Articles
Predicting California Demand Response
How do customers react to hourly prices?
Public Utilities Fortnightly, July 1, 2003 by Chris King and Sanjoy Chatterjee
As California embarks on a Statewide Pricing Pilot (SPP) for residential and small commercial (200 kW) customers, policymakers and participants in the proceedings are asking several questions:
- What elasticity estimates should we expect the pilot to produce?
- Will a voluntary SPP program produce less load reduction than a mandatory program?
- What is the likely response of the participants, and will that response differ as a function of usage level, appliance holdings, or other non-price factors?
The answers to those questions can be anticipated somewhat by reviewing the existing literature on price responsiveness of the residential and small commercial customer segment. (1) The data, summarized in this article, is based on experiments and studies run by the Department of Energy, academic researchers, and utility companies around the world during the past three decades.
Standard economic theory holds that customers react to changes in prices by adjusting their demand for the goods in question. As prices rise, customers reduce the quantity demanded. As prices drop, customers increase the quantity demanded. The responsiveness of customers to price changes is called their price elasticity of demand.
One measurement of elasticity is the customer’s change in consumption in the same time period in which the price change occurs. Another measurement is the customer’s shift in consumption across time periods—such as peak to off-peak—in response to price changes that alter the price relationship between the two time periods (for example, changing the price ratio from 1:1 to 2:1). These two measurements are called the own price elasticity and elasticity of substitution, respectively (see “Flavors of Elasticity of Demand: A Reference,” p. 32).
Elasticity Estimates
What elasticity estimates could we expect the California pilot to produce?
In 1984, Jan Acton and Rolla Park reviewed 34 published studies of residential electricity use drawn from North America. Acton and Park estimated overall “short-run” and “longrun” elasticities with respect to electricity. Short-run is a period in which consumers make no changes in appliance holdings to respond to price changes, while long-run is long enough to make changes in appliance holdings. Acton and Park summarized their findings as shown in Table 1, with “low” and “high” bracketing the 80 percent confidence band.
To update these results, we reviewed the results of 56 papers published since 1980 (see Figure 1 and Table 2). These studies expanded on earlier studies by utilizing additional methodologies for analyzing price response and by examining the effects of additional rate structures, such as critical peak prices. Four studies included automated thermostat controls.
Several researchers, including Caves, Christensen, and Herriges (1983), have looked at the transferability of elasticity estimates from one geographical area to another. Caves et al. use a modeling approach in the spirit of a hybrid demand system. The key price effects are estimated as an elasticity of substitution between peak and off-peak periods. Table 3 shows the consistency of elasticity results calculated form the data collected in five residential time-of-use experiments.
Other researchers have looked at how price response for a time-of-use or critical peak pricing program varies when weather becomes more extreme. Aigner and Lillard (1984) found that time-of-use customers behave in a similar fashion on system peak days as on average summer weekdays and often reduce peak load even more on system peak days (see Figure 2).
Voluntary Versus Mandatory Programs
California’s SPP is a voluntary program. At least two groups of researchers have looked at whether volunteers in time-of-use programs self-select into the program because they tend to have more off-peak consumption and, therefore, receive discounts (from lower off-peak prices) without shifting any additional load from peak to off-peak hours. (3) Some observers call such customers “free riders,” while others say such customers “are finally being relieved of the burden of subsidizing free riding customers.”
Baladi, Herriges, and Sweeney (1994) found that the initial decision of residential customers to volunteer for the Midwest Power Systems (MWP) TOU tariff was largely unrelated to any observable pattern in the household’s electricity usage or appliance ownership. They wrote, “Volunteers, on average, have essentially the same appliance holdings and usage patterns as non-volunteers. Thus, self selection does not lead to revenue erosion for the utility.”
Further, like customers who were automatically placed on TOU rates in other programs, participants in MWP’s voluntary TOU tariff were found to significantly reduce their on-peak usage (by roughly 24 percent during the first summer season). Just as Aigner and Lillard found in their analysis, the MWP changes were larger on days when the system peaked, with on-peak usage dropping by more than 28 percent. In addition, at MWP, the reductions in on-peak kilowatt-hours were not concentrated in any one on-peak hour, but were distributed proportionately throughout the onpeak period. Off-peak usage changes also tended to be proportional, with the exception that households tended to avoid shifting usage to those hours immediately adjacent to the on-peak period (i.e., shoulder hours). This latter result alleviated the concern that the TOU rates might create a secondary needle peak just outside of the on-peak hours.
