| Many industries experience periods of excess capacity. Explanations include firms deterring rivals, anticipating future demand, or the lumpiness of investment. Chapter 1 develops an approach that identifies capacity investment intended to deter rivals separately from investment driven by other factors. First, I estimate a new model of spatial price competition and embed it in a dynamic game of investment. Then, I isolate investment intended to deter by comparing the paths of investment under two distinct equilibrium concepts: one which allows firms to deter rivals, and one which does not. I apply my model to the United States portland cement industry from 1949 to 1969, a period of considerable investment and growth, and find that deterrence explains almost all of the industry's excess capacity.;Chapter 2 draws on the pricing model of Chapter 1 to develop a test for collusion in the cement industry. We address an ongoing debate over whether anti-competitive provisions of the 'Codes of Fair Conduct,' developed under the National Industrial Recovery Act (NIRA) in 1933, had any effect. We use a combination of narrative evidence and a novel plant-level dataset from 1929, 1931, 1933, and 1935 to study how the NIRA affected the cement industry. We find strong evidence that before the NIRA, the costs of a plant's nearest neighbor had a positive effect on a plant's own price, suggesting competition. After the NIRA, this effect is completely eliminated, with no correlation between a plant's price and its neighbor's cost.;In Chapter 3, we suggest one possible coordinating mechanism for firms colluding during the time of the Great Depression. We find that multi-market contact fosters tacit collusion and higher prices based on a new measure of contact that accounts for capacity constraints. A one standard deviation increase in our measure of contact increases prices by around 4.3% across the years in our dataset. The finding is amplified when allowing for a separate effect of the NIRA, and we conclude that the effects of the codes were most strongly felt in markets with the highest level of multi-market contact. |
