. In equilibrium analysis the designers of the new institutions try to study or guess as best they can the properties of the exiting institutional structures as well as the plausible alternatives, and select the one that is in their interest. This is the reason why as I said the study of institutions as independent variables is logically prior to the study of institutional change. The fact that different actors try to do their best and the selected institution is the one that commanded the required majority (simple, qualified or unanimous) does not mean that all the consequences of different structures were anticipated. In a study on the adoption of the conditional agenda setting, Tsebelis and Kreppel (1997) find little evidence that most of the participants knew what they were getting into (although people like Hallstein, Spaak, and de Gaulle had an accurate understanding).

To conclude, institutional analysis disagrees with intergovernmentalism with its fundamental belief that national actors (governments) alone determine the development of the EU. In terms of the importance of unintended consequences (another major difference between neo-functionalism and intergovernmentalism), institutional analysis (of rational choice variety) claims that unintended consequences will exist only under conditions of incomplete information. If information is complete, governments will take the results of institutional analyses as input in their bargaining over institutions. In this respect, under conditions of complete information (or some reasonable approximation) institutional analysis has to be used as the basis of intergovernmentalism. Under conditions of incomplete information, institutional analyses will lead to unintended consequences and will look similar to neo-functionalist analyses, with one significant difference: the rules of what constitutes a rational choice explanation are explicitly stated, and therefore explanations can be easily tested against data, while neofunctionalist arguments are so vague that are almost impossible to falsify.

1. The jury is still out in terms of the comparison between conditional agenda setting and unconditional veto; see Tsebelis and Garrett (1997), Garrett and Tsebelis (1997), and Scully (1997).

George Tsebelis is professor of political science at the University of California Los Angeles.
 

Thomas Risse

FOR THE LAST TEN years or so, wars have raged in many U.S. political science departments leading to hostile takeovers and the disappearance of entire subfields of political science. A particular breed of scholars has turned a legitimate meta-theoretical approach to the study of social phenomena into an ideology. Some try to establish rational choice "to occupy a central position within the discipline" and to provide a general framework "which transforms ethnography and narratives into theory-driven claims, amenable to refutation" (Bates 1997: 168-169). The implicit claim is, of course, that only rational choice theory provides the tools by which "area studies narratives" can be transformed into generalizable propositions and in legitimate social science.

Fortunately, such attempts to achieve intellectual and disciplinary hegemony have not yet reached the study of the European Union (EU). Fortunately, scholars of the EU are still busy with empirical and theoretical debates about liberal inter-governmentalism, neofunctionalism, multi-level governance, and the like (for an excellent review see Hix 1998). They remain far less inclined to engage in meta-theoretical controversies about the relative merits of rational choice versus other theories of social action. The controversy between rational choice and constructivism which has dominated the field of international relations during the last ten years (Checkel 1998; Adler 1997; Price and Reus-Smit 1998) has not yet preoccupied EU studies (but see Jorgensen 1997).

The dominant "puzzle"- rather than "theory"-driven approach to Europeanization  has its merits, but there are also some dangers. The advantages are quite clear. Problem-driven research tends to keep the feet of scholars on the ground. It prevents them from engaging in fruitless meta-theoretical controversies, unless these debates can be engaged to furthering our understanding of real-world puzzles. Moreover, research starting from empirical puzzles can be fruitfully used to evaluate the relative merits of competing propositions stemming from various theoretical approaches. Thus, good problem-driven research should actually lead to theory-testing.

The risk of "puzzle driven" approaches to the study of Europeanization is to replicate the caricature of "area studies versus general social science (read: rational choice)" that seems to preoccupy so many political science departments in the U.S. I do not deal here with single case studies or the "n of 1" problem in EU studies which has been successfully laid to rest in a recent ECSA Review Forum (X: 3, Fall 1997). Europeanization is a unique social process, but it can certainly be compared to other social processes. The inherent danger of "problem driven" research on the EU even if it employs the comparative case study method, is to buy into meta-theoretical commitments without being aware of them. This begins with the question of what counts as an empirical "puzzle"? Such puzzles do not fall from heaven, but usually emerge when an empirical case proves to be surprising in light of existing explanations. In other words, previous theoretical commitments define what is being regarded as an interesting empirical question to be researched. Moreover, our meta-theoretical commitments largely determine the range of potential answers to our empirical questions including the research strategy.

