Sep 21, 2018
Jolley Hall, Room 309
"Computational Explorations of Information and Mechanism Design in Markets"
Adviser: Sanmay Das
Markets or platforms assemble multiple selfishly-motivated and strategic agents. The outcomes of such agent interactions depend heavily on the rules, regulations, and norms of the platform, as well as the information available to agents. This thesis investigates the design and analysis of mechanisms and information structures through the "computational lens'' in both static and dynamic settings. It both addresses the outcome of single platforms and fills a gap in the study of the dynamics of multiple platform interactions.
In static market settings, we are particularly interested in the role of information, because mechanisms are harder to change than the information available to participants. We approach information design through specific examples, i.e., matching markets and auction markets. First, in matching markets, we study the situation where the matching is preceded by a costly interviewing stage in which firms acquire information about the qualities of candidates. We focus on the impact of the signals of quality available prior to the interviewing stage. We show that more "commonality'' in the quality of information can be harmful, yielding fewer matches. Second, in auction markets, we design an information environment for revenue enhancement in a sealed-bid second price auction. Much of the previous literature has focused on signal design in settings where bidders are symmetrically informed, or on the design of optimal mechanisms under fixed information structures. Here, we provide new theoretical insights for complex situations like corporate mergers, where the sender of the signal has the opportunity to communicate in different ways to different receivers.
Next, in dynamic markets, we focus on two dimensions: (1) the effects of different market-clearing rules on market outcomes and (2) the dynamics of multiple platform interactions. Considering both dimensions, we investigate two important real-world dynamic markets: kidney exchange and financial markets. Specifically, in kidney exchange, we analyze the performance of different market-clearing algorithms and design a competing-market model to quantify the social welfare loss caused by market competition and exchange fragmentation. Here, we present the first analysis of equilibrium behavior in these dynamic competing matching market systems, from the viewpoints of both agents and markets. To improve the performance of kidney exchange in terms of both social welfare and individual utility, we analyze the benefit of convincing directed donation pairs to participate in paired kidney exchange, measured in terms of long-term graft survival. We provide the first empirical evidence that including compatible pairs dramatically benefits both social welfare and individual outcomes.
For financial markets, in the debate over high frequency trading, the frequent call (Call) mechanism has recently received considerable attention as a proposal for replacing the continuous double auction (CDA) mechanisms that currently run most financial markets. We examine agents' profit under CDA and frequent call auctions in a dynamic environment. We design an agent-based model to study the competition between these two market policies and show that CALL markets can drive trade away from CDAs. The results help to inform this very important debate.