| This dissertation accomplishes three main tasks. First, it surveys the literature on innovation and diffusion, and proposes a new and more complete theoretical framework for the analysis of technological diffusion of process innovations, based on recent diffusion theory.;Second, it derives a dynamic model of adoption and strategic firm behavior from the neo-classical profit maximization problem. This model integrates four main theoretical streams of models of diffusion identified earlier. The dynamic behavior model developed here includes technological change, learning-by-doing, and lumpy capacity, and assumes closed-loop strategies and equilibria, leading to a dynamic model of strategic interaction which is richer and more appropriate than traditional models based on open-loop or precommitment strategies and equilibria. The results indicate that the optimal time of adoption depends on the specification of the fixed cost function, that firms in imperfectly competitive industries will build capacity which may remain idle for some period of time, and that in slowly growing markets, the incumbent will produce at full capacity while the learning curve is steep, and reduce the capacity utilization rate as the learning curve flattens. In rapidly growing markets, the incumbent may choose to build capacity which remains idle initially, and may set price below cost, so as to move down the learning curve faster. This is true even for a monopoly which does not face the threat of entry. Thus, predatory intent cannot be inferred from below-cost pricing.;Finally, this study provides an empirical test for the integrated approach to diffusion with data for 136 worldwide semiconductor manufacturing firms. The empirical results support the existence of vintage effects, specifically size and product-market effects, indicate the existence of strong information asymmetry effects due to differences in knowledge attributable to learning, suggest the existence of strategic behavior effects associated with the order of adoption, and provide evidence supporting some of the recent theoretical models of adoption. Despite the lack of evidence on imperfect foresight effects, this research supports an integrated approach to studies of adoption, and suggests that future empirical studies of adoption should seek to identify and model these four categories of effects. |
