Are subsidies the best way to achieve smart grid goals?
Dr. Sam Newell is principal at The Brattle Group. He acknowledges the contributions of Ahmad Faruqui, Peter Fox-Penner, Attila Hajos, and Kathleen Spees of The Brattle Group. The views in this article are Newell’s and not necessarily those of The Brattle Group or its clients.
In a notice of proposed rulemaking (NOPR), the Federal Energy Regulatory Commission (FERC) proposes to require all regional transmission organizations (RTOs) to pay demand response (DR) providers the full locational marginal price (LMP) for load reductions.1 It shouldn’t be a matter of economic debate that this would allow DR providers to earn more than the value they directly provide to the market. Because customers pay their retail providers only for the energy actually consumed, their ability to sell load reductions as DR into the wholesale market is like reselling energy they never bought in the first place. Thus, full LMP compensation would allow them to earn the wholesale energy market price plus the avoided retail rate for every kWh reduced.
The embedded subsidy would inefficiently distort behavior and would provide compensation that isn’t comparable to the compensation generation receives.
Paying DR the LMP minus the avoided retail generation rate (LMP – G) would be more efficient, so that the customer’s total savings, including avoided retail payments, would be the full LMP. For example, a customer with a $100 per MWh retail generation rate would be willing to run a $250 per MWh backup generator or load reduction when LMPs are $250 or higher because that customer would earn $100 in avoided retail payments plus $150 (=$250 - $100) LMP-G compensation from the wholesale market; by contrast, wholesale compensation of full LMP will induce the customer to use its $250 resource when the LMP is only $150, thus displacing $150 production with $250 production.2
This isn’t to say that DR is worth less than generation, which it isn’t. The full LMP should be paid out, but with only LMP-G going to the customer and G going to the LSE. This creates the same efficient incentives and the same financial outcome as if the customer had bought the energy from the LSE before reselling it.3
Notably, even LMP-G can provide excessive compensation in some cases, for example, if the customer’s apparent load reduction merely is shifted to another high-LMP hour in which the customer pays only the retail rate for incremental consumption—and thus earns a net payment of LMP minus the retail rate for having accomplished nothing. Customers would have to be exposed to market prices in every hour in order to avoid this problem.
The question we should be debating isn’t whether paying DR the full LMP constitutes a subsidy, which it clearly is. The questions should be whether FERC should require regional transmission organizations (RTOs) to subsidize DR, and whether other policy interventions could expand the number of price-responsive customers and achieve the larger policy goals more efficiently.
The answer to these questions begins by articulating the public policy goals that DR subsidies possibly might support. Policy goals that DR contributes to, or that have been associated with, in FERC’s NOPR or elsewhere include: 1) reducing the amount of generation capacity needed for resource adequacy; 2) suppressing energy prices; 3) improving electricity market efficiency; 4) enhancing electricity market competitiveness; 5) reducing emissions; 6) facilitating wind integration; and 7) promoting the smart grid and its associated transformative capabilities for customers, distribution companies, and transmission operators.
• Reducing the amount of generation capacity needed for resource adequacy: DR flattens the load shape by reducing demand in the highest-load periods, thereby decreasing the amount of generation capacity needed to meet resource-adequacy requirements. This is recognized in the RTOs’ resource-adequacy constructs, with DR generally being paid as capacity supply. DR’s opportunities to earn capacity payments have helped DR to emerge as a major source of low-cost capacity,4 reaching 6 to 8 percent of peak load in PJM, NYISO, and ISO-NE.5 This success seems to have surprised many market observers. It’s a good example of how markets can work well to attract and retain low-cost resources.
Is it possible that, in spite of this success, that DR is being paid too little, and subsidies might be needed in order to attract more DR and achieve a more efficient amount of DR capacity? RTOs and stakeholders have engaged in extensive discussions and tariff revisions to attempt to ensure that DR is receiving just the right price for capacity, with payments and performance obligations comparable to generation. These discussions generally recognize that to the extent DR isn’t being treated comparably to generation, the capacity market rules should be fixed, rather than introducing subsidies. To the extent that DR provides other sources of value besides capacity, it should be compensated for those, as discussed below.