It is not surprising, therefore, that most scholars involved in Europeanization studies share a meta-theoretical approach which one could label "soft rational choice" based on the assumption of utility-maximizing actors whose interests are generally taken for granted.  The prevailing controversies in EU studies between liberal intergovernmentalists, on the one hand, and those using network analysis and multi-level governance approaches, on the other hand, share the same rationalist meta-assumptions. Historical institutionalism has only recently entered Europeanization research (Pierson 1996; Bulmer 1998; Pollack 1996). This is surprising, since the European Union is certainly an institution sui generis. How can we theorize about an institution such as the EU without exploring further what is an essential element of any institution, namely the inter-subjective quality of its norms as collectively shared standards of appropriate behavior (Kratochwil and Ruggie 1986)? Sociological institutionalism focussing on rule-guided behavior and the logic of appropriateness as a mode of social rationality which differs substantially from the instrumental rationality of rational choice has not yet generated much empirical research in EU studies (but see Wind 1998; Kohler-Koch 1997). The same holds true for argumentative rationality, i.e., the role of arguing and deliberation whereby actors' interests and their definition of the situation are subject to challenges and counter-challenges (but see Joerges and Neyer 1997). Interestingly enough, monetary integration, on the one hand, and citizenship issues, on the other, are about the only areas of Europeanization where substantial theory-guided empirical research beyond rational choice has been conducted (McNamara 1998; Verdun forthcoming; Marcussen 1998; Risse et al. 1998; Checkel 1997; Wiener 1997).

Why is it that sociological institutionalism or social constructivism have not yet generated much empirical research in Europeanization studies? They certainly pose interesting questions which are worthwhile to research: How do European law and European regulations achieve their taken for granted nature in the domestic settings? Which socialization effects emerge from the fact that more and more domestic political and bureaucratic actors are continuously involved in European policy-making and implementation? How do member state institutions change in response to the adaptational pressures emanating from European rules and regulations? Can we explain European policy-making "by subterfuge" (Héritier 1997) and informal networking without taking the logic of arguing into account? How does the ongoing process of Europeanization affect collective systems of meanings and understandings including national identities?

I propose two answers why these questions have rarely been tackled so far in Europeanization studies. First, as argued above, the prior meta-theoretical commitment of most puzzle-driven scholars is the assumption that actors pursue instrumentally defined interests and try to maximize given utilities. Second, most of Europeanization research has so far implicitly taken a "bottom up" approach to the problem. I.e., most scholars have focused on how to explain European integration as such. No wonder that strategic bargaining played such a prominent role! The more Europeanization research shifts the perspective and tries to understand the effects of European integration on what we used to call the domestic level, the more it is likely that scholars will be more open-minded toward sociological institutionalism which offers theoretical insights in processes of institutional adaptation. In addition, the more Europe "hits home" and becomes subject of normal public debates and controversies, the more it is likely that research on meaning constructions and public discourse will increase. In sum, there is hope for social constructivism in Europeanization studies!

1. I prefer the term "Europeanization" to "EU studies," since the European structure of governance entails far more than just processes and events at the EU level in Brussels.
2.  As opposed to "hard" rational choice using explicit game-theoretical analysis.

Thomas Risse is professor in the Dept. of Political and Social Sciences, European University Institute, Florence, Italy.
 

Fritz W. Scharpf

MUCH OF THE policy-oriented research conducted within the ECSA community is of the case-study variety, producing detailed historical accounts of policy processes and outcomes in one or a few historical cases, with explanations emphasizing the importance of situational and institutional conditions and/or of the perceptions, preferences and strategies of the (individual and corporate) actors involved. But since contemporary social science is thought to demand more than accurate descriptions, plausible narratives, and perceptive interpretations, most authors of case studies feel the urge to use their findings as a basis for either formulating or testing theoretical propositions of supposedly universal application. As a consequence, most of these studies begin with a discussion of competing general "approaches" (current favorites are the three institutionalisms, rational choice, historical and sociological), and they then move on to competing "theories" of European integration (with "liberal intergovernmentalism" and "neofunctionalism" alias "supranationalism" presently competing against each other as well as against "multi-level" and "network" representations of European policy processes and their variants). If the study is well done, it will then typically come to the conclusion that none of the competing approaches is entirely adequate for framing the research design, and that none of the competing theories will fully account for the empirical findings (S. Schmidt 1997). If the author is ambitious, she will then try to improve on the state of the art by formulating new generalizations that perfectly fit her own case and others that she happens to be thinking of. But of course this new "theory" will inevitably be rejected as well by the author of the next empirical study that is focusing on another set of cases.

