Posted prices vs. negotiations: an asymptotic analysis

Full revelation of private values is impractical in many large-scale markets, where posted price mechanisms are a simpler alternative. In this work, we compare the asymptotic behavior of full revelation auctions to posted price auctions in a Bayesian model. We show that posted-price auctions that use discriminatory (i.e., personalized) prices can be asymptotically equivalent to optimal full revelation auctions with the right choice of prices. On the other hand, posted price auctions with one symmetric price are asymptotically inferior to optimal full revelation auctions. Our results are given for general independent distribution functions under standard weak conditions (called the von Mises conditions) that correspond to inherent properties of the distribution functions (e.g., if the support is bounded or unbounded, the shape of the distribution tail, etc). Our results apply to other settings like online algorithms and secretary problems.