Optimal auctions with signaling bidders

We study optimal auctions in a symmetric private values setting, where bidders' care about winning the object and a receiver's inference about their type. We reestablish revenue equivalence when bidders' signaling concerns are linear, and the auction makes participation observable via an entry fee. With convex signaling concerns, optimal auctions are fully transparent: every standard auction, which reveals all bids yields maximal revenue. With concave signaling concerns there is no general revenue ranking. We highlight a trade-off between maximizing revenue derived from signaling, and extracting information from bidders. Our methodology combines tools from mechanism design with tools from Bayesian persuasion.