This study uses a random coefficient estimation procedure to analyze the U.S. gasoline market from 1960-1995 with three main objectives: (1) provide an empirical methodology that can estimate a gasoline demand function capable of performing well in prediction; (2) evaluate the elasticities of the models presented to determine which model is more accurate at capturing supply shocks that impacted gasoline demand; and (3) evaluate the behavior of the elasticites of the beta coefficients.

This research will show that the variation from historical economic patterns was a result of supply shocks. I argue that when the OLS model of the gasoline market developed by William H. Greene is used supply shocks are not well captured because the coefficients are fixed. If the random coefficient model developed by P.A.V.B. Swamy is introduced, the coefficients vary over time, and thereby, enable supply shocks to be included in the model and more accurate forecasts are produced, as well as, meaningful time patterns in the beta coefficients.