Does urbanization affect energy intensities across provinces in China?Long-run elasticities estimation using dynamic panels with heterogeneous slopes

Although there has been extensive debate in the literature that addresses the impact of urbanization on total energy use, the relative magnitude of each impact channel has not been empirically examined and urbanization's effects on energy transition dynamics in China remains unknown. Using panel datasets at the provincial level from 1986 to 2011, this paper employs dynamic models to investigate both the long-run and short-run elasticities of urbanization on energy intensities and the most significant impact channel is identified. Coal intensity and electricity intensity are also modeled to reveal energy transition dynamics driven by urbanization. A set of newly developed regression techniques, namely well-performed common correlated effects mean group (CCEMG) and augmented mean group (AMG) estimators, are used to treat residual cross-sectional dependence, nonstationaty residuals, and unlikely-to-hold homogeneous slope assumptions. The results obtained verify that the net effects of urbanization on overall energy intensity and electricity intensity are statistically positive, with long-run elasticities of 0.14% to 0.37% and 023% to 0.29%, respectively, whereas China's urbanization does not significantly increase coal intensity. The fact that short-run elasticities account for a majority of corresponding long-run values indicates that the short-run effect that is, indirect energy use induced by urban infrastructures is the most significant impact channel of urbanization on energy use in China. An energy transition from high-pollution coal to clean electricity is also present in China, although the fundamental transition to renewable energy is still in its infancy. From a regional perspective, urbanization exerts asymmetric impacts on provincial energy use so that energy policies associated with urbanization should be province-specific The findings also illustrate that for a panel dataset on regional dimension within large and fast-growing economies such as China, error cross-sectional dependence and residual nonstationarity must be tested and properly treated to avoid size distortion and biased estimators. (c) 2015 Elsevier B.V. All rights reserved.