Date of Award
Doctor of Philosophy (PhD)
Mechanism design has found various applications in today's economy, such as ad auctions and online markets. The goal of mechanism design is to design a mechanism or system such that a group of strategic agents are incentivized to choose actions that also help achieve the designer’s objective. However, in many of the mechanism design problems, the theoretically optimal mechanisms are complex and randomized, while mechanisms used in practice are usually simple and deterministic. The focus of this thesis is to resolve the discrepancy between theory and practice by studying the following questions: Are the mechanisms used in practice close to optimal? Can we design simple mechanisms to approximate the optimal one? In this thesis we focus on two important mechanism design settings: multi-item auctions and two-sided markets. We show that in both of the settings, there are indeed simple and approximately-optimal mechanisms. Following Myerson's seminal result, which provides a simple and revenue-optimal auction when a seller is selling a singleitem to multiple buyers, there has been extensive research effort on maximizing revenue in multi-item auctions. However, the revenue-optimal mechanism is proved to be complex and randomized. We provide a unified framework to approximate the optimal revenue in a fairly general setting of multi-item auctions with multiple buyers. Our result substantially improves the results in the literature and applies to broader cases. Another line of works in this thesis focuses on two-sided markets, where sellers also participate in the mechanism and have their own costs. The impossibility result by Myerson and Satterthwaite shows that even in the simplist bilateral trade setting (1 buyer, 1 seller, 1 item), the full welfare is not achievable by a truthful mechanism that does not run a deficit. In this thesis we focus on a more challenging objective gains from trade --- the increment of the welfare, and provide simple mechanisms that approximate the optimal gains from trade, in bilateral trade and many other two-sided market settings.
Zhao, Mingfei, "Simple vs. Optimal Mechanism Design" (2021). Yale Graduate School of Arts and Sciences Dissertations. 145.