³ÉÈË´óƬ

This paper conducts an empirical analysis of eBay online auctions for computers. The theory of common value auctions has a number of implications. In equilibrium, bidders must account for the winner's curse. The winner's curse changes with the dispersion of information about the common value, the number of bidders, and the credibility of the seller. A novel feature of my analysis is that I supplement the bidding data with an original survey, in which an average of 46 people provided estimates of the value of each computer in my sample. From my survey, I construct measures that should be correlated with the mean and dispersion of bidders' signals in my sample of eBay auctions. I then test whether bidding in these auctions is consistent with predictions from common-value, Nash equilibrium behavior. My results indicate that they are: prices decrease with dispersion, and the rate of decrease depends on the level of information dispersion, the number of bidders, and sellers reputation (credibility). I them impose the common-value, Nash equilibrium price function to estimate the biases in my survey measures. After adjusting the survey measures of common value and information dispersion, I then calculate the effect of reputation on prices and information dispersion and quantify the winner's curse present in these auctions. I find that an interaction effect between reputation and information dispersion creates an incentive for credible sellers to provide detailed information in their auction descriptions, reducing uncertainty and promoting efficient trade.