Interfuel Substitution: A Meta-Analysis

Interfuel substitutability has been of longstanding interest to the energy economics and policy community. However, no quantitative meta-analysis has yet been carried out of this literature. This paper fills this gap by analyzing a broad sample of studies of interfuel substitution in the industrial sector, manufacturing industry or subindustries, or macro-economy of a variety of both developed and developing economies. Publication bias is controlled for by including the primary study sample size in the meta-regression. Results for the shadow elasticities of substitution between coal, oil, gas, and electricity for forty-six primary studies show that at the level of the industrial sector there are easy substitution possibilities between all the fuel pairs with the exception of gas-electricity and coal-electricity. Substitution possibilities seem more constrained at the macro level and less constrained in sub-industries. Estimates also vary across countries and there seems to be less substitutability in more developed economies. Publication bias does not seem to be present, but model and data specification issues very significantly affect the estimates derived by each individual study. Estimates from cross-section regressions are generally largest, fixed effects panel estimates intermediate in magnitude, and time-series estimates are mostly much smaller. Econometric research suggests that the fixed effects estimates are likely the best among the existing studies, though biased downwards.