Learning through Oil and Gas Exploration

I investigate the importance of learning in oil and gas exploration. I developed a tractable dynamic structural model of oil and gas exploration in which firms gradually learn about the productive qualities of different regions through exploratory drilling. Exploratory drilling is modelled as an information-gathering process in which each new exploratory well provides information concerning the profitability of drilling additional wells in a given area. The model is geographically based and accounts for the heterogeneity in the characteristics of oil and gas deposits that can exist across large regions. The model is estimated using proprietary data acquired from a number of different
sources which I organized in a geographic information system. The geographic information system contains records for over 400,000 wells drilled in the province of Alberta, Canada between 1930 and 2006. My estimates support the notion that spatial heterogeneity is important. I find considerable variation in the cost and revenue characteristics across different regions in Alberta. I
estimate a second structural model, one without learning, and show that the model with learning does a much better job at generating the observed geography of exploratory drilling. The broader implications of my model indicate that the structure of information has important effects on drilling
behaviour, and that these effects vary, depending on the specific characteristics of the regions being explored.