• Suppressing energy prices: Many regulators have expressed interest in promoting DR to suppress energy prices and lower customer costs. Indeed, energy price suppression can benefit customers in the short term, and it does so by transferring wealth from suppliers. However, price suppression can’t persist in a long-term competitive market equilibrium.6 Suppliers eventually will respond to depressed prices by retiring more, or building less, generation capacity and/or by increasing their capacity market offers because of their reduced margins in the energy market. Until the suppliers fully respond, customers may enjoy temporary savings, a prospect that might be attractive to regulators considering supporting DR subsidies. However, those potential savings need to be weighed against the economic losses from behavioral distortions induced by subsidies,7 the cost to customers of funding such subsidies, and the likely increase in risk premiums that suppliers will need to charge for operating in a market in which regulators provide subsidies to suppress prices. FERC appears to be soliciting comments on such tradeoffs in its supplemental NOPR issued on Aug. 2, 2010.8
• Improving market efficiency: without DR or dynamic pricing, customers pay fixed rates and are insensitive to changes in spot prices. This describes the status quo for most customers, and it isn’t economically efficient. For example, when spot prices are high, customers have no incentive to cut back on relatively low-value uses of energy. Dynamic retail pricing is the most direct and natural way to address this problem by charging more when marginal costs to serve are high and charging less when marginal costs are low. Dynamic retail pricing is economically more efficient than fixed retail pricing.
However, wholesale DR is an alternative mechanism to expose retail customers to the right price signal at the margin, even if they are on fixed rates. Even if they pay a fixed amount for energy they actually consume, customers can sell load reductions at wholesale energy spot prices. Their participation as DR can increase short-term economic efficiency, so long as their wholesale payment is LMP minus the avoided retail rate (not full LMP). Adding a subsidy to the wholesale payment is less efficient than with no subsidy and possibly less efficient than pure fixed rates with no DR at all (i.e., because the subsidy is distortionary).9
DR’s relationship to market efficiency becomes more interesting in the long-run. Imagine if a large fraction of customers became responsive to wholesale market prices. The peak load might be reduced by 10-20 percent,10 and if that resulted in less generation capacity being built, energy price volatility and the frequency of shortage conditions could increase.11 In shortage conditions, energy and operating reserves prices would have to rise until sufficient customers responded for supply to equal demand. Prices in scarcity conditions would be set according to the customers’ marginal willingness to pay for energy.12
In effect, by responding to high prices during scarcity conditions, the customers would sort themselves into various levels of non-firmness instead of everyone having to pay for the same level of reliability. There would be less need for administratively determined scarcity prices. There still might be administratively determined planning-reserve margins to ensure a certain level of reliability for the fraction of load that prefers to be firm, but the market price of capacity to meet such a requirement might be very low or zero due to the higher energy prices. Compare such an ideal, market-based outcome to today’s: Billions of dollars are paid for capacity in order to meet administratively-determined reliability levels uniformly across almost all customers.
