DRAFT                                    5-24-11 Version               (Only some recommendation changes from 5-17 version...
TABLE OF CONTENTS I. Overview of this Report                                            1II.   Power Needs and SG – A Very...
I.   Overview of this Report     Electricity pricing has resulted in consumption patterns that poorly match electric     p...
and therefore lower electric rates (these approaches primarily address          consumption on peak and have only a small ...
been forming around an integrated approach combining peak capacity expansion         with ways to manage peak demand that ...
governments, and will thereafter be a smaller, albeit still very substantial,        market.”8 Finally, they predict that ...
Next was RTP. Then, three different forms of CPP each produced even higher         overall cost savings.12 This is consist...
Hard constraints are defined as ones that cannot be violated by any price.”17 In          this situation, they propose tha...
and customer relations on the one hand, and revenue management, on the         other.20 They argue that when supply is con...
is, if the pricing scheme contains prices that are allowed to vary with time, then        that pricing scheme is dynamic.2...
use patterns. Prices that vary across markets, such as those described in the          Wharton School paper, are also not ...
set of rates apply, the closer to real time those rates should be established. For         example, an hourly price become...
If the rate design under consideration is TOU, economic theory proposes setting         prices equal to marginal costs. Fo...
This raises a thorny design problem of how to design rates to encourage the         needed efficiency improvements and sim...
of gasoline, he suggested. Motorists content to charge overnight, when power        prices are lowest, might pay the equiv...
higher electric utility bills gets reallocated to spending on cloths, food,        entertainment and the like. Analysis of...
resources devoted to electric production and thereby reduces the environmental           impacts associated of expansions....
identification problem since that program reaches only about 40 percent of those        who are eligible.44 Finally, she a...
percent of low-income customers would benefit depending on rate design.49         They also concluded that low income cust...
group‟s price responsiveness is much less than that for higher-income customers        (pilot results appear more mixed to...
VIII. Dynamic Pricing is Demand Response58         All DP is DR. Some of that DR may not be realized for any number of rea...
concise overview of this linkage was provided by Rick Bush, where he notes that         over 70 utilities offer RTP as eit...
on a voluntary basis.”70 These huge differences do underscore the importance of          this question – is program partic...
For example, in the PowerCents DC study in the Appendix, the addition of a         smart thermostat programmed to respond ...
These types of problems are important because they illustrate cases where           individual customers acting in their o...
3. Design two-part rates with a fixed and uniform rate for a fixed amount of    consumption followed by a second step with...
strategies in this list, (b) the potential for cost shifting, (c) it raises the issue ofAdverse Selection since those most...
Policy Recommendations1. Since at least one time-based rates experiment showed that CPP provided   savings that exceeded R...
5. A smart-meter roll out should be coupled with AMI, or some other way to             implement DP.75 Since studies indic...
Appendix - Summary of Selected Pricing ExperimentsThis appendix is a survey of DP experiments. There are at least 70 DP pi...
29
30
31
Below are summaries of a few other experiments that are not                summarized in the Faruqui and Sergici report.  ...
myPower Sense educated participants about the TOU/CPP tariff and they were        notified of a CPP event on a day-ahead b...
Smart Pricing program is an hourly pricing program for residential customers. In        this case, the electricity prices ...
smart meters installed on their premises. The smart meters provideconsumption data in fifteen minute intervals, enabling t...
customer participation included having a smart meter installed and enrollment            in Gateway‟s variable rate plan.7...
residential participants saved $15.21 over the three-month pilot span, while C&I        customers averaged $15.45 in savin...
Under SmartRate, there can be up to 15 event days during the summer season,which runs from May 1st through October 31st. P...
a program through which enrolled, low income consumers receive lower ratesthan do non-CARE customers). The analysis also i...
event sequence is higher than on the first day. Response on the third day isabout the same as on the first day.Results ind...
41
Upcoming SlideShare
Loading in …5
×

A national perspective on using rates to control power system costs (recommendations version) compatable 5-24-11

571 views
504 views

Published on

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
571
On SlideShare
0
From Embeds
0
Number of Embeds
21
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

A national perspective on using rates to control power system costs (recommendations version) compatable 5-24-11

  1. 1. DRAFT 5-24-11 Version (Only some recommendation changes from 5-17 version) A National Perspective on Using Time-Differentiated Rates to Control Power System Costs By Robert J. Procter, Ph.D._____________________________The views expressed in this paper are solely the professional views of the author. Nothing in this papershould be viewed as representing staff or the commissioners of the Oregon Public Utility Commission. Allrights reserved. 1
  2. 2. TABLE OF CONTENTS I. Overview of this Report 1II. Power Needs and SG – A Very Broad Overview 2III. Use of Dynamic Pricing to Manage Costs 5IV. Defining Dynamic Pricing 7V. Dynamic Pricing and TOU Rate Design Issues 9VI. Winners and Losers 13VII. Concerns about Vulnerable Populations 15VIII. Dynamic Pricing is Demand Response 19IX. Opt In, Opt Out, Mandatory Participation 20X. Time Sensitive Rates and Supporting End-User Technology 21XI. Social Cost Arguments Supporting Mandatory Program Participation (or setting DP as the default with an Opt out) 22XII. Transitioning from Fixed Rates to Dynamic Rates 23 Policy Recommendations 26 Appendix - Summary of Selected Pricing Experiments 28 1
  3. 3. I. Overview of this Report Electricity pricing has resulted in consumption patterns that poorly match electric production and delivery cost patterns. One result is the looming national problem of massive spending to build power plants to meet growing demands. We‟ve been selling the proverbial Cadillac at Yugo prices. Given that the electric industry is characterized by increasing costs to meet growing needs, it will be difficult to continue these historical pricing policies. The subsidies embedded in fixed rates (flat, or inverted, or TOU) will likely become increasingly untenable. For example, one subsidy of particular interest is how flat fixed rates under-price on-peak consumption. This subsidy to on-peak consumption leads to even higher peak capacity needs when that same installed capacity goes unused for many hours of the day, and in some cases, used for only a few hours a year. One key question is how much longer can we continue to afford the hidden costs embedded in the existing regime of fixed power rates? In the future, the current structure of fixed rates (flat, slightly inclining rates, or Time-of-Use (TOU)) will become increasingly untenable in many but probably not all parts of the U.S. Individual utilities, or states, or regions will face different circumstances (e.g. may not face near-term capacity constraints or have very small cost differences in meeting peak versus off-peak consumption). These varying circumstances explain some of the differences in approach taken by utility commissions, legislatures, and utilities in different states and regions towards Dynamic Pricing (DP), Demand Response (DR), and Smart Grid (SG). Considering the potential efficiency benefits of DP, why has its adoption been so slow in coming? What is DP, and what is its role in SG and DR? What does the literature on DP studies indicate about the potential effectiveness and important issues in DP and Peak-Time Rebates (PTR) designs? What are the different ways DP is defined? What are the consumer-level impacts of DP and PTR? How do system level benefits argue for mandatory program participation, or at least setting DP as the default with an opt-out option? Experiments indicate that DP and DR can significantly reduce peak consumption. Reducing peak consumption with cost-effective DR will lower future electric costs 1
  4. 4. and therefore lower electric rates (these approaches primarily address consumption on peak and have only a small impact on total electric consumption). Turing to the distribution of benefits of DP, experiments indicate that DP tends to reward customers with relatively flat load curves, those whose consumption is lower than average, and those who are more able to shift their consumption. Losers from DP are generally those with peaky loads, higher overall consumption, and less ability to shift consumption. Successful DP and DR programs must account for these distributional impacts in their design. A great deal has been written on this topic on a theoretical level, an applied level, and summarizing various time-based pricing pilots and programs. The goal of this paper is to provide an overview of these issues from a national perspective at the ‟40,000 - 60,000 foot level.‟ II. Power Needs and SG – A Very Broad Overview Beyond the desires and spending priorities of the Obama Administration, are there systemic issues underlying SG initiatives? One report suggesting the answer is an emphatic yes is titled “The Power of Five-Percent, How Dynamic Pricing Can Save $35 Billion in Electricity Costs.”1 This paper was written prior to 2007, and focusing on the national situation, they projected electric demand to grow by 19 percent over the next decade while capacity was projected to grow only by six percent. Also on a national level, the Energy Information Administration (EIA) projects the average growth rate of grid-based electric demand to exceed the average growth of electric supply to the grid over the period 2009-2035. 2 It is highly unlikely that we can afford to continue building plants and power lines to solve the peak demand – peak supply imbalance. Rather, a consensus has1 Ahmad Faruqui, Ryan Hledik, Sam Newell, Johannes Pfeifenberger Principal, “The Power of Five-Percent, How Dynamic Pricing Can Save $35 Billion in Electricity Costs,” by - The Brattle Group, May16, 2007.2 Energy Information Administration, “Electricity Supply, Disposition, Prices, and Emissions, AEO2011Reference Case.” 2
