The Limits of Design: Explaining Institutional Origins and Change

PAUL PIERSON* Political scientists have paid much more attention to the effects of institutions than to issues of institutional origins and change. One result has been a marked tendency to fall back on implicit or explicit functional accounts, in which the effects of institutions explain the presence of those institutions. Institutional effects may indeed provide part of such an explanation. Yet the plausibility of functional accounts depends upon either a set of favorable conditions at the design stage or the presence of environments conducive to learning or competition. Exploring variability in the relevant social contexts makes it possible to both establish the restricted range of functional accounts and specify some promising lines of inquiry into the subject of institutional origins and change. Institutions now stand at the heart of much theorizing and explanation in political science. Analysts working from a variety of perspectives have produced compelling work on the significance of institutional arrangements for political and social outcomes (Hall and Taylor 1996). By contrast, we have made far less progress in treating institutions as themselves important objects of explanation. The origins of institutions, as well as the sources of institutional change, remain quite opaque. As David Kreps has observed, the sophisticated economic literature on the effects of institutions leaves open the question, where did the institutions come from? . . . Having a theory about how institutions arise and evolve could be more informative than theories of equilibrium within the context of a given set of institutions (Kreps 1990, 530). Perhaps the dominant response to this challenge in political science is the resort (explicitly or implicitly) to functionalist reasoning: the explanation of institutional forms is to be found in their functional consequences for those who create them. Such an approach, however, has serious shortcomings. The goal of this essay is to explore some of the limitations of functionalist mechanisms of institutional development, and to use this critique to point out some fruitful lines for theorizing about institutional origins and change. The discussion is limited to formal political institutions, which are the products of conscious design and redesign.

*Harvard University Governance: An International Journal of Policy and Administration, Vol. 13, No. 4, October 2000 (pp. 475499). 2000 Blackwell Publishers, 350 Main St., Malden MA 02148, USA, and 108 Cowley Road, Oxford, OX4 1JF, UK. ISSN 0952-1895

476

PAUL PIERSON

The purpose, it should be emphasized, is not to deride functionalist thinking. On the contrary, thinking in functionalist terms about an existing institution, policy, or social organization is one good way to derive causal hypotheses about institutional origins and change. Functional accounts, however, are far from being the only plausible ones. The task is to identify alternative mechanisms of institutional origins and change, and to begin to specify the conditions that make one or another account more probable. The argument proceeds in three stages. The first briefly outlines the functionalist tendencies prominent in much of the work political scientists have done on institutions. The second examines the reasons why one might expect political actors to be only moderately prone to design functional institutions. The third considers at greater length two possible mechanismslearning and evolutionwhich might generate functional institutions over time and discuss the limits of each. The conclusion elaborates some implications and offers suggestions for further research.
I. FUNCTIONALIST ARGUMENTS ABOUT INSTITUTIONS

Political scientists have had much more to say about institutional effects than about institutional origins and change. Both a cause and consequence of these lacunae has been a turn to functionalist reasoning. Although frequently such reasoning is not explicitly stated, functionalist treatments of institutions are prevalent in political science. They are common, for instance, among those who emphasize the rational choices of individual actors that underlie political activity, and the reasonably efficient nature of collective responses to social needs. To take some prominent examples from different subfields of the discipline: international regimes facilitate agreements through issue linkage and the reduction of monitoring costs (Keohane 1984); congressional committees prevent cycling (Shepsle 1986), enable gains from trade among legislators with different priorities (Weingast and Marshall 1988), or rationalize the flow of scarce information (Krehbiel 1991); the European Court of Justice offers a neutral monitoring and enforcement agency for the European Unions member states (Garrett 1995). These works vary in the extent to which they explicitly address issues of institutional origins and change, but in each case one is left with the impression that institutional functioning in large part explains the presence of particular institutional arrangements. Much of this theorizing builds on the new institutional economics, especially Oliver Williamsons work on transaction costs (Williamson 1975; Moe 1984). Williamson is quite explicit in arguing that the development of a particular organizational form can be explained as the result of the efforts of rational actors to reduce transaction costs. More generally, functionalist arguments take the following form: outcome X (an institution, policy, or organization, for instance) exists because it serves the function Y. This is not a crazy place for a social scientist interested in

THE LIMITS OF DESIGN

477

institutional formation to start. If actors are purposive, it will often be the case that the effects of an institution have somethingperhaps a lotto do with an explanation for its emergence, persistence, or change (Keohane 1984). The problem, however, is that much functionalist thinking in political science acts as an end-point rather than a starting point for analysis. Functionalism offers an easy rationale for evading the thorny issues of institutional emergence and change. Rather than directly examining these topics, one simply begins with existing institutions. The assumption is that these institutions exist in the form they do because they are functional for social actors. The task of the analyst is to lay bare the particular function (generally, the resolution of some sort of collective choice problem) the institution serves. Given a reasonable amount of intellectual creativity and a moderately flexible analytical tool kit, this task of unmasking functions often turns out to be a fairly easy one. Rational choice analysts, for instance, possess just such a tool kit, making it possible to reconcile virtually any observed outcome with a functionalist account (Green and Shapiro 1994). If such an account is not readily at hand, one can introduce it by incorporating side payments, or nesting one game inside another (e.g., Tsebelis 1990; Lange 1993). Yet it is one thing to demonstrate (or, more often, speculate) that an institution is doing something for social actors. It is quite another thing to jump to the conclusion that this accounts for the institutions presence. Such a functionalist logic requires one of two supporting lines of argument: either functional institutions must be products of rational design, or they emerge over time through mechanisms of institutional enhancement.
II. THE LIMITS OF RATIONAL DESIGN

A simple vision of institutional design focuses on the intentional and far-sighted choices of purposive, instrumental actors. By this account, institutional effects should be seen as the intended consequences of their creators actions. Although most political scientists would probably reject a strong form of this argument, it is often a hidden assumption in the analysis of institutions. In any event, exploring the potential pitfalls of such a premise highlights some important challenges for a more sophisticated approach to issues of institutional origins and change. Each component of what might be termed the rational-instrumental approach to institutions deserves serious interrogation. Actors may not be instrumental in the sense implied by this framework. They may not be far-sighted. Finally, institutional effects may not be intended. Each of these dimensions exposes a possible limitation on the effectiveness of institutional design, and is therefore an important topic for further theoretical and empirical investigation.

478

PAUL PIERSON

Limitation 1. Institutional Designers May Not Act Instrumentally Research on institutions often stresses or assumes the instrumental behavior of designers. Institutions are constructed along particular lines because actors expect that particular features will produce specific consequences. Features of institutions are believed to hold significance only to the extent they help actors achieve their goals. There is, however, a strong tradition in sociological theory, recently consolidated in sociologys version of the new institutionalism, that challenges this premise (Meyer and Rowan 1977; March and Olson 1989; Powell and DiMaggio 1991). In structuring institutional arrangements, actors may be motivated more by conceptions of what is appropriate than by conceptions of what would be effective. Peter Hall and Rosemary Taylor summarize this line of argument as follows:
[M]any of the institutional forms and procedures used by modern organizations were not adopted simply because they were most efficient for the tasks at hand, in line with some transcendent rationality. Instead, they . . . should be seen as culturally-specific practices, akin to the myths and ceremonies devised by many societies, and assimilated into organizations, not necessarily to enhance their formal means-end efficiency, but as a result of the kind of processes associated with the transmission of cultural practices more generally (1996, 946947).