Caves, Herriges, and Kuester (1989) reported similar results for Pacific Gas and Electric Co.’s voluntary TOU program.
Diversity of the Demand Response to Pricing
Several researchers have found that the strength of a customer’s response will depend on total consumption, appliance holdings, weather, and other socio-demographic factors. For example, the MWP study found that the ability/willingness of individual households to respond to the TOU tariff was significantly influenced by their ownership of major electrical appliances, such as dishwashers, central air conditioners, and dehumidifiers; having more appliances meant a higher percentage reduction in on-peak kilowatthours induced by TOU pricing.
Reiss and White (2001) performed an extensive, statewide analysis on California households, estimating their model results using data from the Department of Energy’s Residential Energy Consumption Survey (RECS). (4) As Baladi et al. found at MWP, Reiss and White, who did not use TOU data, did a particularly thorough job of documenting the relationship of appliance holdings and income levels to residential price response, as seen in Tables 4 and 5.
Reiss and White also looked at how price elasticities varied as a function of income, as shown in Table 5. The findings confirm intuition that consumers with lower income levels are more responsive to price increases.
Reiss and White suggest that there are effectively two types of households with respect to electricity demand behavior: those whose use exhibits some electricity price elasticity, and those who do not and are evidently price insensitive. They developed the chart shown in Figure 3 to show the distribution of price elasticities for the California population. Figure 3 brings into focus the substantial heterogeneity in households’ price sensitivities.
The striking feature of this distribution is its asymmetric, negatively skewed form. This pattern indicates that about half of all households will make very little change in their electricity consumption in response to a price change. The other half respond actively to price changes, with many consumers—those on the far right— taking aggressive load reduction action when prices go up.
This heterogeneity is consistent with findings of recent focus groups in California associated with the SPP, where residential customers fell into three categories, depending on their receptivity to dynamic pricing: indifferent, concerned, and enthused. (5) Presumably, in a voluntary program, few of the first group would participate—including reducing peak demand—some of the second, and most of the third.
Our goal here is to inform policymakers of the wealth of data that has been gathered on residential price response. We invite the reader to draw his or her own conclusions from the data. The California SPP will add another important data point to the existing body of knowledge. As discussed in the adjacent article by Faruqui and George (see article, p. 33), the experimental design methodology will be familiar to those who have been associated with these types of programs conducted over the past three decades in the United States and abroad. These researchers have learned a lot about the appropriate design parameters over these three decades. Similarly, looking to the findings of this body of literature, we can get some guidance and insight into the nature of and the extent to which these customers will respond in a voluntary program that gives them an opportunity to make informed energy consumption decisions in response to the price signals received based on the time varying cost of energy.
An Open Letter From 11 Energy Experts: The Time Has Come for Real-Time Pricing
We write to express our strong support for dynamic pricing of electricity to retail customers. An important missing ingredient in all wholesale electricity markets in the United States is active demand-side participation in the price-setting process. Dynamic pricing at the retail level provides retail customers with the incentive and ability to make efficient consumption and risk-management decisions reflecting their own individual preferences. It is the least-cost way to achieve active demand-side participation in the wholesale market. Active demand-side participation limits price volatility in a competitive wholesale electricity market. In an imperfectly competitive wholesale market, it also limits the ability of suppliers to exercise unilateral market power, and thereby leads to significantly lower average wholesale electricity prices that benefit all retail customers.
We define dynamic prices as retail prices that vary with hourly system conditions. There are a number of ways to accomplish this, but the essential feature of all such pricing plans is that during hours when wholesale prices are high, the price that final customers pay for some or all of their electricity consumption should be high, and when wholesale electricity prices are low, the price that final customers pay for some or all of their electricity consumption should be low. Both of these conditions should apply regardless of the time of day or season of the year when the consumption takes place.
For this reason, it is important to distinguish between time-of-use pricing, where prices change depending on the time of day regardless of system conditions, and dynamic prices that respond to current system conditions. Time-of-use retail prices do not allow active demand-side participation in the wholesale market because the retailer does not have the ability to reduce demand during hours with high wholesale prices by sending a price signal to final customers indicating this increased cost of wholesale electricity
In contrast, dynamic pricing allows electricity retailers to send prices to final customers that signal current conditions in the wholesale market. All final customers equipped with meters that can record their consumption during each hour of the month can benefit from shifting their electricity consumption away from these high-priced hours to other hours of the day. Customers without hourly meters cannot realize this benefit, because the retailer can measure only their total monthly consumption, rather than their consumption during each hour of the year.
