THE RESIDENTIAL DEMAND FOR ELECTRICITY BY TIME-OF-DAY

Abstract

The use of time-of-day (TOD) pricing as a load management tool for electric utilities has recently gained wide interest. Although utilities have successfully used TOD pricing for some industrial customers, its applicability in the residential sector is untested. The Federal Energy Administration (and now the Department of Energy) has funded several experiments to test the implications of TOD pricing for residential customers. The major objective of this study is to analyze the Arizona TOD pricing experiment. Data from the first six months of the experiment had been analyzed previously in several different studies. Summaries of their methodologies and results are presented in Chapter 1. Many of these earlier analyses were unable to identify significant TOD price responses. A major deficiency in all was their failure to account properly for participation incentive payments. Consequently, meaningful inferences regarding residential responses to TOD prices cannot be drawn from these misspecified models. Chapter 2 contains a description of the experiment and the data is generated. The basic observation is monthly kilowatt hours consumed by each household in three time periods. Special attention is given to the derivation of the incentive payment inherent in the experimental design. This payment depends on experimental rates and patterns of pre-experimental usage. Specific adjustments to the data are required due to variations in billing cycle lengths and days. Details of these procedures and information on how the samples were edited are discussed in Chapter 3. The conclusion of this chapter presents data which reveal that households did significantly shift consumption from high to low cost periods. Chapter 4 contains a description of the hypothesized models and statistical methodology. Since this study focuses on household responses to TOD prices while controlling for impacts of experimental design, theoretically derived models are not tested. Income, TOD prices, heating or cooling degree-days, the electricity-using capacity of the households' appliance stocks, and incentive payments are the major determinants of consumption investigated. Ordinary least squares techniques are used to estimate TOD demand models for each month, for the summers of 1976 and 1977, and for the winter of 1976/77. Since the experimental design was modified in May, 1977, an analysis of covariance was done to test for structural changes. The results presented in Chapter 5 emphasize the importance of including the incentive payment in the TOD models. Without this term, no TOD price is significant. With it, TOD prices and the other independent variables are shown to be significant determinants of consumption. Statistical results are very impressive for the models estimated from the 18 months of cross-sectional data. Since the incentive payments depended partially on the rates to which customers were assigned, calculation of price elasticities had to be modified accordingly. Simple elasticities measured price responses which ignored the impact of the incentive payment. Since the incentive did not depend on experimental usage, it is the appropriate measure of household responses to TOD prices. Total price elasticities are used to measure TOD price responses under the specific Arizona experimental environment. A number of important conclusions are discussed in Chapter 6. The most important deal with the treatment of the incentive payment. When it is properly modeled, meaningful price coefficients can be estimated. Also, the results strongly suggest that households earmarked this payment for electricity purchases. Partial derivatives of the incentive were much larger than those for income. Misleading billing information may have produced this unexpected result. In May, 1978, billing procedures were improved. An analysis of these data should shed more light on this important matter.