On the IPE@UNC blog a few days ago, Kindred Winecoff compellingly argued that much of the theory-testing done in international relations (IR) and international political economy (IPE) in recent years rests on the false assumption that outcomes across cases are independent of each other. Paraphrasing here, he points out that “almost all” of the big theoretical traditions in IR and IPE—neorealism, liberal institutionalism, and Marxism among them—identify ways in which outcomes across cases are strongly interdependent, but the research designs we usually adopt to test those theories implicitly assume they are not. In other words, “in the typical case, our empirical design does not match our theoretical structure.”
I think he’s right, and I think the same can be said of theories of political development, which is really most of what comparative politics is about. Two cases of current interest illuminate how it’s really impossible to understand persistence and change in national political institutions without thinking about how those institutions are embedded in a larger global context.
Let’s start with Myanmar. Conventional theories meant to apply to the reforms occurring there focus our attention on domestic processes, like socioeconomic modernization or economic inequality, as the likely impetus behind these changes. At best, though, these processes are structural conditions that have shifted little in Myanmar in recent years, and at worst they’re close to irrelevant. Myanmar is currently experiencing a rush of “modernization,” but much of it’s happening as a consequence, not a cause, of the regime-initiated liberalization. Any effort to understand why this liberalization is occurring now has to consider the growing fears of Burmese elites about their dependency on China, the bite of U.S. sanctions, and the opportunity costs of remaining isolated in a global economy that sees the country as an untapped trove and under-served market. If you try to estimate the effects of income or education or inequality on these trends in a model that ignores these wider forces, you’re probably going to get a misleading result.
Or take Bahrain. It’s impossible to explain the start of the popular uprising in Manama in the spring of 2011 without talking about diffusion, and it’s impossible to understand the outcome (so far) without looking at the material and diplomatic support the monarchy receives from powerful patrons—support that is itself rooted in those patrons’ regional geopolitical (counterbalancing Iran) and global economic (oil) concerns.
If you want to get really silly, imagine trying to infer the effects of income or oil wealth or inequality on the propensity for democratization from a data set composed only of Panama and Iraq. Talk about omitted-variable bias…
I don’t mean to imply that of scholars of comparative politics are oblivious to these issues. Interpretive studies of political development often reference international forces, and over the past 20 years, we’ve increasingly tried to incorporate these ideas into our statistical models as well. Steven Levitsky and Lucan Way’s thoughts on linkage and leverage are an example of the former. Other studies have nibbled at the problem by looking for evidence of diffusion in patterns of democratization, or at the marginal effects from participation in international organizations and other treaty regimes. Studies on the relationship between oil wealth and the survival of authoritarian regimes also lean in this direction, although it’s telling that newer research suggests that these effects really aren’t about oil per se so much as the specific role that commodity has played in a particular (and likely fleeting) realization of the global political economy. Dependency theory also operated at this level, although the results were a bit cartoonish and the long-term predictions have now been proved flat wrong.
What’s still missing from comparative politics, I think, is the one-two punch of theories that are more explicitly systemic combined with methods that suit those theories. Right now, we’ve got little bits of each, but nothing that really brings the two together. We’re stuck in a complex adaptive system that doesn’t really distinguish between national and international, political and economic, human and natural, and our theories of stability and change in political institutions should take that whole more seriously.
Instead of thinking of the international environment as something we incorporate into our models by tacking one or two covariates onto the tail ends of our country-level equations, we should think more carefully about country-level institutions as middle-range manifestations of processes occurring in a global system. The simplifying assumption that states are separable units certainly has its uses, but we shouldn’t conflate that utility with causal relevance. Like maps, all models are simplifications, but those simplifications aren’t useful if they ignore the very causes they’re meant to locate. That’s true in a metaphorical sense, but as Winecoff calls out in the blog post that sparked this ramble, it’s also true in the more literal sense that badly misspecified models produce unreliable results.
I’ll wrap this ramble up by noting that the “development” metaphor itself helps illuminate the problem, and might even contribute to it by reinforcing a certain frame of mind. In many fields of study, “development” is a process that happens to individuals and follows a certain arc. It connotes directional growth and maturation, and it has a beginning, middle, and end. When we apply this metaphor to politics—comparing “fledgling” and “mature” democracies, for example, or talking about the “international community” as if it were something like a gathering of people in a room—we get stuck in a rut from which it’s hard to see the other, arguably richer, aspects of that world.