This project is inspired by two important aspects of the Internet: the growth in the use of auctions online, and the potential of the Internet as an experimental research tool. It involves several experiments to perform new tests of auction theory. It also involves replications of previous experiments on auctions, to see whether past results are robust to the environment in which most commercial online auctions operate. The Internet is an increasingly important tool for facilitating commerce, and one of the exciting potential uses of this tool is in implementing auctions and other pricing mechanisms. Companies such as Onsale and eBay have recently established themselves as online auction houses for items as diverse as refurbished computer equipment, ostrich burgers, Nissan trucks, and vacation packages. Currently, these sites sell over $15 million in goods per week via auction. Even some airlines have begun to auction a limited number of passenger tickets on their Web sites. Networked computer technology appears to be bringing auctions to a wide variety of new markets. The increased use and variety of auction mechanisms raises new research questions. It is important to understand how well the existing body of auction theory applies to the new market situations where auctions are being applied, and to learn how efficiently these auctions allocate resources in practice. In this project, experiments with actual bidders investigate the properties of various auction selling formats, using the Internet itself as a research tool. This research involves two main components. First are new tests of the theory of the effects of reserve prices (minimum bids) in single-unit auctions with varying numbers of bidders. Reserve prices are significant features of almost all auctions, from art to wine to computers. Their effects have been investigated theoretically by a number of authors, but have received very little empirical attention to date. Second are new comparisons of revenues between auction formats, such as English (ascending-price), Dutch (descending-price), and sealed-bid auctions. Comparisons of revenues between auctions have previously been studied extensively in the laboratory, but the author's recent field experiments suggest that past results from laboratory sessions are not always robust to Internet-based auctions for real goods. The proposed work will replicate several laboratory experiments, checking their results' sensitivity to (1) the use of real goods rather than induced valuations, and (2) variations in the rules and timescales of the auctions. As part of the research program, the author develops significant methodological innovations through the construction of a `virtual laboratory`. The ability to conduct experiments with participants over the Internet will be a boon to researchers in the social, behavioral, and economic sciences. This project aims to improve research technology in Internet-based experiments, opening up possibilities for effective data collection on human behavior from participants in all walks of life, and in various types of interaction with each other.
