In this paper, a U.S. per capita gasoline demand function is estimated using a stochasticexogenous trend model with time varying parameters (TVP). Structural Time SeriesModelling with TVP is employed for annual data over the period 1949-2008 allowing forboth asymmetric price responses (for technical progress to impact endogenously) and anunderlying energy demand trend (UEDT) for gasoline (for technical progress and otherfactors both in a linear and non linear way to impact exogenously). As far as is known this isthe first attempt to incorporate a stochastic UEDT and asymmetric price responses within aTVP framework. The results suggest that income, price max and price recovery and theUEDT are important drivers of U.S. gasoline demand. Furthermore, this study illustrates that,although income and price elasticities fluctuate over time, these differences are notsignificant (when reduced to 2 decimal places they become identical). The study suggests thatdifferent price movements have different effects on U.S. gasoline demand; historicalmaximum prices have a greater impact than price recoveries. The estimated income, pricemax, price recovery, and price cut elasticities are around 0.42, -0.31, -0.17, and 0respectively. Moreover this study, estimates the stochastic UEDT, which has an importantimpact on gasoline consumption, and advocates the necessity of considering the UEDT formeasuring the elasticities. Moreover, this paper provides valuable information such as priceand income elasticities and the UEDT, for the policy makers both for evaluating currentpolicies and for implementing new ones.
展开▼
