I analyze the coarse revenue guarantee, defined as the minimum expected revenue across all Bayes Coarse Correlated Equilibria (BCCE), in the First Price Auction (FPA). This approach generalizes the standard revenue guarantee—based on Bayes Correlated Equilibrium (BCE)—established by Bergemann, Brooks and Morris (2017, ECMA) in the FPA. Because BCCE characterizes the limit points of no-regret learning dynamics, this metric provides a more robust lower bound for markets populated by algorithmic or AI agents.

I characterize the coarse revenue guarantee using a reduction to “identical play equilibria,” in which bidders adopt identical strategies regardless of their private signals. I prove that while the coarse revenue guarantee is strictly lower than the BCE-based guarantee for any finite number of buyers, the two measures converge asymptotically as the market grows. Furthermore, I develop a coarse revenue guarantee ranking over standard auctions, generalizing the logic of Bergemann, Brooks and Morris (2019, AER). I also extend the coarse analysis to all value distributions with a fixed expected value.

• UCSD TBE workshop
• GAIMSS'24 (Flash talk and poster)
• ACM EC24 (poster)
• Stony Brook Game Theory Conference 2024
• Caltech Student Theory Conference 2025
• Southwest Economics Theory Conference 2026
• Econometric Society Asian-China Meeting 2026 (Scheduled)

A homogeneous good is for sale to a large number of buyers with multi-unit demand. There are both common and idiosyncratic components to preferences. Trade occurs via a market order mechanism: Each buyer places an order for a number of units; the orders are filled in a random sequence; and buyers pay a price per unit that is a function of the aggregate order. We study the limit of such economies as the numbers of buyers and units grows large. If the pricing rules have vanishing price impact and span, then regardless of the sequence of information structures and equilibria, the limiting market outcome is guaranteed to be competitive: The goods are sold to the buyers with the highest values, at an expected price that is equal to the expected marginal value for the good. It is impossible to achieve similar guarantees for information aggregation.

• Triangle micro conference 2024^
• UChicago^
• Yale^
• Cambridge^
• UCL^
• Stony Brook Game Theory Conference 2025
• UC Berkeley^
• Stanford^
• NBER Decentralization Conference 2026^
• 2026 Asian School in Theory (Scheduled)

^: presented by a co-author