Coarse Revenue Guarantee in First Price Auction and Beyond

I study the coarse revenue guarantee (CRG) of the single-unit first-price auction with common values. The CRG is the minimum expected revenue over all information structures and all Bayes coarse correlated equilibria (BCCE), the equilibrium notion to which no-regret learning leads. Identical play, in which all bidders submit the same bid, attains the worst-case revenue, and this reduction yields a closed-form characterization. The coarse guarantee is positive and increases with the number of bidders $N$. It lies strictly below the revenue guarantee of Bergemann, Brooks and Morris (2017) at every finite $n$, and the gap between the two closes at rate $\log N\mathrm{/}N$. I extend the analysis to reserve prices, symmetric and asymmetric priors, other standard auctions, and an unknown value distribution.