AbstractsComputer Science

The thesis investigates a number of auction formats both theoretically and empirically. The effect of different auction rules on the final price and on bidder valuations is analyzed. Results from an experimental sale of real goods, testing revenue equivalence of the open and sealed-bid second-price auction do not conform to theoretical predictions: the open auction leading to significantly lower prices than the sealed-bid auction. It turns out that the open auction format allows bidders to satisfy a tendency to "stick together" with their valuations. The empirical results motivate a dynamic bidding model of interdependent valuations, bidders being uncertain about their valuations and learning from the exit-prices of their rivals. Furthermore, bidding behavior on the Internet is investigated in the hard close and the automatically extended auction. Late bidding is shown to be a rational strategy in the hard close auction, but not in the automatically extended auction. Theoretical results show that the expected final price and seller revenue is lower in the hard close auction than in the automatically extended auction, where prestige-concerns can lead to an explosive final price. Moreover, Yahoo auction data confirms the strong presence of late bidding in the hard-close auction and the seller's preference for the automatically extended auction. (author´s abstract)