What explains this endless cycle of generalizations and falsifications that does not seem to advance the field at all? It is, I suggest, a basic misunderstanding of the role of theory in empirical social science research of the case study variety —which is always plagued by the discrepancy between the number of the variables that play a role in the cases studied and the limited number of cases that can be studied by the case method.

Take the example of the Common Agricultural Policy which I had used to illustrate my (rational-choice institutionalist and intergovernmentalist) analysis of the "joint decision trap" (Scharpf 1988). Comparing CAP policy outcomes to similar ones in German federalism, I had concluded that a multi-level governing structure in which decisions at the center depended on the (near unanimous) agreement of sub-central governments (i.e., a governing structure with many veto positions in the sense defined by Tsebelis 1995) would produce policy immobilism. I still think that conclusion was a good one, for the purposes and the range of constellations that I had in mind. In the meantime, however, a number of contributions in the literature have shown that in a variety of policy areas the "joint decision trap" did not prevent significant progress not only in substantive European policy but also with regard to the decision rules themselves (Pierson 1996; Pollack 1994). In other words, my original argument was overgeneralized in the sense that its formulation, though true in its proper domain, seemed to cover constellations where it would not hold. What I might have said instead is that EC decision rules, like those of German federalism, made policy changes more difficult by giving member governments defending the status quo an asymmetrical advantage, and that under the specific (but unspecified) situational conditions prevailing in CAP and in the other policy areas which I had discussed, this advantage could not be overcome by the proponents of change.
The tendency to overgeneralize is so pervasive because generalizations must focus narrowly on a limited set of variables, and because no author of a case study is able to provide a comprehensive list of those unspecified background conditions whose presence is also necessary to produce the specified outcome. By the same token, however, any general theory will be too "narrow" (i.e., focus on too few variables) to account for a concrete individual case. That explains why the use of individual cases as "test" of general theories is likely to produce mostly "falsifications."

For an example, take Tsebelis’ contribution to the present issue. I doubt that he truly meant to say that institutions do "determine the choices of actors." Institutions create opportunity structures and constraints. But what happens within these constraints is very much influenced, first by the characteristics of the external situation in which the actors are interacting with each other, second by the characteristics of the actors (their capabilities and cognitive and normative orientations) and, most importantly, by the problem- and policy-specific constellations of actor preferences and options—all of which need to be explicitly conceptualized in any framework that is able to account simultaneously for the influence of institutions and for the fact that situations differ and that actors are purposeful and resourceful agents, rather than rule-following automata (Scharpf 1997).

Different external situations will activate specific actor interests that would otherwise remain dormant, and may thus create situation-specific alliances and conflict lines. Similarly, in some policy areas—e.g., in product-oriented technical standardization—the economic interests of member states will favor uniform European rules, whereas in other areas—e.g., the harmonization of national tax or social security systems—interest constellations are much less conducive to agreement (Scharpf 1999, chapter 3). Moreover, specific political or ideological preferences of a national government (think of Thatcher on social policy) may make a difference, and ultimately the outcome will also be affected by the strategic and tactical skills of different actors, and their ability to design acceptable compromise solutions and package deals.

In other words, the outcomes of specific European decision processes are not only affected by the institutional rules on which Tsebelis focuses but also by a variety of contingent (and interacting) factors which in combination define the "game" that is being played. Hence, even if we restrict attention to factors that are clearly relevant in rational-choice analyses—thus leaving aside the further contingencies introduced by Thomas Risse’s "constructivist" claims—any individual case that is interesting enough to merit being studied in depth will be characterized by a complex and nearly unique constellation that defies the propositions of any theoretical generalization.

