Price Response in Connecticut Light & Power's Variable Peak Pricing Option

Abstract

In 2007, the Connecticut General Assembly passed Public Act No. 07-242, An Act Concerning Electricity and Energy Efficiency, requiring electric distribution companies to offer voluntary critical peak pricing or real-time pricing to all customer classes. In 2008, Connecticut Light & Power implemented Variable Peak Pricing (VPP), a rate option that allows time-of-use rate customers to purchase on-peak electricity at daily-varying prices linked to the clearing of New England’s day-ahead wholesale electricity market. This study uses regression analysis to estimate the hourly price elasticity of demand for electricity of the 24 residential VPP customers. Using average hourly load data from 2008 to 2011, hourly electricity demand is modeled as a function of price, temperature, relative humidity, and hourly, weekday and household effects. A modified difference in differences estimation technique is also employed to in an attempt to make causal inference about the “treatment effect” of variable on-peak prices using a larger control group of residential time-of-use rate customers. The fully-specified demand models for the VPP subset indicate that consumers increase their hourly demand when prices increase; in contrast, the difference in differences technique produces a small but statistically significant hourly price elasticity of demand of -0.1. In other words, a 10% increase in hourly price results in an approximately 1% reduction in hourly consumption. This estimate compares favorably with the larger control group, with an estimated price elasticity of demand between -.05 and -0.1. Low observed price elasticities of demand comport with previous research on dynamic pricing pilots indicating that large on- and off-peak price differentials are necessary to encourage load shifting. Causal inference is rendered difficult by the non-experimental nature of this study and the small number of VPP customers. To the extent that the estimates are biased, the results are not generalizable to the residential customer class as a whole, given the voluntary nature of the program.