Inter-Fuel Substitution Possibilities: A Translog Application to Intercountry Data

One of the pressing issues in current energy policy debates in both the U. S. and other consuming countries is the feasibility of substantially reducing the use of crude oil. In some energy consuming sectors suchl as transportation, the shortrun mechanism to reduce consumption is through lower utilization of energyconsuming equipment while in the long-run, the equipment can be redesigned to achieve greater fuel efficiencies. In electricity generation and heatilng uses the primary mechanism to reduce consumption is inter-fuel substitution rather than capital/energy substitution. This paper is concernied with estimating the interfuel substitution relationships between fossil fuels (coal, gas, and residual fuel oil) in the generation of electricity. It is important to consider fossil fuel substitution possibilities in electricity generation both because considerable technical flexibility exists2 and because electric utilities are large consumers of fossil fuels. For example, electric utilities consumed 25% of primary energy inputs in the U. S. in 1970. Although not as high as the U. S., the other developed countries exhibit a similar pattern. In the 20 0. E. C. D. coun-tries, electric utilities consumed 50% of the coal, 17% of the gas, and 11% of petroleum products.3 High substitution elasticities among fuels in electricity generation indirectly increase subsequenit substitution of electricity for petroleum as a fuel source in residential, commercial, and industrial sectors. Thus the nature and extent of inter-fuel substitution possibilities in electricity generation are fundamental to the question of reducing dependency on crude oil. There have been several econometric studies of inter-fuel substitution in electricity generation using a variety of estimation procedures and samples. MacAvoy [15] used cross sectional data for U. S. power regions to examine substitution between nuclear fuel and fossil fuel. Griffin [10] estimated dynamic price and cross price elasticities among fuels utilizing annual time series data. More recently, Joskow and Mishkin [13] used individual plant data to analyze fuel choice in a logit model framework. But probably the approach capturing the most attention is the application of