The purpose of the proposed research is to explore the feasibility of an economic and statistical approach to the problem of forecasting the conditional means of variables used to predict energy demand. Refinements in the modeling and forecasting of energy demand will result in better forecasts which can be used by utility companies and government agencies to more effectively plan for the future. The proposed technique recognizes the necessity of using conditional means (based on a subset of customers) of relevant variables and it takes into account changing distributions of these variables. The first step in the approach is to econometrically estimate probability distributions of appliance ownership, conditional on certain explanatory economic or demographic variables such as income. Statistical relationships are then used to derive the probability distribution of the explanatory variables, given ownership of an appliance such as air-conditioning or poolheating. The conditional means of the explanatory variables can then be calculated on the basis of this conditional probability distribution. Phase f of the research will further develop the theoretical framework for estimating conditional means and will apply the technique to energy usage data from Southern California Gas Company.The potential commercial application as described by the awardee: The ultimate goal is a computer based forecasting model which will aid utilities in more accurately forecasting the demand for energy. The forecasting package could also be used at both a state and national level as an aid in formulating energy policies to deal with future energy needs of the country. Applied Econometrics, Inc., is a consulting firm specializing in statistical economics.
