SMU SOE Seminar Series (November 12, 2026): Both Sides Now: Identification of Buyer and Seller Values in Ascending Auctions

TOPIC:

ABSTRACT

Auctions are inherently two-sided, yet empirical auction models typically recover bidder values while treating sellers' preferences as known, fixed, or irrelevant. We study identification of buyer and seller primitives in ascending auctions in which a sale occurs only when the highest buyer value exceeds a seller-specific threshold. Using variation in bidder counts, we characterise identification across data environments and modelling assumptions. When sale outcomes are observed but rejected bids are not, buyer value and seller threshold distributions are point identified. When only successful auctions are observed, identification reduces to a one-dimensional fixed-point problem in the ratio of sale probabilities across bidder counts. We show that the set of fixed points is finite, and provide an upper-tail support condition guaranteeing uniqueness. A conduct assumption maps seller thresholds into values; weaker conduct restrictions generate bounds. Monte Carlo experiments validate our approach and quantify biases from ignoring rejected offers.

Keywords: Auctions, Identification, Set Identification.

JEL: C57, D44.

Click here to view the paper.