The Impact of Dynamic Pricing on Residential and Small Commercial and Industrial Usage: New Experimental Evidence from Connecticut.

AuthorFaruqui, Ahmad
  1. INTRODUCTION

    1.1 Overview of the Issues

    Electricity cannot be stored economically in large quantities, and has to be consumed instantly on demand. The load duration curve for most utility systems is very peaky, with some eight to eighteen percent of annual peak load being concentrated in the top one percent of the hours of the year. These two factors, taken in conjunction with the time-variation in marginal energy and capacity costs that characterizes different generation technologies, mean that the optimal way for pricing electricity is to institute time-varying rates. (1) Not only would this increase economic efficiency, it would also eliminate inter-customer cross-subsidies that are embedded in flat rates. (2) Of course, dynamic pricing can only be carried out once smart meters are in place. As of this writing, about a quarter of U.S. households are on smart meters and the number is projected to rise by the end of the decade to nearly a hundred percent. However, only one percent of the households are on any type of time-varying rate and only one percent of that one percent are on any form of dynamic pricing rate (Federal Energy Regulatory Commission (2012)). Commissions and utilities continue to study the potential rollout of dynamic pricing. Well-designed experiments in which customers are randomly placed on different rates provide an important avenue for gaining insights into the likely impact of those rates. (3)

    1.2 A New Experiment in the New England Region

    The extant literature discusses experiments with dynamic pricing that have been carried out in Australia, Europe, North America and New Zealand during the past decade. However, most of them have been located in regions with hot and humid summers such as the District of Columbia, Florida, Illinois, Maryland, Michigan and Oklahoma. (4)

    It is uncertain whether the results observed in these pilots would apply to regions with milder climates, such as New England, where the saturation of central air conditioning (CAC) systems is under 30 percent. An earlier experiment, carried out in California in the 2003-04 time frame, found that customer response in the mildest climate zone (the coastal regions and mountains) was less than half the size of response in the strongest climate zone (the central valley) (Charles River Associates (2005)). (5)

    This paper presents an impact evaluation of a dynamic pricing pilot that was carried out in New England by the Connecticut Light & Power Company (CL&P). Called the Plan-It Wise Energy Pilot (PWEP), it was designed to test if time-varying pricing could lower future power costs by curtailing peak demands during critical periods or by shifting them to other periods.

    Although other dynamic pricing pilots had published their findings prior to the execution of the PWEP, they had been carried out in different geographies and it was not clear whether the results from these pilots would be transferable to New England, given differences in socio-demo-graphic and climatic conditions (Faruqui and Sergici (2009) and Faruqui and Sergici (2011)). The PWEP was intended to provide results that would support the execution of a cost-benefit analysis of advanced metering infrastructure (AMI).

    Unlike other pilots, which only included residential customers, the PWEP also included small C&I customers. Around 2,200 customers were included in the experiment, equally divided between the residential and small C&I classes. The pilot featured three rate designs: critical-peak pricing (CPP), peak-time rebates (PTR) and standard time-of-use (TOU) rates. Low and high values of each rate design were included in PWEP to allow precise estimation of price elasticities. Each variant was designed to be revenue neutral for the class as a whole relative to the existing tariffs. The time-varying rates were also tested with and without enabling technologies. Four types of technologies were considered in the PWEP: In-Home Displays which show how much electricity is being used at different times of day and the associated cost, the Energy Orb which changes color as prices change, a Smart Thermostat that raises the temperature setting as prices rise and a switch to cycle the compressor unit of central air conditioning systems during critical peak hours.

    The pilot ran from June 1, 2009 through September 30, 2009. Ten critical peak days were called during June, July and August. Hourly usage was recorded for both the treatment and the control customers during the pilot period to determine if the treatment group used less electricity during the more expensive periods. In addition, to assess for any pre-existing difference in the groups, hourly usage was also recorded during a pre-pilot phase. Econometrically, a difference-in-differences estimation procedure was applied to an unbalanced panel for estimating the treatment effects.

