Rev Int Org (2008) 3:201–206 DOI 10.1007/s11558-008-9037-2 BOOK REVIEW
Xinyuan Dai, International Institutions and National Policies Cambridge: Cambridge University Press, 2007 Lisa L. Martin
Published online: 3 May 2008 # Springer Science + Business Media, LLC 2008
The theoretical and empirical study of how international institutions exert their effects is a thriving field of international relations (Morrow 2007; Kelley 2007; Von Stein 2005). Xinyuan Dai contributes to this field in International Institutions and National Policies by directing attention to the ways that the victims of state policies can use international institutions to change those policies. Dai’s analysis provides a new way to think about the mechanisms linking institutions and state policy, and should stimulate further study of the role of domestic actors in the enforcement of international agreements. Her conceptualization of the causal mechanism has much in common with the theory of “fire alarm” monitoring in the relationship between legislatures and agencies (McCubbins and Schwartz 1984). The core of Dai’s book consists of three related arguments. First, she considers the type of monitoring arrangements that international regimes will adopt. She draws attention to two explanatory variables: whether the victims of state policy have interests that are aligned with some regime members, and whether these victims have easy access to information about state policy. When victims cannot serve easily as monitors, monitoring will either take place by the international organization itself (in the case of interest alignment), or by nongovernmental organizations (NGOs) when states cannot agree among themselves to monitor. Victims who are able to monitor directly will complement either state or NGO monitoring activities. Dai provides a nuanced discussion of the different types of potential victims of state policies and what determines whether they can effectively serve as “low-cost monitors.” Types of actors that might be victims include other states as well as nonstate actors either in one’s own or other states. A victim can only serve as a low-cost monitor if two conditions are met: the effect of state policies is apparent, not latent; and victims are able to discern the source of their harm. When these conditions are not met, the fire-alarm model of monitoring will not work. In that case, monitoring
L. L. Martin (*) Weatherhead Center for International Affairs, Harvard University, 1737 Cambridge Street, Office N207, Cambridge, MA 02138, USA e-mail:
[email protected]
202
L. L. Martin
will only occur if states either set up an explicit monitoring system, or if NGOs take on the cause of monitoring. One unusual aspect of Dai’s argument in this part of the book is that she generally assumes that NGOs will step in when no other monitoring arrangement emerges. In fact, it is possible that no monitoring at all occurs, as Dai acknowledges in a brief discussion of “obstacles to monitoring.” Her predictions about monitoring arrangements thus seem to be missing a crucial category: “no monitoring.” A look at regimes including the General Agreement on Tariffs and Trade, environmental agreements, and others suggests that Dai’s model has explanatory power, although more systematic research would be necessary to provide a rigorous test. When interest alignment is low and victims find it difficult to discern the source of harm, as in environmental regimes, NGOs tend to step in. When interests are aligned but again discerning the source of harm is difficult, regimes set up explicit monitoring arrangements, as in the International Monetary Fund or Nuclear Nonproliferation Treaty. Interest alignment combined with victims who can monitor effectively, as in trade regimes, leads to states and victims themselves monitoring. Finally, when interests are not aligned but victims can monitor effectively, as when states are violating human rights, the victims work with NGOs. The evidence here is promising enough to merit more sustained empirical inquiry, especially if the category of “no effective monitoring” is included as part of the dependent variable. Otherwise studies of monitoring arrangements will be subject to selection bias, as only those regimes with effective monitoring will become part of the dataset. Second, Dai looks directly at “compliance mechanisms,” with a game-theoretic model of how governments will respond to interest groups with a stake in whether state policy meets the terms of international agreements. The model highlights the electoral power of pro- and anti-compliance interest groups, and their access to information regarding state policy. When governments care about re-election, groups with more electoral power and better informational endowments will be able to shape state policy. Dai’s emphasis on the role of interest groups in elections, and on their ability to determine the content of the government’s policy, makes this model of a domestic constituency mechanism a welcome addition to the literature. Dai’s treatment of the government’s decision is somewhat less satisfactory than her treatment of the interest groups. The government is assumed to care both about public welfare and its probability of re-election. Public welfare is defined as the average of the two interest groups’ ideal points. While Dai states that not all voters will be part of the two interest groups, those who are not are neglected in the model; they do not enter the government’s utility function either as part of the public welfare function or the probability of re-election. The government is also assumed to have no independent substantive policy preferences. Thus, even in the absence of an electoral constraint, interest groups’ attitude toward compliance with the agreement determines government policy as the government will simply choose a policy that is the average of the groups’ ideal points. The weight given to each of the groups shifts when the electoral constraint is introduced, so that groups with better information and more electoral weight pull policy in their preferred direction. Dai then applies this model to the reduction of sulfur emissions in Europe, showing that the degree of domestic environmental activism is positively correlated with reductions in sulfur emissions. While the empirical analysis is limited by endogeneity concerns and lack
Xinyuan Dai, international institutions and national policies
203