Attaining a more efficient, customer preference-based market depends primarily on achieving deep penetration of DR or dynamic pricing, e.g., 60 to 75 percent in the FERC potential study.13 The limiting factors primarily are the development of advanced metering infrastructure (AMI) and the introduction of dynamic retail rates. Arguably, the direct funding of AMI infrastructure (i.e., roughly $200 per meter before counting operational savings and other benefits, discussed below) should be viewed as competing with DR subsidies for scarce ratepayer dollars. Funding AMI is a more direct way to expand the number of customers that are price responsive. Full LMP subsidies for load reductions certainly stimulate more frequent activity by existing DR resources, but it’s unclear how many new DR resources they attract. When PJM eliminated the full LMP subsidy in 2008, DR participation in energy markets decreased, as expected, but the total amount of DR capacity increased.14 Capacity payments have been the primary source of revenue for DR, which suggests that LMP-based energy payments have not been the primary reason for new DR to enter the market with or without the full LMP subsidy.15
Funding AMI directly could enable a large fraction of customers to be price responsive and enable increased DR participation in both energy and capacity markets. With AMI in place, state commissions could further promote price responsiveness by making dynamic retail rates the default rate structure and allowing customers to opt out. Dynamic rates have an advantage over fixed rates with DR opportunities: They avoid having to define a hypothetical customer baseline load. Customer baselines potentially can indicate some false reductions since they are based on customers’ past volatile consumption patterns and can’t provide a perfect forecast of future consumption that would occur absent DR. Baseline definition has been controversial in all RTOs that have relied on them, and they likely would become more controversial if a large fraction of customers were relying on them for settlement purposes.16
• Enhancing market competitiveness: The Electric Power Supply Association (EPSA) argued in its recent comments on the FERC NOPR that markets already are competitive and that any exercises of market power are mitigated.17 However, markets are never perfectly competitive structurally, and market monitors never can mitigate with perfect precision, given their imperfect information about suppliers’ costs. Moreover, markets that are competitive under normal circumstances may become less so under very tight market conditions with transmission constraints. The presence of DR does indeed make markets structurally more competitive, especially under extreme conditions. Thus, it’s important to enable more customers to be responsive during such conditions. This might be accomplished more effectively by enabling large amounts of load with smart meters and dynamic pricing (or interruptible rates) rather than subsidizing existing DR providers to reduce or shift load under any market conditions.
• Reducing emissions: DR may increase or decrease emissions slightly, depending on whether the load is eliminated or shifted and depending on market conditions.18, 19 In either case, the effect on cumulative emissions is likely to be very small since DR typically is active in so few hours. Energy efficiency and clean generation, however, clearly reduce emissions, and they do so over many more hours than DR operates.
• Facilitating Wind Power Integration: The integration of large amounts of unpredictably intermittent wind resources to meet state renewable portfolio standards (RPS) poses substantial challenges for maintaining the supply-demand balance on the system. Inevitable fluctuations in wind output will necessitate other resources to provide compensating increases or decreases. Traditionally, adjustments to system fluctuations have been provided by generation resources on various timeframes from day-ahead scheduling to real-time dispatch. Generation also has provided ancillary services, which require it to be prepared to respond in contingencies (e.g., operating reserves) or to respond to 5-second dispatch signals (e.g., regulation).
Now, however, the expansion of wind resources is increasing the demand for these ancillary services and for flexibility in the system generally. Numerous RTOs and utilities are analyzing how much of each ancillary service product (or new products) will be needed.20
DR could be one way to satisfy the additional needs, potentially providing fast ramping reserve response or other ancillary services. However, DR provides little ancillary services in the RTOs. DR provides a substantial amount of spinning reserves in PJM and responsive reserves in ERCOT,21 and some regulation in MISO.22 There are pilot programs in other RTOs. For example, ISO-NE recently completed its demand response reserve pilot to determine if small generation and DR resources can provide ancillary services. DR currently can provide ancillary services as a dispatchable asset-related demand (DARD). NYISO’s demand-side ancillary services program (DSASP) enables retail customers that can meet telemetry and other qualification requirements to provide ancillary services. DR provision of ancillary services has been limited primarily by the lack of adequate controls and meters at customer sites (e.g., the RTO has to have direct control and real-time telemetry on any resource or load-providing regulation). In addition, most RTOs are at an early stage of developing the best market designs and operational procedures for using DR for ancillary services. Expanding DR provision of ancillary services will require addressing these challenges directly (i.e., by creating a market for the right ancillary service market products). It’s difficult to see how subsidizing DR activity in the energy market would help.
