Optimal online overlapping auctions: Analytical and empirical insights

We examine how online retailers should optimally determine an auction based liquidation schedule of a given inventory over a fixed time period. In contrast to purely simultaneous or sequential auctions that have received attention in the prior literature, we emphasize a more general overlapping pattern of auctions that an auctioneer can benefit from. Overlapping auctions induce information spillovers from one auction to another, and therefore can be expected to endogenously influence the arrival patterns and distribution of bidders that are attracted to a given auction. We find empirical evidence that mechanism and market level control factors such as auction duration and degree of overlap significantly affect the arrival intensity and distribution of bidder willingness-to-pay. We use these empirical insights to develop an analytical model that yields the optimal degree of overlap, i.e. the optimal liquidation schedule, for a retailer facing a given inventory holding cost. Our computational analysis suggests that auctioneers can significantly benefit from using the results of the analytical model, especially under dynamic market environments characterized by exogenously varying arrival patterns.