Econometrics of Ascending Auctions by Quantile Regression

This paper suggests an identification and estimation approach based on quantile regression to recover the underlying distribution of bidders' private values in ascending auctions under the IPV paradigm. The quantile regression approach provides a flexible and convenient parametrization of the private values distribution, with an estimation methodology easy to implement and with several specification tests. The quantile framework provides a new focus on the quantile level of the private values distribution and on the seller's optimal screening level, which can be both useful for bidders and seller. The empirical application on timber auctions suggests that using policy recommendations from seller's expected payoff may be sometimes inappropriate from a seller's point of view due to the low probability of selling the good. This seems to be an important issue specially in auctions with strong heterogeneity among the bidders, since the seller has incentive to screen bidders' participation by setting a high reservation price, which in turn leads to a low probability of selling the good.