Sociologists, for instance, have emphasized the pervasive diffusion of particular institutional forms, even in widely divergent contexts. Rather than revealing the focus of actors on efficiency, they suggest that this institutional isomorphism reflects the sensitivity of actors to the need to legitimate their activities. If institutional arrangements are adopted because they are perceived to be appropriate, not because they serve a means-end instrumentality, then such arrangements may actually be dysfunctional for the particular local context. The extent to which such logic of appropriateness behavior motivates institutional design remains controversial among sociologists. At a minimum, however, they have marshaled considerable empirical material casting doubt on the validity of assuming that institutional designers, as a rule, are motivated exclusively or even predominantly by instrumental concerns. Limitation 2. The Problem of Short Time Horizons A statement attributed to David Stockman, budget director during the Reagan administration, is unusual among political decisionmakers only for its candor. Asked by an adviser in 1981 to consider pension reforms to combat Social Securitys severe long-term financing problems, Stockman dismissed the idea out of hand, exclaiming that he had no interest in wasting a lot of political capital on some other guys problem in [the year] 2010 (quoted in Greider 1982, 43). The question of actors time horizons constitutes a central issue for analysts of institutional design. This problem is of particular relevance to

THE LIMITS OF DESIGN

479

political scientists. Long-term institutional consequences may be the by-products of actions taken for short-term political reasons. The evolution of the congressional committee system in the USAan important institutional feature of contemporary American governanceis a good example. As Kenneth Shepsle notes, Henry Clay and his supporters introduced the system to further their immediate goals without regard to long-term consequences: The lasting effects of this institutional innovation could hardly have been anticipated, much less desired, by Clay. They were by-products (and proved to be the most enduring and important products) of self-interested leadership behavior (Shepsle 1989, 141). In this case, the systems long-term functioning was not the goal of the actors who created it. By the same token, an explanation for the institutions creation cannot be derived from an analysis of its long-term effects. Many of the implications of political decisionsespecially complex policy interventions or major institutional reformsonly play out in the long run. Yet political actors, especially politicians, would often seem most interested in the short-term consequences of their actions; long-term effects may be heavily discounted. One major reason is the logic of electoral politics. John Maynard Keynes once noted that in the long run, we are all dead; for politicians in democratic polities, electoral death can come much faster. Because the decisions of voters, which determine political success, are taken in the short run, elected officials may employ a high discount rate. They generally will pay attention to long-term consequences only if these become politically salient, or when they have little reason to fear short-term electoral retribution. If politicians often have short time horizons, this has important implications for theories of institutional origins and change. Assume that the crucial decisionmaker is a politician up for reelection in two years. In this context, effects after the election cycle might not count for much. If political decisionmakers face many decisions where short-term and long-term benefits of institutional choices diverge, and if their time horizons tend to be short, this would seem to diminish the prospect that institutions will be designed to achieve functional outcomes over the long term. Political scientists are beginning to pay considerable attention to the issue of time horizons. The efforts of rational choice theorists, who recognize the possible implications of short time horizons for their assumptions about institutional origins, have been particularly significant. How, these theorists ask, might the shadow of the future be made more relevant for political actors making choices today? There are in fact two distinct but related problems here: (1) Actors may have genuinely short time horizonsthat is, they lack incentives to think about the long term; and (2) they may care about the future but for a variety of reasons believe they cannot reliably influence itin such cases, short-term considerations will again predominate. If time horizons can be lengthened, then a claim that the long-term consequences of institutions in some sense explain their origins becomes

480

PAUL PIERSON

more plausible. In fashioning responses to this problem, rational choice theorists have drawn heavily on earlier work in economics. Economic actors would seem to confront similar problems related to time horizons, but much of the literature produced by the new institutional economics stresses how particular social arrangements can overcome these difficulties. The first response stresses that political actors will not focus exclusively on immediate considerations if they are accountable to actors with longer time horizons. If politicians are agents for principals (such as voters or interest groups) who take a long view and can effectively monitor politicians behavior, then the problem is effectively mitigated. Even the politician interested only in winning the next election will need to give the future its due. Because of difficulties with this line of argument (discussed below), theorists have also turned to more subtle claims to establish the relevance of the long term for current decisionmakers. Thus a second line of argument involves the adoption of mechanisms of credible commitment (North and Weingast 1989; Shepsle 1991; North 1993). Shepsle (1991, 246) has summarized the key insight of this work: [T]he ability to commit often (not always) expands ones opportunity set, whereas the capacity to exercise discretionwhich includes the latitude to renege or behave opportunisticallyreduces it." The crucial problem here is what economists call time inconsistency. It may be rational for an actor to make an agreement, but equally rational to subsequently break it. Since others will recognize this danger, the initial bargain will not be possible. In such contexts, which are common in social affairs, actors can make beneficial long-term bargains if they can tie their hands (or those of their successors), increasing the confidence of other participants that agreements will not be exploited. Institutions can be designed to disable discretion, making commitments credible and therefore lengthening the relevant time horizons for all concerned. Douglas North and Barry Weingasts (1989) frequently invoked example concerns the ability of British monarchs to increase their revenue-generating capacity by ceding significant authority over financial matters to Parliament, thereby reducing the prospect that they would unilaterally renege on accumulated debts. Recent research suggests that particular institutional designs (such as independent central banks), empowering particular kinds of political actors (e.g., bankers), may succeed in lengthening time horizons in politics (Persson and Tabellini 1994). International relations scholars have developed similar arguments about the consequences of international regimes for increasing the credibility of interstate bargaining and thereby lengthening the relevant time frame for decisionmakers. Galvanized by the work of North, Weingast, and others, a new generation of rational choice scholars is canvassing the political universe in search of credible commitment mechanisms.

