Ulric B. and Evelyn L. Bray Social Sciences Seminar

Abstract: We revisit the problem of single-agent mechanism design with limited commitment. We show how to write the principal's problem as that of designing a mechanism for today and an information structure over the agent's types for her future self, which determines what mechanism is chosen in the following periods. Contrary to information design problems, today's principal (the sender) takes an action—designs a mechanism—and only information structures that are compatible with the agent's incentive constraints are feasible. Like in information design problems, today's principal treats her future self as a receiver that selects the optimal continuation mechanism given her posterior belief. The representation allows us to maximize separately in the allocation and the distribution of posterior beliefs, which greatly simplifies the mechanism design with limited commitment problem. We show the power of this tool by applying it to a version the allocation problem considered in Skreta (ReStud 2006) and to the regulation problem in Fiocco and Strausz (2017).

Research performed jointly with Vasiliki Skreta.