Can We Use an Equilibrium Model to Interpret the Auction Data?

• Activity Number: 656
• Type: Invited
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Xiaoming Huo*+
• Companies: Georgia Institute of Technology
• Keywords: auction ; equilibrium ; game theory

Abstract

We consider modeling the internet auction data. Our model is rooted on the efficient market assumption and the game theory. In principle, market equilibrium can exist when every bidder adopts the optimal strategy to maximize their profit. Empirically, it is interesting to find out that the actual bidding prices are not far away from what the equilibrium predicts. We introduce two important latent variables: expected returns and ratio of clicks from neighboring positions. By looking at the estimated values of these latent variables, we have some interesting findings as well as conjectures on how the auction evolves.