| The dissertation examines the dynamics of firm behavior using computational methods, with the goal of modeling mergers more realistically than previous studies in order to examine policy implications. In this model, mergers, investment, entry and exit are endogenous choice variables rationally chosen by firms in the industry in order to maximize expected future profits. The model differs from previous studies in two ways: first, it directly specifies a process through which firms can merge and predicts which mergers will occur at different states; second, it incorporates dynamic determinants of firm behavior (namely entry, exit and investment). In Chapter 1, I discuss criteria for designing an endogenous merger process and I detail the processes chosen and corresponding results. The results show that by not incorporating the effects of investment and entry on the merger behavior of firms, the static literature is overpredicting the probability of mergers; this type of result has not been shown before since it can only be shown with an endogenous model. In Chapter 2, I examine antitrust policy implications using the dynamic model of endogenous mergers and compare these to the planner and colluder solutions. The results show first that a wide range of intermediate antitrust policies leads to a higher total welfare than either a complete merger prohibition or no antitrust law. In addition, they show that with production cost differences, merger synergies and barriers to entry, the implications of different antitrust policies are surprising and very different than those predicted by static non-endogenous merger models. Finally, the model uses its rational framework to explain the puzzle, discussed in the empirical literature, of why acquired firms seem to capture all of the gains from mergers. In Chapter 3, I discuss the issues surrounding the design and computation of the model. In particular, I examine when equilibria to the merger processes are likely to exist, the choice of specifications necessary to allow for computation, and the computational details of the processes. This discussion is of interest in order to further develop computational merger models and to design computational models of other oligopolistic games. |