  5. 5. been forming around an integrated approach combining peak capacity expansion with ways to manage peak demand that enhance customer‟s ability to better control their electric consumption.3 DP is seen as an important piece of the overall approach to peak demand management, and one element to help bridge this peak demand – peak supply imbalance and also help manage overall power supply costs.4 DP does not require that the utility invest in an expensive AMI system. Getting price signals to customers with no AMI is fairly easy. In Illinois, part of the answer is the belief that for the vast majority of customers, it is counter- productive to send them prices every hour. Rather, their approach is to post hourly prices on a web-site, educate customers about the general price pattern, and only send them alerts when prices are going to be exceptionally high5. Turning to other elements of SG, a short article by Pike Research issued in December 2009, argued that smart meters, while the most visible part of SG, aren‟t where the real benefits exist. They argue that grid infrastructure projects (transmission upgrades, substation automation, and distribution automation) will likely find the best return on investment.6 They further predict that grid automation (i.e., the actions noted in the prior paragraph) will capture 84 percent of global SG investment through 2015, compared to just 14 percent for AMI, and 2 percent for electric vehicle (EV) management systems.”7 According to their research, “…revenues [from the sale of SG equipment] will peak in 2013 after several years of a strong push by key3 Ibid.4 It should be noted that AMI isn‟t necessary for DP. If AMI has been installed, and is operational, then itis relatively easy to implement DP. DP can still be implemented in the absence of AMI. Doing so doesrequire targeted meter exchanges. For example, the RTP program in Illinois does not rely on AMI.ComEd uses interval meters read once per month and Ameren uses a one-way Automated Meter Reading(AMR) for some customers. Implementing meter data management systems may however be as importantas the technology choices for the meters themselves.5 When there is one or more hours the following day over a certain threshold (currently $0.13/kWh, energyonly, but this may be lowered to $0.10/kWh), customers are notified automatically via email and telephonecalls.6 “Smart Grid Investment to Total $200 Billion Worldwide by 2015,” Pike Research, See:http://www.pikeresearch.com/newsroom/smart-grid-investment-to-total-200-billion-worldwide-by-20157 Pike Research. 3
  6. 6. governments, and will thereafter be a smaller, albeit still very substantial, market.”8 Finally, they predict that grid automation upgrades and smart metering will be in the range of $200 billion in worldwide investment between 2008 and 2015.9 There is no lack of competing definitions for what is and is not SG. Or, what SG is supposed to enable or accomplish. There isn‟t any reason to summarize that material here. The Pike Research report takes a different approach and looks at the key factors driving SG investment. They identify investment as SG if they fall into one or more of the following four categories based on results or goals: 1. Improved reliability and security, 2. Improved operating efficiencies (with associated lower costs), 3. Balancing power generation supply and demand, and 4. Reducing the overall electrical system‟s impact on climate change.10 These four rationales for SG all relate to reduced future spending to meet customer needs for electricity. If cost savings is the primary driver behind SG, why isn‟t SG being implemented faster? Barriers to this transformation go well beyond pure technical and economic issues. They note that the slow progress can also be attributed to a lack of common vision and standards, outdated and fragmented business and regulatory models, and lack of awareness, and often the trust, of the consuming public.11 Later in this paper, the reader will see that rates are an integral part of SG. Turning to rates alternatives using SG, one study that included an examination of the Net Present Value (NPV) of utility cost savings compared various forms of variable pricing, concluded that TOU rates produced the lowest overall savings.8 Ibid.9 “Smart Grid Technologies,” Pike Research, See: http://www.pikeresearch.com/research/smart-grid-technologies.10 Ibid.11 Ibid. 4
  7. 7. Next was RTP. Then, three different forms of CPP each produced even higher overall cost savings.12 This is consistent with results from other studies summarized in the appendix to this report. III. Use of Dynamic Pricing to Manage Costs Taking a step back and looking more broadly at the use of DP, we see that it‟s been with us since the beginning of economic exchange, be that barter exchange or market-based trade. One writer notes that "Dynamic pricing has always been with us…the classic hagglers in the market of a Middle East bazaar [is one example]. People will pay very different prices for the same bolt of fabric. This is more the norm in transactions than fixed pricing. Fixed pricing is a much later phenomenon and its an artificial one.”13 It‟s only been since the Industrial Revolution that DP was replaced by standardized pricing schemes. DP has come back into favor in various industries over the last 20 – 30 years. Some applications of DP are e-commerce, eBay being one example. It‟s also been reported that IBM uses it to establish prices for some of its computers.14 Auto and home purchases are two additional examples where timing affects prices (as we will see, prices that vary by time are not necessarily dynamic). Another place where more standardized pricing has been replaced by DP is the airline industry.15 The authors note that airlines previously used what they refer to as “…an allocation based fare-class model” and business success was measured by load factor - the number of passengers per available seat on a single leg trip.16 They argue that “The primary driving reason behind the utilization of a variable pricing policy is capacity limitation or “hard constraints.”12 Ahmed Faruqui and Lisa Wood, “Quantifying the Benefits of Dynamic Pricing in the Mass Market,”Edison Electric Institute, January 2008, Table 7, p.18.13 “What Consumers -- and Retailers -- Should Know about Dynamic Pricing,” in Knowledge@Wharton.See: http://knowledge.wharton.upenn.edu/article.cfm?articleid=1245.14 Nick Wreden, “Advantages of Dynamic Pricing,” Direct Marketing News, May 13, 2003. See:http://www.dmnews.com/advantages-of-dynamic-pricing/article/80877/.15 Leslie Anne Palamar and Victoria Edwards ,“Dynamic Pricing Friend or Foe, A report on the state ofdynamic pricing in the contracted, corporate rate segment of the North American hospitality industry,”2007, bte tourism training and consulting, Buckhiester Management.16 Ibid, p.4. 5
  8. 8. Hard constraints are defined as ones that cannot be violated by any price.”17 In this situation, they propose that there are limited ways to essentially ration this limited supply. These ways are: (1) allow available supply to be sold on a first come – first serve basis, (2) allocate limited supply to specific customers, and (3) slowly raise prices until demand falls to meet supply.18 Hard constraints are certainly an issue in the electric industry, at least nationally, in the short-run. If they weren‟t an issue, the context for SG laid out earlier in this paper would be different, and SG‟s rationale would shift to other objectives. It is likely true that in some regions, for example the Pacific Northwest, where interconnections with California and British Columbia allow for power purchase contracts to substitute for building more power plants, that demand-supply imbalances are often more economically met through power purchase contracts. When a utility relies on the wholesale market to economically meet customer needs, this exposes customer to the wholesale power market. It often appears that both the utility and the customer reaction to the potential for higher market prices overwhelms the potential for lower than expected power costs. At least with flat rates, power purchase expenses are spread across time and customers whereas DP pricing schemes potentially exposes customers to the full variability in those markets. However, it needs to be stressed that there is a wide range of options between flat rates on one extreme and full RTP on the other extreme. This anxiety over one side of the distribution of wholesale market prices is one reason why there‟s a tension between using DP to ration a scare inventory, on the one hand, and manage revenues and customer relations, on the other hand. John Burns, president of Hospitality Technology Consulting captures this thusly, “We tell ourselves that relationships are important but dynamic pricing is being driven by revenue management at the expense of good customer relationship management. Sales and marketing is constantly finding themselves in the middle”19 This conflict may require balancing the competing goals of marketing17 Ibid.18 Ibid.19 Ibid, p. 5. 6
