49 Pages Posted: 8 Jan 2009
Date Written: April 10, 2006
Bidders' psychological biases can have important effects on behavior in auctions and the principles of auction design. I consider two potential sources of the winner's curse in common values auctions. The first, overconfidence, is here defined for the first time in a fully general, distribution free manner by bidders believing their signals are more informative than they really are. The other bias, disregard, is intended to capture the spirit of an alternative explanation of the winner's curse proposed by Eyster and Rabin (2006) and is present when bidders underestimate the informativeness of other bidders' signals. When the value of the object is normally distributed, the two hypotheses can be formulated in an intuitive manner. Both explain the winner's curse, but they offer sharply different predictions about bidder behavior and prescriptions for auction design. When bidders are very overconfident, the first-price auction is revenue preferred to the second-price auction. Furthermore, even when the effects are relatively mild, they can reverse Milgrom and Weber (1982)'s Disclosure Principle as it can be optimal for auctioneers to induce speculation.
Keywords: auctions, overconfidence, winner's curse
JEL Classification: D44
Suggested Citation: Suggested Citation