Forecasting demand for electricity in Maryland: An econometric approach

Most of the models for forecasting demand for energy are based on simple extrapolations of past trends or on a simple regression equation with price of the energy and the stock of appliances as explanatory variables. In this paper, an attempt has been made to derive static and dynamic multiple regression equations from economic theory of consumption and production (Section II). Historical data were fitted to these theoretical constructs to test the equations in terms of econometric theory and forecast the demand according to “higher order conditional interval forecasts”. The residential demand for electricity is a function of its price, price of its substitute, per capita income and a lagged demand variable for dynamic adjustment of actual demand to equilibrium demand for electricity. The forecasts of residential demand to 1990 are based on projections of exogenous variables such as residential price of electricity, per capita income and the estimated long run elasticity of demand (Section III). The nonresidential demand for electricity is a function of employment in that sector, sectoral prices of electricity and the lagged sectoral demand. The forecasts of nonresidential electricity demand are also based on projections of its independent variables (Section IV). The last section converts the total demand for electricity into the required generating capacities and juxtaposes them against the estimates of expected supplies available from the forecasts of the utilities. The paper concludes that the eighties will be faced with excess supply of electricity in Maryland, in case the assumptions of projections of independent variables hold good. The misallocation of resources inherent in such excess supplies could be avoided if realistic scenerios of future demand, as attempted in this paper, could be predicted.