Identification and Estimation in Sequential Dutch Auctions

Within the independent private-values paradigm, assuming asymmetric bidders,using the principle of dynamic programming, and employing the notion of Bayes-Nash equilibrium, we demonstrate nonparametric identification of the data-generating process of the sequence of winning prices at multi-unit, sequential, oral, descending-price (Dutch) auctions. Subsequently, we propose estimators of the distribution of latent valuations. We implement our framework using a sample of data from sequential, Dutch auctions of flowers held in Aarhus, Denmark. Using the estimated distributions in conjunction with simulation methods, we then undertake the policy experiment of comparing the outcomes at Dutch auctions with those that would obtain were Myerson's optimal auction employed