• Promoting the Smart Grid: It’s unclear how subsidizing DR participation in energy markets might fit into a coherent vision of the smart grid, since DR is only one component of the smart grid. The smart grid refers to communication and control enhancements all along the power system, from customer devices and meters, to the distribution and transmission systems, to the RTO control room. Because the smart grid includes AMI, it will enable widespread DR and dynamic pricing. But the effects of AMI can be greatly enhanced by in-home displays and feedback that “inform, engage, empower, and motivate people” to reduce their annual energy usage (not just peak)23 by 3 to 13 percent.24 Other components of the smart grid could enable RTOs and transmission operators to manage a higher penetration of distributed technologies that may be part of state or federal energy strategies, such as distributed renewable generation and plug-in hybrid electric vehicles (PHEVs). The RTOs will be able to adjust PHEVs and other flexible loads in response to imbalances and price signals in response to variations in the output of large amounts of intermittent renewables. In the long run, the smart grid also could reduce some transmission and distribution costs. The effects on costs, the environment, operations, utility business models, and regulations could be profound.25 Thus, policy makers should be discussing overall policy direction and the best market and regulatory strategies for getting there.
There are more efficient ways to achieve various policy objectives than by subsidizing DR on a volumetric, full LMP basis. Emissions reductions are best achieved through programs and incentives to promote energy efficiency and clean generation. Energy price suppression might be achieved through DR subsidies, but it should be understood that the resulting customer energy savings come at the expense of supplier profits and only be can temporary (i.e., although capacity savings are likely to be permanent). Market efficiency and competitiveness are best achieved without distortionary subsidies; a preferable approach would be funding AMI directly and introducing dynamic retail rates to make the market more efficient and competitive by enabling more customers to be price-responsive. Finally, full LMP subsidies alone are unlikely to stimulate the development of the smart grid and shouldn’t be confused with the smart grid or the far-reaching effects it might achieve.
In short, whether full LMP constitutes a subsidy isn’t a serious economic question. How best to enable customers to respond to spot prices and how to promote the broader policy goals that DR supports are serious questions. Full LMP subsidies don’t appear to be the best answer.
1. For the purposes of this article, DR refers to load reductions that are treated as supply resources and subject to payments by the RTOs, not dynamic retail pricing.
2. Comments of Samuel Newell, Kathleen Spees, and Philip Hanser, Docket RM10-17-000 (filed May 13, 2010 in response to FERC’s NOPR on DR compensation). See also Hung-po Chao, “An Economic Framework of Demand Response in Restructured Electricity Markets,” Feb. 8, 2009, Retrieved from: http://www.hks.harvard.edu/hepg/Papers.
3. See Supplemental comments of Samuel Newell, Kathleen Spees, and Philip Hanser, Docket RM10-17-000, filed October 5, 2010 in response to FERC’s Supplemental NOPR on DR compensation.
4. In some ISO programs, DR capacity providers also receive payments for the energy value of load reductions they actually provide during emergency events, although these payments tend to be much smaller than the capacity payments.
5. The higher numbers sometimes reported by ISO-NE include passive DR, which refers primarily to energy efficiency.
6. Frank A. Felder and Samuel A. Newell, “Quantifying Demand Response Benefits in PJM,” report prepared for PJM and the Mid-Atlantic Distributed Resources Initiative, Jan. 29, 2007.
7. Hung-po Chao, “An Economic Framework of Demand Response in Restructured Electricity Markets,” Feb. 8, 2009, Retrieved from http://www.hks.harvard.edu/hepg/Papers.
8. Supplemental NOPR and Notice of Technical Conference, Re: Appropriate Compensation for Demand Resources (RM10-17-000).
10. FERC staff found that expanded participation in the current mix of DR programs could lead to a 9-percent reduction in peak demand by 2019, relative to a baseline forecast with no demand response. With full participation in mostly price-based demand response programs, this impact could be as much as 20 percent. (FERC Staff, “A National Assessment of Demand Response Potential,” prepared by The Brattle Group, Freeman, Sullivan & Co., and Global Energy Partners, June 2009.)
