Dondena Interviews

With a tenure that may last only as long as it takes to plan a coup against them or plot to take their lives, we might assume an African dictator keeps his enemies closer to him than his friends. Wrong, says Francesco Trebbi, of the University of British Columbia. In fact, even under threat of an abbreviated grip on order, these rulers distribute cabinet ministerships generally in proportion to the size of the country’s various ethnic groups. More than a quirky area of research, studying this topic reveals the conditions under which a half billion of the world’s people live. And when they are under the most brutal dictators in human history, it behooves us to know how it is their governments operate.


Tell us about African dictators.

Autocracies are a very common form of government in human history and even today, and surprisingly much less well-understood relative to democracies. For democracies we have lots of data, we have elections, we have polls, we do have some degree of transparency in what these politicians do. With autocracies typically you don’t. So the goal of my research program is to provide some empirical guidance on the organization of these African dictatorships.

As you can imagine, it was kind of hard for us to find a good way of getting at this. But then, you start studying the way government is organized in Africa, and you realize parliaments don’t matter, they’re rubber-stamping. The judiciary: non-existent. But there’s one thing that matters, and it’s the executive branch. Typically, you would think of the personal ruler—Mobutu Sese, these kinds of guys—big figures, and then you say, ok there is a literature on dictators, they’re easy to follow, it’s just one individual. Period. But what do you really learn about how these organizations are internally organized? So we started collecting information about ministers, the national cabinets. Some positions are particularly prominent, like defense, treasury, these kinds of positions. And others are a little bit more difficult to trace: tourism, communications, et cetera. We started collecting information about the ministers—their identities, their ethnicities.

In the first paper of a series of projects that I have [How Is Power Shared in Africa? – with Patrick Francois and Ilia Rainer], we studied how the dictators allocated to certain ethnic groups different ministerial positions. In Africa ministerial positions are essentially pots of gold. The state is this huge thing that has to be used for patronage distribution. They do split the cake, and they allocate it to different groups. So the first question we had was: How do these guys do it? Do they keep it all for their co-ethnics or do they actually allocate it in a different way?


This is an interesting topic. I immediately thought, Do you keep your enemies closer to you than your friends?

That’s a typical thing you would say, specifically with respect to people who can kill you. We were talking about a very peculiar coalition formation situation in which the guys that you get inside the palace can potentially kill you. The guys that you leave out of the coalition potentially can stage revolutions against you.

So the first questions were: How is power allocated and how is power shared in these dictatorships? And, it turns out, very proportionally. Large ethnicities get a large chunk of the seats in government and small groups, small allocations. The reason is not that these countries are Sweden or Finland, the reason is that these dictators are fragile and therefore they have to be generous with other groups. They have to buy them, they have to give them enough so that they don’t stage coups or revolutions against them. Proportionality is a sign, essentially, of instability within these governments. In that paper we show interesting regularities, like the effect of the Cold War. When the U.S. and Russia moved out of African politics, you have essentially a drop in the cost of revolutions and much more inclusive coalitions form within these governments. So that paper studied these coalitions. Today’s paper [The Dictator’s Inner Circle – with Patrick Francois and Ilia Rainer] instead studies individual ministers.

So what you’re going to see today is the analysis at the individual level, which tells us, What happens to the career of a minister? The careers of ministers in Africa are very different from the careers of ministers in England, for all sorts of reasons, but one in particular is that these ministers are prominent potential threats to the leader. This means the leader monitors them very closely, and when they become too powerful kicks them out. This in turn implies that the risk of being terminated from a cabinet position is increasing over time for these ministers. The likelihood that you make it to the next year in government as a minister decreases and decreases and decreases, and conversely the hazard that you are kicked out increases with tenure. This is completely different from what you would expect from a normal job, where typically, if you’re not good at it you drop it, and if you are good you keep it and that will deliver you, automatically, decreasing hazard rates for the risk of being terminated in your job. So this is completely different.

In addition, for certain countries, you see that the ministerial hazards of termination, up to a certain point increase and then they decrease, so you have this nice hump. What’s happening there is that after a while you become a safe entity for the leader, your hazard eventually falls, and why? Because maybe at that point you are so entrenched in the system that you’re making enough for yourself that starting the government as a leader, which you would do if you were to stage a coup and kill the leader, is not so convenient, because it’s very risky to become a new leader relative to a minister of a safe old leader. In fact, a typical thing you see in the data is that, initially in a new dictatorial regime, there is a high likelihood of being killed or terminated as a leader, and then that fragility decreases. The leadership looks a lot like your standard job. Mobutu faced diminishing hazard functions. But for the ministers it’s a completely different ballgame.


Why is this interesting? Why study this?

