A century of interfuel substitution

We investigate the demand for energy and the degree of substitutability among fossil fuels in the United States using the Normalized Quadratic (NQ) expenditure function and (to our knowledge) the longest span prices and quantities that have ever been studied before, from 1919 to 2012. In doing so, we merge the empirical energy demand systems literature with the recent financial econometrics literature, relaxing the homoskedasticity assumption and instead assuming that the covariance matrix of the errors of the flexible demand system is time-varying. We generate inference, in terms of a full set of elasticities, consistent with neoclassical microeconomic theory and the data generating process. Our results are in line with earlier findings in the literature based on shorter duration samples and different methodologies. They have important implications for climate policy intervention designed to reduce greenhouse gas emissions. We show that there is a small but statistically significant substitution possibility between crude oil and natural gas, but not between crude oil and coal. We also provide evidence that natural gas is a substitute for coal when the price of coal changes, but coal is not a substitute for natural gas when the price of natural gas changes.