Department of Industrial Engineering & Decision Analytics [Joint IEDA/ISOM] seminar - The Role of Prescreening in Auctions with Predictions

Sellers often prescreen potential bidders to limit participation to a group of serious and capable individuals. Advances in machine learning and GenAI have increasingly enabled this strategy by cost-effectively identifying high-valuation bidders. However, this practice deviates from standard auction theory, which typically favors broad competition over selective exclusion. In this paper, we examine whether and under what conditions bidder prescreening can be economically justified. We analyze a setting in which bidders have independent and identically distributed private valuations, and the seller observes noisy signals generated by a valuation predictor. Based on these signals, the seller decides how many top bidders to admit. We demonstrate that an auction with prescreening is equivalent to a standard auction (i.e., without prescreening) but with statistically correlated valuations. Our findings indicate that while admitting fewer bidders leads to revenue losses in both second-price and first-price auctions, employing a more accurate predictor can effectively mitigate or entirely offset these losses. Conversely, in all-pay auctions, prescreening can substantially enhance revenue; notably, with a perfect predictor, admitting only two bidders is optimal. All our results remain valid when considering reserve prices. We also characterize the joint optimality of prescreening and auction design under certain tractable conditions.