  9. 9. and customer relations on the one hand, and revenue management, on the other.20 They argue that when supply is constrained and demand varies, fixed prices aren‟t sustainable. Flat rates (or slightly inclining block rates) help provide both revenue stability for the utility and bill stability for the customer. This is one reason customers and utilities like them. However, they also encourage over-building of increasingly expensive infrastructure. It seems that this over-building is overlooked by at least some of the stakeholders in the debate over time-based rates. The practice of average cost pricing helps to bury the cost of new infrastructure additions by essentially diluting the impact of the higher incremental cost investments. This is accomplished by adding the higher incremental cost investments to the existing rate base and spreading the total bundle of costs across both time and sales. In contrast, both RTP and CPP can help avoid or reduce overbuilding. Paul Centolella, Commissioner on the Public Utilities Commission of Ohio, sounds a cautionary note that seems to get lost in the fight over rate impacts today. His message is clear that customers will face higher electric rates without innovation in electric rate structures.21 He notes the fear of driving up monthly power bills in at least some months can be addressed through „work-arounds‟. IV. Defining Dynamic Pricing Within economics, dynamic analysis explicitly includes time as a variable. This is in contrast to a static analysis which excludes time as an explicit variable in the analysis. A dynamic analysis studies the path between two discrete points in time. From this vantage point, any approach that includes time as an explicit variable is a dynamic analysis. Turning to pricing schemes, some writers use an approach to defining DP that is consistent with how economics has historically defined dynamic analysis. That20 The article by Palamar and Edwards points out that these two functions have been the province ofseparate part of the organization.21 Paul Centolella, “The Smart Grid Needs Smart Prices to Succeed,” Harvard Business Review, October14, 2010. See: http://blogs.hbr.org/cs/2010/10/smart_prices_are_key_to_smart.html. 7
  10. 10. is, if the pricing scheme contains prices that are allowed to vary with time, then that pricing scheme is dynamic.22 One example of this is a report by the Public Utility Commission of Texas (PUCT) to the Texas Legislature defines “Dynamic pricing [as] either time-of-use or real-time pricing…”23 In contrast, the National Action Plan on Demand Response24 explicitly excludes TOU rates from DP. They reject TOU as one form of DP since the peak period and rates do not change in response to changes in system conditions. Yet a third approach is contained in a report out of the Wharton School which lumps DP in with other forms of flexible pricing.25 One example they use is drug companies setting lower prices for low-income customers. In a paper surveying the results of 17 DP programs, the authors define dynamic pricing as prices that reflect the wholesale market prices.26 They also reject TOU as a form of dynamic pricing since the peak period and the rate(s) are set in advance. However, they do include CPP as a form of dynamic pricing even though the rates are set in advance. They note that the CPP critical days are called based on wholesale market conditions. Because of this feature, CPP reflects power system conditions. Given the confusion about what is and what is not DP, the clearest and cleanest approach is to define a pricing scheme as dynamic if it reflects system conditions at the moment the prices are established or the critical day is called. Therefore, RTP is dynamic as is CPP. PTR may be dynamic if either the rebate amount or when it is effective varies with system conditions. Day ahead pricing (DAP) is also a form of DP since prices reflect system conditions at the time they are posted. TOU rates are not dynamic pricing even though it‟s a scheme with different prices at different points in time that have some relationship to historical22 For example, Faruqiu argues that „Dynamic pricing is a form of time-of-use (TOU) pricing.” See:Ahmad Faruqui, “The Ethics of Dynamic Pricing,” The Energy Journal, July 2010, p. 13.23 “A Report on Advanced Metering as Required by House Bill 2129 Public Utility Commission of TexasSeptember 2010,” Public Utility Commission of Texas September 201024 National Action Plan on Demand Response, FERC, June 17, 2010, footnote 15 on pg. 4.25 “What Consumers -- and Retailers -- Should Know about Dynamic Pricing”26 Ahmad Faruqui and Sanem Sergici, “Household Response to Dynamic Pricing of Electricity a Survey ofSeventeen Pricing Experiments,” November 13, 2008. 8
  11. 11. use patterns. Prices that vary across markets, such as those described in the Wharton School paper, are also not dynamic. V. Dynamic Pricing and TOU Rate Design Issues This section is split into four sub-sections. Each sub-section addresses an issue that should be addressed in DP and/or TOU rate design. Length of Time Price is Stable This issue applies to both DP and TOU. Referring first to RTP, experiments that I‟ve seen use hourly day-ahead price, sometimes referred to as DAP, as the price signal. Some, but not all, then use the actual RTP for a given hour for billing. One type of DP that has received consideration and that is the focus of a pilot program development by PGE is CPP. One report argues that CPP is essentially an alternative approach to pricing peak and off-peak energy differently.27 The IEEE Whitepaper correctly note that CPP is an attempt to send price signals to customers that accurately captures the actual cost of providing electricity on a small number of hours on a few critical days during the year. They suggest that CPP is “…particularly effective when high wholesale prices are limited to about 100 hours of the year, and their onset is somewhat predictable.”28 CPP is better than TOU since from an efficiency perspective “…the additional charges are based on consumption when the [electric] system is actually constrained. Since CPP effectively allows retail prices to vary with some movements in the wholesale market, it captures some of the efficiency aspects of RTP. How far in advance prices are set Borenstein distinguishes between what he calls the „granularity‟ of prices and how far in advance prices are set, which he refers to as the „timeliness of prices.‟29 It makes little sense to set hourly prices a year in advance, for example. The implication here is that the shorter the timeframe to which a given27 IEEE Whitepaper, p. 3.28 Ibid.29 Ibid, p. 8. 9
  12. 12. set of rates apply, the closer to real time those rates should be established. For example, an hourly price becomes a better signal on cost when the rate is set nearer to the hour to which the rate applies. Michael Jaske identifies three ex ante approaches to RTP: (1) day-ahead, (2) hour-ahead, and (3) near real-time based on ancillary service market.30 At this point, I‟ve not yet seen a rate design that uses hourly prices set an hour ahead for price signaling, though that design may exist. Nor have I seen prices set using the ancillary service market. In the previous section, I referred to a design employed in Illinois that notifies customers of hourly prices set the day ahead but bills using actual prices at the hour of use. Turning to TOU rates, these rates are set far in advance. The regulatory process determines how far in advance they will be set. For example, in Oregon, when an IOU submits a rate filing with the Commission, or the Commission proposes the company make a rate filing, the issue of how far in advance rates are set does not arise. Rather, the rates that are established are a by-product of using a test year for determining revenue requirements. Once a set of rates are approved by the Commission, they are in place until they are changed. How to Set the Prices for Each Time Period This section applies to CPP, PTR, and TOU rates only. The smaller the difference between the rates for on-peak and off-peak use, there is less incentive to shift consumption. What might not be quite so obvious is the connection between differences in rate levels at different times and the choice of the default rate.31 Since we know that many more customers will choose to stay in a program if the rate is opt out than will choose to opt into the program if the spread between on and off peak narrows, it will be more important to use RTP as the default rate and allow opt out.30 Severin Borenstein, Michael Jaske, and Arthur Rosenfeld, “Dynamic pricing, Advance Metering andDemand Response in Electricity Markets,” Center for the Study of Energy Markets, October 2002, pp. 33-34.31 In this paper, the term „default rate‟ refers to the rate design customers will be on if they do not exercise aopt out option. 10
  13. 13. If the rate design under consideration is TOU, economic theory proposes setting prices equal to marginal costs. For a period in the middle of the night, theory suggests using a short-run price of a variable input into generation (SRMC), such as, off-peak wholesale energy prices with no distribution or transmission costs included. For on-peak consumption, theory suggests setting it equal to the long- run incremental fully allocated cost (LRIC) of the marginal resource, usually a combined-cycle combustion turbine, with marginal distribution and transmission costs included. Theory doesn‟t provide clear guidance for prices for time periods between these two periods. Several caveats to the marginal cost guidelines for a TOU rate are (a) there may be reasons to use Ramsey pricing,32 and (b) there may be other policy considerations that support an even larger difference between the on-peak and off-peak rates than result from applying the aforementioned marginal cost principles. If this is the case, it is important to set the on-peak rate above LRIC, rather than lowering the middle of the night rate below SRMC. As for CPP and PTR, the rate should be set using either the long-run incremental fully allocated cost of a simple cycle combustion turbine, or using an average of wholesale energy and capacity purchase expenses during critical hours, whichever is lower. Equity Considerations This issue pertains to both DP and TOU. One result of a shift from fixed flat rates to DP or TOU is some customers will confront the hidden costs they have been imposing on other parties. When those customers who face higher DP or TOU rates include vulnerable populations, some, though not all, consumer advocates call for special consideration of these impacts or oppose any shift from flat fixed rates33. However, if this concern results in resistance to DP or TOU, we are left with a rate design that encourages over-investment in generating and distribution investments (and maybe transmission). Over time, that will result in higher rates for everyone, including vulnerable populations.32 This will lead to a divergence between the rates and what they would be using strict MC principles.33 Two notable exceptions are the legislatively mandated consumer advocates in both Illinois and D.C. whoboth went to their respective legislative bodies requesting that utilities be directed to explore and offer DP. 11