THE LIMITS OF DESIGN

481

A final line of analysis has involved applications of overlapping generations models (Soskice, Bates and Epstein 1992; Bates and Shepsle 1997). Political actors may be short-lived, occupying key decisionmaking positions for relatively brief periods of time. Yet the organizations within which they act persist over extended periods of time, and are thus made up of overlapping generations of short-lived actors. Often these actors are embedded in career trajectories where advancement depends on adherence to organizational expectations. In such settings, organizational designs may transform the ambition for advancement within the institution into constraint, thereby generating enduring regularities in the choices of individuals (Soskice, Bates and Epstein 1992, 548). Similarly, actors at the end of their careers may care only about the short term (e.g., party leaders placing priority on their immediate political prospects at the expense of the long-term reputation of the party), but they may depend on the support of younger actors further down in the organization. Because these younger members are interested in the character of the party whose leadership they will inherit in the future, they may make their support of current leaders conditional on the attentiveness of the latter to long-term considerations. These last two strands of argument about the shadow of the future are complementary. Arguments about overlapping generations may explain why particular (collective) actors might care about the long term. Arguments about credible commitment may account for how these actors can successfully pursue these long-term considerations in contexts where one might expect problems of time inconsistency to be overwhelming. To be very convincing, in other words, both arguments (or some equivalent) will need to be true. For the shadow of the future to matter, actors must both care about the future and feel capable of influencing it. That these arguments have real force seems unquestionable. Even a brief examination of political life suggests that organizations and political actors often do attend to the long term. Indeed, modern societies would be inconceivable if this were not the case. Governments, for instance, do not generally seek to maximize short-term revenue extraction by engaging in systematic confiscation of private assets. Politicians do not vote themselves spectacular pay increases. Interest groups generally act as if they care about their reputationsa long-term asset (Hansen 1991). Usually they will not squander their future credibility by lying to legislators they seek to influence. In general, however, there are good reasons to expect that efforts to lengthen time horizons are likely to prove less effective in politics than in economics. The marketplace possesses some strong mechanisms for lengthening time horizons. Crucially, the existence of property rights provides particular individuals with a clear entitlement to income streams from particular assets, and therefore creates an incentive to attend to the long-term value of those assets.

482

PAUL PIERSON

The mechanisms operating in politics are generally far weaker. Especially problematic is the fact that each of the time-lengthening devices discussed above relies heavily on the capacity of some actors to adequately assess the behavior of others in order to detect opportunistic actions, and then to bring the guilty parties to heel. Yet such monitoring behavior is often exceptionally difficult in politics. Indeed, as we shall see, many of the biggest problems with efforts to translate functionalist theories of institutional design originating in economics to politics stem from the complexity and uncertainty that characterizes the political realm. It is often very hard to establish accountability in political environments. Outcomes themselves are frequently difficult to measure. And, if we believe that a system is not performing well, it is still more difficult to determine which elements in these highly complex systems are responsible and what kinds of adjustments would lead to better results. There are often long lags and complex causal chains connecting political actions to political outcomes. The complexity of the goals of politics, and the loose and diffuse links between actions and outcomes, render politics inherently ambiguous. As North has argued, [P]olitical markets are far more prone [than economic markets] to inefficiency. The reason is straightforward. It is extraordinarily difficult to measure what is being exchanged in political markets and in consequence to enforce agreements (1990b, 362). Even if failures in politics are relatively apparent and the culpability of agents can be established, efforts of principals to sanction their agents may be difficult. Many participants in politics (voters, members of interest groups) engage in activities only sporadically. Their tools of action are often crude, such as the blunt instrument of the vote, and their actions may have consequences only when aggregated with those of other actors in circumstances where coordination is difficult or impossible. Thus both monitoring and sanctioning difficulties place serious limitations on techniques for lengthening actors time horizons. It is no accident, for instance, that much of the generally optimistic rational choice discussion of credible commitments in politics has focused on relatively transparent financial issues (e.g., budget deficits, monetary policy). In these settings, performance indicators are clear and lines of accountability are unambiguous. Hence, behavior is relatively easy to monitor. While these issues are obviously important, it must be stressed that for reasons already noted they are fundamentally atypical of the kinds of matters dealt with in politics. An additional limitation of these techniques for lengthening time horizons is particularly relevant for the specific issue of institutional design. Even if some of the mechanisms discussed by rational choice theorists are operative in everyday politics, they will often be especially fragile or absent altogether precisely at moments of institutional formation. At founding moments, when crucial new rules are put in place, one often cannot count on the operation of well-instutionalized contexts to frame

THE LIMITS OF DESIGN

483

and structure the actions of political decisionmakers (Elster, Offe and Preuss 1998). My point is not that actors always have short time horizons, but that they often do. Thus the issue of actor time horizons should be treated as a variable with real implications for questions of institutional origins and change, and therefore as a subject deserving serious study. Where we would expect relatively short time horizons to be operative, functional claims about institutional design become more suspect. Limitation 3. Institutional Effects May Be Unanticipated Even if institutional designers do act instrumentally, and do focus on long-term effects, unanticipated consequences are likely to be widespread. Anyone engaged in empirical research in the social sciences knows that the most instrumental and canny of actors still cannot hope to anticipate adequately all the consequences of their actions. Institutions may not be functional because designers make mistakes. Unanticipated consequences are likely to be of particular significance in modern polities. Over time industrial societies have become much more differentiated, involving increased interactions among increasing numbers of people. The historical process has been elegantly summarized by Norbert Elias:
The network of human activities tends to become increasingly complex, far-flung, and closely knit. More and more groups, and with them more and more individuals, tend to become dependent on each other for their security and the satisfaction of their needs which, for the greater part, surpass the comprehension of those involved. It is as if first thousands, then millions, then more and more millions walked through this world with their hands and feet chained together by invisible ties. No one is in charge. No one stands outside. Some want to go this way, others that. They fall upon each other and, vanquishing or defeated, still remain chained to each other (1956, 232).

The profound implications of increasing social complexity need to be underlined. As the number of decisions made and the number of actors involved grow, relations of interdependenceamong actors, organizations, and institutionsexpand geometrically. This growing complexity has two distinct consequences. First, it generates problems of overload. More prevalent and complex political activity places growing demands on decisionmakers. In this context, time constraints, scarcities of information, and the need to delegate decisions may promote unanticipated effects. The second consequence of social complexity is growing interaction effects: the tendency of initiatives to have important consequences for realms outside those originally intended. As Garrett Hardin puts it, We can never do merely one thing (1963, 7980). Instead, we should expect that social processes involving large numbers of actors in densely institutionalized societies will almost always generate elaborate feedback loops and significant interaction effects which decisionmakers cannot hope to fully anticipate.

484

PAUL PIERSON

Nor is it just that social contexts are extremely complex; the difficulties are exacerbated by the fact that the abilities of individuals to draw inferences and judgments from their experiences have systematic biases. Barbara Levitt and James March provide an excellent summary:
[I]ndividual human beings are not perfect statisticians. . . . They make systematic errors in recording the events of history and in making inferences from them. They overestimate the probability of events that actually occur and of events that are available to attention because of their recency or saliency. They are insensitive to sample size. They tend to overattribute events to the intentional actions of individuals. They use simple linear and functional rules, associate causality with spatial and temporal contiguity, and assume that big effects must have big causes. These attributes of individuals as historians all lead to systematic biases in interpretation (1988, 323).

