Industrial dynamics: An evolutionary model for an interactive simulation.

摘要

This dissertation presents a formal model for simulating the rise and fall of industries. The model is the foundation of an interactive simulation teaching management of innovation. Students compete in a product differentiated industry where bounded rationality necessitates problem framing. Holland's schemata, from his genetic algorithms, provides a formal representation. With the simulation a student learns the patterns of industrial dynamics; a firm's influence on these dynamics; that the unit of focus is not a product line but Schumpeter's creative destruction; and that exploiting new technology is a path-dependent, unpredictable process.;The model is evolutionary, dynamic, and has a functional role for management (thereby confronting two limitations of the Neoclassical theory of industrial organization). The model complements Heiner's diagnostic reliability condition; rather than hypothesizing a structure and adding uncertainty, uncertainty is hypothesized and structure is added. This contrasts with the predominant methodology of evolutionary economics, based upon Nelson and Winter's work, which instead of offering an alternative to optimization relies upon ad hoc modeling.;Uncertainty fosters behavioral constraint, thereby influencing entrepreneurship and competition. Structural influences--the nature of technology and consumer behavior--produce the topography of a product space searched by firms. Competitive search yields knowledge, as suggested by Hayek. A product space well correlated locally, uncorrelated over larger distances, and with diminishing returns to search (all characteristics of Kauffman's nk-landscapes) produces dynamics similar to the pattern of industrial development proposed by Abernathy and Utterback. Investigations of competition's effect on search predict punctuated dynamics; a shift from radical to incremental innovation as industries mature; and the poor response of established firms to new technologies. Most importantly, the model predicts a time horizon and its influence on the risks of missed opportunities and lost investment.;Heiner's reliability theory plays a fundamental role in the analysis. Furthermore, the model's combinatorial structure is compatible with Shannon's information theory. Its focus upon decision making, categorization, and covariation detection, permit application of cognitive psychology. All this knowledge can enrich economic theory and management training, thereby closing the gap between economic theory and management practice.