Regression analysis of historic oil prices: A basis for future mean reversion price scenarios

We propose price forecasting algorithms based on regression analysis of historic oil prices over 150 years (1861-2012). From 1986 onward daily market prices allow more detailed analyses of the principal crude oil benchmarks (West Texas Intermediate [WTI] and Brent). The mean reversion price for a given time period corresponds to the marginal cost of supply. When supply and demand are out of equilibrium, spot prices move in a bandwidth bound at the bottom by cash cost of supply and at the top by the concurrent price of demand destruction. Short-term elasticity of demand is 0.015 (highly inelastic), and long-term elasticity of supply changed from 0.99 (highly elastic) during 1965-1983 to 0.39 (less elastic) during 1984-2012. We derive functions for the long-term equilibrium price and expand them into scalable equilibrium price functions for forecasting future price scenarios if “business-as-usual” is assumed. We also consider how two hypothetical black swan events (“unknown unknowns”) may affect the mean equilibrium price.