Information Sharing in Common Value Auctions An Experimental Investigation

A number of conflicting conjectures have been proposed about the role of information and collusion in common-value auctions. This study investigates these hypotheses in an experimental setting by conducting a series of auctions with an equal number of signals (estimates of the object’s value) but a varying number of bidders among whom those signals are concentrated. Specifically, two types of common-value auctions are examined, each characterized by participants bidding on objects of a common, but unknown value. Collusion and mergers among bidders have two potential effects. First, since collusion decreases the number of active bidders, competition is lessened, leading to lower prices paid at auction. Second, recent theoretical results suggest that the pooling of information within a bidding syndicate, implying more precise estimates of the object’s value, lead to more aggressive bidding, and higher prices. Our results are consistent with the latter hypothesis that bidding is more aggressive as information is concentrated among fewer bidders. Furthermore, we find that this effect is dominated by the reduction in the number of bidders, leading to lower revenues. JEL Classification Numbers: D44: Auctions, L41: Horizontal Antitrust Policy, C92: Laboratory Experiments