Though some express concern that exposure to hourly price changes creates more price risk for customers, such pricing would not preclude individual customers from reducing their price risk by purchasing forward contracts at fixed prices to cover some or all of their expected consumption. Indeed, dynamic pricing would give customers the appropriate incentives to weigh the costs and the benefits of risk-management products on the basis of their individual preferences.
Because of the sizes of their monthly electricity bills and their sophistication as purchasers of electricity, large industrial and commercial electricity customers are likely to realize the largest dollar savings from dynamic pricing. Consequently, these customers should be able to justify the cost of purchasing the hourly metering technology necessary for dynamic pricing. As the price of interval-metering technology falls, we expect that it will become economic for residential and small business customers to purchase according to dynamic pricing plans.
We strongly urge state public utility commissions to support the widespread adoption of hourly metering technology and dynamic-pricing plans. For large industrial and commercial customers, we see no reason to delay implementing these programs. We also encourage extending these pricing programs to residential and small business customers as soon as it is economic. Even though the dollar benefits per customer may be lower for these customers, a number of studies suggest that residential and small business customers are able to shift their electricity demand in response to time-varying prices. Consequently, the aggregate price response from residential and small business customers may be greater than what could be achieved from large industrial and commercial customers.
Sincerely,
Severin Borenstein
E.T. Grether Professor of Business Administration and Public Policy, Haas School of Business, University of California; director, UC Energy Institute; Former Member, Board of Governors of California Power Exchange
James Bushnell
Research Director, UC Energy Institute; Member, Market Surveillance Committee of the California ISO; Former Member, Market Monitoring Committee of the California Power Exchange
Peter Cramton
Professor of Economics, University of Maryland; President, Market Design Inc.; Member, California Power Exchange Blue Ribbon Panel
Benjamin F. Hobbs
Professor and Chairman, Department of Geography & Environmental Engineering, Whiting School of Engineering, The Johns Hopkins University; Member, Market Surveillance Committee of the California Independent System Operator
Paul Joskow
Elizabeth and James Killian Professor of Economics, Massachusetts Institute of Technology; Director, MIT Center for Energy and Environmental and Policy Research
Alfred E. Kahn
Robert Julius Thorne Professor of Political, Economy, Emeritus, Cornell University; Chair, California Power Exchange Blue Ribbon Panel
Edward Kahn
Managing Principal, Analysis Group Inc.
Alvin K. Klevorick
John Thomas Smith Professor of Law and Professor of Economics, Yale University; Member, Board of Directors of ISO New England Inc.; Former Chair, Market Monitoring Committee of the California Power Exchange)
Carl Shapiro
Transamerica Professor of Business Strategy, Haas School of Business; Director, Institute of Business and Economic Research, University of California at Berkeley; Former Member of the Market Surveillance Committee of the California Independent System Operator
Frank Wolak
Professor of Economics, Stanford University; Chairman, Market Surveillance Committee of the California Independent System Operator
Catherine Wolfram
Assistant Professor of Business Administration, Haas School of Business, University of California; Research Associate, UC Energy Institute
Endnotes
- This article focuses on the literature for residential and small commercial customers only. For a discussion of price elasticity for large commercial and industrial customers see Lafferty et al, “Demand Responsiveness in Electricity Markets”, Office of Markets, Tariffs and Rates, Federal Energy Regulatory Commission, Jan. 15, 2001.
- Canada, the U.K., France, Switzerland, and Denmark.
- Comments of the California Office of Ratepayer Advocates in Rulemaking 02-06-001, April 28, 2003.
- The RECS is conducted every three to four years by the U.S. Department of Energy to collect information on household appliances and energy use. The survey is a nationally representative probability sample of households, with representative subsamples for several large states. The study used the California subsamples of the 1993 and 1997 survey waves, which were the most recent available. Together they provide information on 1,307 California households.
- Hiner, H., “Market Response to Time-Varied Pricing,” Presentation at AESP/EPRI Pricing Conference, Chicago, Illinois, May 14, 2003.
