Manipulative auction design

Abstract

This paper considers a manipulative auction design framework in which the designer in addition to choosing the auction format(s) is free to choose how much information feedback about the distribution of bids observed in previous auctions to report to bidders. A feedback equilibrium is proposed to model the long run interactions of bidders in such environments with partial feedback. It is shown that the first-price auction in which bidders get only to know the aggregate distribution of bids (across all bidders) generates more revenues than the second-price auction while achieving an efficient outcome in the asymmetric private values two-bidder case with independent distributions. It is also shown how by using several auction formats with coarse feedback a designer can always extract more revenues than in Myerson's optimal auction, and yet less revenues than in the full information case whenever bidders enjoy ex-post quitting rights and the assignment and payment rules are monotonic in bids.