  14. 14. This raises a thorny design problem of how to design rates to encourage the needed efficiency improvements and simultaneously address the needs of vulnerable populations. As a result, it is important to pay close attention to rate designs and rate levels that both reduce the pressure for expansion of the existing power system and that also have a greater chance of reducing a customer‟s monthly bill. Those two goals may be mutually exclusive in the near- term and compatible only in the long-run. In the near-term this will probably require some type of „work around‟ that will help to blunt the impact on vulnerable populations. This issue is addressed more fully in section VII. Interoperability34 has been identified as a factor affecting DR adoption and it can also be seen as one aspect of equity. Interoperability eases implementation as it removes concerns about equipment being able to talk with each other. The customer needn‟t worry if the software and hardware can communicate effectively. This removes some risk from investment decisions, especially when standards are not yet in place. Since more affluent customers will be less concerned about this issue than will the more vulnerable customers, this requirement is especially important as one part of managing the impact of DP (and maybe TOU) on vulnerable populations. Degree of Market Segmentation If some rates are optional, one question is how to design of those rates to make them attractive option relative to current rates. Lewis characterizes rate choice by customers as a risk management issue. What is striking is the shift in focus away from focusing solely on cost recovery to rate designs that offer customers options to individual risk-reward profiles. This isn‟t a new concept even when applied to regulated electric utilities, though the industry is on the cusp of a sea-change in product and price offerings. Mohler put it thusly, "We should have [the] ability to differentially price [EV charging services],".The fast-charge price could be equivalent to $20 for a gallon34 Interoperability is the notion that different equipment potentially from different suppliers be able to „talk‟to each other. For example, the communication protocols allow for information, data, to be passed throughthe system. Imagine that you could use your cell phones with any carrier? If that were the case, cellphones and communications infra-structure would be interoperable. 12
  15. 15. of gasoline, he suggested. Motorists content to charge overnight, when power prices are lowest, might pay the equivalent of 75 cents a gallon -- a bargain price. "Until we can give them a way to painlessly respond to that price signal, I dont know how we get to where we need to go,"35 VI. Winners and Losers In an article titled “Dynamic Pricing is Smart Grid’s Secret Sauce,” Kiesling argues that DP is one of the most valuable direct consumer benefits provided by SG. It provides these benefits by making the customer aware of the cost of their energy use and thereby helping the customer compare that to his/her value.36 Kiesling further argues that DP benefits consumers whose consumption is flexible while not harming customers with less flexibility. She argues that less price responsive customers can benefit from DP since it reduce the quantity of peak power demanded, thereby reducing system costs and average prices paid by these customers. Using economic jargon, an opportunity cost of not adopting DP (and perhaps TOU) are the higher system costs that otherwise could have been avoided. There will be customers in vulnerable populations who will be disadvantaged by DP and TOU, at least in the near-term. However, rather than lose the potential cost savings to all customers, we need to find tools that work to cushion these impacts. Commissions and legislatures seem reluctant to adopt DP and TOU that return benefits to many, but not all, at least partly to „protect‟ non-price responsive customers. The irony is that all customers will likely face even higher future electricity rates due to higher system costs absent better managing of peak usage. In turn, money that could have gone to supporting other businesses goes to paying electric bills. This suggests there are benefits from DP and TOU that accrue to the economy generally as money that would have been spent on even35 “Consumer Response a Lingering Riddle for Backers of Smart Grid,” by Peter Behr of ClimateWire,October 23, 2009, as published in The New York Times, See:http://www.nytimes.com/cwire/2009/10/23/23climatewire-consumer-response-a-lingering-riddle-for-bac-52276.html36 “Dynamic Pricing is Smart Grid’s Secret Sauce,” by Lynne Kiesling, May 13, 2008. See:http://www.smartgridnews.com/artman/publish/article_441.html. 13
  16. 16. higher electric utility bills gets reallocated to spending on cloths, food, entertainment and the like. Analysis of the magnitude of foregone economic growth should be counted as a cost of maintaining fixed electric rates. Turning to RTP, the benefits of RTP have been discussed in various forums.37 Borenstein argues that while RTP can reduce peak consumption by smoothing out the peak, there will also be some energy savings, but not much.38 Its clearer what the benefits are to the utility instituting a CPP or RTP rate structure, for example, but it isn‟t so clear what the benefits are to the individual customer.39 Studies of DP and TOU generally indicate that DP returns benefits to customers as a group.40 It is harder to reach conclusions about individual customers since numerous factors combine to determine how DP affects an individual customer. However, these same studies do show that there are benefits for all customer classes.41 Customers with flatter load profiles and those who are more able to shift their consumption will tend to benefit from DP. Customers with peakier loads, those less able to shift consumption to lower cost periods, and those with higher overall consumption levels tend to be disadvantaged by DP. Some other categories of benefits include, but are not necessarily limited to, reduced disruptions associated with the permitting, siting, and actual construction of generation, distribution, and transmission facilities, and less demand pressure on wholesale power market prices. While state statutes sometimes limits the environmental benefits and costs that a commission may consider in decision- making, some of these impacts do indirectly filter into Commission decisions. For example, DP that achieves greater use of existing power system will reduce the need for, and the size of, any increments to that system. This reduces37 One example is the work of Severin Borenstein. For example see his presentation titled “Issues inImplementing Dynamic Electricity Prices,” CITRIS Research Exchange, April 2007. See:http://www.youtube.com/watch?v=LdD4sYvDa08.38 He‟s used the number 10 percent for the amount of total energy savings.39 For example, see a short piece by Chris Lewis, “The Evolution of Dynamic Pricing,” CogneraCorporation. See: http://www.electricenergyonline.com/?page=show_article&mag=64&article=502.40 The Appendix contains summaries of some DP and TOU pricing experiments.41 This „composition problem‟ is common in economics and is partly addressed in section X below. 14
  17. 17. resources devoted to electric production and thereby reduces the environmental impacts associated of expansions. Faruqui suggests targeting those most likely to benefit while avoiding those most likely to be harmed by DP. He argues that the benefits of DP can be achieved without having all customers participate. While there‟s isn‟t adequate room in this paper to layout his argument with the accompanying graphs, suffice it to say that a customer with a load profile flatter than the class average will benefit immediately from DP and should enroll. VII. Concerns about Vulnerable Populations Earlier, I touched on the very hot issue of how a move from fixed rates impact vulnerable populations. While there is no lack of opinion on this issue, there is a lack of solid data upon which to reach conclusions about how best to protect vulnerable populations while also designing prices to better reflect costs. This concern about harming some ratepayers has its roots in a belief about what constitutes a fair price. Faruqui references Vickery who, in a paper on responsive pricing, proposed there was a sense of a just price as an ethical norm. This belief in a fair price on an ethical basis was echoed by Eric Hirst who wrote [regarding DP], “The greatest barriers are legislative and regulatory, deriving from state efforts to protect retail customers from the vagaries of competitive markets.42 It goes without saying (but I‟ll say it) that flat electric rates while a time-honored design for residential customers in particular, result in higher system costs and higher overall rates than need be the case. Another issue is who to include as part of a vulnerable population. Often these populations are identified as the elderly, or low-income customers, or the medically fragile. As Alexander points out, it can be very difficult to find these customers. She argues that utilities don‟t typically gather demographic date, such as, a customer age and household income.43 She also argues that reliance on the Low Income Home Energy Assistance Program doesn‟t solve this42 Eric Hirst, “Price Responsive Demand in Wholesale Markets: Why is so Little Happening?”43 Ibid, p. 44. 15
