Information Dispersion and Auction Prices

Do bidders behave as auction theory predicts they should? How do bidders (and thus, prices) react to different types of information? This paper derives implications of auction theory with respect to the dispersion of private information signals in an auction. I conduct a survey of non-bidders to construct a measure of information dispersion that is independent of bidding data. This permits joint tests of Bayesian-Nash equilibrium bidder behavior and information structure (common vs. private value) in a sample of eBay auctions for computers. The measure also allows me to separately estimate the price effects of seller reputation and product information. eBay prices appear consistent with Bayesian-Nash common value bidding behavior. Uncertainty about the value of goods due to information dispersed over auction participants plays a larger role than uncertainty about the trustworthiness of the sellers, but both are significant drivers of price. Thus, seller reputation complements, rather than substitutes for, information provided in the auction descriptions by lending credibility to that information, creating an incentive for sellers to reduce uncertainty in their auctions