Pilot paralysis: Why dynamic pricing remains over-hyped and underachieved

Electricity Journal

Volumn 26, Issue 4, 2013, Pages 8-21

  Flaim, Theresa ^a   Neenan, Bernie ^b   Robinson, Jen ^c  

^a Cornell University ^*  (United States)

^b UNIVERSITY OF FLORIDA   (United States)

^c UNIVERSITY OF WATERLOO   (Canada)

Author keywords

[No Author keywords available]

Indexed keywords

'CURRENT; DYNAMIC PRICING; FIELD TRIAL; KEY ISSUES; PERFORMANCE METRICES; RESEARCH RESULTS; SIMPLE++; TESTING PROGRAMMES;

COSTS;

EID: 84879889472     PISSN: 10406190     EISSN: None     Source Type: Journal    
DOI: 10.1016/j.tej.2013.04.002     Document Type: Article

References (33)

1. Electric Utility Rate Design Study, Load Management, Issues, Objectives, and Options, Final Report Electric Power Research Institute (EPRI), Palo Alto, CA, Feb. 1982. Available from the National Technical Information Service.
2. Among other things, the 1978 Act required that state regulatory authorities require utilities consider time-of-day rates. ''The rates charged by any electric utility for providing electric service to each class of electric consumers shall be on a time-of-day basis which reflects the costs of providing electric service to such class of electric consumers at different times of the day unless such rates are not cost-effective with respect to such class, as determined under section 115
3. See: http://www.usbr.gov/power/legislation/purpa.pdf
4. R. Hemphill and B. Neenan, Characterizing and Quantifying the Societal Benefits Attributable to Smart Metering Investments, 1017006, EPRI, Palo Alto, CA, 2008, and Hunt Allcott, Social Norms and Energy Conservation, J. PUB. ECON., Vol. 95,No. 9-10 (2011), at 1082.
5. Department of Energy funding authorized through the American Recovery and Reinvestment Act. See http://www.smartgrid.gov/recovery_act/consumer_behavior_studies
6. The arguments for and against customer participation in wholesale markets were summarized in: Federal Energy Regulatory Commission (FERC), Demand Response Compensation in Organized Wholesale Electricity Markets, Docket No. RM10- 17-000 Order No. 745, Final Rule, March 15, 2011.
7. FERC, 2011 Assessment of Demand Response and Advanced Metering, Staff Report, Dec. 2012, at 1.
8. Id. at 1.
9. The early objections of Larry Ruff raised these issues. See L. Ruff, Demand Response: Reality versus Resource, ENERGY J., Dec. 2002, at 10-23.
10. FERC, supra note 6.
11. A glimmer of hope for the efficacy of RTP for residential customers comes from ComEd's Residential RTP program, an assessment of which issummarized herein.
12. B. Neenan, J. Robinson, TheresaFlaim and Richard Boisvert, Understanding ElectricUtility Customers, What We Know and What We Need to Know: Summary Report, 1025856, EPRI, Palo Alto, CA, Nov. 2012.
13. Some distinguish this (randomized encouragement design) from an RCT because randomization takes place before recruitment to treatments or assignment to the control group, and therefore an adjustment must be made to eliminate control group bias. None of the studies evaluated made such an adjustment.
14. See: G. Imbensand J. Wooldridge, RecentDevelopments in the Econometrics of Program Evaluation, J. ECON. LITERATURE, (47) 2009, at 5-86.
15. T. Flaim, J. Robinson, C. Holmes and B. Neenan, A System for Understanding Retail Electric Rate Structures, EPRI, Palo Alto, CA, 1021962, 2011, at 3-6.
16. Id. at 3-6.
17. Id. at A-9.
18. A CPP or PTR which can be called 15 times per year for 4 hours per event would apply to 60 hours per year. A TOU rate that had peak periods that were four hours in duration and applied to weekdays during six months out of the year would apply to 520 hours. Peak prices are relatively lower for TOU than for CPP or PTR because they apply to many more hours in the year.
19. M. Wakefield, G. Horst, B. Neenan, S. Braithwait, D. Hansen, M. Hilbrink, L. Kirsch and R. Boisvert, The Effect on Electricity Consumption of the Commonwealth Edison Customer Application Program Pilot: Phase 1, 1022703 (2011) and Phase 2, Supplemental Information, 1023644 (2012), EPRI, Palo Alto, CA.
20. For example, a 0.1 EoS means that a 1 percent change in the ratio of peak to off-peak prices would lead to a 0.1 percent change in the ratio of off-peak to peak consumption.
21. When the program involves controlling a specific device, like the HVAC system, during events and that control is exercised by the utility or other agent, not the customer, the EoS provides a powerful measure of the extent of load modification, but its inferential value is limited.
22. One would expect that the own price elasticity would be very small since the dynamic rates tested in these pilots only affect usage prices on a very few hours of the year and therefore induce an event response, but not an overall change in electricity demand since prices in all other hours are constant and substantially lower. (The exception is the ComEd Residential RTP program.) Do customers really modify their electricity demand even when the high event prices are not in effect? Or, is a different structural specification of the measurement of own-price elasticity of demand required?
23. The pilots are described in more detail in EPRI 1025856, supra note 11.
24. As this study represents multiple Opower experiments, only this Opower study plus one additional evaluation (chosen because it was the longest running) were selected for review.
25. Notes to Table 2: (1) Results reported here are for feedback-only treatments. (2) A percentage savings was reported only for the ''credited'' savings, which are the measured savings minus the savings attributed to other utility energy efficiency programs.
26. A meta-analysis refers to an analysis approach whereby data from multiple studies with similar research questions are pooled and analyzed together.
27. For a comprehensive overview of the types of rates that are offered in different jurisdictions, see Federal Energy Regulatory Commission, 2010 Assessment of Demand Response and Advanced Metering, Staff Report. Washington, DC, Feb. 2011.
28. An alternative (and opposite) approach - adopted by some utilities in states which opened retail markets to competition - is to make the basic rate for firm service more highly time differentiated and allow customers to buy optional hedged services offered retail providers, if retail access is available For example, Niagara Mohawk Power Corporation (which became a National Grid Company in 2002) provides default service to retail customers priced at the load-weighted day-ahead hourly wholesale price of electricity in their load region (adjusted for losses and capacity charges).
29. The focus on this report is on studies of how retail customers have responded to dynamic pricing when it is offered. It does not review retail customer response to demand response programs offered by independent system operators (ISOs) who operate wholesale markets because their effects are not well studied except at the system impact level.
30. One exception is that the CA-SPP found that customers with central air conditioning had a statistically significantly larger price response than customers without central AC. However, that fact is confounded by the impact of climate - customers without central AC tend to live in climate zones which are more temperate and therefore do not need it.
31. Paul L. Joskow, Creating a Smarter U.S. Electricity Grid, J. ECON. PERSPECTIVES, Winter 2012, at 45.
32. Id. at 45. The source of the results for 109 pilot studies is: Ahmad Faruqui, 109 and Counting: An Update on Dynamic Pricing Experiments, The Brattle Group, PowerPoint Presentation, July 12, 2011.
33. An example of this approach is DOE's Smart Grid Investment Grant (SGIG) consumer behavior study pilots whose designs were developed with the assistance of technical advisory groups whose mission was to help direct individual field trials to address key behavioral uncertainties, and to do so in a way that might produce convincing verifications of a root cause and effect relationship. See: http://www.smartgrid.gov/recovery_act/consumer_behavior_studies