Three essays on applied energy econometrics with policy implications

This thesis examines the implications of econometric estimation of energy demand in three separate empirical chapters. In particular; the issues addressed are: (i) the extent in which inappropriate modelling techniques could impact energy demand estimates, (ii) the relationship between energy demand estimates and carbon emissions and (iii) the relationship between the decomposition of derived energy input and carbon emissions. The research begins with the estimation of industrial energy demand across 29 European countries over the period 1995 2009 using both the generalised method of moments (GMM) and the dynamic multilevel model (DMM) that accounts for the hierarchical structure of the data used. The main results indicate that the long run income and price elasticities of the standard dynamic model, that is, the GMM, which does not account for the hierarchical structure of the data used, are overestimated. The second empirical chapter carries out an exploratory investigation on the impact of energy demand elasticities on carbon emissions across Chinese sectors. The study allows for a structural change by dividing the period under consideration into period before (1995 2001) and after (2002 2009) China s accession to WTO. This chapter estimates/demonstrates how to compute a range of elasticities by estimating a translog model, and then examines the impact of these elasticities on industrial carbon intensity. Findings suggest that there is a moderately negative relationship between energy substitution and carbon emissions, more especially after the structural change. The third chapter combines the first two chapters into a single study by adopting a two-stage procedure to measure the implications of inappropriate energy modelling technique/energy demand estimates on carbon emissions. The study is based on industry level data across Europe over the period 1995 2007. Firstly, the study decomposed energy estimates into substitution and output effects with a multilevel model and iterated seemingly unrelated regression (iSUR). The second stage examines the impact of the decomposition effects with other competing forces on carbon emissions. Findings reveal that the substitution effect dominates the output effect and is inversely related to the carbon emissions. For the output effect, the results derived from both techniques differ, as the output effect from the iSUR show a positive sign; however, the output effects from the multilevel model show a negative relationship with carbon emissions, which is more consistent with the ideal practice of a cost minimising firm.