Why do international institutions vary so widely in terms of such key institutional features as membership, scope, and flexibility? We argue that international actors are goal-seeking agents who make specific institutional design choices to solve the particular cooperation problems they face in different issue-areas. In this article we introduce the theoretical framework of the Rational Design project. We identify five important features of institutions—membership, scope, centralization, control, and flexibility—and explain their variation in terms of four independent variables that characterize different cooperation problems: distribution, number of actors, enforcement, and uncertainty. We draw on rational choice theory to develop a series of empirically falsifiable conjectures that explain this institutional variation. The authors of the articles in this special issue of International Organization evaluate the conjectures in specific issue-areas and the overall Rational Design approach.

Barbara Koremenos, Charles Lipson, and Duncan Snidal conjecture that the conditions of membership in international institutions will grow more restrictive as a response to uncertainty about state preferences. Membership criteria will act as a signaling device—states more committed to cooperation will be willing to meet the criteria, whereas those less committed to cooperation will not. The recent enlargement of NATO to include the former Warsaw Pact members Poland, Hungary, and the Czech Republic illustrates this logic. The potential candidates for admission had to meet standards with respect to democratization, civilian control over the military, and the resolution of border and ethnic disputes with neighbors. These criteria served to identify the more cooperative potential members and to encourage cooperative behavior among those who aspired to membership. However, NATO enlargement came at a price. Although trust was built and cooperation fostered between the East European states that gained membership, trust was broken and cooperation harmed between NATO and Russia. This unfortunate outcome represents a dilemma that arises in the expansion of a security community: While expanding the security community enlarges the zone of peace and mutual trust, it may generate fear among those still on the outside, who view it as a potentially hostile alliance. I present a game-theoretic analysis of this dilemma and analyze the conditions under which it arises.

International institutions that include an escape clause generate more durable and stable cooperative international regimes and are easier to achieve ex ante. The escape clause is endogenous in a model of repeated trade-barrier setting in the presence of symmetric, two-sided, political uncertainty. They permit, along the equilibrium path, countries to temporarily deviate from their obligations in periods of excessive, unexpected political pressure at some prenegotiated cost. The architects of international agreements optimally choose a cost so that escape clauses are neither too cheap to use (encouraging frequent recourse, effectively reducing the benefits of cooperation) nor too expensive (making their use rare and increasing the chance of systemic breakdown). The international institution's crucial role is to provide information, verifying that the self-enforcing penalty has been paid (voluntarily), rather than to coerce payment. Escape clauses also make agreements easier to reach initially. Their flexibility reassures states that the division of the long-term gains from the agreement is not immutable.

Though substantively important, centralized negotiations have received less theoretical attention than problems of centralized monitoring and enforcement. I address this gap by examining variation in a particular form of centralized negotiations that I call “clustering.” Clustering occurs when a state negotiates with several other states at the same time. Clustering enables states to avoid having to make concessions on the same issue to one state after another, and therefore has important distributional advantages. Clustering also centralizes bargaining within a regime, especially when several states cluster simultaneously in a “macro-cluster.”

I propose several hypotheses about clustering. First, most-favored-nation (MFN) clauses are a necessary condition for clustering. They link the distributional conflicts among many pairs of countries and make centralized bargaining more likely. Second, increasing membership in the trade regime makes clustering more likely. This relationship between membership and centralization echoes Rational Design conjecture C3, centralization increases with number, though the causal mechanism differs significantly. Third, clustering provides distributional advantages to those who cluster. A state that clusters, such as France under the Méline tariff or Germany under Chancellors Leo von Caprivi and Bernard von Bülow, will make fewer concessions than one that does not.

States experiencing negative externalities caused by other states' behaviors have incentives to devise international institutions to change those behaviors. The institutions states create to counter incentives to defect vary in whether and how they expand institutional scope to accomplish that goal. When facing symmetric externalities, states tend to devise narrow institutions based on issue-specific reciprocity. When facing asymmetric externalities, or upstream/downstream problems, states tend to broaden institutional scope using linkage strategies. When victims of an externality are stronger than its perpetrators, the resulting institutions, if any are devised, are likely to incorporate the negative linkage of sanctions or coercion. When victims are weaker, exchange institutions relying on the positive linkage of rewards are more likely. We illustrate the influence of situation structure on institutional design with three cases: international whaling, ozone-layer depletion, and Rhine River pollution.