    The PWEP, formulated as a scientific experiment, was designed to test five major hypotheses: (1) Do customers exhibit similar price responsiveness (as measured by elasticities of substitution) to the CPP, PTR and TOU tariffs? (2) Are the enabling technologies employed in the pilot effective in increasing customers' price responsiveness? (3) Does dynamic pricing elicit lower response in a mild climate compared to a warmer climate? (6) (4) Do customers respond to longer peak windows? And (5) Do the residential and small C&I customers respond differently to price signals?

    Section 2 of this paper describes the experimental design of the PWEP. Section 3 summarizes the analytical methods and data used in the estimation of the load impacts. Section 4 reports on the empirical findings and Section 5 concludes the paper.

  2. PWEP EXPERIMENTAL DESIGN

    2.1 Rate Design

    CL&P's standard rate is a flat, seasonal, volumetric rate that includes a fixed customer charge. During the PWEP period, the control group customers paid the standard rate which, on an all-in basis, amounts to $0.201/kWh for residential customers and $0.203/kWh for small C&I customers. These rates applied to all customers in those classes, regardless of their load profile.

    The treatment customers were placed on one of the three following rate designs which included low and high rate variations. Under the Critical Peak Pricing (CPP) rate design, the hours between 2 pm through 6 pm on non-holiday weekdays were designated as the peak period and were priced between $0.17/kWh and $0.19/kWh for residential customers and between $0.15/kWh and $0.19/kWh for small C&I customers. On the ten critical peak days that were called on a day-ahead basis, the peak hours would become the critical peak hours and be priced between $0.86/kWh and $1.82/kWh for residential customers and between $0.86/kWh and $1.80/kWh for small C&I customers. On non-critical weekdays and weekends, the treatment customers faced an off-peak price between $0.17/kWh and $0.19/kWh for residential customers and between $0.15/kWh and $0.18/kWh for small C&I customers, respectively. The rates were designed so that customers whose load profiles corresponded to the load profile of their class would see no change in their bills, in the absence of load shifting. Thus the off-peak price was lower than the standard tariff.

    Under the Peak Time Rebate (PTR) rate design, the PWEP participants were still subject to the standard CL&P rates. However, on the ten critical peak days, between the hours of 2 pm and 6 pm, they had the opportunity to receive a rebate between $0.78/kWh ($0.86 all-in rate) and $1.74/kWh ($1.82 all-in rate) for residential customers and between $0.78/kWh ($0.86 all-in rate) and $1.73/kWh ($0.80 all-in rate) for small C&I customers, if they reduced their consumption below their typical usage during these hours.

    Finally, under the Time-of-Use (TOU) rate design, the hours between 12 pm through 8 pm on non-holiday weekdays and on critical days were designated as the peak period and were priced, for both the residential customers and small C&I customers, between $0.27/kWh and $0.34/kWh. All the remaining hours were designated as the off-peak period and priced between $0.14/kWh and $0.17/kWh. Additional details on these rates are presented in the Appendix 1.

    2.2 Technology

    The PWEP program also tested the effectiveness of enabling technologies in facilitating demand response when offered in conjunction with dynamic rates. In order to distinguish the impacts of enabling technologies from that of prices alone, each rate design was tested with and without enabling technologies.

    The PWEP involved four types of technologies: In-Home Displays, Energy Orb, Smart Thermostat and a Control Switch to cycle the CAC compressor. The In-Home Display provided real-time electricity usage and cost information. This was intended to enable customers to lower peak usage and/or shift it to off-peak hours. The Energy Orb, a small sphere, emitted different colors to notify participants of changes in electricity prices. The Smart Thermostat allowed CL&P to adjust the "normal" central air conditioner temperature setting during peak demand periods. And the Control Switch, placed on the compressor of the CAC, enabled CL&P to cycle the compressor of the central air conditioner during peak hours. Of course, the smart thermostat and the control switch required the customer to have a central air conditioner and were not applicable to those customers who did not have central air conditioners. A combination of three time-varying rate designs and four different technologies yielded a rich tableau of 44 treatment cells.

    2.3 Sample Design

    The PWEP featured 1,251 residential customers of which 1,114 customers were the program participants and constituted the treatment group while 137 customers constituted the control group. The pilot also featured 1,186 small C&I customers which 1,123 participants and 63 participants made up the treatment and the control groups, respectively. CL&P identified a random sample of customers that represent the residential and...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT