Buyer Signaling Games in Auctions
Weiran Shen,Pingzhong Tang,Yulong Zeng
DOI: https://doi.org/10.5555/3306127.3331878
2019-01-01
Abstract:We consider an auction setting where a seller sells one item to several buyers. Before a buyer's type is realized, he can commit himself to a so-called signal scheme. Mathematically, a signal scheme can be regarded as a linear decomposition of his prior type distribution into a probability distribution over a set of posterior distributions, each of which the seller can use a revenue optimal auction tailored for that distribution. It is known, from the literature of Bayes persuasion, that such signal schemes can lead to utility increase for both the seller and the buyers. Our goal, is to analyze how a buyer should signal his distribution, given that other buyers may also signal their distributions. In other words, we want to find an equilibrium profile of signal schemes. We obtain the closed-form solution for the single buyer case with regular distributions, and the multiple buyers case with symmetric type distributions under certain conditions. To prove our technique results, we also obtain some interesting intermediate results. In particular, we show that, if each buyer's signal scheme is to decompose his prior distribution into a set of posteriors that has the same virtual value function (in the exact sense of Myerson's virtual value function), his expected utility is equal to his utility in a first price auction game where his bidding function is always his virtual value function. Furthermore, perhaps surprisingly, we show that, certain distributions, including the uniform distribution, satisfy the property that every buyer's optimal signal scheme is indeed to decompose the prior into a set of posteriors that has the same virtual value function. As a result, we give the closed-form of an equilibrium profile of signal schemes for these cases.