Actuarial Science and Financial Mathematics seminar: Xiaosheng Mu

Friday, January 23, 2026 10:30 am - 11:30 am EST (GMT -05:00)

Joint seminar with Department of Economics

Xiaosheng Mu
Princeton University

Room: HH 334


Privacy Preserving Auctions

In many auction settings, the auctioneer must disclose the winner's identity and payment. We characterize the auction that minimizes the winner's privacy loss among those that maximize total surplus or seller revenue and are strategy-proof. Privacy loss is defined by what an outside observer learns from the disclosed price, measured as the mutual information between the price and the winner's valuation. When only interim individual rationality is required, the most privacy-preserving auction uses stochastic ex-post payments. Under ex-post individual rationality and a monotone hazard rate, privacy loss is minimized by the second-price auction with deterministic payments.