There is a lot of literature in economic development that looks at Africa and says this continent is an economic and a political failure. Why? Because some of these politicians are myopic and rapacious, they’re basically stripping the country of its resources and they have zero time horizons, and why? Maybe the leader would like to have ministers with longer time horizons and growth-oriented policies, but the risk of being killed or being the object of a coup is too high that he has to terminate them. By terminating them he is shortening their time horizon and that produces the policies that we see every day on this continent. There is a sense that this is consequential from an economic perspective. Clearly this is the political leadership of the country and they set policy. It’s beyond the quirkiness of studying the most brutal political systems in human history, it’s also because there are a half-billion people in Africa and many more under other autocracies.


How does this fit in with your broader research and what you do in your career?

My career as a political economist has maintained two parallel paths. One is the political economy of poor countries, or more precisely, the political economy of development. Institutions are what makes countries grow. How are economic outcomes related to the design of checks and balances within the government? That’s the central question. I also work on the political economy of rich countries, like lobbying, special interest politics. These are apparently completely separate, but there are some fortunate—or unfortunate—analogies in the way in which these things work. But today’s talk is straight up in my institutionalist stream, and it’s basically the result of the investment in collecting these data.


Tell me what the data is like and how did you go about collecting it and working with it?

So the data collection has been a nightmare. Fortunately, a co-author of mine [Ilia Rainer] is a very competent Africanist, and he was the person who made it all possible. I tried my best, but really it was him. The way the data collection worked is that you have to go and collect information about each cabinet position, every year. And you have to find a couple places where you can check [Europa Yearbooks]. Once you’ve found the names, that’s only stage one. You have to discover their ethnicity and some individual characteristic. It’s easy to find information on the internet or in the library about Mobutu Sese. It’s much harder to find the Undersecretary of Defense in Nigeria or Tourism in Kenya in 1962. We collected this information with the help of 7 research assistants. I think that was also the fun part—we cross checked these with local political scientists and journalists in each of these countries. Our sample is 15 countries so far, we are expanding it. We have covered the African equatorial belt from Kenya all the way to Sierra Leone. For this area, we contacted in each country three or four former ministers, political scientists at local universities, journalists that work in local politics, and asked them to cross check. The dataset, at the end, is unique because it’s very clean, very complete, very few missing observations. And honestly, it almost bankrupted me, in terms of the sheer amount of money I had to spend on this. I’ve used funds from the University of Chicago, the University of British Columbia, Social Science and Humanities Research Council grants…


Was it to send people over there and pay them by the hour and turn the pages at every place they needed to go to find it?

No, no. We didn’t send research assistants to Africa. The political scientist, the local consultants, they were African, they were paid to work from there. The research assistants were at the University of Chicago, the University of British Columbia, typically undergraduate students. A lot of them, for a lot of hours. And that’s essentially the human resource cost of the operation. But I think at the end we have something like this that’s probably going to be useful for economists and political scientists in the future. Once our first paper is published, we are going to release the first wave of data, and when the paper of today’s presentation is published we’re going to release the individual ministerial level data.


Tell me about, in your field and more broadly speaking about the social sciences, what do you like about what you see going on research-wise, and what would you like to see a lot more of?

I’m an empirical guy. I work with data. All my papers have models and I estimate models, but first of all you need to start from the facts. There is a tendency in economics, and a little bit less in political science but for sure in economics, of over-theorizing and under-appreciating statistically what is going on. So when we are talking about political economics and the political economy of macro-development, I would like see much more data work done with much more care with respect to the theoretical side. It’s kind of the opposite at the micro-development level instead. What I’m trying to say is that it’s completely fine, in my view, to run experiments, but there is a sense in which this is only a substitute for more macro-ish empirical analysis that has to be done at the regional/aggregate level. That cannot necessarily be done in an experimental setting, it has to be done approaching the specific part of the analysis with an open mind to the theory. So trying to collect new data with an idea, a theory about how you’re going to use it, as opposed to just showing statistics.

In this case, before embarking upon this project, we had an idea of what we wanted to do with this with this type of information, because we were essentially exposed to a large literature that studies these allocations in parliamentary democracies, so you want to have a little bit of this kind of theoretical modeling and present a logical chain where you have the facts that motivate the model and then you take the model for further validation for counterfactual simulations, for policy assessments, and I think that’s really where the literature is going. We’re moving away from just the facts, cross-sectional correlations, or brute-force randomized control trials and kind of going through a more structural approach. You do want to have a model that can tell you what can happen in certain areas, you want to have at least facts in general (not solely conditional on the specifics of your treatment), you want to have a little bit of generality there because of course we’re interested in bigger questions as economists. I think all this comes together nicely in a series of papers I see coming out these days.


Learn more about Francesco Trebbi here. Learn more about the Dondena Seminar Series here.
















Last updated 03 March 2014 - 12:50:31