Exit, sunk costs and geographic diversification in selected United States manufacturing industries

The gradual development of establishment and firm-level micro data has led to a large body of empirical work in industrial organization. This dissertation extends this work by incorporating firm and market heterogeneity into the empirical analysis. Using the Census of Manufactures, we construct new datasets based on regional markets for a subset of manufacturing industries that offer some advantages over previous data sources. Our resulting measures of firm experience take into account both industry and geographic dimensions, and regional variation in cost and demand conditions more accurately controls for market level heterogeneity.; Chapter two estimates two empirical models of plant exit from an industry. Using data from seven regional manufacturing industries, we find that characteristics of the plant at the time of entry affect the decision of the plant to exit, even after controlling for differences in plant, firm, and market characteristics. The second empirical model allows different methods of exit from the industry. The findings suggest that experience at the time of entry not only affects the exit decision, but it also helps to determine the method of exit.; Chapter three examines the micro-level dynamics of the oil equipment industry during the period surrounding the rise in oil prices during the late 1970's and early 1980's. While industry output is virtually the same before and after the demand shock, the structure of the industry at the micro-level is quite different. According to our findings, large numbers of small producers enter the market during the boom period but persist in the industry during the contractionary period. Industry output becomes concentrated at large, relatively efficient producers. However, these micro-level changes have little effect on overall industry productivity.; Chapter four develops a model of participation in geographic markets and estimates the model using a random effects probit model for multi-unit firms in three regional food industries. The estimation controls for initial conditions, market factors, and observable firm characteristics. We find that sunk costs are an important factor in the decision of a firm to participate in a market, and these costs vary by the prior industry and geographic experience of the firm.