What Goes Up Must Come Down? An Economic Analysis of Peak Oil∗

While world oil production and reserves have been increasing since 1859, ‘peak oil’ analysts claim that production is on the cusp of a period of sustained decline. I first subject the peak oil model to a number of robustness tests. The peak oil assumption of a linear relationship between the ratio of production to cumulative production and cumulative production is rejected using data from different regions, time periods, and commodities. The peak oil model predicts a single peak in discoveries, followed by peak in proved reserves, then a peak in production. Yet, world discoveries have had four distinct peaks since 1950, and world proved reserves have continued to rise. Second, I derive an economic model of oil supply in which scarcity occurs both in total reserves and in the quality of those reserves. Depletion raises drilling costs and reduces the size of undiscovered pools. Technological change driven by learning-by-doing offsets Ricardian depletion effects. The peak oil predictions are shown to be a special case of the more general model.