Learning in Sequential Auctions

Abstract

This is the working paper version of my paper by the same title published in the Southern Economic Journal, 65(1), July 1998, 98-112.

The importance of information is studied in a model of a sequential auction in which bidders have independent private values. In the course of the auction information about the bidders’ values becomes available, as winning bids are revealed. From this bidders learn about their opponents’ types. A more subtle effect of information is that bidders anticipate the generation of information and take this into account in the first auction.

The equilibrium in this model is contrasted to a scenario in which bidders are unaware of informational effects. It is shown that bidders who are aware of informational effects have higher payoffs. However, a static auction, in which informational effects are absent, yields the same expected final allocation as the equilibrium. Finally, properties of the equilibrium price path are studied. Regardless of the outcome of the first auction, the second price is expected to be equal to the first price. Despite this, the probability of a decreasing price sequence depends on the information generated in the first auction.

Keywords: Sequential Auctions, Learning

JEL Classification: D44, D83