State censorship

We characterize a ruler's decision of whether to censor media reports that convey information to citizens who decide whether to revolt. We find: (i) a ruler gains (his ex ante expected payoff increases) by committing to censoring slightly less than he does in equilibrium: his equilibrium calculations ignore that censoring less causes citizens to update more positively following no news; (ii) a ruler gains from higher censorship costs if and only if censorship costs exceed a critical threshold; (iii) a bad ruler prefers a very strong media to a very weak one, but a good ruler prefers the opposite.