This PhD thesis is concerned with buyers' strategies in sequential and concurrent auctions. It deals with both the theoretical viewpoint and data analysis of online consumer auctions. The first chapter contains a newly developed model of sequential auctions with overlapping generations of bidders. The emphasis is on the existence of learning from observed past prices. With the addition of overlapping generations the learning happens through two channels: updating on valuations and expectation of composition of bidders with different horizons lengths. The model shows how this happens on the micro level, where expected distributions of bids are updated. In the following chapter, the predictions of theoretical models of sequential auctions together with learning are tested empirically. It is shown that bidders adjust their bids as a consequence of learning as predicted by the model. Bid discounting is also observed in the data. The following empirical chapter uses the bids data from online auctions to perform multinomial logit estimations. Individual choice model allows to analyze the aspects that attract bidders to particular auctions out of many very similar ones available. A unique dataset that contains data from many auctions for the same product is used in this new way. Dynamic aspects of auctions such as the number of bidders and bids are shown to play a role in auction choice. Overall, there are three approaches to the empirical analysis of bidders strategies, based on the same dataset. It is shown that with appropriate adjustments the data collected from online auctions can be used in different formats to answer various questions.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:752509 |
| Date | January 2018 |
| Creators | Wojciechowska, Olga |
| Publisher | University of Warwick |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://wrap.warwick.ac.uk/105556/ |
Page generated in 0.0012 seconds