| In this dissertation, I develop a dynamic model of online auctions for standardized goods. The model incorporates two key environment features: first, there is an infinite series of auctions for an identical item; second, those auctions are naturally ordered over time.;In Chapter 1, a theoretical model is presented which incorporates the buyers' decision of what auctions to bid, as well as how much to bid. It is shown that the optimal strategy of risk-neutral, forward-looking bidders is to bid below their valuation. By contrast, the dominant strategy of myopic bidders is to bid their valuation.;In Chapter 2, the model is applied to a rich dataset of online auctions for a 60 GB Apple iPod Video. An identification and estimation strategy is proposed that allows me to evaluate the efficiency of the current mechanism compared to a perfect competition benchmark. I consider separately the cases of forward-looking and myopic bidders. Unfortunately, the identification strategy does not allow me to distinguish between these two behavioral assumptions based on the field data alone.;The results from Chapter 2 motivate the analysis in Chapter 3, where I report on a laboratory experiment of sequential auction bidding. The purpose of the experiment is to evaluate whether in a dynamic laboratory environment subjects adjust their bids according to the theory developed in Chapter 1. The results of the experiment are negative: I find no significant difference between the subjects' behavior in a two-period environment with respect to a one-period environment. In other words, I cannot reject the hypothesis of myopic bidding behavior. |
