In the Continent of International Law: Explaining Agreement Design, Barbara Koremenos sets out to empirically demonstrate that there is considerable variation in the way that international treaties are designed, and that the variation reflects states making rational decisions based on the specific cooperation problems they are facing when drafting agreements. The book is the culmination of a research program that Koremenos—a political science professor at the University of Michigan—initially launched fifteen years earlier when she coauthored the seminal article “The Rational Design of International Institutions.”Footnote ¹ That article tried to move beyond the debate about whether international law is an exogenous constraint on state behavior by focusing on the variation in how international institutions are set up. Now a staple on seminar syllabuses in international relations classes, the article inspired a line of research that sought to answer questions like: why do some international agreements have dispute resolution provisions, while others do not specify what happens when a state fails to comply with their obligations?Footnote ² In her new book, Koremenos goes beyond her own prior work by both clearly developing a theoretical framework to explain why states design agreements the way they do and by empirically testing the hypotheses derived from her theory on an original dataset that codes the design of a random sample of international agreements. Koremenos thus makes both an important theoretical and empirical contribution to the study of international law.

Koremenos's project begins with the observation that, although a handful of international treaties get the lion's share of attention from politicians, the press, and academics, there are a huge number of international agreements currently in force. As Koremenos notes, the United Nations Treaty Series documents nearly 200,000 agreements in force as of 2015 (p. 1). These treaties lay out international commitments on topics ranging from regulating the flow of guest workers to restricting the development of nuclear weapons. In addition to covering a wide range of substantive topics, these is also considerable diversity in the way that these agreements are designed. For instance: some agreements continue for an indefinite period of time, others last for just a few years; some agreements create a formal process for dispute settlement, others do not lay one out; and some agreements cover a large number of topics, others focus on a single issue.

Koremenos's project sets out to explain the variation in the design, and not the substance, of international agreements. More specifically, Koremenos argues that there is variation in the way that international agreements are designed because different kinds of countries are trying to solve different kinds problems when they draft agreements. As Koremenos puts it, “the variation we see [in the design of international agreements] is a sign that states care, which is why they take the time and effort to negotiate specific treaty provisions that fit the demands of the situation” (p. 4, emphasis in original).

Koremenos builds on this basic premise by outlining a theory regarding the determinants of why states choose different designs for their agreements. This theory is based on three assumptions (pp. 28–29). First, Koremenos assumes that the central players in international politics are still states—and not other groups like NGOs, corporations, or regional organizations. Second, Koremenos assumes that states are rational actors that use international law to pursue their own self-interest. Third, Koremenos assumes that states use international law to “pursue Pareto-improving agreements with minimum cost” (p. 29, emphasis in original). For instance, the assumption is that states would not create an expensive international organization to monitor compliance with agreements when non-compliance is unlikely. By starting with these assumptions, Koremenos situates her book squarely within the rational choice tradition in international law and international relations.Footnote ³

Using this rational choice framework, Koremenos argues that states enter into international agreements to solve cooperation problems with other states. More specifically, Koremenos argues that states are trying to solve one of five cooperation problems created by their divergent interests or one of three cooperation problems created by constraints that emerge because of underlying uncertainty (pp. 31–40). The cooperation problems created by divergent interests are: (1) Enforcement Problems (present when states have incentives to not cooperate while others cooperate); (2) Distribution Problems (present when states have divergent preferences over substantive terms in an agreement); (3) Commitment Problems (present when states have time-inconsistent preferences); (4) Coordination Problems (present when states have to agree on exactly one outcome); and (5) Norm Exportation (present when states want to spread norms to other states). The cooperation problems created by constraints are: (6) Uncertainty About Behavior (present when states are uncertain about the actions taken by other states); (7) Uncertainty About Preferences (present when states are uncertain about what negotiating partners are actually trying to achieve); and (8) Uncertainty About the State of the World (present when states are uncertain about the consequences of their potential cooperation).

