Might electricity consumption cause urbanization instead? Evidence from heterogeneous panel long-run causality tests

Abstract

The share of a population living in urban areas, or urbanization, is both an important demographic, socio-economic phenomenon and a popular explanatory variable in macro-level models of energy and electricity consumption and their resulting carbon emissions. Indeed, there is a substantial, growing subset of the global modeling literature that seeks to link urbanization with energy and electricity consumption, as well as with carbon emissions. This paper aims to inform both modelers and model consumers about the appropriateness of establishing such a link by examining the nature of long-run causality between electricity consumption and urbanization using heterogeneous panel methods and data from 105 countries spanning 1971–2009. In addition, the analysis of the time series properties of urbanization has implications both for modelers and for understanding the urbanization phenomenon. We consider total, industrial, and residential aggregations of electricity consumption per capita, three income-based panels, and three geography-based panels for non-OECD countries. The panel unit root, cointegration, and causality tests used account for cross-sectional dependence, nonstationarity, and heterogeneity – all of which are present in the data set. We cannot reject pervasively Granger causality in the urbanization to electricity consumption direction. However, the causality finding that is both the strongest and most similar across the various panels is that of long-run Granger causality from electricity consumption to urbanization. In other words, the employment and quality of life opportunities that access to electricity afford likely encourage migration to cities, and thus, cause urbanization. Also, nearly all countries’ urbanization series contained structural breaks, and the most recent post-break annual change rates suggested that nearly all countries’ rates of urbanization change were slowing. Lastly, future modeling work on energy consumption or carbon emissions should consider subnational scales of analysis, and focus on measures of urban density or urban form rather than national urbanization levels.