Information Revelation and Random Entry in Sequential Ascending Auctions

We examine a model in which multiple buyers with single-unit demand are faced with an infinite sequence of auctions. New buyers arrive on the market probabilistically, and are each endowed with a constant private value. Moreover, objects also arrive on the market at random times, so the number of competitors and the degree of informational asymmetry among them may vary across from one auction to the next. We demonstrate by way of a simple example the inefficiency of the second-price sealed-bid auction in this setting, and therefore assume that each object is sold via ascending auction. We then characterize an efficient and fully revealing periodic ex post incentive compatible equilibrium for the game in which the objects are sold via ascending auctions. We show that each buyer's bids and payoffs depend only upon their rank amongst their competitors and the (revealed) values of those with lower values. Furthermore, strategies are memoryless -- bids depend only upon the information revealed in the current auction, and not on any information that may have been revealed in earlier periods. We then demonstrate that the sequential ascending auction serves as an indirect mechanism that is equivalent -- in our setting -- to the dynamic marginal contribution mechanism introduced by Bergemann and Välimäki (2007) and generalized in Cavallo, Parkes, and Singh (2007).