An empirical investigation of investment under uncertainty with sunk costs: Oil production in Oklahoma

We observe a very distinct pattern of entry and exit in the oil industry. Significant numbers of well openings and closings are only observed during periods of very high oil prices and very low oil prices, respectively. A band of inaction lies between the high entry and the low exit price. Recent economic theory finds that the combination of sunk costs and uncertainty of future market conditions leads to this pattern of hysteresis.; This dissertation develops an empirical model of firm investment that can measure the role of sunk costs and uncertainty on investment and production. In my dynamic discrete choice model firms make the decision to produce oil, mothball, or shutdown oil wells. Switching between the states production, mothballing, and shutdown involves non-recoverable cost. The firm's decision is affected by current and expected future market conditions, including the price of oil, oil price volatility, and the well's characteristics. The model, which allows me to test if sunk costs and price uncertainty are determinants of the investment choice, is used to explain the supply response of oil producers to fluctuations in the market price of oil.; The model is estimated using a recently developed panel-probit simulation estimator that allows me to control for several sources of correlation in the long well history data, including persistent time-invariant random effects and serial correlation. Using a new micro-level data set on oil production units in Oklahoma it is found that sunk costs are an important determinant of the production status. Although the probability of being a producer increases in the price of oil, mothballed or shut down leases only slowly respond to price changes because of non-recoverable switching costs.