I’m back from a stimulating workshop on autocracy at Indiana University. Regina Smyth, Armando Razo, and Michael Alexeev have some exciting plans to institutionalize the study of nondemocracy at IU, and as part of the planning process they asked Konstantin (Kostya) Sonin and me to provide some perspective on the field. Armando joined us for a freewheeling discussion at Indiana’s famous Ostrom Workshop, where we proceeded to violate Elinor Ostrom’s one inviolable principle of seminar protocol: we went over time.
Where does the literature on nondemocracy stand? What have we learned, and what are the open questions? What’s next?1 Here’s what I said…
Much of the literature on nondemocracy falls into one of two broad categories: the analysis of formal institutions and the study of information control.2 There is some obvious overlap between the two: one purpose of autocratic institutions, for example, is to manage the flow of information among elites. But it’s nonetheless a useful distinction.
Let’s look first at the study of institutions. Here, I see a largely successful and reasonably complete research program. A short list of accomplishments would include:
- The identification and analysis of regime types characterized by particular clusters of institutions. The term “nondemocracy”—by definition, any regime that is not a democracy—is arguably too encompassing to be useful. Recognizing this, scholars have narrowed their focus to autocratic regimes with a common set of characteristics. Levitsky and Way’s study of competitive authoritarianism is the ideal type here.
- The elucidation of the logic of important institutions. Models of autocratic elections illustrate elections’ value to the ruler in a) signaling strength, and b) gathering information about the regime’s popularity. Similarly, models of parties and legislatures show how these institutions enable credible commitments and serve to co-opt the opposition.3
- A frequently useful dialogue between theory and empirics. Consider, for example, the theory that autocratic legislatures serve as a mechanism to co-opt the political opposition. This argument has its roots in the empirical observation that autocracies with legislatures live longer. Theoretical work suggests that legislatures play this role by co-opting potentially dangerous social groups or elites. That work, in turn, has prompted renewed empirical inquiry, as scholars examine the behavior of authoritarian legislators to validate, support, and refine the co-optation thesis.
That is a lot of progress, but I don’t think we’re done. Existing work helps us to understand why autocracies are frequently populated by superficially democratic institutions, but it doesn’t say enough about why we observe these institutions in some autocracies and those in others. There is a story in the literature to support nearly any institutional arrangement one might observe in a dictatorship; we need a better understanding of why autocrats choose one set of formal institutions over another.
Now to the study of information control: Models of autocratic elections fit here as well, as manipulating election outcomes is one way to confuse elites and citizens about the regime’s popularity. But there is also a booming theoretical and empirical literature on media in autocracies that largely abstracts from formal institutions.
I love this work. To my mind, some of the most innovative research in political science today examines the control of media—especially but not only social media—in nondemocracies. It takes chutzpah, creativity, and skill to set up your own social network for the sake of figuring out how censorship works. And yet, much of this research feels geographically bounded. Reading these papers, I don’t always have a good sense of where in the parameter space we are. Why state media here and co-opted private media there? Why Russian bots but a Great Firewall of China? Kostya and I took a crack at this with our model of government control of the media—the size of the advertising market and the “mobilizing character” of the government do the work in our theory—but we did not offer a full theory of the role of information control in autocratic survival.
One paper that ties this all together is Guriev and Treisman’s “How Modern Dictators Survive.” The basic idea is that “informational autocracies” survive by manipulating citizens’ beliefs about the competence of the ruler. In one equilibrium, the autocrat manipulates beliefs by investing in state media; in another, he censors private media. (Think of this as Maoist broadsheets vs. the Great Firewall of China.) Repression is off the equilibrium path—citizens believe the ruler to be competent, so they do not rebel.
This is a promising theoretical approach, with some complementary empirical work on the horizon. I suspect we’ll see more of the same as a second generation of research on information control attempts to integrate what we’ve learned from rich country studies (not a pejorative) in recent years.
Where else can the field go from here? I see at least three possible directions:
- Informal institutions and norms of autocratic rule. My sense is that Putin was able to consolidate control so quickly, and so completely, because he triggered protective responses learned during decades of Soviet rule. Work in this area would help not only to understand the mechanisms of autocratic rule, but also to illuminate threats to democratic survival—a focus of John Carey, Gretchen Helmke, Brendan Nyhan, and Sue Stokes’ Bright Line Watch initiative.
- The relationship between the polity and the economy. It’s not that this has gotten no attention—think of what we know about economic growth in Mexico, not to mention the numerous papers and books on the Chinese growth miracle—but theoretical models of autocracy too often ignore the economy entirely or consider only economy (polity) ==> polity (economy). We need general-equilibrium theories. Janos Kornai’s The Socialist System is the model here.
- The experience of non-contemporary autocracies. I have in mind especially pre-WWII cases, which have received comparatively less attention in the literature. I can say from personal experience that working on history is fun. But the real reason to study earlier autocracies is because the range of empirical phenomena is so much greater. This creates opportunities to test key theoretical propositions, and it gives us a sense of the choices that contemporary autocrats could have made, but for various reasons have not. Kudos to those who have already taken the historical turn: you’ve been an inspiration to me. I hope we’ll see more.
There is a lot to do!
- Thanks to the many friends who joined a Facebook discussion of these questions earlier this week.
- Kostya’s interesting work with Daron Acemoglu and Georgy Egorov is situated in an essentially institution-free environment—an important benchmark. The large literatures on protest and regime change sit just adjacent.
- There isn’t room in a blog post to give proper credit to the vast literature discussed here. For a recent (and already slightly dated) review of formal models of nondemocratic politics, see my ARPS piece with Sonin and Svolik.