| This dissertation investigates commercially feasible automated negotiation in electronic commerce. Three types of automated negotiation are investigated: automated bargaining, the online auction, and the online broker. A three-layered framework explains why multi-dimensional automated bargaining is not currently possible with existing electronic commerce technology. The framework also explains why automated price negotiation through the online auction is a thriving sector of electronic commerce, and why, despite some difficulties, online brokered marketspaces are partially feasible.; An overview of online auction activity during 1997–1998 is presented, and different business strategies are outlined. Within the framework of the online auction, a novel Markov chain model of the progression of the price vector of an online auction is presented. Extensions to multiple-item auctions and applications to dynamic inventory management are also given. The predictions of the model are compared to the results of actual online auctions, using data from a leading online auctioneer, Onsale, Inc.; The online brokered marketspace is also investigated. Using a continuous double auction as the matchmaking mechanism, a simulation of a dynamic brokered marketspace is created. This marketspace is used to investigate the effect of search costs and brokerage fees upon marketspace dynamics. A key finding is that when search costs are moderate, the online broker executes the most transactions, yields the highest sale prices, and can charge the highest brokerage fees.; To summarize, automated negotiation in electronic commerce is an emerging business phenomenon which is, in places, highly quantitative. Online auctions and online brokerages offer buyers, sellers, and middlemen many more degrees of freedom than do traditional channels, and hence new optimization problems arise. This dissertation applies the tools of operations research to help better understand and manage the online auction and the online broker. |