Thus, as a number of careful analysts have concluded, social activity even when undertaken by highly knowledgeable and instrumental actors possessing long time horizonsshould routinely give rise to significant unintended effects (Hayek 1973; Hirsch 1977; Schelling 1978; Van Parijs 1982; Perrow 1984; Jervis 1997). Nor is there any reason to exempt the task of institutional design from this general tendency. Two brief illustrative examples may underscore the point, though most students of politics would have no trouble generating many additional instances. The first concerns the changing institutional position of state governments in the United States (Riker 1955). Because approval of the American Constitution required state ratification, the interests of states received considerable attention in the process of institutional design. The framers intended the Senate to serve as a strong support of state interests. State legislatures were to appoint senators, who were expected to serve as delegates representing states in the formation of policy. Over time, however, senators seeking greater autonomy were able to gradually free themselves from state oversight. By the early 1900s, the enactment of the Seventeenth Amendment requiring popular election of senators only ratified the result of a lengthy erosion of state legislative control. The development of Canadian federalism provides another example (Watts 1987). The designers of the Canadian federation sought a highly centralized form of federalismin part as a reaction to the ways in which decentralization contributed to the horrors of the Civil War. Yet the Canadian federation is now far less centralized than the American one. Among the reasons: the Canadian federation left the provinces with sole responsibility for many activities that were then considered trivial. With the growing role of government in social policy and economic management, however, these responsibilities turned out to be of tremendous importance. The prevalence of unanticipated consequences raises a difficult challenge for social scientistsafter all, if savvy social actors cannot anticipate mistakes, what do social scientists have to offer? This is a very big topic. The above discussion, however, points to two productive lines of

THE LIMITS OF DESIGN

485

investigationboth prominent in recent sociological researchwhich seek to pin down the conditions where unintended consequences are likely to be most significant. The first focuses on problems of cognition the ways in which we are systematically error-prone in our judgments of the social world, in particular those related to questions of cause and effect. For instance, since individuals tend to overweight the significance of recent, highly visible events in their understandings of the social world, we might hypothesize that institutional designers will systematically err in focusing on dramatic failures in the immediate past. A second line of analysis would focus on distinctive social contexts. Social settings vary in the extent to which they are tightly coupled, involving dense and intensive connections among multiple domains (Perrow 1984; Jervis 1997). Such connections generate complexity and multiply the number of consequences flowing from any single intervention. The prevalence of unintended consequences in institutional design should stem in part from this aspect of social contexts. In political science, however, the dominant response to the issue has been avoidance. Better altogether if this problem could be made to go away. There are four plausible ways of doing so; I discuss them in ascending order of helpfulness. Often, the reaction of the analyst is to treat unanticipated consequences, perhaps only implicitly, as an error term. In practice this means to ignore the matter. But without at least some further discussion, theorists cannot know how serious the issue is that they are skirting. It would be nice to at least have some idea how big the error term is. This is especially pressing since there seems good reason to believe that it is rather large indeed. A second option is to treat unintended effects as noise and to assume that such effects will be randomly distributed and will therefore tend to wash out, leaving the systematic, intended effects behind. For some problems and settings, this is a reasonable way to proceed. It makes a good deal of sense, for instance, in studying certain aggregative social processes involving very large and atomized populations, such as the stock market or public opinion (Page and Shapiro 1992). However, it is a far less helpful approach for the investigation of institutional design and reform. Here, single unintended effects may be quite large. Furthermore, as I will discuss below, if institutionalization is path-dependent, then we cannot expect accidents to cancel out; instead, early accidents may be self-reinforcing (Arthur 1994). A third and more helpful option is to acknowledge the high potential for unanticipated consequences and to argue that this itself constitutes a crucial organizing principle of institutional design. As Robert Goodin puts it, Accidents happen: but the frequency and direction of accidents can be significantly shaped by intentional interventions of social planners. . . . Insofar as the social world is accident-prone, we might want to design around the risk of accidents, seeking robust institutions that can withstand the various shocks that will inevitably befall them (1996, 29). This

486

PAUL PIERSON

is not simply a normative argument about how institutions should be designed; the claim is that an awareness of the potential for unintended consequences itself shapes the activities of institutional designers. More broadly, the new institutional economics (and some of its political offshoots) argue that uncertaintyboth about the behavior of others and about unforeseen contingenciesplays a major role in shaping the optimal design of institutions. An efficient institution should be able to absorb, or adapt to, the predictable (if unspecifiable) bumps in the road. Yet while this line of argument helpfully incorporates unintended consequences into the discussion of institutional design, it does not directly address the issue of how common such unintended outcomes will be. We are still left wanting to know the extent to which accidents are likely to lead institutions to function in ways at odds with the expectations of designers. Nor does this general resort to correction-prone or adaptive design specify what such a design might be, or how corrections can be made in reaction to unanticipated consequences. These concerns are addressed in a final response to the problem of unintended consequences, which pushes the analysis away from a focus on the capacities of institutional framers. This line of argument, not inconsistent with those already discussed, suggests various selection mechanisms that will squeeze unanticipated consequences out of institutional settings over time. Claims about mechanisms of institutional enhancement are sufficiently interesting and important to warrant separate and extended discussion. I take them up in Section III. That ones attention shifts to such mechanisms, however, reflects the evident limitations of institutional design. There are strong grounds for challenging any presumption that institutional effects will reflect the expectations and desires of institutional designers. Designers may not be thinking primarily in instrumental terms; they may be thinking instrumentally, but be preoccupied by short-term considerations; or they may simply make mistakes. Convincing treatments of institutional design must take all these possibilities into account.
III. EMERGENT FUNCTIONALISM? EVALUATING MECHANISMS OF INSTITUTIONAL ENHANCEMENT

If the preceding arguments have merit, then institutional effects cannot be assumed to derive in a straightforward way from the intentions of far-sighted, goal-oriented actors. Yet this is not enough to demonstrate the limits of functionalist reasoning about institutions. A second line of functionalist argument could accept the preceding criticisms of rational institutional designers, while nonetheless asserting that functional institutional designs will result over time. That economists have rarely worried about the possibilities of inefficient institutional outcomes stems less from a nave faith in the capacities of human designers than from a confidence in the potential for

THE LIMITS OF DESIGN

487

institutional enhancement. Winter (1986, 244) terms this the as if principle: given the presence of certain adaptive mechanisms, one can act as if individuals were rational, highly knowledgeable, and instrumental actors, because the results over time will be the same. Markets, economists argue, provide two powerful mechanisms for generating efficiency: competition and learning. Competitive pressures in a market society means that new organizations with more efficient structures will develop, eventually replacing suboptimal organizations (Alchian 1950). Learning processes within firms can also lead to correction. According to Williamson, one can rely on
the far-sighted propensity or rational spirit that economics ascribes to economic actors. . . . Once the unanticipated consequences are understood, these effects will thereafter be anticipated and the ramifications can be folded back into the organizational design. Unwanted costs will then be mitigated and unanticipated benefits will be enhanced. Better economic performance will ordinarily result (1993, 116117).

