The degree of endemic volatility in the number of firms and establishments varies considerably across industries. Examining the within-industry range of variation (max.-min.) of the number of firms over our sample period, the low and high values across U.S. manufacturing industries are 4 and 3,500 firms respectively, with a mean value of about 324 firms. This reveals that (1) the typical industry experiences significant fluctuations in the number of firms and (2) there are large cross-industry differences in this dimension. Theory suggests several potential factors that might explain this dispersion of firm volatility across industries: for example, sunk capital costs, uncertainty about profits and technological change. An advantage of the manufacturing industry dataset we have assembled for this study is that it combines the annual time-series data from the Annual Survey of Manufactures with data from the five-yearly Census of Manufactures. The former allows us to measure uncertainty about profits and technological change, while the latter enables us to obtain information on the industry-specific size distribution of establishments, the number of establishments per firm and construct proxies for sunk capital costs. Our empirical findings show that: (1) industries with higher sunk capital costs and profit uncertainty have significantly lower variability of the number of firms; and (2) these relationships are non-linear as suggested by theory with initial increases in sunk costs or uncertainty having relatively greater effect on firm volatility. The effects of technological change appear to be mixed. We explore the implications of our findings for antitrust analysis.
firm and establishment volatility sunk capital costs profit uncertainty technological change antitrust