Drawing on the analytical framework developed by Barbara Koremenos, Charles Lipson, and Duncan Snidal in the Rational Design project, I seek to shed light on the striking institutional differences among the various methods of international commercial dispute resolution for private parties. These methods include recourse to public courts and more frequently to private international courts, such as the International Court of Arbitration of the International Chamber of Commerce or the London Court of International Arbitration, as well as recourse to so-called ad hoc arbitration and alternative dispute-resolution techniques, such as conciliation and mediation. The key institutional dimensions along which these methods of international dispute resolution vary are (1) procedural and adaptive flexibility, and (2) centralization of procedural safeguards and information collection. I explain why different methods of international commercial dispute resolution are selected. I argue that these methods respond to the varying institutional needs of different types of disputes and disputants. Such needs can be explained in terms of the severity of the enforcement problem, uncertainty about the preferences or behavior of contractual partners, and uncertainty about the state of the world.

Europe's postwar shift to multilateral trade and payments arrangements was complicated by three factors. Distributional problems and uncertainty about the state of the world made European governments reluctant to adopt multilateral arrangements without financial support from the United States. An enforcement problem made U.S. policymakers reluctant to finance a European multilateral trading system. The severity of these problems was reduced by institutional designs that combined flexibility, centralization, and particular decision rules. Centralization and flexibility reduced uncertainty and softened distributive conflict. Centralization and particular decision rules solved the enforcement problem that U.S. policymakers faced.

During the twentieth century states negotiated and ratified formal treaties on the treatment of prisoners of war (POWs). These treaties have created a system for the treatment of POWs with universal and detailed standards and decentralized enforcement. I explain the form of the POW system as a rational institutional response to four strategic problems the issue of POWs poses: monitoring under noise, individual as opposed to state violations, variation in preferred treatment of POWs, and raising a mass army. In response to these four problems, neutral parties help address the problem of monitoring the standards. The ratification process screens out some states that do not intend to live up to the standards. The two-level problem of state and individual violations is addressed by making states responsible for punishing the actions of their own soldiers. By protecting POWs, the treaties help states raise armies during wartime. The POW case supports many, but not all, of the Rational Design conjectures. In particular, it suggests other strategic logics to explain variation in the membership and centralization of international institutions.

In the aftermath of World War II, states created a complex set of bilateral and multilateral institutions to govern international aviation markets. National governments concluded bilateral agreements to regulate airport entry and capacity and delegated to the airlines, through the International Air Transport Association (IATA), the authority to set fares and the terms of service in international markets. The resulting mixture of public and private institutions produced a de facto cartel that lasted for more than thirty years. Consistent with the Rational Design framework put forth by Barbara Koremenos, Charles Lipson, and Duncan Snidal, I argue that the institutions states created reflect the bargaining and incentive problems generated by international aviation markets. This case provides support for four of the Rational Design conjectures and slightly contradicts three others.

The Rational Design project is impressive on its own terms. However, it does not address other approaches relevant to the design of international institutions. To facilitate comparison I survey two “contrast spaces” around it. The first shares the project's central question—What explains institutional design?—but addresses alternative explanations of two types: rival explanations and explanations complementary but deeper in the causal chain. The second contrast begins with a different question: What kind of knowledge is needed to design institutions in the real world? Asking this question reveals epistemological differences between positive social science and institutional design that can be traced to different orientations toward time. Making institutions is about the future and has an intrinsic normative element. Explaining institutions is about the past and does not necessarily have this normative dimension. To avoid “driving with the rearview mirror” we need two additional kinds of knowledge beyond that developed in this volume, knowledge about institutional effectiveness and knowledge about what values to pursue. As such, the problem of institutional design is a fruitful site for developing a broader and more practical conception of social science that integrates normative and positive concerns.

In this article we summarize the empirical results of the Rational Design project. In general the results strongly support the Rational Design conjectures, especially those on flexibility and centralization; some findings are inconclusive (in particular, those addressing scope) or point toward a need for theoretical reformulation (in particular, the membership dimension). We also address the broader implications of the volume's findings, concentrating on several topics directly related to institutional design and its systematic study. First, we consider the trade-offs in creating highly formalized models to guide the analysis. Second, our discussion of the variable control is a step toward incorporating “power” more fully and explicitly in our analysis. We also consider how domestic politics can be incorporated more systematically into international institutional analysis. Finally, we initiate a discussion about how and why institutions change, particularly how they respond to changing preferences and external shocks. We conclude with a discussion of the forward-looking character of rational design.