I will leave aside the question of whether these economists are justified in these confident assertions about economic organizations. For political scientists, the issue is the relevance of these mechanisms to political settings. Each of these mechanisms can be seen as part of an evolutionary argument for the gradual refinement of institutional design. As Axelrod has noted, [T]he evolutionary approach is based on a simple principle: whatever is successful is likely to appear more often in the future (1984, 169). Competition and learning constitute two alternative mechanisms, akin to Darwinian natural selection, which might account for the elegance of institutional arrangements even in contexts marked by short time horizons, unintended consequences, and so on. Yet there are good reasons to believe that each of these mechanisms of enhancement, while clearly important, is likely to be less effective when one shifts from firms in private markets to the world of political institutions (Moe 1984; 1990). At a minimum, clear arguments need to be made concerning the circumstances under which each mechanism is likely to operate. Again, this is fruitful terrain for theoretical and empirical work. The possible limitations of Williamsons rational spirit in politics are perhaps clearest for mechanisms of competition. Here one has to ask: Do political institutions, like firms, confront a dense environment of competing institutions? Will competitors be able to capitalize on inefficient performance, swooping in to carry off an institutions customers and drive it into bankruptcy? The answer is sometimes, but usually not. Models of political competition are probably most helpful for understanding international relations, although the extent to which states have needed to rationalize or die has varied considerably over time (Kahler 1999). Even in the harsh environment of early modern European state formation, competition was not the only mechanism at work in selecting survivors (Spruyt 1994). By contrast, the contemporary international

488

PAUL PIERSON

environment offers far more extensive niches for weak and/or retrograde state structures to survive. In democratic polities, party systems also involve a clear competitive dynamic. One could argue with some justification that parties must adapt their organizations and platforms in the face of such competition or confront the prospect of decline. In general, however, there can be little doubt that political environments are typically more permissive than economic ones (Krasner 1989; Powell and DiMaggio 1991). Another way to put this is that barriers to entry are often extremely high. As has already been discussed, poor performance in politics is not always easy to identify or attribute to a particular source. Even where failure can be identified and attributed, major difficulties arise if correction requires collective action in pursuit of outcomes that have the characteristics of public goods (as is usually the case in politics). In such settings, actors have to overcome very difficult coordination problems in order to generate an institutional competitor of the poorly performing institution. In many cases, however, political institutions are not really subject to direct competition at all.1 Instead, single institutional arrangements, or sets of rules, often have a monopoly over a particular part of the political terrain. Consider the two examples of political competition introduced above. Competition occurs among nation-states, which are territorial units, each of which contains a wide array of political institutions. Electoral competition occurs among parties, all of which operate under a single set of electoral rules. Thus, while political parties and nation-states face competition, it is not clear that the notion of competition between electoral institutions is a meaningful one. Thus, where competition operates in politics, it often operates above (interstate conflict) or below" (among contending organizations) the level of institutions. At a minimum, analysts would need to carefully specify the conditions for competitive pressures to generate institutional efficiencies. In practice, such conditions are generally going to be absent or weak in nonmarket settings. More detailed discussion is probably warranted for the issue of learning. Again, however, there are reasons to doubt that learning dynamics will generally provide a reliable mechanism for institutional enhancement. The central reason is the intense complexity and ambiguity of the political world. Although this issue has already been discussed, a brief contrast with economics may be helpful in clarifying the difficulties confronting learning-based mechanisms of institutional enhancement. Economics is built in large part around the useful and plausible assumption that actors seek to optimize and are relatively good at it. Firms operate to maximize profits. The metric for good performance is relatively simple and transparent. Prices send strong signals which facilitate comparisons. One can analyze how various features of the economic environment contribute to or detract from firm performance. Observable, unambiguous, and often quantifiable indicators exist for many of these

THE LIMITS OF DESIGN

489

features. Workers can easily obtain fairly good information on the wages and working conditions on offer from different firms. Consumers, too, are reasonably adept at navigating most aspects of the economic world. Links between choices and outcomes are generally clear: I take a new job and my income rises; I buy a car and my savings account balance shrinks. The quality of goods is generally evident in relatively short order, and repeated purchases allow consumers to sample alternatives. Of course, one could add many complications to this simple picture of the economic realm. The market is obviously highly complex and often confusing. Yet the central clarifying role of prices, the prevalence of repeated interactions, the absence of a need to coordinate many of ones economic decisions with those of other actors, and the presence of relatively short causal chains between choices and results make it comparatively easy for economic actors to correct mistakes over time. In other words, these features improve the prospects for learning. By contrast, politics lacks anything like the measuring rod of price, despite some reductionist efforts to make the search for votes the equivalent of the search for dollars. Political actors frequently pursue a range of goals. While politicians often will be focused on reelection, others (e.g., bureaucrats, interest groups) have different ambitions. Thus, it is difficult to say what an effective political system would look likewhat it would optimizeeven in theory. Political activity is often intermittent rather than regularly repeated. Causal chains between actions and outcomes are often very long. Politics is simply a far, far murkier environment. This murkiness exacerbates the limitations of human cognition considered in the earlier discussion of unintended consequences. In part because of this opacity, even mistaken understandings of the political world are often self-reinforcing rather than corrective. To introduce a concept explored in depth below, our basic conceptions of the political world, of what works and what does not, tend to be path dependent. As Douglass North (drawing on research in both cognitive psychology and organizational theory) has argued, actors operating in contexts of high complexity and opacity are heavily biased in the way they filter information into existing mental maps (North 1990b; Denzau and North 1994). Confirming information tends to be incorporated, while disconfirming information is filtered out. Social interpretations of complex environments like politics are subject to positive feedback. Norths work here converges with the long-standing views of those studying political culture as well as with the recent contributions of cognitive science and organization theory. Once established, basic outlooks on politics, ranging from ideologies to either understandings of particular aspects of governments or orientations toward political groups or parties, are generally tenacious. Complexity of context and limits of human cognition mean that mistaken understandings in politics often do not get corrected. It may be appropriate in some circumstances to argue that politics involves learning

490

PAUL PIERSON

processes, in which responses to public problems proceed in a trial-and-error fashion (Lindblom 1959; Heclo 1974; Hall 1993). There is little reason, however, to think that this acts as a selection mechanism with anything like the efficiency-enhancing properties of market competition in economics or Darwinian natural selection in biology. Because political reality is so complex and the tasks of evaluating public performance and determining which options would be superior are so formidable, such self-correction is often limited. My point is not that competition and learning are implausible mechanisms of institutional enhancement. On the contrary, it would be hard to deny that political institutions may be modified and enhanced through each of these processes. Rather, the point is that each of these mechanisms exhibits considerable limitations in the political world. Thus, any tendency toward evolved functionalism should be treated as a variable. Instead of assuming the efficacy of such mechanisms, political scientists should engage in sustained investigation of the circumstances under which they can be expected to operate reasonably well. Yet formidable as these limitations of enhancement mechanisms may be, the problems do not stop there. In addition to this internal critique of evolutionary arguments, one needs to consider an external critique. Assume for a moment that competitive pressures are significant, or learning is considerable. Does it follow that institutional refinement must result? The answer is no. Even where these mechanisms are clearly operative, they face additional hurdles: in Williamsons words, learning or competitive pressures must still be folded back into the organizational design. Here, two fundamental obstacles to institutional correction need to be considered: institutional stickiness and path dependence. These obstacles operate even in contexts where learning effects and/or competitive pressures may be significant. Institutional Stickiness in Politics Institutional arrangements in politics are typically hard to change. Again, the greater adaptability of the economic organizations studied by Williamson offers a useful point of comparison. An individual with a new idea for a product need only secure the finance to put it on the market. If enough consumers (choosing independently) find it sufficiently appealing, the product will be a success. Change can be engineered through competition against existing products. Similarly, those with property rights over a firm are generally in a relatively strong position to remake their organizations as they choose. Lines of authority are clear, and the relevant decisionmakers are likely to share the same broad goal of maximizing profits. By contrast, formal political institutions are usually change-resistant. As Robert Goodin puts it, stability and predictability are achieved through a system of nested rules, with rules at each successive level in

