Threats of mass revolts could effectively constrain a dictator's public policy if it were not for the collective-action problem. Mass revolts nevertheless happen, but they follow a stochastic pattern. We describe this pattern in a threshold model of collective action and integrate it into an agency model which demonstrates how mass revolts can impact on a winning coalition's incentives to keep backing an incumbent dictator. Having observed public policy and found a sufficiently high posterior probability of the dictator to be of a "bad" character, the winning coalition's members may exploit an incidentally happening mass revolt for escaping a loyalty trap that had otherwise prevented them from switching to disloyalty. While this explains why mass revolts sometimes happen to oust a dictator, the arising policy constraints in dictatorships may nevertheless be weak in practice.
Autocracy Revolutions Threshold Models Selectorate Theory