Design of the Los Angeles peak-load pricing experiment for electricity

Abstract

The statistical design and planned analysis of an experiment in peak-load pricing for residential customers served by the Los Angeles Department of Water and Power are described. A broad range of questions was studied including: is seasonal pricing more efficient than existing tariffs; are the efficiency gains from time-of-day pricing sufficient to cover the additional metering and billing costs; and what is the effect of the level and structure of electricity prices on residential patterns of electricity use. The experiment was designed so that demand curves, rather than analysis of variance models, could be estimated. A recently developed statistical design method, the Allocation Model, was used to select those tariffs that minimize the variance of the answers to the policy questions of peak-load pricing. A related method, the Finite Selection Model, guarantees that the subsample on each tariff is similar to those on other tariffs, and permits certain precision balance and robustness gains over simple random sampling. The design selected should yield estimates of consumer demand functions and responses to policy-relevant tariffs of considerable statistical significance. The results from using the Allocation Model indicate that particular parameters on price and other variables of interest will be estimated with a standardmore » error generally less than 10% of the standard error of the equation. The application of these methods and the associated design principles result in the statistical design for the study, and involves approximately 2100 customers in four types of electric power rate charges. The design is stratified by four intervals of preexperimental annual energy consumption (kwh) and three climate zones. Time-of-day tariffs are further differentiated according to whether peak prices do, or do not, apply on the weekend.« less