THE LIMITS OF DESIGN

491

the hierarchy being increasingly costly to change (1996, 23). Thus, in many national settings nested rules created by ordinary legislation must pass through multiple veto points, often requiring broad supermajorities. In other political settings, such as the European Union, hurdles to change are even greater. And in most polities, efforts to change rules higher in the hierarchy (e.g., constitutions) require even greater levels of consensus. Indeed, many constitutions effectively prohibit short of a revolutionary rejection of the existing regimecertain types of revision altogether (e.g., by providing veto power to those who would lose protections or privileges as a result of possible reforms). There are two broad reasons political institutions are usually designed to be change resistant, each of which has already been discussed. First, in many cases, designers seek to constrain themselves. The key insight of the credible commitments literature is that actors can often do better if they remove certain alternatives from their future menu of options. Like Ulysses preparing for the Sirens, political actors often bind themselves, restricting their own freedom in order to achieve some greater goal. Second, and probably more important, those who design institutions and policies may wish to bind their successors. Terry Moe terms this the problem of political uncertainty" (Moe 1990). Unlike economic actors, political actors lack property rights. Designers know that continuous control over institutions is unlikely. This lack of continuous control has implications both for how institutions are designed and for the prospect of changing institutions once they are created. In particular, those designing institutions must consider the possibility that their political rivals will one day be in power, and will be eager to overturn their designs, or to turn the institutions they create to other purposes. To protect themselves, these actors therefore create rules that make preexisting arrangements hard to reverse. As Moe puts it, designers
do not want their agencies to fall under the control of opponents. And given the way public authority is allocated and exercised in a democracy, they often can only shut out their opponents by shutting themselves out too. In many cases, then, they purposely create structures that even they cannot control (1990, 125).

This is, of course, perfectly sensible. Confronting the twin problems of time inconsistency and political uncertainty, designers may reasonably decide to make political institutions change-resistant. Nonetheless, this characteristic of political institutions has clear implications for functional explanations grounded in processes of institutional enhancement. When actors at a later time attempt to make institutional reforms, they will often face considerable obstacles. The Consequences of Path Dependence It is not just that institutional arrangements may make a reversal of course difficult. Individual and organizational adaptations to previous arrangements may also make reversal unattractive. When actors adapt

492

PAUL PIERSON

to the new rules of the game by making extensive commitments based on the expectation that these rules will continue, previous actions may lock in options that actors would not now choose to initiate. Put another way, social adaptation to institutions drastically increases the cost of exit from existing arrangements. Rather than reflecting the benefits of institutionalized exchange, institutional continuity may reflect the rising costs over time of adopting previously available alternatives (Pierson 2000). Recent work on path dependence has emphasized how initial institutional decisionseven suboptimal onescan become self-reinforcing over time (Krasner 1989; North 1990a). These initial choices encourage the emergence of elaborate social and economic networks, greatly increasing the cost of adopting once-possible alternatives and therefore inhibiting exit. Major institutional arrangements have major social consequences. Individuals make important commitments in response to these institutions. These commitments, in turn, may vastly increase the disruption caused by institutional reforms, effectively locking in previous decisions. Research on technological change has revealed some of the circumstances conducive to path dependence (David 1985; Arthur 1994). The crucial factor is the presence of increasing returns or positive feedback, which encourages actors to focus on a single alternative and to continue movement down a particular path once initial steps are taken. Large set-up or fixed costs are likely to create increasing returns to further investment in a given technology, providing individuals with a strong incentive to identify and stick with a single option. Substantial learning effects connected to the operation of complex systems provide an additional source of increasing returns. Coordination effects (or network externalities) occur when the individual receives increased benefits from a particular activity if others also adopt the same option. Finally, adaptive expectations occur when individuals feel a need to pick the right horse because options that fail to win broad acceptance will have drawbacks later on. Under these conditions, individual expectations about usage patterns may become self-fulfilling. As North has argued, all of these arguments can be extended from studies of technological change to other social processes, particularly to the development of institutions. In contexts of complex social interdependence, new institutions often entail high fixed or start-up costs, may involve considerable learning effects, and generate coordination effects and adaptive expectations. Established institutions create powerful inducements which reinforce their own stability and further development. In short, North concludes, the interdependent web of an institutional matrix produces massive increasing returns, making path dependence a common feature of institutional evolution (1990a, 95).

THE LIMITS OF DESIGN

493

Lock-in arguments have received little attention within political science, in part because these processes have a tendency to depoliticize issues. By accelerating the momentum behind one path, they render previously viable alternatives implausible. The result is often not the kind of open conflict over alternatives that political scientists would quickly identify, but the absence of conflict. Lock-in leads to what Peter Bachrach and Morton Baratz (1962) called non-decisions. This aspect of politics can probably be identified only through careful, theoretically grounded historical investigation of how social adaptations to institutional and policy constraints alter the context for future decisionmaking (Hacker 1998). Over time, as social actors make commitments based on existing institutions, the cost of exit rises. Williamsons confident assertion that learning allows firms to adjust to unanticipated consequences applies far less well to an analysis of politics, and especially to the study of formal political institutions. Learning from past events may lead actors to act differently in launching new initiatives. Recapturing ground in previously institutionalized fields of activity, however, will often be quite difficult. Actors do not inherit a blank slate that they can remake at will when their preferences shift or unintended consequences become visible. Instead, actors find that the dead weight of previous institutional choices seriously limits their room to maneuver. Previous institutional choices may be sticky, and they may be path dependent. Thus, even if learning and competitive mechanisms are present, it is far from self-evident that these pressures will translate into institutional enhancement.
IV. SOME NEXT STEPS

This essay has advanced two core arguments: (1) Functionalist premises about institutional origins and change should be replaced by functionalist hypotheses; and (2) functionalist hypotheses should be supplemented and contrasted with hypotheses stressing the possible nonfunctionalist roots of institutions. We should expect the prevalence of functional outcomes in the construction and reconstruction of political institutions to be highly variable. More subtle theories of institutional origins and change must be built around careful argument about the preconditions for functional outcomes to occur. This requires specification of where such claims might break down and the circumstances that make the presence of such unfavorable conditions more or less likely. Rather than assuming that institutions originate and change to meet functional requirements, we need to make this a target for investigation. This is wide open terrain for systematic research. It may be true, as Richard Nixon said of Keynesianism, that we are all institutionalists now. Yet political scientists have produced very little sustained empirical work, organized around clear competing hypotheses, comparing institutional origins and change across different settings. Without this research, we are in no position to evaluate the impact of particular contextual features on