  18. 18. identification problem since that program reaches only about 40 percent of those who are eligible.44 Finally, she argues that the elderly seldom seek out or apply for low-income programs, and they usually aren‟t well represented when these rate decisions are made.45 Alexander also argues that high cooling and heating costs contribute to “food insecurity.”46 For example, she quotes from a U.S. Department of Agriculture study that concluded that the odds of food insecurity are 43 percent lower in the summer than in the winter in those states that have relatively high-heating requirements.47 Regarding low-income customers, there are two basic hypotheses about how DP affects these customers. One approach argues that low income customers benefit immediately since they use relatively less energy during air conditioning peaks than more affluent customers with larger dwellings. In turn, it‟s argued that the low-income customer has a less peaky load profile than the class average. The second hypothesis is that low income customers generally use less energy than more affluent customers (smaller dwellings, fewer electric consuming appliances etc.) and they are much less able to shift load from peak to off-peak periods and/or to curtail peak period usage. Hence, they would be harmed by dynamic pricing.48 In an effort to determine which of these two hypotheses are closer to the truth, a study was designed using data from a load research sample of a large urban utility. Next, they reviewed the empirical evidence from five recent utility projects. They concluded that a majority of low income customers do benefit from DP because they use relatively less energy during the peak hours compared to the average residential customer. They determined that between 65 percent and 7944 Ibid.45 Ibid.46 Ibid, p. 41.47 Ibid.48 Ahmad Faruqui and Lisa Wood, “Dynamic Pricing and Low Income Customers -- Can they Co-Exist?”July 9, 2010. See: http://www.smartmeters.com/the-news/1079-dynamic-pricing-and-low-income-customers-can-they-co-exist.html. 16
  19. 19. percent of low-income customers would benefit depending on rate design.49 They also concluded that low income customers do shift their load in response to price signals.50 Faruqui separately argues that 80 percent of low-income customers would benefit from DP and that increases to 92 percent with a “…modest amount of …” of DR achieved.51 Faruqui‟s conclusion that low-income consumers benefit because they have flatter consumption profiles and use less electricity on average than wealthier customers may not persuade some consumer advocates to support a move to some form of time-variable rates. However, those opposing these moves seem to believe that rates will otherwise not go up as a result of keeping the existing rate structure. This implicit assumption in arguments opposing DP or TOU is likely to be false, at a minimum for the country as a whole, for reasons that were discussed earlier. Alexander summarizes the political side of this argument about rate design and vulnerable populations indicating that the American Association Retired People and the National Association of State Utility Consumer Advocates are opposed to mandatory DP programs and call for cost-effective DP programs with voluntary participation.52 She also summarized the experience of various utilities with TOU rate structures. According to her, Central Maine Power implemented a mandatory TOU rate structure but abandoned it after a few years in the face of what she termed „vociferous‟ opposition especially from the elderly. She also notes that Puget Sound Energy implemented a mandatory TOU rate in 2001 but had abandoned it late in 2002.53 She argues that while there is a dearth of analysis of DP‟s impacts on low-income populations, they indicate that this49 Ibid.50 Faruqui and Wood noted that two studies, carried out by Connecticut Light & Power Company (CL&P)and BGE, find that low income customers were equally price responsive to the average customers, whilethe California Statewide Pricing Pilot (SPP) carried out jointly by the state‟s three investor-owned utilitiesand the SmartRate program offered by Pacific Gas & Electric Company (PG&E) found that they were lessresponsive. The Pepco DC results, on the other hand, showed that low income customers were much moreresponsive than other customers.51 Ibid.52 Alexander, p. 42.53 Ibid. 17
  20. 20. group‟s price responsiveness is much less than that for higher-income customers (pilot results appear more mixed to me that she suggests).54 55 Turning to the issue of ways to provide some bill protection for vulnerable populations, there are a number of options, including but not limited to, 1. Capping their power bills for some period of time and gradually removing that cap. 2. Using a tiered-pricing scheme where a fixed and flat rate applies to an amount of energy purchases that are considered „essential‟ and allowing consumption above that amount priced at some higher rate (note that there are many ways to design this tiered- rate approach). 3. Create an account where bill reductions are added to the account and they are used to balance bill increases which are debits to that account, and any remaining savings at the end of a year is refunded to the customer. 4. Placing these customers on fixed and flat rates. A recently released report on PG&E‟s various time based pricing tariffs includes evidence on the impacts on vulnerable populations of bill protection to reduce the risk of higher monthly bills. They found that bill protection reduced peak energy savings induced by DR by about 25 percent, and that it reduced program attrition. They report that the average load impact for customers under bill protection was around 12.8 percent, compared to 18.1percent for customers not under bill protection.56 They also report not having data to assess how bill protection affects the decision to enroll in the program.5754 Barbara A. Alexander, p. 44.55 The PowerCentsDC experiment (summarized in the appendix) shows similar results for low-income andall other participants.56 Stephen S. George, Josh L. Bode, Elizabeth Hartmann, 2010 Load Impact Evaluation of Pacific Gas andElectric Companys Time-Based Pricing Tariffs, Final Report, April 1, 2011, p. 66.57 Ibid. 18
  21. 21. VIII. Dynamic Pricing is Demand Response58 All DP is DR. Some of that DR may not be realized for any number of reasons; but, that doesn‟t diminish the fact that DP is DR. The National Action Plan on Demand Response defines DR as “…the ability of customer to rely on a reliability trigger or a price trigger…to lower their energy use.”59 They further differentiate between dispatchable and non-dispatchable DR.60 Dispatchable DR is defined as planned DR that is not controlled by the customer which includes, but is not limited to, direct load control.61 They define non-dispatchable DR as DR that the customer controls62. With non-dispatchable DR, the customer may or may not respond to price changes. They note that this latter form of DR is also referred to as price-responsive DR.6364 One design challenge of non-dispatchable DR programs is the need to determine the baseline from which demand reductions are measured. Borenstein argues they are difficult to set for several reasons.65 The issue of baselines also arises with the PTR. He argues that PTR is a poor substitute for either CPP or RTP.66 67 He rightly points out that if the PTR payments come from other customers, then those costs will be reflected in rates. One reason DR is attractive is it offers the potential to meet the need for peak energy faster and at lower cost than building more generation. A short and58 One demand management program omitted from this discussion is the use of interruptible contracts thatallow the system operator to curtail loads and provide for very large penalty payments if the customer doesnot curtail loads.59 National Action Plan on Demand Response, FERC, June 17, 2010, p. 3.60 Ibid.61 Ibid.62 Ibid.63 Ibid.64 PTR is one form of non-dispatchable DR.65 For his argument, see: Severin Borenstein, “Time-Varying Retail Electricity Prices: Theory andPractice,” p. 1766 In a separate paper, Borenstein argues that PTR, which he refers to as Real-Time Demand ReductionPrograms, are fairly blunt instruments in which the system operator announces the program is in effect andthe rate offered for voluntary curtailment is usually set in advance. See: Severin Borenstein, Michael Jaske,and Arthur Rosenfeld, “Dynamic pricing, Advance Metering and Demand Response in ElectricityMarkets,” Center for the Study of Energy Markets, October 2002, p. 16.67 Several of the studies summarized in the appendix to this paper show that PTR are not as effective asRTP or CPP but are more effective than TOU. 19
  22. 22. concise overview of this linkage was provided by Rick Bush, where he notes that over 70 utilities offer RTP as either a pilot or a permanent program. In a cautionary note from the telecom industry, Bush comments that consumers appear to desire choice until a large bill arrives, and then choice isn‟t seen in such favorable light. Realizing DR requires more than just adopting some form of DP. It also requires that enabling technology be installed that can take the price signals and automatically change consumption (as pre-determined by the building tenant/owner). One report on enabling technology, based on a review of 57 different residential sector initiatives performed between 1974 and 2010, concludes that to realize higher program savings, smart meters must be used in conjunction with real-time (or near-real time) web-based or in-home devices and enhanced billing approaches and well-designed programs that successfully inform, engage, empower, and motivate customers.68 Numerous studies demonstrate that DR potential varies from modest to substantial, largely depending on the data used in the experiments and the availability of enabling technologies. Across the range of experiments examined, TOU rates induced a drop in peak demand that ranged between three to six percent. By comparison, CPP tariffs led to a drop in peak demand of 13 to 20 percent. When enabling technologies were employed, reductions in peak demand from CPP rates range from 27 to 44 percent.69 IX. Opt In, Opt Out, Mandatory Participation Earlier in this paper, I made reference to a report titled “The Five Percent Solution, How Dynamic Pricing Can Save $35 Billion in Electricity Costs.” The authors of that paper used Monte Carlo simulation to estimate participation rates of opt out versus opt in program designs. They concluded that “…about 80 percent [of customers] would stay on dynamic pricing if it is offered as the default rate and that a substantially smaller number, perhaps 20 percent, would select in68 Ehrhardt-Martinez, Karen, Kat A. Donelly and John A. “Skip” Laitner, “Advanced Metering Initiativesand Residential Feedback Programs: A Meta-Review for Household Electricity-Saving Opportunities,”June 2010, available at http://www.aceee.org/pubs/e105.htm.69 Faruqui and Sergici. 20