But how could we escape this vicious cycle where empirical studies tend to produce overgeneralized hypotheses, only to see them falsely falsified by the next study, which will proudly present its own overgeneralized propositions, and so on? One solution might be to turn the vice of "selection bias" into a methodological virtue: if test cases are carefully selected so as to hold the unspecified antecedent conditions of a generalized proposition as constant as possible, the risks of false falsification can be minimized. An example is Andrew Moravcsik’s decision to test his fairly general "liberal-intergovernmentalism" hypothesis (Moravcsik 1993) in a study that focuses exclusively on the "great" decisions shaping and changing the constitution of the European Community through (of course) intergovernmental negotiations (Moravcsik 1998). Of course, if the author of the hypothesis is not also in control of case selection, confirmation depends on a sympathetic reconstruction of unspecified antecedent conditions which may overtax the normal supply of human kindness in competitive academic environments. But that is not the only price. In effect, Moravcsik seems to be squandering the wealth of information produced by his supremely well researched case studies on the effort to prove the narrow proposition that, by and large, national governments are in control of intergovernmental negotiations over treaty revisions.

A more standard solution in the literature treats narrow hypotheses as being merely "probabilistic" laws which, by definition, could not explain or predict a single case and hence could also not be tested in single-case studies. The appropriate procedures are statistical tests on large data sets in which the causal influence of idiosyncratic factors in the individual case would be neutralized by the large-numbers effect. Much of the current work in comparative politics and comparative political economy is of this type (Garrett 1998; Iversen 1999; M. Schmidt 1997). Its limitation arises, first, from the large-numbers condition itself which restricts the amount of information that can be obtained (or, if obtained, processed) for each individual case and, second, in the distortions introduced when theoretically defined variables must be represented by only those empirical indicators for which quantitative measures or plausible proxies can be obtained. But even if these hurdles are overcome, the question is how much the confirmation of a probabilistic law will add to our ability to explain (let alone, predict) the outcome of individual cases? It is of course important to know that Tsebelis’ analytically derived veto points hypothesis held up in a multivariate statistical study that explains 57 percent of the variance in the adoption of important laws (Tsebelis 1998). But how much does that help us in explaining an individual case when we also know that, within exactly the same constitution with the same number of veto points, the Dutch political system was totally blocked in the 1970s but capable of major policy reforms in the 1980s and 1990s (Visser and Hemerijck 1998)?

This is not a plea against the search for general laws in the social sciences. But it is a plea for separating the activities of theory development and theory testing from research aiming at the explanation of concrete cases. Theory development, while profiting from familiarity with historical and contemporary real-world conditions, is best performed under the discipline of logical analysis and formal modeling— which provide some protection against the temptations of sloppy thinking and ad hoc interpretations by requiring the explicit and precise specification of assumptions, causal mechanisms, and rules of interaction. Theory testing, to the extent that it is possible at all, is best done under conditions that allow the application of either experimental or statistical controls. Empirical case studies, at any rate, are not the best environment for either activity.

But if general theories do not explain concrete cases, what is their use in empirical research? Here, the normal understanding of "probabilistic theory" is misleading. If it represents a causal mechanism, it makes no sense to think that this mechanism is present 35% of the time, and absent in the remaining 65% of cases. If the causal relationship exists at all, we must assume that it is always present. Thus it makes more sense to consider empirically confirmed causal mechanisms as partial theories that define what we should expect in a specific case. If the predicted effect is not manifest, this suggests that it is being counteracted by another mechanism, presumably specified in another partial theory, which we should then try to identify in our attempt to explain the complex real-world constellations of our case studies. Thus, by separating the activity of theory formulation and testing from the activity of explaining real-world cases, and by accepting the fact that any theory will at best explain a partial aspect of such cases, we might finally get both, better explanations and more cumulative theory development.

Fritz W. Scharpf is a scholar at the Max Planck Institute in Köln, Germany.
 

Bibliography

Adler, Emanuel (1997) "Siezing the Middle Ground: Constructivism in World Politics," European Journal of International Relations 3: 3, 319-363.

Bates, Robert et alia (1998) Analytic Narratives. Princeton: Princeton University Press.

