| Network effects exist when there are benefits to aligning one's behaviour with the behaviour of others. There is a large literature on network effects, as issues such as technology adoption, fads, and many others revolve around network effects. Typically, such models are specified in a manner so that multiple purely coordinated and Pareto efficient equilibria exist, which introduces an equilibrium selection problem. Strangely, this selection problem has largely been ignored, and little effort has been made to examine how agents form expectations supporting coordinated outcomes. This thesis attempts to fill this gap by introducing a generalized, dynamic network effect model in which agents make their decisions sequentially and use their observations of previous decisions to form expectations about future decisions. Several results are proven regarding the likelihood and extent of coordination, and numerical examples are provided to complement the formal theory. The results show that even under strong network effects, purely coordinated outcomes are unlikely to occur, and some coordinated outcomes may actually be impossible, even under nontrivial model specifications. |