494

PAUL PIERSON

institutional outcomes, or even to establish how prevalent such features are in the political world. The contrast with sociology, where there is a thriving tradition of theoretically informed empirical research on related topics, is striking (Clemens 1999). It is worth stressing two common threads in many of the arguments advanced in this essay. The first is the major contribution of sociological research to an understanding of functionalisms limitations. Whether the issue is the noninstrumental behavior of actors, the limits of learning or competitive processes of institutional enhancement, or the cognitive and social sources of unintended consequences, sociologists have had a lot of interesting things to say. Political scientists should continue to borrow insights from other disciplines, with appropriate modifications. In doing so, however, greater attention could well be paid to recent theoretical and empirical work in sociology. The second common theme linking the various issues discussed is that all of them require that we look closely at the intertemporal aspects of politics, rather than take a snapshot view of political processes and outcomes. To see where functional accounts might come up short one needs to look not just at the moment of institutional origins, or at a current institution (deducing origins from current functioning). Instead, one must consider dynamic processes that can highlight the implications of short time horizons, the scope of unintended consequences, the emergence of path dependence, and the efficacy or limitations of learning and competitive mechanisms. This requires genuinely historical research. By genuinely historical research I mean work that carefully investigates processes unfolding over time. This work needs to be distinguished from efforts to mine history for illustrations of essentially static deductive arguments. The application of tribal labels is too widespread in American political science, but one could see the line of argumentation I am recommending as a variant of historical institutionalism (Hall and Taylor 1996). It rests on the basic premise that the explanation of many aspects of politics requires diachronic as well as synchronic analysis. Highlighting the benefits of more sociological and genuinely historical work implies that the new institutional economics is unlikely to provide sufficient basis for theories of institutional origins and change in politics. Indeed, I have stressed throughout that efforts to translate theoretical arguments from the economic realm to the political one are more perilous than is often recognized. Such translations need to be done with care, and with an appreciation for the limits of the analogy. This is not to dismiss the new institutional economics and its rational choice offshoots in political science. These lines of inquiry have generated important insights, particularly through a partial but significant effort to show how various obstacles to functional dynamics might be overcome.2 To emphasize both insights and limitations is not to advocate that we split the difference among alternative approaches. It is to recognize that distinct bodies of theory may provide greater leverage for analyzing

THE LIMITS OF DESIGN

495

particular contexts and dynamics. As Jepperson (1996) has put it, seemingly antagonistic theoretical imageries may sometimes (but only sometimes) be more complementary and less competitive than we realize. They may cover different aspects of processes, with different theories contributing modules that can potentially be linked to produce more complete accounts (Scharpf 1997). They may be discussing discrete phenomena. They may possess poorly articulated boundary conditions (x will hold under conditions a, b, and c, but y will hold under conditions d, e, and f). Such would seem to be the case here. Figure 1 summarizes my discussion, presenting conditions that make functionalist outcomes more or less likely. Two broad contexts are more favorable to functionalist accounts. The first is where conditions are such that we can expect initial designers to behave instrumentally, to focus on long-term institutional effects, and to be relatively accurate in their projections concerning those effects. The second context is where learning and/or competitive mechanisms are strong and obstacles to reform stemming from path dependence and institutional veto-points are modest.
FIGURE 1 Social Context and Functionalist Explanations

496

PAUL PIERSON

I would not suggest that circumstances where these various combinations of favorable conditions hold will be rare in politics; indeed, they are probably fairly common. On the other hand, it seems highly unlikely that they would be so common that theorists could safely operate from functionalist premises. Functional explanations of institutional origins and change are not wrong-headed, but they are radically incomplete. As a consequence, they suggest a world of political institutions that is far more prone to efficiency and continuous refinement, far less encumbered by the preoccupations and mistakes of the past, than the world we actually inhabit. Acknowledgments I am grateful for the support of the John Simon Guggenheim Foundation and the European University Institutes Robert Schuman Center. Alan Jacobs offered very helpful comments on an earlier draft. Andrew Rudalevige, Effi Tomaras, and Jeremy Weinstein provided valuable research assistance. Notes
1. 2. Thanks to Alan Jacobs for suggesting this line of argumentation. Again, the target of my criticisms in this essay is loose functionalist explanations of institutional origins and change. While I stress that many rational choice analyses fall into this pattern, Gary Miller quite rightly notes in his thoughtful commentary in this issue that they need not do so. Like Miller, I do not believe that analysts face an either/or choice between rational choice and some alternative(s). Indeed, some of my own analysis (e.g., the discussion of path dependence) draws heavily on rational choice theory. Yet it remains the case that rational choice theorists have focused primarily on the effects of extant institutions. By contrast, many of the dimensions of social processes relevant to a discussion of institutional origins and change have received limited attention. Randall Calvert, for instance, acknowledges that the institutions-as-equilibria approach . . . says nothing directly about the emergence of institutions (1995, 80). As Jack Knight (1995, 111) has argued, theories of institutional dynamics, including those grounded in rational choice, turn in part on different claims about the actual empirical conditions under which social institutions emerge and change. How much uncertainty do actors face in different contexts? How long are actors time horizons? How difficult is it to detect and correct institutional failures? How competitive is a particular institutional environment? Analysts studying these crucial issues can usefully draw on bodies of theory outside of rational choice traditions.

References
Alchian, Armen A. 1950. Uncertainty, Evolution and Economic Theory. Journal of Political Economy 58:211221. Arthur, W. Brian. 1994. Increasing Returns and Path Dependence in the Economy. Ann Arbor: University of Michigan Press. Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books.

