Yuki Takagi - Concessions and Information Cascades in Autocracies (With Daiki Kishishita and Hans H. Tung)

Motivated by Tocqueville’s long-held insight that, “The most dangerous moment for a bad government is when it begins to reform.” (1856), this paper examines the effects of concessions on mobilization and the optimal level of concession for the regime. We develop a formal model in which a dictator decides whether to make a costly concession, and then two citizens sequentially decide whether to revolt. Each citizen receives a private signal about the level of grievance against the regime, and then decides whether to revolt. Each citizen revolts if she believes that the level of grievance is sufficiently high. If both citizens revolt, the revolution happens. The revolution is successful if and only if the grievance level is sufficiently high. We first show that there are two effects of concession for citizen 1 on mobilization: concession for citizen 1 reduces citizen 1’s incentive to revolt but increases citizen 2’s incentive to revolt. When the regime makes a concession to citizen 1, it reduces citizen 1’s incentive to revolt because citizen 1 is treated better under the current regime, which makes a revolution less attractive to her. This is the benefit of concession for the regime and usually expected. The interesting result is the effect on citizen 2’s incentive. The concession for citizen 1 increases the probability of citizen 2’s revolt conditional on citizen 1’s revolt, i.e., causes information cascade. As a result, after a concession, citizen 1 is less likely to revolt but citizen 2 is more likely to revolt conditional on citizen 1 revolting. Thus, concession is not necessarily beneficial to the regime. Second, our comparative statics of the equilibrium show that, as the cost of suppression increases, the optimal level of concession for citizen 1 increases, and thus the information cascade is more likely to happen.