On the probability of coups

"I hope the Jedis haven't studied any game theory."

Last week, Paul Collier upset some people by using the crisis in Côte d’Ivoire as an opportunity to bring up his election insurance-by-coup idea first-introduced in Wars, Guns, and Votes. The basic premise is that the international community should provide countries that hold (free and fair) democratic elections with coup insurance: if you are fairly elected but are subject to a coup later on, then there will be some sort of external (possibly military) intervention to set things right. Conversely, if you are an incumbent who just lost, but you refuse to give up your position, then the international community withdraws its protection-from-coup guarantee. The idea isn’t to actually increase the incidence of coups, but to create steep incentives for losing incumbents to consider the advantages of retiring and working on those mémoires over ending up against the wall.

Collier isn’t advocating an actual coup, he’s advocating creating enough of a credible threat so that Gbagbo leaves before a coup happens. This requires that Gbagbo be completely rational, and can he see to the end of the strategic game. Unfortunately, leaders often don’t match these criteria: we can think of many ways in which Saddam Hussein would have been much better off if he accepted George W. Bush’s ultimatum on the eve of the Iraq invasion, and got the hell out of Dodge.

The other issues are obvious: the elections probably weren’t entirely free or fair, and Côte d’Ivoire never decided they wanted to be part of this banana republic insurance scheme in the first place. So it is not surprising that most of those who heard about the scheme recoiled in horror. Ranil rightfully pointed out that coups tend to be pretty messy, and if Gbagbo did decided to dig in and a coup resulted, what’s the guarantee that the next step would be a return to democracy?

Bill Easterly was harsher, arguing that the crisis is a complex system, and that a coup (or even the threat of a coup) could have largely negative unintended consequences. Many of those involved in Chris Blattman’s thoughtful discussion of the scheme suggested that Collier was being incredibly irresponsible.

But unfortunately (or perhaps fortunately) credible coup threats don’t only occur when Oxford professors wish they would. All sorts of factors influence the probability of a coup – and very few people have asked: how do the alternative strategies impact this probability?

To risk boiling down an extremely complex problem to a simple inequality (I can hear Ranil grinding his teeth already), consider the case of the first successful `couper’. This is the officer or division of the military that, when the planets are aligned, will be the first to mount a successful coup. For the first successful moving couper, the decision to mount a coup d’état and (hopefully) turn over power to Ouattara depends on the following inequality:

Projected welfare if he does start a coup > Projected welfare if he doesn’t start a coup.

We can argue of what welfare means in this inequality (I’d argue that it includes things like loyalty, ideology, etc), and how reliable the `projections’ are, but let us stick with this basic framework. Now, let’s think about the commonly suggested alternatives for the crisis:

  1. External military intervention
  2. Making Gbabo and friends `pariahs’ (sanctions, financial crackdowns, declaring debt odius etc).
  3. Letting Ggbabo stay

The first two of these both decrease the first successful couper’s welfare in the event that he decides not hold a coup, so they both, to some unknown extent, increase the probability of a coup. It could be by only a small margin – politics, history, and the unknowns of life under an Ouattara government dictate that, but I’d be surprised if the increase was  only negligible.

Unless you are arguing for letting Ggbabo remain in power, you’re closer to Collier’s camp than you might have thought.

One thought on “On the probability of coups

Comments are closed.