THE LIMITS OF DESIGN

497

Bachrach, Peter and Morton Baratz. 1962. Two Faces of Power. American Political Science Review 56:947952. Bates, Robert, and Kenneth Shepsle. 1997. Intertemporal Institutions. In The Frontiers of New Institutional Economics, John Drobak and John Nye, eds. New York: Academic Press. Calvert, Randall L. 1995. Rational Actors, Equilibrium, and Social Institutions. In Explaining Social Institutions, Jack Knight and Itai Sened, eds. Ann Arbor: University of Michigan Press. Pp. 5793. Clemens, Elisabeth S. 1999. Politics and Institutionalism: Explaining Durability and Change. Review of Sociology 25:441466. David, Paul. 1985. Clio and the Economics of QWERTY. American Economic Review 75:332337. Denzau, Arthur D. and Douglass C. North. 1994. Shared Mental Models: Ideologies and Institutions. Kyklos 47(1):331. Elias, Norbert. 1956. Involvement and Detachment. British Journal of Sociology 7(3):226252. Elster, Jon, Claus Offe and Ulrich K. Preuss. 1998. Institutional Design in Post-Communist Societies: Rebuilding the Ship at Sea. Cambridge, UK: Cambridge University Press. Garrett, Geoffrey. 1995. The Politics of Legal Integration in the European Union. International Organization 49:171181. Goodin, Robert E. 1996. Institutions and Their Design. In The Theory of Institutional Design, Robert E. Goodin, ed. Cambridge, UK: Cambridge University Press. Green, Donald and Ian Shapiro. 1994. Pathologies of Rational Choice Theory: A Critique of Applications in Political Science. New Haven, CT: Yale University Press. Greider, William. 1982. The Education of David Stockman and Other Americans. New York: Dutton. Hacker, Jacob. 1998. The Historical Logic of National Health Insurance: Structure and Sequence in the Development of British, Canadian, and U.S. Medical Policy. Studies in American Political Development 12(1):57130. Hall, Peter. 1993. Policy Paradigms, Social Learning, and the State: The Case of Economic Policymaking in Britain. Comparative Politics (April):275296. Hall, Peter and Rosemary C.R. Taylor. 1996. Political Science and the Three New Institutionalisms. Political Studies 44:936957. Hansen, John Mark. 1991. Gaining Access: Congress and the Farm Lobby, 19191981. Chicago: University of Chicago Press. Hardin, Garrett. 1963. The Cybernetics of Competition. Perspectives in Biology and Medicine 7:5884. Hayek, Friedrich. 1973. Law, Legislation, and Liberty: Rules and Order. London: Routledge. Heclo, Hugh. 1974. Modern Social Politics in Britain and Sweden. New Haven, CT: Yale University Press. Hirsch, Fred. 1977. The Social Limits to Growth. Cambridge, MA: Harvard University Press. Jepperson, Ronald L. 1994. Relations Between Different Theoretical Imageries (with Application to Institutionalism). Unpublished manuscript. Stanford University. Jervis, Robert. 1997. System Effects: Complexity in Political and Social Life. Princeton, NJ: Princeton University Press. Kahler, Miles. 1999. Evolution, Choice, and International Change. In Strategic Choice and International Relations, David A. Lake and Robert Powell, eds. Princeton, NJ: Princeton University Press. Keohane, Robert O. 1984. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, NJ: Princeton University Press. Knight, Jack. 1995. Models, Interpretations, and Theories: Constructing Explanations of Institutional Emergence and Change. In Explaining Social Institutions,

498

PAUL PIERSON

Jack Knight and Itai Sened, eds. Ann Arbor: University of Michigan Press. Pp. 95119. Krasner, Stephen. 1989. Sovereignty: An Institutional Perspective. In The Elusive State: International and Comparative Perspectives , James A. Caporaso, ed. Newbury Park, CA: Sage. Krehbiel, Keith. 1991. Information and Legislative Organization. Ann Arbor: University of Michigan Press. Kreps, David. 1990. A Course in Microeconomic Theory. New York: Harvester Wheatsheaf. Lange, Peter. 1993. The Maastricht Social Protocol: Why Did They Do It? Politics and Society 21:536. Levitt, Barbara, and James G. March. 1988. Organizational Learning. Annual Review of Sociology 14:319340. Lindblom, Charles E. 1959. The Science of Muddling Through. Public Administration Review 19(2):7988. March, James and Johan Olson. 1989. Rediscovering Institutions: The Organizational Basis of Politics. New York: Free Press. Meyer, John W. and Brian Rowan. 1977. Institutionalized Organizations: Formal Structure as Myth and Ceremony. American Journal of Sociology 83(2):340363. Moe, Terry. 1984. The New Economics of Organization. American Journal of Political Science 28:739777. _____. 1990. The Politics of Structural Choice: Toward a Theory of Public Bureaucracy. In Organization Theory: From Chester Barnard to the Present and Beyond, O. E. Williamson, ed. Oxford: Oxford University Press. North, Douglass C. 1990a. Institutions, Institutional Change and Economic Performance. Cambridge, UK: Cambridge University Press. _____. 1990b. A Transaction Cost Theory of Politics. Journal of Theoretical Politics 2(4):355367. _____. 1993. Institutions and Credible Commitment. Journal of Institutional and Theoretical Economics 149(1):1123. North, Douglass C. and Barry R. Weingast. 1989. Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in Seventeenth Century England. Journal of Economic History 49:803832. Page, Benjamin and Robert Shapiro. 1992. The Rational Public: Fifty Years of Trends in Americans Policy Preferences. Chicago: University of Chicago Press. Perrow, Charles. 1984. Normal Accidents. New York: Basic Books. Persson, Torsten and Guido Tabellini, eds. 1994. Monetary and Fiscal Policy. 2 vols. Cambridge, MA: MIT Press. Pierson, Paul. 2000. Increasing Returns, Path Dependence, and the Study of Politics. American Political Science Review 94(2)251267. Powell, Walter W. and Paul DiMaggio, eds. 1991. The New Institutionalism in Organizational Analysis. Chicago: University of Chicago Press. Riker, William H. 1955. The Senate and American Federalism. American Political Science Review 49:452469. Scharpf, Fritz W. 1997. Games Real Actors Play. Boulder, CO: Westview Press. Schelling, Thomas. 1978. Micromotives and Macrobehavior. New York: Norton. Shepsle, Kenneth A. 1986. Institutional Equilibrium and Equilibrium Institutions. In Political Science: The Science of Politics, Herbert F. Weisberg, ed. New York: Agathon Press. _____. 1989. Studying Institutions: Lessons from the Rational Choice Approach. Journal of Theoretical Politics 1:131147. _____. 1991. Discretion, Institutions and the Problem of Government Commitment. In Social Theory for a Changing Society, Pierre Bourdieu and James Coleman, eds. Boulder, CO: Westview Press.

THE LIMITS OF DESIGN

499

Soskice, David, Robert Bates, and David Epstein. 1992. Ambition and Constraint: The Stabilizing Role of Institutions. Journal of Law, Economics, and Organization 8(3):547560. Spruyt, Hendrik. 1994. The Sovereign State and Its Competitors. Princeton, NJ: Princeton University Press. Tsebelis, George. 1990. Nested Games: Rational Choice in Comparative Politics. Berkeley and Los Angeles: University of California Press. Van Parijs, Philip. 1982. Perverse Effects and Social Contradictions: Analytical Vindication of Dialectics? British Journal of Sociology 33:589603. Watts, Ronald L. 1987. The American Constitution in Comparative Perspective: A Comparison of Canada and the United States. Journal of American History 74:769791. Weingast, Barry R. and William J. Marshall. 1988. The Industrial Organization of Congress; or, Why Legislatures, Like Firms, Are Not Organized as Markets. Journal of Political Economy 96:132163. Williamson, Oliver E. 1975. Markets and Hierarchies: Analysis and Antitrust Implications. Chicago: Free Press. _____. 1993. Transaction Cost Economics and Organization Theory. Industrial and Corporate Change 2:107156. Winter, Sidney G. 1986. Comments on Arrow and on Lucas. In Rational Choice: The Contrast between Economics and Psychology, Robin M. Hogarth and Melvin W. Reder, eds. Chicago: University of Chicago Press.