Nonparametric estimation of English auctions with selective entry: An application to online judicial auctions

Abstract

This paper proposes an estimation approach following the constructive identification strategy of Athey and Haile, and Gentry and Li with adaption in the context of ascending auctions with selective entry. Our estimators are shown to be consistent in a large sample and to perform well in a finite sample by a simulation study. We apply our estimation approach to the Alibaba online judicial auctions of used cars to recover the bounds of conditional value distribution and the entry cost. The bounds estimates of both conditional value distribution and entry cost are quite tight (resp., relatively wide) for middle-valued (resp., low-valued or high-valued) signal, and the cumulative distribution functions of conditional value distribution given signal comply with the law of ordered dominance. Finally, our counterfactual analysis indicates that (i) the ascending auction yields a higher revenue than the first-price sealed bid auction, and (ii) the revenue can be improved significantly when the entry cost is cut by half.