U.S. petroleum consumption behavior and oil price uncertainty: Tests of cointegration and parameter instability

Abstract

This study utilizes cointegration theory to correctly characterize U.S. petroleum consumption behavior. Initial estimates show the absence of any long-run, unique relationship among petroleum consumption, real income, and relative prices. However, the introduction of oil price uncertainty into this relationship shows the presence of a cointegrating relationship. Oil price uncertainty was introduced in two ways, namely, as an exogenous I(1) variable and as a regressand. Estimates of the cointegrating relationship are obtained using a variety of techniques such as the Johansen system, the Phillips-Hansen, the Stock-Watson, the Park canonical cointegrating regression, the Phillips spectral, and the Engle-Granger test procedures. Parameter instability of the cointegrating relationship is tested using methods discussed in Hansen [1992] and Hansen and Johansen [1993]. While previous studies in this literature have yielded mixed results on the issue of cointegration and ignored tests for parameter instability of the cointegrated systems, this study has presented new evidence on an empirically stable petroleum oil demand function.