Doctoral Dissertation in Economics: Informational Asymmetries and Auction Design: An Experiment

New York University, New York NY

Investigators

Abstract

Auctions are an important mechanism used to allocated goods and raise revenue. Oil-lease and spectrum auctions raise large amounts of revenue for the government and have been the focus of much economic research. Many of these auctions are for goods which have an uncertain value which is the same for all bidders (e.g., the amount of oil in a lease). Traditional theoretical models assume that bidders have the same amount of information concerning the value of the good. However, this assumption ignores the fact that often bidders may be asymmetrically informed, which has been empirically shown to affect bidding behavior. Understanding how asymmetries in bidders' information affects their bidding behavior is important for predicting the revenue an auction will generate and for designing auctions to increase the revenue they raise. The PIs expand the traditional auction framework to allow for bidders to differ both in their estimates of a good's value and the precision of their estimates. They study how bidding behavior changes when bidders can and cannot observe the precision of opposing bidders' information. In many sealed-bid or online auctions, the identity of opposing bidders is hidden and each bidder may not be perfectly informed about how precise his opponent's information is. With the rise of internet auctions, however, sellers have a large degree of control of the information bidders possess about one another. How much information about bidders' characteristics an auction designer should release is an important consideration when designing auctions and is one that has not previously been studied. The current research project shows how sellers can optimally design the release of this information. By studying the impact of changing how much information bidders possess about one another, the PIs find a new avenue by which auction design can improve revenue generation. The theoretical predictions show that publicly revealing the precision of each bidder's information decreases the expected revenue an auction will generate. This finding suggests that controlling how much information bidders have about each other is an important factor to consider when designing auctions. The PIs seek to answer the following question by performing a series of experiments: how do bidders behave when they know that their information about the good is more (or less) precise than other bidders? How do their bidding strategies change when they do not know the precision of their opponents information? The proposed study intends to answer these questions, which will contribute to a better understanding of how much information to disclose in auctions. This experiment will provide a better understanding of how to implement information design into auction formats.