The motivation for this research is derived from a currently active research topic in both marketing and economics, Internet auctions. A sequence of bids in this market mechanism can be viewed as record breaking events with participants (bidders) having dynamically changing valuations for the auctioned item, but the latent number of bidders "competing" in those events is unseen.
Our approach to addressing this problem is through data augmentation in which we stochastically draw the number of latent bidders, and then model the observed bid amounts and bid submission times conditional on the non-seen events.
In addition, as our research is among the first attempts to take a descriptive and exploratory modeling approach to auction data by allowing for dynamically changing (bid-by-bid) bidder valuations, we provide a model that is suggestive for those doing theoretical and/or theoretically grounded empirical work in this area.
Meet the speaker in Room 212 Cockins Hall at 4:30 p.m. Refreshments will be served.