Correcting for Truncation Bias in the Analysis of Experiments in Time-of- Day Pricing of Electricity

This paper applies methods for analyzing data from samples that have been chosen by restricting the target population in discernible ways to date from a 1976 experiment in time-of-day pricing of electricity for residential customers in Arizona. We find that whereas conventional estimation methods lead to the conclusion that the peak price elasticity of demand is larger (in absolute value) than either the corresponding midpeak or offpeak elasticity, once truncation bias is accounted for, the peak elasticity is smaller than the other two. This finding accords with a preliminary analysis of data from Wisconsin, where no such sample truncation was present.