This award funds research in two different areas of economic theory. The first project develops a new method for understanding an important problem in mechanism design: how to assign an indivisible object to one person when several different people would benefit from owning or using the object. The research generalizes an existing theorem for implementing random assignments; the new generalized theorem applies to a broad range of circumstances and various practically important constraints. The research identifies a condition on the constraint structure that is sufficient for implementation, and develop a computable algorithm for implementation. The research also establishes the condition to be maximally sufficient in two-sided matching. Applications of the theory include improving the ex post fairness of assignment mechanisms, adapting existing mechanisms to be more broadly applicable, and developing new mechanisms. Finally, the research establishes the welfare, fairness, and incentive properties of these mechanisms.
The second project examines behavior in a generalized second price auction (GSP), which is now the industry standard used to conduct auctions for internet search advertising. The goal is to understand rich, dynamic interactions among bidders using a lab experiment. Finally, the researchers compare the GSP method with alternative auction formats based on per-impression bids and analyze a broader class of auction mechanisms.
Random assignment of indivisible goods comprises an important class of real-world problems such as assigning children to public schools, low-income housing, offices, tasks, and queuing for services. The theory and the computable algorithm developed here enhance the real world applicability of this class of mechanisms. Understanding the effects of alternative ways to conduct interent search auctions will benefit providers, purchasers, and policymakers who consider the degree of competition in this increasingly important industry.
