Measuring Peak Load Pricing Response From Experimental Data: An Exploratory Analysis

Abstract

Beginning in March 1979 and continuing for 24 months, approximately 480 Southern California Edison (SCE) residential customers faced time-of-use (TOU) rates for electricity. These sample households were spread over three temperature zones and five consumption levels, and faced two definitions of the peak (off-peak) period and four peak/off-peak price ratios. To supply a reference case for subsequent comparisons to SCE's current domestic rate, a statistical control group of 120 customers facing standard rates was also selected. The two main rate groups in the experiment, Rate A, whose peak period definition is 10 A.M.−8 P.M. PST nonholiday weekdays, and Rate B, whose peak period definition is 12 P.M.−10 P.M. PST nonholiday weekdays, are always treated separately. In this article we use the 18 months of data on peak period, off-peak period (including weekends), and total consumption spanning May 1979–October 1980, thereby encompassing two complete summers and one complete winter. We pool the monthly data in seasonal ANCOVA models but allow the months to have separate additive influences. Weather effects are accounted for explicitly. We conclude that there are significant conservation effects in the peak period as a result of TOU pricing. This result holds for the overall sample and in particular for high consumption groups (410–740 kWh/mo.) and (>740 kWh/mo.). Some shifting of usage to the off-peak period does take place, but frequently this amount does not equal the peak period reduction, and hence overall conservation takes place. This conservation effect, when observed, is limited to roughly 5% in the overall sample, while in some cases peak period reductions in the summer range up to 19.2%. Increased interest in “direct” load management techniques such as remote air conditioner cycling and demand subscription service led us also to address the question of the impact of TOU pricing on system peak days. If customers choose to ignore even significant peak/off-peak price differentials on system peak days though they may be few in number, the hoped for capacity savings from peak-load pricing may not materialize. The main findings are that contrary to the skepticism often expressed, TOU customers behave in a similar fashion on system peak days and on average summer weekdays. Indeed, there are several instances of even higher relative peak period savings on system peak days.