Failed coups in democracies do not seem to depress economic growth

Here is my first column for Bloomberg View (more on that transition soon), on the research of Erik Myersson:

In autocracies, successful coups often improve economic performance, perhaps by replacing an incompetent or malevolent leader. In democratic countries, however, a successful coup is associated with lower per capita growth rates by an average of 1 to 1.3 percentage points per year over the following decade. On average, these coups reverse beneficial economic reforms, especially for the financial sector.

When a coup does overthrow a democratically elected government, it tends to bring a military leader and significant changes in policy, and not usually for the better. There are long-run correlations of such successful coups against democracies with lower investment, lower schooling and higher infant mortality.

…for failed coups in democracies the more general historical results are quite different. In fact, they are difficult to distinguish from no economic growth effects at all. Given the various imprecisions of statistics, this does not prove that failed coups will have no growth effects, but it can be said that the numbers give us no clear reason to be worried, at least not over the 10-year time horizon chosen by Meyersson. This may be one reason why asset markets do not seem to be panicking over the failed Turkish coup attempt.

To be sure, there are some possible or even likely short run effects of the recent turmoil, such as declines in tourism or foreign investment. Still, the data as a whole are showing that the long-run fundamentals of democracies with failed coups tend to reassert themselves within the 10-year time horizon, and those short-run disruptions end up mattering less than we might think.

Do read the whole thing.  You will note that shares of the Turkish closed end mutual fund are still up about thirteen percent for the year (FT link), though down 2.5 percent at Friday’s close.

Now if Turkey had left the European Union, that would be a different matter altogether…