Dynamic behavior of the U.S. oil industry

Abstract

This paper considers two alternative models of firm dynamics and uses a nonparametric test to evaluate their empirical relevance to the U.S. oil industry. I am interested in the class of models that control for the selection induced by firm liquidation and that explain how similar firms operating under the same conditions can exhibit very different activity paths. Such variability is generally associated with a source of uncertainty that is specific to the firm, generating idiosyncratic differences in behavior over time. In particular, I consider Jovanovic's (1982) [Jovanovic, B., 1982. Selection and evolution of the industry. Econometrica 50: 649–670.] Bayesian learning model and Ericson and Pakes' (1995) [Ericson, R., Pakes, A., 1995. Markov perfect industry dynamics: a framework for empirical work. Review of Economic Studies, 63: 53–82.] active exploration model. I found that the data on a sample of U.S. oil firms is consistent with the Bayesian learning model.