  23. 23. on a voluntary basis.”70 These huge differences do underscore the importance of this question – is program participation mandatory or voluntary and if voluntary, what is the default rate structure? Michael Godorov, the manager of smart meter operations for Pennsylvania Power and Light (PPL), indirectly addresses the question of opt in, opt out, or mandatory participation arguing that the peak-off peak difference must be high enough to induce the consumer to change their behavior.71 For example, if opt-in is chosen, and the alternative is flat rates, we can expect customers to opt in who expect to benefit, even if they don‟t shift their use pattern. This likely also increases the revenue recovery risk the utility faces since those who opt in are more likely to be those who expect to benefit. The customer‟s existing use pattern will play a large role in determining who wins and who loses from DP. If program participation is voluntary, this raises a concern about Adverse Selection. Borenstein argues that a RTP can be implemented using opt in as long as there is no cross subsidization between customers who select the RTP and those who choose to remain on the flat rate.72 While he may be right as a matter of theory, there probably isn‟t a rate design in existence at a real-world utility that has no cross-subsidization in it. Anyone with actual rate case experience knows that rate design and setting rates is a sausage-making process. By definition, rate design is all about who pays how much. X. Time Sensitive Rates and Supporting End-User Technology Another policy question is what requirement, if any, to include that require energy management devices at the point of end-use. This is a policy question since studies show that automatic control devices, even in the residential sector, substantially increase peak savings.70 “The Five Percent Solution…,” p. 4.71 “Consumer Response a Lingering Riddle for Backers of Smart Grid,” p.2.72 Severin Borenstein, “Time-Varying Retail Electricity Prices: Theory and Practice,” p. 30. 21
  24. 24. For example, in the PowerCents DC study in the Appendix, the addition of a smart thermostat programmed to respond to price signals about doubled the percentage savings from both CPP and PTR (called CPR in that study). One example of the impact that automatic controls have on peak savings is indicated by how much higher peak savings are when an automatic thermostat is present. For example, participants with all electric homes reduced peak use 22 percent without an automatic thermostat and by 51 percent with an automatic thermostat. Another study that showed significant savings increase when automatic control is introduced was California‟s experiment with various rates designs. One result of that experiment showed a residential peak load reduction for the average critical peak day of 23.5 percent without automatic controls and 34.5 percent with automatic controls.73 XI. Social Cost Arguments Supporting Mandatory Program Participation (or setting DP as the default with an Opt out) In light of the peak savings achieved by the experiments in the appendix, it‟s safe to say that customers who either opt out of (or fail to opt in to) DP rates, also benefit from those customers who do participate. These non-participating customers are known as „free riders‟ since they benefit without participating in the program. The FRP is a classic issue in the Natural Resources literature within economics. Basically, a FRP arises when it is prohibitively costly to exclude people from program benefits if they are not program participants. This type of problem overlaps with the economic discussion of externalities, Public Goods, and the property right theory. Some common examples of Public Goods are clean air and traffic congestion. Public Goods type problems arise when property rights are either not well defined (e.g., clean air), or they are well defined but very costly to enforce (e.g., illegal downloading).73 “California Statewide Pricing Pilot (SPP) Overview and Results 2003-2004,”p. 21. See:http://sites.energetics.com/madri/toolbox/pdfs/pricing/pricing_pilot.pdf. Also see “Retail Rate Options forSmall Customers, The California Statewide Pricing Pilot,”www.raabassociates.org/Articles/Levy_10.28.05.ppt 22
  25. 25. These types of problems are important because they illustrate cases where individual customers acting in their own self interest make decisions that are not the best for the customer base or the utility as a whole. When there is a Public Good present, the policy conclusion is that there is too little of that good being produced and consumed. There are more than economic arguments at stake in this decision. There are equity issues involved and perceptions of what individual choice means. Choice for whom? Under what circumstances? There are political issues involved arising out of perceptions about the sanctity of individual choice, among other factors, including but not limited to, how to define fairness. As we‟ve seen in California lately, some communities have gone so far as to try and criminalize smart meter installations. XII. Transitioning from Fixed Rates to Dynamic Rates Its one thing to talk about some far-off destination, and it‟s often quite another to plot the route and actually take the journey. A journey from a vanilla-type electricity rate designs, especially for residential customers, to one with different flavors is likely to have fits and starts. If the benefits are sufficient for that journey, what are some important considerations in planning our route? The IEEE whitepaper proposes a five-step process that makes general sense,74 1. Create customer buy-in by educating them about why their rates are changing, how DP helps the community (improve reliability, prevent even higher rates, help the environment), how they can use them to reduce their monthly bills. 2. Offer Supporting Technology, such as in-home displays that information such as, (a) how much electricity is being used by various end-uses, (b) real- time consumption information, and (c) that are married to enabling technologies like programmable thermostats, appliances, and home area networks,74 IEE Whitepaper, pp. 29-30. 23
  26. 26. 3. Design two-part rates with a fixed and uniform rate for a fixed amount of consumption followed by a second step with DP.4. Provide bill protection and bill comparisons by guaranteeing that bill would be no higher than what it would have been under the pre-existing rates (this shifts some risk to the utility) and phase the bill protection out over future years.5. Give customers choices by allowing customers to shift between different time varying rates and/or between time varying and fixed rates.These five steps do provide a starting point for designing such a transition.Clearly, they shouldn‟t be read as being sequential. Rather, they are key steps inthe transition.The two-part rate proposal is consistent with one of several possible approachesto managing monthly electric bills for vulnerable populations. Two-part rates andbill protection may help gain the support of consumer advocates wary of theequity impacts of time-varying rates on vulnerable populations. As long as thereare still net benefits after offering that protection, the overall system cost will belower than would have otherwise been the case. In the language of economics,that set of policies would be Pareto optimal and better than what currently exists.Few, if any, would be harmed and many would benefit substantially compared tothe status quo.There are a variety of ways to provide protection against higher monthly bills.Four different approaches were identified in Section VII above. When designing abill protection strategy, preference should be given to an approach that includessome exposure to price variability since doing otherwise also limits the customer‟sability to take advantage of price reductions.Allowing customers to switch between fixed and time-varying rates, warrants closeexamination to better gauge (a) how this fits into the other risk management 24
  27. 27. strategies in this list, (b) the potential for cost shifting, (c) it raises the issue ofAdverse Selection since those most likely to switch believe they will beadvantaged, and (d) if it is allowed, how to define the rules to offer meaningfulchoice with an eye towards how this choice impacts other customers. 25
  28. 28. Policy Recommendations1. Since at least one time-based rates experiment showed that CPP provided savings that exceeded RTP, there is a question about the need to adopt RTP to achieve peak use reductions. This raises a question about the rate design to start with that customers may opt out of (opt out is preferred to opt in for reasons discussed in the paper), or that a utility mandates. Three options are (a) flat rates with CPP, (b) TOU rates with CPP, or (c) full RTP. The selection should reflect your expectation of which of these alternatives better correlates with your overall policy goals.2. If RTP with opt out option is the rate option, TOU with CPP should be preferred to flat rates with or without CPP for customers who opt out of the RTP rate.3. Considering that (a) studies indicate CPP produces at least the same amount of reduction in peak use as Peak-Time Rebates (PTR), and (b) problems in setting PTR baselines, CPP should be preferred to PTR, perhaps except as part of monthly bill risk management for vulnerable populations.4. Off-peak rates should only reflect fuel cost of the marginal resource. On- peak power rates should reflect the fully allocated cost of the incremental resource. If wholesale power market prices are used, the on-peak price should include both energy and capacity. If there is a shoulder period, moving it closer to the off-peak rate will allow for a larger difference between the off-peak rate and the on-peak rate and avoid exceeding the revenue requirement constraint. Transmission and distribution costs should only appear in the on-peak rate. If other policy goals warrant a larger difference between on-peak and off-peak rates, increase the on-peak rate rather than lowering the off-peak rate. 26
