Revenue Maximization in Platforms

The goal of the project is to understand revenue maximization policies for a platform that can control the assortment of items shown to a prospective buyer, but not the transaction prices. This set of problems is motivated by the common observation that most online platforms do not manufacture goods or provide services, but simply enable buyers and sellers to discover each other; the buyers and sellers jointly determine the price at which a transaction occurs. In such an environment, how should the platform decide which subset of sellers to display to a prospective buyer? This general question can be thought of as a dynamic combinatorial optimization problem in which the platform chooses, at each time step, the subset of players who then compete in a price-setting game. There is a small, but growing, literature on such models, with several interesting open questions. The summer project will be focused on a careful study of the literature and developing an efficient algorithm that achieves near-optimal revenue for a specific instantiation of this model.