_____ (1997) "Area Studies and the Discipline: A Useful Controversy?" PS: Political Science & Politics 30: 2, 166-169.

Bulmer, Simon J. (1998) "New Institutionalism and the Governance of the Single European Market," Journal of European Public Policy 5, 365-386.

Burley, Anne-Marie and Walter Mattli (1993) "Europe Before the Court: A Political Theory of Legal Integration," International Organization 47: 41-76.

Cameron, David R. (1992) "The 1992 Initiative: Causes and Consequences," in  A. Sbragia (ed.)  Euro-Politics. Washington, D.C.: Brookings.

Cederman, Lars Erik and Gerald Schneider (1994) "The Change of Tide in Political Cooperation: A Limited Information Model of European Integration," International Organization 48: 633-62.

Checkel, Jeffrey T. (1998) "The Constructivist Turn in International Relations Theory," World Politics 50: 2, 324-348.

_____ (1997) "International Norms and Domestic Politics: Bridging the Rationalist-Constructivist Divide," European Journal of International Relations 3: 4, 473-495.

Crombez, Christophe (1997) "Policy Making and Commission Appointment in the European Union," Aussenwirtschaft 52: 63-82.

_____ (1996) "Legislative Procedures in the European Community," British Journal of Political Science 26: 199-228.

Garrett, Geoffrey (1998) Partisan Politics in the Global Economy. Cambridge: Cambridge University Press.

_____ (1995) "From the Luxembourg Compromise to Codecision: Decision Making in the European Union," Electoral Studies 14: 3, 289-308.

_____ (1992) "International Cooperation and Institutional Choice," International Organization 46, 533-560.

_____ and George Tsebelis (1997) "More on the Co-decision Endgame," Journal of Legislative Studies 3 (4): 139-143.

_____ and George Tsebelis (1996) "An Institutional Critique of Intergovernmentalism," International Organization 50: 269-300.

Green, Donald and Ian Shapiro (1994) The Pathologies of Rational Choice. New Haven: Yale University Press.

Haas, Ernst (1961) "International Integration: The European and Universal Process,"  International Organization 15: 366-92.

Hall, Peter and Rosemary Taylor (1996) "Political Science and the Three New Institutionalisms," Political Studies 952-973.

Héritier, Adrienne (1997) "Policy-making by Subterfuge: Interest Accommodation, Innovation ..." Journal of European Public Policy 4: 171-189.

Hix, Simon (1998) "The Study of the European Union II: The ‘New Governance’ Agenda and Its Rival," Journal of European Public Policy 5: 38-65.

Iverson, Torben (1998) "Wage Bargaining, Hard Money, and Economic ..." British Journal of Political Science 28, 31-61.

Joerges, Christian and Jürgen Neyer (1997) "Transforming Strategic Interaction into Deliberative ..." Journal of European Public Policy 4: 609-625.

Jorgensen, Knud Erik (ed.) (1997) Reflective Approaches to European Governance. London: Macmillan.

Katzenstein, Peter J. (1996) Cultural Norms and National Security: Policy and Military in Postwar Japan. Ithaca, NY: Cornell University Press.

Kohler-Koch, Beate (ed.) (1997) Interaktive Politik in Europa: Regionenen im Netzwerk der ... Opladen: Leske & Budrich.

Kratochwil, Friedrich and John G. Ruggie (1986) "International Organization: A State of the Art ..." International Organization 40: 4, 753-775.

Kreppel, Amie and George Tsebelis (forthcoming) "Coalition formation inside the European Parliament," Comparative Political Studies.

Lange, Peter  (1992) "The Politics of the Social Dimension," in  A. Sbragia (ed.) Euro-Politics. Washington D.C.: Brookings.

Levi, Margaret (n/a) "Producing an Analytic Narrative," in Bower and Peterson (eds.), Critical Comparisons in Politics and Culture. New York: Cambridge University Press.

Marcussen, Martin (1998) "Central Bankers, the Ideational Lifecycle, and the Social Construction of EMU," Florence, Italy: European University Institute.

McNamara, Kathleen (1998) The Currency of Ideas: Monetary Politics in the European Union. Ithaca, NY: Cornell University Press.

Moravcsik, Andrew (1998) The Choice for Europe: Social Purpose and State Power from Messina to Maastricht. Ithaca, NY: Cornell University Press.

