Modelling Gasoline Demand in the United States: A Flexible Semiparametric Approach

Using the most recently available data, this paper estimates the price and income elasticities of gasoline demand in the United States from a semiparametric smooth coefficient model. The econometric approach provides more flexibility by allowing functional coefficients to accommodate heterogeneity in gasoline demand. Instrumental variables are used to correct for the potential endogeneity of the price of gasoline that has often been ignored in the literature. A formal model specification test rejects the parametric translog model, and suggests strong evidence of heterogeneous gasoline demand across states and over time. The results illustrate a significant income effect and price effect on gasoline demand elasticities. The dynamics of the gasoline price is the driving force behind the time variation of elasticities, and the demand for gasoline is more sensitive to unpredicted price shocks than gradual fluctuations. State level attributes, such as government expenditure on public transportation, contribute to the cross-state variation in gasoline demand elasticities. JEL Classification: C14, Q41, Q54.