This paper starts with the observation that almost all military dictatorships that democratize become presidential democracies. I hypothesize that military interests are able to coordinate on status-preserving institutional change prior to democratization and therefore prefer political institutions with strong veto players. Parallel civilian interests conversely suffer from coordination failure by being more diverse and les cohesive.
The hypothesis therefore implies that most military democratizations are partially planned while most democratization events from civilian autocracy are either unforeseen or poorly planned. Exploring the characteristics of 111 democratization episodes between 1950 and 2015, I find a number of features broadly consistent with further theoretical predictions.
