Thomas R. Covert
University of Chicago
“Learning to be productive and learning to produce in the North Dakota Shale Boom”
Abstract:The learning-by-doing literature shows that firms with more experience make more efficient choices about unobserved factors of production. That is, firms learn to be productive. In this paper, I argue that experience may also help firms learn to make improved choices about observed factors, so that firms learn how to produce. In administrative data documenting the use of hydraulic fracturing technology by firms in North Dakota’s shale oil boom, I find that these firms primarily learned how to produce. While productivity of the average well is stable across cohorts, firms made more efficient choices about observable inputs in later cohorts than they did in earlier cohorts. To determine whether this can be explained by learning, I measure the efficiency of input choices using production function estimates based on data about fracking technology that was available to firms when they made those choices. These ex ante measures of efficiency are more stable than ex post measures, and suggest that the continual arrival of publicly available data on fracking inputs and oil production helped firms learn how to optimize the fracking production function.