Studies in the Foundations of Game and Auction Theory

Princeton University, Princeton NJ

Investigators

Abstract

This project consists of a methodological and an auction-theoretic section. The former extends the literature on the foundational theory of extensive games by introducing the notion of epistemic priorities and analyzing its implications. The latter applies the foundational approach to game theory to the study of standard and novel auction mechanisms; hence, it helps clarify the connections between the predictions of auction theory, and its underlying behavioral and epistemic assumptions. The notion of epistemic priority reflects the idea that players may entertain various hypotheses about their opponents, and rank them according to their plausibility. Epistemically prioritized hypotheses form the basis for an axiomatization of solution concepts incorporating forward induction, but are also useful in understanding the role of belief consistency in equilibrium analysis. The analysis of auctions employs normal- and extensive-form rationalizability, augmented with appropriate assumptions on bidders' mutual beliefs. This project develops general techniques to solve a variety of "standard" auction mechanism under different assumptions about payoffs and information. This methodology is then employed to assess the robustness of "classical" results in auction theory, such as revenue equivalence / revenue ranking of different auction formats, bid shading in first-price auctions, and demand reduction in uniform-price auctions.