Koremenos then argues that the relative prevalence of these cooperation problems influences the design of international agreements along four dimensions: the flexibility of the agreements; the centralization of tasks proscribed by the agreement; the scope of issues covered by the agreement; and the “control” rules (e.g. decision-making processes) established by the agreement. In other words, the eight cooperation problems are independent variables that have predictive relationships with the design dimensions that serve as Koremenos's dependent variables.

Koremenos develops a series of hypotheses about how the specific cooperation problems that states are trying to solve will influence the design decisions they make along these four dimensions. As an example, Koremenos hypothesizes that when Enforcement Problems are relatively high, it is likely that states will design international agreements to increase their scope by incorporating explicit provisions that provide for punishment for non-compliance (p. 54). Throughout the book, Koremenos develops dozens of specific hypotheses along these lines about how states should rationally choose to design institutions.

In order to test these hypotheses, Koremenos built an original dataset that codes the design provisions of 234 international agreements. The agreements included were randomly drawn from the United Nations Treaty Series (UNTS) and cover one of four topic areas: economic, environment, human rights, and security. As Koremenos acknowledges, although the UNTS is the largest database of international agreements, it does not include the complete universe of agreements (p. 69). Although it is possible that the agreements included in the UNTS systematically differ from agreements that are not recorded in it, relying on the UNTS is a reasonable and practical way to build a random sample of international agreements.

Each agreement was then coded in two ways: first, for the cooperation problems that the parties were trying to address (the independent variables); and second, for the design decisions outlined in the agreement (the dependent variables). For instance, if an agreement in the sample was a bilateral investment treaty, the first group of coders would be asked to code whether the states that negotiated the treaty were facing each of the eight cooperation problems (e.g. Was a major concern when the agreement was signed uncertainty about the other state's behavior?) and the second group of coders would be asked to code each of the design dimensions (e.g. Does the treaty have a finite duration?). Although the coding of the design dimensions is relatively straightforward, I was not always entirely certain how the cooperation problems were coded. Despite an entire Appendix on the topic (pp. 371–76), it is not clear how things like the relatively high presence of Commitment Problems would be reliably captured by even well-trained coders. That said, Koremenos is upfront about this concern (p. 74), and should be given credit for taking the first step to quantify these illusive concepts.

The balance of the book tests whether the design of the sample of agreements is consistent with Koremenos's predictions. To continue with an earlier example, Chapter 8 is dedicated to examining when agreements include punishment provisions. Agreements are coded as having punishments provisions if they have clauses allowing for things like the curtailing of membership rights or financial sanctions for non-compliance (p. 228). Regressions are then presented showing that agreements coded as having enforcement problems are twenty-five percentage points more likely to contain a punishment provision (p. 234). It is worth noting that this is not the only empirical finding related to punishment provisions; as with all the design provisions, the book lays out and tests a number of specific predictions. In this case, the results also support the hypotheses that agreements will be more likely to include punishment provisions when they are characterized by an underlying enforcement problem, when there is uncertainty about behavior, and when the agreements are bilateral.

Throughout, the results largely support Koremenos's central claim that “even though international law is widely varying, it is not wildly varying—that is, it is neither random nor uniform but instead follows identifiable regularities” (p. 320). Based on these findings, Koremenos argues that the rational design of international agreements is evidence of the importance of the agreements. After all, if states do not believe that others will change their behavior because of international law, why would they spend years carefully negotiating agreements that are designed to address the cooperation problems the agreements are trying to solve? This leads Koremenos to conclude that, “[f]or those who still want to assert that international law does not matter, this book makes clear that there is no scientific basis for this belief” (p. 345).

