Time Dynamics of User Behavior in a Series of Overlapping Electronic Mechanisms

This paper investigates the time dynamics of online user behavior in a series of overlapping electronic auction mechanisms. An auction is a dynamic game where the valuation and strategic space of bidders determine the final outcomes. When those mechanisms exist concurrently or in a series, multiple sources are visible and accessible, and there is likely to be a form of interdependency. While heterogeneous bidder behavior has been studied in some literature, the focus is mainly on individual auction level and the underlying dynamics regarding the interdependency among the auctions in the market has not been explained. We use two-phased approach. First, we classify user strategy using k-means clustering. Then we characterize the transition pattern of heterogeneous clusters using dynamic systems framework. Long-term behavior of the system is effectively and efficiently predicted using system parameters. Our model can serve as a decision support framework for optimal market design.