Show simple item record

EXPLORING AND MODELING OF BIDDING BEHAVIOR AND STRATEGIES OF ONLINE AUCTIONS

dc.contributor.advisorRand, Williamen_US
dc.contributor.advisorJank, Wolfgangen_US
dc.contributor.authorGuo, Weien_US
dc.date.accessioned2013-06-28T06:56:13Z
dc.date.available2013-06-28T06:56:13Z
dc.date.issued2013en_US
dc.identifier.urihttp://hdl.handle.net/1903/14125
dc.description.abstractInternet auctions, as an exemplar of the recent boom in e-commerce, are grow- ing faster than ever in the last decade. Understanding the reasons why bidders be- have a certain way allows invaluable insight into the auction process. This research focuses on methods for modeling, testing and estimation of bidders' behavior and strategies. I start my discussion with bid shading, which is a common strategy bidders believe obtains the lowest possible price. While almost all bidders shade their bids, at least to some degree, it is impossible to infer the degree and volume of shaded bids directly from observed bidding data. In fact, most bidding data only allows researchers to observe the resulting price process, i.e. whether prices increase fast (due to little shading) or whether they slow down (when all bidders shade their bids). In this work, I propose an agent-based model that simulates bidders with different bidding strategies and their interaction with one another. The model is calibrated (and hence properties about the propensity and degree of shaded bids are estimated) by matching the emerging simulated price process with that of the observed auction data using genetic algorithms. From a statistical point of view, this is challenging because it requires matching functional draws from simulated and real price processes. I propose several competing fitness functions and explore how the choice alters the resulting ABM calibration. The method is applied to the context of eBay auctions for digital cameras and show that a balanced fitness function yields the best results. Furthermore, in light of the discrepancy find from the model in bidders' be- havior and optimal strategies proposed from online auction literature. I extract empirical bidding strategies from auction winners and utilize the agent based model to simulate and test the performance of twenty-four different empirical and theo- retical strategies. The experiment results suggest that some empirical strategies perform robustly when compared to theoretical strategies and taking into account other bidders' ability to learn. In addition, I expended the online auction framework from single auction to multiple auction simulation, which acts as a platform for investigating and test- ing more complicated situations that involves the competition among concurrent auctions. This framework facilitates my investigation of bidders' switching behavior and enables me to answer a series questions. For example, is it beneficial for auction website to promote bidders' switching behavior? Will bidders and even sellers get any advantage from bidders' switching? What is the best auction recommendation strategy for online auction website to obtain higher profit and/or a better customer experience? Through careful experiment design, it has been showed that higher switching frequency leads to higher profit for auction website and reduces the price dispersion, which leads to reduced risk for both bidders and sellers. In addition, the best auction recommendation strategy is providing the five earliest closing auctions so that bidders can choose the lowest price auction.en_US
dc.titleEXPLORING AND MODELING OF BIDDING BEHAVIOR AND STRATEGIES OF ONLINE AUCTIONSen_US
dc.typeDissertationen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.contributor.departmentMathematicsen_US
dc.subject.pqcontrolledStatisticsen_US
dc.subject.pqcontrolledMarketingen_US
dc.subject.pqcontrolledBusinessen_US
dc.subject.pquncontrolledAgent-based modelen_US
dc.subject.pquncontrolledBidding strategyen_US
dc.subject.pquncontrolledOnline auctionsen_US
dc.subject.pquncontrolledParameter estimationen_US
dc.subject.pquncontrolledRecommendation strategyen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record