Do Wrongfully Convicted Defendants Receive Higher Punishments?

This paper explores theoretically and empirically whether wrongfully convicted defendants receive different punishments than correctly convicted defendants. I develop a theoretical model in which judges observe a signal -- the level of incriminating evidence -- and then engage in `compromise verdicts' by adjusting the punishment size to the probability that the defendant is guilty. Due to this adjustment, wrongfully convicted defendants systematically receive lower punishments, as long as judges' intrinsic cost of punishing the innocent is sufficiently high. Type I errors (wrongful convictions) and type II errors (wrongful acquittals) will then generally affect crime deterrence asymmetrically. I test the model's prediction in an empirical analysis of criminal sentences in the U.S and find that exonerated defendants receive lower punishments compared to non-exonerated defendants. In line with the theoretical prediction, I also find that the punishment of exonerees increases in the level of the incriminating evidence