| Although the concept of network externalities has been widely discussed in the literature, the empirical work in this area is still limited. The first part of this paper analyzes field data on adoption decisions of Internet browser software. Since the early 1995, Microsoft employed many strategies in order to diminish the popularity of Netscape's Navigator and Communicator. Based on the GVU WWW User Survey data in 1997 and 1998, the results from logit models imply that the success of Internet Explorer was driven by Microsoft's free distribution and bundling strategies, not by the product itself nor by network externalities.; The evidence from the theory and the empirical observations from the browser war shows the importance of predatory pricing strategies, although previous experiments that did not model network externalities did not find any evidence of predatory pricing. Chapter three and four report data from two experiments confirming the existence of frequent predatory pricing in the presence of network externalities. Subjects were recruited and assigned the role of either a superior seller (seller A) or an inferior seller (seller B) in a two period duopoly game. Sellers A and B were paired randomly, and had to make an entry decision in period 2, as well as a price and a quantity decision in both periods 1 and 2. Market share gained in period 1 determines the extent of network externalities in period 2. Apart from network externalities, buyers do not differentiate between the two goods. The observations overwhelmingly support the presence of predatory pricing. We find, however, that predatory pricing does not result in eliminating competitors.; The second experiment extends the model from the first one by allowing a longer time horizon. Sellers start out with perfectly identical demand and supply conditions and network externalities evolve over time. On the demand side, buyers have differentiated preferences over the two goods, but the ratio of the valuation is distributed uniformly and symmetrically around one. The data strongly supports the presence of predatory pricing, especially in later rounds. |
