Linos Book Symposium: Response to Eric Posner and Ryan Goodman

[Katerina Linos is an Assistant Professor of Law at Berkeley Law]

I’m honored to receive comments on the Democratic Foundations of Policy Diffusion from two preeminent scholars in international law. Eric Posner has written thought-provoking work in countless fields, but I’ll highlight one article, “The Law of Other States” for its rich insights on what an ideal policy diffusion process might resemble.  Ryan Goodman’s work has changed the way we think about human rights, the law of war, and interdisciplinary scholarship in international law more generally. Goodman’s path-breaking article “How to Influence States: Socialization and International Human Rights Law” has lead many international lawyers to focus not only on only political science and economics, but also on sociology. It inspired me to write this book.

Their comments invite debate on several empirical issues, as well as on two major theoretical questions:

• 1)   Do diffusion studies imply that “international law is weaker than generally recognized”?
• 2)  How does my theory of diffusion through democracy connect to theories of state socialization, and more generally to research on constructivism and sociological institutionalism? Which exact mechanism do my experimental results support?

Eric Posner’s post begins with a provocative question: Does evidence of policy diffusion imply that “international law is weaker than generally recognized”? More specifically, does the fact that states mimic one another inflate our estimates of how influential international law is? I don’t think so. However, diffusion studies suggest that international law might exert its influence through somewhat different pathways than we often emphasize. Individuals comply with domestic laws for a variety of reasons – some are deterred by the possibility of sanctions, others see the law as a guide for appropriate behavior, while still others observe their neighbors’ behavior, and only follow the law to the extent that they see their neighbors comply. If international law operates only in part through the threat of sanctions, and in part by influencing citizen’s beliefs about appropriate government behavior, as I suggest in my book, this does not make international law less influential overall. Relatedly, if states decide to comply (or violate) international law depending on the compliance choices of their neighbors, this could increase or decrease overall compliance levels.

In fact empirical studies of compliance might underestimate the impact of international law. If ILO Conventions influence both countries that ratify these Conventions, and countries that do not (as I suggest), comparing these two types of countries will bias results downwards. Relatedly, if countries often change their legislation slightly before (rather than slightly after) ratifying key Conventions, as happened in some of the cases I studied, this too would lead us to underestimate of the impact of international law.

Eric Posner’s post also raises four major questions about the empirical study of policy diffusion.  As these are fundamental questions to the field, I offer brief answers here, and point interested readers to the parts of the book that address these questions more extensively.

First, does information from countries with very different values matter? Yes, and in some circumstances politicians find it especially helpful to reference the choices of very different countries, for example to emphasize a policy’s broad appeal. Consider Olympia Snowe’s argument in favor of the introduction of the FMLA: “Until recently the United States was alone among industrialized nations, with that well-known center of enlightened government, South Africa, in lacking a family leave policy. Now even South Africa has adopted a more progressive policy than we have, leaving us in shameful isolation” (p. 64; pp. 32-33 and p. 197 discuss this issue theoretically).

Second, how do we distinguish learning from herd behavior? Regression models allow us to study whether policies associated with good results spread quickly (e.g., health policies associated with low infant mortality), or whether countries follow other countries in their peer group regardless of the success of their policies (see pp. 14-17, 77-85). I also studied the argumentation politicians used, and compared arguments related to policy success (e.g., “countries with universal health care spend less on health care”) to emulation claims (e.g., “every other industrialized country in he world has universal coverage”) (see, e.g., pp. 61-65).

Third, how does diffusion research avoid “selecting on the dependent variable” – i.e. studying only policies that spread quickly? I selected on the independent variable, and compared an issue area (family policy) in which there were strong international organization templates, to another area (health policy) where international organization involvement was more limited. In addition, within family policy, I compared maternity leave, where there were clear ILO templates, to family benefits, where there was less ILO involvement (see pp. 33, 67-69, 128-29).

Fourth, how do we know whether foreign models only influence laws on the books, or also influence the law as implemented on the ground? To try to address this concern, I looked government expenditures on social policies, among other dependent variables (see e.g., pp. 144-45)

How does my theory of diffusion through democracy connect to theories of state socialization, and more generally to research on constructivism and sociological institutionalism? Which exact mechanism do my experimental results support?

Ryan Goodman’s comments invite me to situate my work vis-a-vis related theories on state socialization, constructivism, and sociological institutionalism. My work is deeply influenced and generally very supportive of arguments in these traditions. What always impressed me was the strong empirical evidence, at the country-level, that countries in fact mimic one another. I find this pattern in the areas I study.

My major departure from these literatures concerns the question of agency.  Who learns? Who is socialized? Is it the state itself? Is it technocrats? Is it liberal elites? Is it ordinary voters? Prior work in the constructivist vein is often unclear on this question. And the work that is clearest in this regard, top-notch research by leading scholars such as Anne-Marie Slaughter and Kurt Weyland, emphasizes elite socialization rather than voter socialization. Instead I propose a theory in which elites do not operate in isolation, but are constrained by ordinary voters’ views.

My experiments were designed to test two basic claims of my proposed theory: whether Americans respond positively or negatively to information from abroad, and whether these effects are concentrated among elites, and liberal elites in particular, or whether they occur among diverse groups of ordinary voters. The experiments speak clearly to both of these claims – they confirm both that the effects are positive, and that they are widespread.

The experiments do not get to the bottom of the issues Ryan Goodman’s post raises, namely the reasons why voters respond to foreign models. While I have a slightly different interpretation of the experimental findings than Ryan Goodman does, my key conclusion is that further research is needed to understand exactly why ordinary voters care about what happens abroad. I propose that information about international models allows voters to situate their own government’s proposal, and evaluate whether it is a radical, ill-thought out experiment, or a mainstream, tried-and-true solution. Ryan Goodman proposes that a more general tendency towards social conformity could also drive these results. I think that it is very possible that both of these mechanisms are at work – I see them as complements, rather than as alternatives. I find that less informed people responded more than well-informed people to foreign countries’ maternity leave proposals. I did not find this information differential in health care policy, perhaps because health care was extensively debated at the time the experiments were fielded, and thus all respondents had relatively high levels of information already. Ryan Goodman is correct that I find that even highly informed people respond to foreign models, and that this piece of evidence suggests that information differentials alone might not be driving the results.   My data does not allow me to compare Republicans with conformist tendencies to Republicans with non-conformist tendencies, but further experiments that measure individuals’ general tendencies to conform would be very helpful to move diffusion scholarship forward.