Testing functional forms in energy modeling: An application of the Bayesian approach to U.S. electricity demand

In the estimation of demand functions for energy resources, linear, log-linear, translog, or AIDS functional forms are commonly assumed. It is frequently debated which is the “best” functional forms to employ in order to accurately represent the underlying relationships between the consumption of various energy resources and explanatory variables such as energy prices, weather variables, income, and other factors.In an example analysis of residential energy demand employing cross-sectional household-level data, we find that the AIDS model is slightly better than the translog model, which is superior to the log-linear model, and which in turn is better than the linear model. This is in sharp contrast to previous findings in this context.