Is timing everything? Assessing the evidence on whether energy/electricity demand elasticities are time-varying

In this review we assess both the reasons/hypotheses for why GDP and price elasticities of energy/electricity demand might change over time, and the evidence from papers that have addressed that issue. We consider papers that have used parametric methods (e.g., endogenous or exogenous breaks, rolling window regressions) and semi- or non-parametric approaches (e.g., structural time series models, local linear estimators). The most convincing reasons for GDP and price elasticities varying over time were: (i) results from reactions to the energy price shocks of the 1970s and early-1980s (that would have affected OECD countries directly and non-OECD countries, possibly, via leapfrogging); (ii) saturation of the demand for energy services (typically at high income levels); and (iii) country-specific events such as policy changes and/or economic crises/recoveries/shocks/transitions. Despite those reasons, the weight of evidence points toward income/GDP and price elasticities of energy/electricity demand that are fairly stable over time. In addition to changes caused by reactions to the energy crises and (other) country specific events, the main exception to effectively constant elasticities is the GDP elasticity of economy-wide electricity demand in OECD countries. This elasticity may have declined substantially over time, perhaps, because the share of energy services provided by electricity has saturated. Lastly, it is recommended that modelers consider a combination of (parametric and nonparametric) approaches and search for consensus among the results.