This paper studies bidding dynamics where values and bidding opportunities follow an unrestricted joint Markov process, independent across agents. Bids cannot be retracted, as is frequently the case in auctions. Our main methodological contribution is that we construct a mapping from this general stochastic process into a distribution of values that is independent of the type of auction considered. The equilibria of a static auction with this distribution of values is used to characterize the equilibria of the dynamic auction, making this general class very tractable. As a result of the option of future rebidding, early bids are shaded and under mild conditions increase toward the end of the auction. Our results are consistent with repeated bidding and skewness of the time distribution of winning bids, two puzzling observations in dynamic auctions. As an application, we estimate the model by matching moments from eBay auctions.
↧