  29. 29. 5. A smart-meter roll out should be coupled with AMI, or some other way to implement DP.75 Since studies indicate that an energy management system significantly increases savings, a rollout of some type of energy management system should be evaluated as part of the AMI (or its substitute) roll-out. The burden of proof for not including it should rest with the utility. 6. Moving to time-based rates discussed in recommendations 1-4, should include strategies to mitigate at least some of the very near-term customer bill risk (and utility revenue recovery risk). There are numerous ways to structure that mitigation and several different approaches were described in this paper.75 Smart-meters are the digital meters that are installed at the point of end-use. AMI requires that SmartMeters be installed but it also includes the hardware and software that operationalizes two-waycommunications between the utility and the Smart Meter and links with the record keeping system used ingenerating customer bills. As noted in the paper, there are alternatives to an AMI roll-out that can supporthourly pricing even in the residential sector. 27
  30. 30. Appendix - Summary of Selected Pricing ExperimentsThis appendix is a survey of DP experiments. There are at least 70 DP pilots andprograms nationwide. I made no attempt to draft a comprehensive list of all theseefforts. Instead, this appendix represents an overview of the more widely citedexperiments. This summary is a broad-brush overview. It is not designed to be athorough and detailed summary of any specific effort. Nor does it compare the efficacyof various efforts. A. Faruqui and Sergici Report76 The report by Faruqui and Sergici referenced in the above provides a survey of seventeen U.S. pricing experiments. The report provides an excellent, and concise, overview of a variety of pricing experiments in the U.S. and other countries. Those seventeen experiments used different pricing strategies (TOU and CPP), were conducted for varying lengths of time, with different number of participants, with and without enabling technology. The overarching conclusion is these pricing schemes can substantially reduce consumption at critical periods. People do respond to the price signals. Table 1 summarizes features of the experiments summarized in that report. Figure 1 illustrates the range of DR impacts from each of those experiments.77 Notes for Table 1 follow that table.76 Rather than replicate their entire report here, you will find the details in their report. What is includedhere is a table summarizing the experiments studied and the percentage reduction in peak load of eachexperiment.77 Both Figure 1 and Table 1 are form the report summarized. 28
  31. 31. 29
  32. 32. 30
  33. 33. 31
  34. 34. Below are summaries of a few other experiments that are not summarized in the Faruqui and Sergici report. B. PowerCentsDC One interesting pilot was PowerCentsDC78. Several reasons makes their pilot unique, 1. It was conceived in part by the official consumer advocate organization for D.C. 2. It tested three different price structures and various information formats, and 3. Limited income customers were recruited to test their price responsiveness. Three pricing plans were studied, Critical Peak Pricing (CPP), Critical Peak rebates (CPR), and Real-Time Pricing (HP for Hourly Pricing) that followed the wholesale electric price. Customers with limited income participated only in the CPR option. It should be noted that summer peak reduction under CPR for the low-income group was 11 percent while it was 13 percent for „regular‟ income customers.79 Among other conclusions, they note that CPP led to the greatest reductions in peak demand while CPR was the most popular option.80 Regarding low-income participants, participation rates were higher than for the regular income group, and the low-income group‟s peak reduction was only slightly less than that for the regular group.81 C. MyPower Pricing Pilot Program82 Public Service Electric and Gas Company (PSE&G) offered a residential TOU/CPP pilot pricing program in New Jersey during 2006 and 2007. The PSE&G pilot had two programs, myPower Sense and myPower Connection.78 PowerCentsDC Program, Final Report, September 2010. See: http://www.powercentsdc.org/ESC 2010-09-08 PCDC Final Report - FINAL.pdf79 Ibid, p. 11.80 Ibid, p. 5.81 Ibid.82 IEE Whitepaper, pp. 17-18. 32
  35. 35. myPower Sense educated participants about the TOU/CPP tariff and they were notified of a CPP event on a day-ahead basis. myPower Connection participants received a free programmable communicating thermostat (PCT) that received price signals from PSE&G and adjusted their air conditioning settings based on previously programmed set points on critical days. There were 1,148 participants in the pilot program; 450 in the control group, 379 in myPower Sense, and 319 in myPower Connection. The TOU/CPP tariff consisted of a base rate of $0.09 per kWh. There were three adjustments to this base rate, (1) a night discount of $0.05 per kWh in both summers, (2) an on-peak adder of $0.08 per kWh and $0.15 per kWh respectively in the summers of 2006 and 2007, and (3) a critical peak adder for the summer months that resulted in a critical peak prices of $0.78 per kWh and $1.46 per kWh, respectively, in the summers of 2006 and 2007. The results from this experiment were as follows,83 myPower Sense customers with Central A/C reduced peak load o by three percent on TOU only days. o by 17 percent on peak days. myPower Sense customers without Central A/C reduced peak load o by six percent on TOU-only days, and o by 20 percent on CPP days. myPower Connection customers (those with the PCT) reduced their peak demand o by 21 percent due to TOU-only pricing o by 47 percent on CPP days D. Power Smart Pricing Program According to discussions with ICC staff, the current ComEd and Ameren Power Smart Pricing program (PSPP) were legislatively created and are optional rates open to anyone.84 According to the company web-site for Ameren, the Power83 Ibid, p. 18.84 The ICC will be opening a docket soon to review the programs and the net benefits they may or may notbe creating for non-participants 33
  36. 36. Smart Pricing program is an hourly pricing program for residential customers. In this case, the electricity prices are set a day ahead by the hourly wholesale electricity market run by the Midwest Independent System Operator (MISO). According to ICC staff, the ComEd RTP uses DAP for advisory purposes but bills used the RTPs. The Ameren program started that way but reverted to using the day ahead prices for billing. ICC staff noted that Ameren now has about the same number of participants as ComEd despite having a customer base one third the size. Follow the link to learn more about Midwest ISO prices compared to flat rate prices.85 86 According to the T&D World column, a survey of 600 residential homes “Nearly 60% of residential energy consumers are willing to change their electricity-use patterns to save money, though many seek savings in return for signing on to a demand-response program.” One study performed by Frost & Sullivan titled “U.S. Smart Grid Market – A Customer Perspective on Demand Side Management,”87 In that study, they noted a significant percent of those surveyed (78 percent) said they would be interested in adjusting their power usage with a one-day notice of prices. A smaller fraction (60 percent) expressed an interest in allowing the utility to cycle their air-conditioner if that resulted in a lower utility bill. E. Texas88 The Public Utilities Commission of Texas (PUCT) staff wrote a report to the Texas legislature last year covering AMS deployment in Texas and efforts, to include DP pilots, outside the state. Among the points made are the following, Demand response programs that rely on dynamic pricing or TOU rates are only just beginning to be offered in Texas. Currently, Nations Power offers prepaid service with RTP. This service is only available to customers with85 See: http://www.powersmartpricing.org/about-hourly-prices/86 ICC staff has indicated that in May both ComEd and Ameren will be filing a variety of reports includingfour year program evaluations that will contain a significant amount of new information and will be thebasis for a docketed proceeding to review the programs.87 This report is quite expensive. I‟ve relied on a separate 15 slide presentation for thesecomments.88 Comments are based on correspondence and phone calls with PUCT staff. 34
  37. 37. smart meters installed on their premises. The smart meters provideconsumption data in fifteen minute intervals, enabling the company to providecustomers RTP. Customers can see their historical and current consumptionand current prices.TXU Energy offers a TOU rate that encourages their residential customers tosave money by shifting demand to off-peak hours. Under this plan, customerspay a higher peak rate during summer afternoons (1-6pm, M-F, May-October)when demand is highest and a lower rate at all other times of the year. Thelower rate applies to 93% of the hours of the year.Reliant Energy also offers a TOU plan that rewards the customer for shiftingdemand to lower priced off peak periods. Reliant‟s plan divides pricingperiods into three categories, off peak, standard and summer peak. Thehigher summer peak hours account for only 3% of the total hours in the year(4-6pm, M-F, April-October). Standard pricing applies to the other periods ofhigh demand and varies by season. Reliant‟s TOU plan is available tocustomers with smart meters.Reliant is also piloting the implementation of in-home displays withconsumers in Texas. This product offers consumers the ability to see realtime consumption and projected bill amounts. In addition, Reliant Energyoffers email alerts that utilize the 15-minute interval consumption data toprovide weekly insights into consumption and projected bill amounts.Gateway Energy Services recently launched the Lifestyle Energy Plan, athree month pilot program to test two different TOU rates. Under the pilot,customers will continue to be billed on their current flat rate structure but willbe able to see their monthly bill based on a TOU rate. Customers will haveonline access to reports detailing their usage and a side-by-side billinganalysis of the TOU rate plan versus their flat rate plan. At the end of thepilot, customers who would have saved money with the TOU rate plan willreceive a credit on their monthly bill equal to that savings. Criteria for 35