11. Adding DR that provides capacity and energy should make prices more volatile in the long run, since it’s like adding a super-peaker to the exclusion of other, lower variable cost generation. However, engaging DR that already counts as capacity (thus already having displaced some generation capacity) in the energy market could reduce volatility.
12. Price-responsive demand would have to be able to set prices in both day-ahead and real-time energy markets in order for energy prices to reflect willingness-to-pay. Currently, very little price-responsive demand or DR is able to set prices.
13. This was the assumed market penetration rate of dynamic pricing in the Achievable Participation scenario of A National Assessment of Demand Response Potential. Additionally, a small fraction of non-participants in dynamic pricing were assumed to participate in non-pricing demand response programs in that scenario.
14. Robert Earle, Sam Newell, Ahmad Faruqui, Attila Hajos, and Ryan Hledik, “Fostering Economic Demand Response in the Midwest ISO,” prepared for the Midwest ISO, Dec. 31, 2008, Figure 6.
15. This has been the case in PJM since the introduction of RPM in 2007. In 2008, capacity payments accounted for more than 80 percent of DR’s wholesale market revenues in 2008. In 2009, capacity payments accounted for more 98 percent of DR’s market revenues. Source: Figure 2-23, 2009 State of the Market Report for PJM, by Monitoring Analytics, LLC, March 11, 2010.
16. See, for example, ISO New England Inc., Docket No. ER08-538-000; Filing of Changes to Day-Ahead Load Response Program; Feb. 5, 2008.
17. “Reply Comments Of The Electric Power Supply Association,” Docket No. RM10-17-000 (filed June 30, 2010)
18. Ryan Hledik, “How Green is the Smart Grid?” The Electricity Journal, April 2009, page 39. The article finds that a portfolio of smart-grid programs could lead to CO2 emissions reductions. However, with respect to DR alone, it finds that there’s little or no impact (and indicates that there could potentially be an increase in emissions).
19. See also Stephen P. Holland and Erin T. Mansur, “Is Real-Time Pricing Green?: The Environmental Impacts of Electricity Demand Variance,” Center for the Study of Energy Markets, University of California Energy Institute, Sept. 21, 2006, available here.
20. Chang, Judy et al., “Renewable Integration Model and Analysis,” Proceedings of Transmission and Distribution Conference and Exposition, 2010 IEEE PES, pp. 1-8. See also The National Renewable Energy Laboratory, “Eastern Wind Integration and Transmission Study,” January 2010.
21. The National Renewable Energy Laboratory, “Western Wind and Solar Integration Study,” May 2010
22. ERCOT has over 2,200 MW of Load Acting as a Resource (LaaRs) registered and qualified to provide ancillary services. LaaRs typically provide up to half of ERCOT’s hourly requirement for 2,300 MW of Responsive Reserves. Furthermore, ERCOT has used LaaRs to manage sudden drops in wind output. See ERCOT Press Release, “Demand Response Helps Restore Frequency Following Grid Event,” Feb. 27, 2008.
23. “Demand Response in the Midwest ISO: An Evaluation of Wholesale Market Design,” by Samuel A. Newell and Attila Hajos, The Brattle Group, Inc., Jan. 29, 2010.
24. Karen Ehrhardt-Martinez, Kat A. Donnelly, and John A. “Skip” Laitner, “Advanced Metering Initiatives and Residential Feedback Programs: A Meta-Review for Household Electricity-Saving Opportunities,” report by the American Council for an Energy-Efficient Economy, June 2010.
25. Ahmad Faruqui, Sanem Sergici and Ahmed Sharif, “The Impact of Informational Feedback on Energy Consumption- A Survey of Experimental Evidence,” Energy 35 (2010), 1598-1608.
26. Peter Fox-Penner, Smart Power: Climate Change, the Smart Grid, and the Future of Electric Utilities, Island Press, 2010.