of attention to anti-compliance constituencies, it is suggestive. Dai’s work on the domestic constituency mechanism opens a new path for empirical analysis of the interaction of international regimes, domestic interest groups, and governments. Third, Dai goes to the heart of the matter by asking how international institutions empower the domestic groups that influence government policy. After all, unless this question is answered the causal effect of institutions is opaque—groups could influence policy even in the absence of any institution or agreement. Dai argues that even “weak” institutions, those lacking direct enforcement power, can provide information to domestic groups, encourage them to organize, and legitimize them so that their ability to influence government policy is enhanced. In addition to the sulfur emissions case, Dai considers the effect of the Helsinki Accords on human rights. In the latter case, the role of the Accords in encouraging the formation and supporting the activities of domestic groups is quite clear, and it is impossible to understand the effect of these Accords without considering the domestic interest group mechanism. This empirical analysis, however, also points out a potential weakness in Dai’s theoretical framework. Her model treats the existence of domestic interest groups as given. However, in the Helsinki case perhaps the most important effect of the international regime was to support the creation of domestic groups that made demands on governments where these groups did not previously exist. Extending the framework to allow for the endogenous emergence of domestic monitoring groups would be valuable. While Dai strives to provide a purely rationalist account of compliance mechanisms, in her analysis of how regimes interact with domestic groups she also brings in concepts such as legitimacy that are not easily captured in purely strategic models. Dai has drawn attention to a mechanism of institutional influence that deserves further systematic study. While domestic interest groups are surely not the only pathway through which institutions shape state behavior, there is little doubt that they play an essential role. Dai’s book is therefore important reading for those interested in the study of institutional effects. Dai’s analysis also reveals, however, a conceptual weakness that is common to nearly all of the recent literature on the effects of international institutions on state policies. Dai conceptualizes “state policy” as “compliance.” In this, she adopts the most common metric of institutional effects used in the contemporary literature (Simmons 2000; Mitchell and Hensell 2007; Chayes and Chayes 1995). However, as I will argue in the rest of this essay, compliance is a fundamentally flawed concept for the purposes of the social-scientific study of institutional effects. It is a legal concept, not a social-scientific one, and it has been imported into the politicalscience literature without adequate thought as to its utility (or disutility). In fact, conceptualizing the effects of institutions as being about compliance has constrained and shaped studies of institutional effects in unproductive ways. A similar point was made over a decade ago by Downs et al. (1996), with their argument that compliance did not indicate cooperation. Even if one does not accept their claim that most international agreements are “shallow,” their critique of the use of compliance as a concept by political scientists studying international agreements remains largely valid. To some extent, the way that compliance has entered the international relations literature is a cautionary tale about the risks associated with celebrating
204
L. L. Martin
interdisciplinarity without careful attention to the different demands and purposes of various disciplines. States have policies—they pollute, reduce pollution, trade, obstruct trade, protect human rights, violate human rights, and so on. They do not necessarily have a strategy about whether or not to comply; that is, state policies are not obviously or primarily “compliance policies.” Our understanding of the causal effects of institutions and agreements will be much stronger if we focus directly on substantive policies rather than compliance strategies. Operationalizing state policy as the degree of compliance implicitly assumes that the agreements are having a causal effect, when in fact states could be technically “in compliance” without actually paying any attention to the agreement, or having made a decision “to comply.” One weakness of thinking about policy as compliance is therefore that it assumes what we wish instead to prove, that agreements cause changes in state actions. Social scientists studying the effect of international institutions on national policies are, as a rule, interested in the causal effect of these institutions. The causal effect is defined as a counterfactual: what the difference is in observed policy with institutions in place as compared to what that policy would have been without the institution. Since of course this counterfactual cannot be directly observed, determining the causal effect of institutions involves looking at changes in policies across time and/or space, while controlling for other factors that would interfere with causal inference. What this means in practice is that when determining the effect of institutions on policies, we should be focusing on variation in substantive policies. When a state joins an institution, or when an institution develops new rules, do that state’s policies change? And can the change in policies reasonably be attributed to the institution? If not, it is difficult to argue that the institution had any causal effect. Consider, for example, a state with a good record of protecting human rights. Let’s say that this state joins a new United Nations human rights convention, whose provisions mirror policies that the state already has in place. This state would be fully in compliance with the convention, but the causal effect of the convention would be zero, as it caused no change in state policies.1 Conversely, consider a state with a poor human rights record that signs the convention. This state moves toward better respect for human rights, but still falls short of the convention’s requirements in significant ways. This state would be out of compliance, but here the causal effect of the convention would be positive, in contrast to the previous state with the perfect compliance record. These examples show how the concepts of compliance and of causal effect can be orthogonal to one another. Learning about compliance may tell us absolutely nothing about the causal effect of institutions, as they are independent concepts that do not necessarily covary. Social scientists adopting the legal concept of compliance and assuming that it allows us to identify the causal effect of institutions are making a logical error, unless a series of stringent underlying assumptions are met. So far, studies of compliance have not generally recognized the distinction between 1
Unless, of course, the state had been planning to move to a policy of abusing human rights and was constrained from doing so by the convention. This example demonstrates Downs et al. (1996) concept of shallow agreements.