_____ (1993) "Preferences and Power in the European Community: A Liberal Intergovernmentalist Approach," Journal of Common Market Studies 31, 473-524.

Moser, Peter (1996) "The European Parliament as a Conditional Agenda Setter: What are the Conditions?" American Political Science Review 90: 834-838.

Pierson, Paul (1996) "The Path to European Integration: A Historical Institutionalist Analysis," Comparative Political Studies 29: 123-63.

Pollack, Mark (1996) "The New Institutionalism and EC Governance: The Promise and Limits of ..." Governance 9: 4, 429-458.

____ (1994) "Creeping Competence: The Expanding Agenda of the EC," Journal of Public Policy 14, 95-145.

Price, Richard and C. Reus-Smit (1998) "Dangerous Liaisons? Critical International Theory and Constructivism," European Journal of International Relations 4: 3, 259-294.

Risse, Thomas et alia (1998) "To Euro or Not to Euro? The EMU and Identity Politics in the European Union," Florence, Italy: European University Institute.

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

_____ (1997) Games Real Actors Play. Boulder, CO: Westview Press.

_____ (1988) "The Joint Decision Trap: Lessons From German Federalism and ..." Public Administration 66: 239-278.

Schmidt, Manfred G. (1997) "Determinants of Social Expenditure in Liberal Democracies ..." Acta Politica 32: 153-173.

Schmidt, Susanne K. (1997) "Sterile Debates and Dubious Generalizations: European Integration Theory ..." Journal of Public Policy 16: 233-271.

Schneider, Gerald (1995) "The Limits of Self-Reform: Institution-Building in the European Union," European Journal of International Relations 1: 59-86.

Scully, Roger M. (1997) "The European Parliament and the Co-Decision Procedure: A Re-Assessment," Journal of Legislative Studies 3 (3): 58-73.

Steunenberg, Bernard (1994) "Decisionmaking under Different Institutional Arrangements,"  Journal of Institutional and Theoretical Economics 150: 642-69.

_____ et alia (1996) "Policy Making, Comitology and the Balance of Power in the European Union," International Review of Law and Economics 16: 329-44.

Stone Sweet, Alec and Wayne Sandholtz (1997) "European Integration and Supranational Governance," Journal of European Public Policy 4: 297-317.

Thelen, Kathleen (1998) "Historical Institutionalism in Comparative Politics," paper presented at APSA (Boston, Sept.3-6).

Tsebelis, George (1998) "Veto Players and Law Production in Parliamentary Democracies ..." Mss., University of California Los Angeles.

_____ (1997) "Maastricht and the Democratic Deficit," Aussenwirtschaft 52: 29-56.

_____ (1994) "The Power of the European Parliament as a Conditional Agenda Setter," American Political Science Review 88: 128-142.

_____ (1990)  Nested Games: Rational Choice in Comparative Politics. Berkeley: University of California Press.

_____ and Geoffrey Garrett (1997) "Agenda Setting, Vetoes and the European Union's Co-decision Procedure," Journal of Legislative Studies 3 (3): 74-92.

_____ and Anastassios Kalandrakis (forthcoming) "European Parliament and Environmental Legislation: The Case of Chemicals," European Journal of Political Research.

_____ and Amie Kreppel (1998) "The History of Conditional Agenda Setting in European Institutions," European Journal of Political Research  3: 41-71

_____ and Jeannette Money (1997) Bicameralism. New York: Cambridge University Press.

Verdun, Amy (forthcoming) "EMU and Epistemic Communities."

Visser, Jelle and Anton Hemerijck (1997) A Dutch Miracle: Job Growth, Welfare Reform and Corporatism in The Netherlands. Amsterdam: Amsterdam University Press.

Wiener, Antje (1997) Building Institutions: The Developing Practice of ‘European’ Citizenship. Boulder, CO: Westview Press.

Wind, Marlene (1998) "IR Theory Meets European Union Law: Constitutional Battles ..." Ph.D. Dissertation, Florence, Italy: European University Institute.

Zysman, John (1994) "How Institutions Create Historically Rooted Trajectories ..." Industrial and Corporate Change 3: 1, 243-283.