Characterizing demand uncertainty in high-tech environments: Models, analysis, and case studies.

摘要

This dissertation studies methods that characterize demand uncertainty in rapidly changing high-tech environments, and is organized as a set research papers and case studies.We first describe a demand characterization methodology that combines information provided by life-cycle growth models and demand leading indicators. Using a Bayesian framework, we propose a method to improve forecast accuracy while reducing the variation in model projections. We show theoretically that the reduction in forecast variance can be achieved with probability one, and demonstrate empirically that the improvement is consistent. In separate chapters, we present two field studies conducted at the semiconductor industry. The first case study focuses on the leading-indicator analysis and its implications in different planning functions. The second case study describes full implementation of the integrated approach at a major semiconductor firm and illustrates its performance in reducing forecast variation.As a separate theoretical study, we analyze the effect of demand uncertainty on capacity procurement decisions using a game-theoretic model. We discuss the trading firms' incentives and procurement decisions under different market characteristics and decision-making structures, and investigate coordination mechanisms that improve the efficiency of the supply chain. We show that the buyer behaves opportunistically in communicating the demand forecast information, and does not have incentive to choose the optimal forecast-improving activities. This strategic dimension of the analysis sheds some light on the interactions among players concerning the forecast sharing.Demand uncertainty has important implications to essentially all aspects of operational decision-making. This dissertation contributes to the theory and practice of operational planning by proposing a demand characterization methodology that focuses on improving the variability in demand as well as its accuracy, and by analyzing the impact of demand uncertainty on the operational decision-making activities of the supply chain.