Analysis of Oil Consumption with Dynamic Panel Data Models From a thesis supervised by Dr Chirok Han and Professor Viv Hall

This paper applies dynamic panel data techniques to a panel of 73 non-oil exporting countries in order to estimate price and income elasticities of oil consumption in three sectors. These sectors are: Transportation; the industrial sector; and “other ” sectors. We focus on the Han-Phillips (2006), Hsiao et al (2002) maximum likelihood and Arellano-Bond (1991) dynamic panel data. Our results illustrate the adverse consequences of high autoregressive coefficients on the Arellano-Bond estimator and provide an empirical comparison of the performance of the Han-Phillips and Hsiao et al estimators. We focus on a large panel over five intervals (1972–1980, 1981–1986, 1987–1990, 1981–1989 and 1991–2000). This approach differs from previous studies. Of particular interest, is our finding of autoregressive coefficients which are close to unity, especially in the transportation sector. These have not been found in previous research.