With this conclusion, one of the central debates about international law has come full circle. To explain why, let me offer an oversimplified account of the relationship between international law scholarship and international relations scholarship during the post-war period. At the start of the period, international law scholars focused on examining the specific details of international treaties and took it as given that international agreements were exogenous constraints on states’ behavior. International relations scholars, on the other hand, argued that international treaties endogenously emerge from states coordinating and cooperating with each other and, as a result, treaties are unlikely to change states' behavior for non-instrumental reasons. At the turn of the century, law professors started to directly build on insights from the international relations scholars to challenge the premise that international law is an exogenous constraint. Legal scholarship has subsequently been heavily influenced by this view of international law that was largely imported from international relations. But Koremenos's book ends by directly calling out law professors (pp. 342–43)—most explicitly Jack Goldsmith and Eric Posner (id.)—for what she argues is an underestimation of the efficacy of international law. To make this claim, Koremenos does not directly test whether international agreements actually act as an exogenous constraint on state behavior. Instead, she presents evidence that agreements are largely designed in the ways that a rational state would have chosen if trying to maximize the chance of changing state behavior. In other words, an international relations scholar is questioning the claims of international law scholars for not taking the specific details of international treaties as evidence that the agreements serve as an exogenous constraint on state behavior.

It is not until this conclusion that I depart from Koremenos. Her rational choice assumptions are easily defendable, her theory is well-developed, her dataset is carefully constructed, and her empirical work is sound. What I question is whether any conclusions can be drawn about whether international law serves as an exogenous constraint on state behavior from Koremenos's evidence. To be clear, Koremenos may well be correct that rationally designed agreements shapes state behavior. But since she does not offer direct evidence on state behavior, her evidence is unable to prove this claim.

Let me use another prominent debate between law professors and political scientists to explain why. Every year, the United States Supreme Court hears cases of national importance. When lawyers argue before the Supreme Court, they carefully draft briefs that take time to discuss technical issues like the appropriate standard of review and they cite to the U.S. Constitution, statutes, and prior Supreme Court decisions. The justices then issue lengthy opinions that go on for dozens of pages, cite scores of cases, and try to engage directly with prior precedent. Yet despite the fact that the attorneys and judges act like the law is an exogenous constraint when arguing and deciding the case, there is still an active debate among scholars about whether political preferences are doing all the work.Footnote ⁴ The constitutional law professors and political scientists studying judicial behavior are all aware that the parties act like—to use Koremenos's phrase—the law “matters”; some just think the “law” does not act as a constraint on decision-making at the Supreme Court when it conflicts with the justices’ political preferences.

In the same way, I assume that the scholars Koremenos situates her project against are aware that negotiators draft treaties as if the design of agreements “matters.” For example, two of the scholars that Koremenos challenges released a book four years ago that offers a rational choice explanation regarding why states may design treaties with many of the design dimensions Koremenos considers, but that still contends that those agreements may not constrain state behavior when it is not otherwise in their interest to comply.Footnote ⁵ To the extent that there is a disagreement between Koremenos and her interlocutors, it is thus about whether international agreements act as exogenous constraints that drive states to comply for non-instrumental reasons—a point which cannot be refuted by Koremenos's evidence that agreements are written rationally based on the cooperation problems the states were facing during the drafting phase.

This begs the question, why would states expend resources to rationally design international agreements if they do not constrain their behavior? There are many possible explanations. One explanation is that the state is not a unitary actor. For instance, the lawyers in the foreign office negotiating a treaty might care about rationally designing the agreement because they hope it will succeed, but the political leaders calling the shots might not care about the agreement when deciding between compliance and non-compliance. Another possible explanation is that states face time inconsistency problems. The states negotiating an agreement might hope that it binds the signatories in the future, but when the future arrives, the political leaders may disregard the earlier commitment (either because they are different people or face different constraints). Another still is that some of the states negotiating a given treaty may hope that it will be effective, but other states may sign the agreement without planning to change their behavior. These are not new arguments. They simply illustrate how evidence that agreements are designed rationally is not necessarily evidence that the agreements “matter”—that is, serve as an exogenous constraint—in the way that Koremenos claims.

But this disagreement about the breadth of the conclusions drawn from the book should not detract from the success of the project. Koremenos's book is a model of how to combine theory and empirical work to advance our knowledge of international law, and I am sure scholars will be busy debating the numerous hypotheses she develops for years to come.