Dynamic Demand Estimation in Auction Markets

Matthew Backus, UC Berkeley, NBER, and CEPR and Gregory Lewis

We study demand estimation in a large auction market. In our model, a dynamically evolving population of buyers with unit demand and heterogeneous and privately known preferences for a finite set of differentiated products compete in a sequence of auctions that occur in discrete time. We define an empirically tractable equilibrium concept in which bidders behave as though they are competing with the stationary distribution of opposing bids, characterize bidding strategies, and prove existence of equilibrium. Having developed this demand system, we prove that it is non-parametrically identified from panel data. We extend the model to consider a random coefficients demand system akin to workhorse demand models in industrial organization, and show that this too is non-parametrically identified. We apply the model to estimate demand and show how large sellers can exercise market power by using persistent reserve price policies, which induce higher bids and, therefore, revenues. Our analysis highlights the importance of both dynamic bidding strategies and panel data sample selection issues when analyzing these markets.