Punishment Menus


Models of optimal criminal penalties generally impose a single punishment scheme. However, when there is unobservable heterogeneity in criminal propensities, welfare can be improved by offering punishment menus to potential criminals ex ante. In doing so, individuals with low criminal tendencies will choose low probabilities of detection and high penalties, while individuals with high criminal tendencies will choose a higher likelihood of detection and low penalties. We present a formal model showing that allowing individuals to choose from a punishment menu improves social welfare relative to a single punishment scheme by simultaneously reducing enforcement costs and enhancing deterrence. This model provides a rationale for some recently proposed citizen trust policies where people can enter their identities in a public registry and be subject to lower audit probabilities.