Xinyuan Dai, international institutions and national policies
205
compliance and causal effect, and have not examined whether the necessary conditions to infer one from the other are met. This conflation of two very different concepts, and the dangers of inferring causal effect from observed compliance, is not merely an abstract concern. It has confounded many otherwise serious empirical studies of institutional effects. Take, for instance, Edith Brown Weiss’ and Harold Jacobson’s (1998) excellent edited collection of studies of international environmental agreements. This wide-ranging and exhaustive study provides a great deal of information about how environmental agreements are implemented. However, one of its major findings is that those states most likely to be “in compliance” with these agreements are those whose policies met the agreements’ provisions before the agreements were in place. In these cases, the causal effect of the agreements is prima facie zero. In other words, in its focus on “strengthening compliance” this major study missed an opportunity to learn about the causal effect of environmental agreements. While this is a particularly stark example of the dangers of relying on a legal concept for causal inference, the same underlying dilemmas and unexamined assumptions affect all social-scientific studies of institutional effects that use compliance as a dependent variable. Given the disutility of studying compliance to learn about institutional effects, why is this practice so common? Dai is far from alone in framing her study as about compliance policies and compliance constituencies. Her study of sulfur emissions is actually less subject to inferential fallacies than many other studies, as the regime called for percentage reductions in emissions from all members. A member therefore could not “comply” simply by continuing an existing level of emissions. There are some superficially attractive features of studying compliance. While legal scholars argue about the technicalities of measuring compliance, in practice many international regimes regularly collect and publicize information about the compliance of their members. Framing the discussion in terms of compliance also provides a handy metric for comparing different states’ actions to one another. Thus the data on compliance are available. However, I would contend that studies of compliance are of practically no use for the purposes of inferring the effect of institutions; and such inferences are exactly what empirical studies of institutions need to provide. Establishing that institutions have a causal effect, and quantifying it, will require different strategies of data collection and estimation. Taking the shortcut of using data on compliance collected by lawyers has in fact been a significant diversion from this effort. Acknowledgement essay.
My thanks to David A. Singer and Jeff Frieden for their comments on parts of this
References Chayes, A., & Chayes, A. H. (1995). The new sovereignty: compliance with international regulatory agreements. Cambridge: Harvard University Press. Downs, G., Rocke, D., & Barsoom, P. (1996). Is the good news about compliance good news about cooperation? International Organization, 50, 379–406. Kelley, J. (2007). Who keeps international commitments and why? The International Criminal Court and bilateral nonsurrender agreements. American Political Science Review, 101, 573–89.
206
L. L. Martin
McCubbins, M., & Schwartz, T. (1984). Congressional oversight overlooked: Police patrols versus fire alarms. American Journal of Political Science, 29, 165–79. Mitchell, S. M., & Hensell, P. R. (2007). International institutions and compliance with agreements. American Journal of Political Science, 51, 721–37. Morrow, J. (2007). When do states follow the laws of war? American Political Science Review, 101, 559–72. Simmons, B. A. (2000). International law and state behavior: commitment and compliance in international monetary affairs. American Political Science Review, 94, 819–35. Von Stein, J. (2005). Do treaties constrain or screen? Selection bias and treaty compliance. American Political Science Review, 99, 611–22. Weiss, E. B., & Jacobson, H. K. (1998). In Engaging countries: strengthening compliance with international environmental accords. Cambridge: MIT.