Peter Norman Sørensen, University of Copenhagen

“Competitive Selection and Information Aggregation in Auctions"

Abstract

Are more competitive markets better at aggregating private market information? Specifically, suppose n identical objects are offered for sale to k symmetric bidders with interdependent values. Do the n winning bids reveal more information about the state of demand for the good when the number of bidders k rises? We find that competition decreases information if bidders’ private signals have a log-submodular reverse hazard rate function – overturning received wisdom. When private signals derive from a location family, and only one object is for sale (n=1), competition harms information aggregation if and only if the signal’s noise component follows a distribution which is less convex than Gumbel’s extreme value distribution. Drawing on extreme value theory, we quantify the amount of information in the perfectly competitive limit.

Contact person: Peter Norman Sørensen