  38. 38. customer participation included having a smart meter installed and enrollment in Gateway‟s variable rate plan.75 F. Baltimore Gas & Electric Company (BGE) BGE recently tested customer price responsiveness to different dynamic pricing options through a Smart Energy Pricing (SEP) pilot. The rates were tested in combination with two enabling technologies: an IHD known as the energy orb, a sphere that emits different colors to signal off-peak, peak, and critical peak hours, and a switch for cycling central air conditioners. Without enabling technologies, the reduction in critical peak period usage ranged from 18 to 21%. When the energy orb was paired with dynamic prices, critical peak period load reduction impacts ranged from 23 to 27%. The ORB boosted DR approximately by 5%. BGE repeated the SEP pilot for the second time in the summer of 2009. Results revealed that the customers were persistent in their price responsiveness across the period. The average customer reduced peak demand by 23% due to dynamic prices only. When the ORB was paired with dynamic prices, the impact was 27%.89 G. The Connecticut Light and Power Company/ Plan-It Wise Pilot Another full scale pilot taking advantage of smart meters and three types of dynamic pricing was recently carried out by Connecticut Light and Power (CL&P). The Plan-It Wise Energy Pilot was designed as both a smart metering and rate plan pilot before the further deployment of smart meters to the 1.2 million metered electric customers in the CL&P service territory.90 Consumers who participated received a smart meter, along with an enabling technology such as a smart thermostat, energy orb or appliance smart switch. Residential customers enrolled in the Peak-Time Price (PTP) rate plan reduced peak demand by 23.3% if supplied with an efficiency enabling device, and 16.1% without such a device. Commercial and industrial (C&I) PTP customers reduced peak demand 7.2% with a device and 2.8% without. On average, Plan-it Wise89 Faruqui, Ahmad, Sanem Sergichi, Effects of In-Home Displays on Energy Consumption: A Summary ofPilot Results, Peak Load Management Alliance Webinar, April 6, 2010.90 Connecticut Department of Public Utility Control‟s Docket No. 05-10-03RE01 Compliance Order No. 4,Results of CL&P Plan-It Wise Energy Pilot, available athttp://nuwnotes1.nu.com/apps/clp/clpwebcontent.nsf/AR/PlanItWise/$File/Planit%20Wise%20Pilot%20Results.pdf. 36
  39. 39. residential participants saved $15.21 over the three-month pilot span, while C&I customers averaged $15.45 in savings.99 In an exit survey, 92% of the residential and 74% of the C&I participants said they would be open to further programs.91 H. Report on PG&E’s Opt In CPP Experiment92 The report contains ex post and ex ante load impact estimates for PG&E‟s residential time-based pricing tariffs. In 2010, PG&E had three time-based tariffs in effect: (1) SmartRateTM1 is a dynamic rate that is an overlay on other available tariffs. SmartRate has a high price during the peak period on event days, referred to as Smart Days, and slightly lower prices at all other times during the summer. Prices vary by time of day only on Smart Days; (2) Rate E-7 is a two-period, static time-of-use (TOU) rate with a peak period from 12 PM to 6 PM. This rate is closed to new enrollment; and (3) Rate E-6 is a three-period TOU rate with a peak period from 1 PM to 7 PM in the summer and from 5 PM to 8 PM in the winter (when partial peak prices are in effect). The report contains ex post load impact estimates for the above rates. It also examines the incremental impact of enabling technology on SmartRate demand response for customers that are enrolled in both PG&E‟s SmartRate and SmartAC programs. Load impact estimates for the SmartAC program are contained in a separate report. PG&E began offering SmartRate to residential customers in the Bakersfield and greater Kern County area in May 2008. This region was the first in PG&E‟s service territory to receive SmartMeters. By the end of the 2008 program year, enrollment in the Kern County area exceeded 10,000 customers. At the start of the 2010 summer season, enrollment had grown to around 24,500 customers and was extremely stable over the summer. In light of the pending termination of SmartRate, PG&E stopped actively marketing the rate in 2010, although enrollment remained open to new customers.91 Ibid.92 This summary is based on the Executive Summary of a report by Stephen S. George, Josh L. Bode,Elizabeth Hartmann, 2010 Load Impact Evaluation of Pacific Gas and Electric Companys Time-BasedPricing Tariffs, Final Report, April 1, 2011. 37
  40. 40. Under SmartRate, there can be up to 15 event days during the summer season,which runs from May 1st through October 31st. Prices only vary by time of day onSmartDays, unless a customer‟s underlying rate is a time-of-use (TOU) rate. Thepeak period on SmartDays is from 2 PM to 7 PM and customers are notified thatthe next day will be a SmartDay by 3 PM on the preceding day. The SmartRatepricing structure is an overlay on top of PG&E‟s other tariff offerings. SmartRatepricing consists of an incremental charge that applies during the peak period onSmart Days and a per kilowatt-hour credit that applies for all other hours fromJune through September. For residential customers, the additional peak periodcharge on Smart Days is 60¢/kWh.There were 13 event days in 2010. The average load reduction across the fivehour event window provided by residential SmartRate customers on each eventday was 0.26 kW, or 14.1%, which is similar in percentage terms to the 2009impact estimate of 15%. The average percent reduction ranged from a low of5.7%6 on June 29th, the first event of the summer, to a high of 22.8% onSeptember 10th. The average load reduction per participant ranged from a low of0.11 kW on the first event day to a high of 0.47 kW on PG&E‟s system peak day,August 24, 2010. On that day, SmartRate participants reduced electricity use by21.3% across the 2 PM to 7 PM event period.Aggregate reductions in peak demand on Smart Days ranged from a low of 2.6MW on the first event day, June 29th, to a high of 11.5 MW on PG&E‟s systempeak day, August 24, 2010. Aggregate load reduction for the summer averaged6.5 MW per event.Due to a notification problem, slightly less than half of all participants werenotified on June 29th, which largely explains the low impact estimate for that day.In addition to meeting the basic load impact protocol requirements, detailedanalysis has been conducted to understand how load impacts vary acrossseveral factors, including: (1) Local capacity area; (2) CARE status; (3) Numberof successful notifications; and (4) Central air conditioning saturation andtemperature (Note: CARE stands for California Alternate Rates for Energy, and is 38
  41. 41. a program through which enrolled, low income consumers receive lower ratesthan do non-CARE customers). The analysis also investigates several importantpolicy questions, including: (1) Attrition rates and the pattern of attrition forSmartRate participants; (2) Persistence of load impacts across multiple years;(3) Whether bill protection affects customer load impacts; (4) Whether loadimpacts vary between structural winners and losers; and (5) The extent to whichautomated load response via thermostats or direct load control switchesproduce incremental impacts over and above what customers with central airconditioning (AC) provide on their own.Key findings from this detailed analysis include, but are not limited to, thefollowing: Consumers do not appear to increase energy use in response to the slightly lower prices afforded on non-event days, nor do demand reductions on Smart Days carry over to other weekdays. CARE customers in aggregate responded less to price signals than other customers. However, after controlling for variations in underlying characteristics, such as air conditioning ownership, event notification and other factors, percent reductions for CARE customers are not significantly different from those of non-CARE customers. Event notification is highly correlated with load reductions. Comparative statistics show that both the average and percentage load reduction roughly triple between customers who are successfully notified through one option and those that receive four successful notifications. Customers that are enrolled in both SmartRate and SmartAC provide significantly greater demand response than those who are on SmartRate alone. There is a very wide range of demand response across customers. 36% of customers provide no load reduction at all, although one quarter of these participants (9% overall) did not receive event notifications. On the other hand, more than one third of all customers provided impacts of 0.2 kW or greater and 9% of all customers provided load reductions exceeding 1 kW. Load reductions do not decline over the course of multiple day event periods. Indeed, demand response on the second day of a two or three-day 39
  42. 42. event sequence is higher than on the first day. Response on the third day isabout the same as on the first day.Results indicate that (1) average load reductions appear to persist over timefor customers that have been on the program for multiple years; (2) There isevidence that first year bill protection mutes price signals to some extent(regression analysis indicates that average load impacts are roughly 25%less when customers are under first year bill protection than when they arenot.); (3) Load impacts for customers on a balanced payment plan are notstatistically significantly different from those of customers who are not onsuch a plan.The vast majority of customers who sign up for SmartRate have stayed onthe program. Attrition is quite low after adjusting for customer turnover that isunrelated to the program (e.g., account closures). The attrition rate is highestduring the first two months a customer is on SmartRate. CARE customershave marginally higher drop-out rates than non-CARE customers, all otherthings being equal. Drop-out rates differ only marginally in months that have alarge number of events compared with months in which fewer events werecalled. On the other hand, high bills are correlated with higher drop-out rates,although less so for CARE customers than for non-CARE customers (whosebills fluctuate much more due to the much steeper increasing block ratestructure faced by non-CARE customers.Load reductions are greater during summer months than in the winter, bothin absolute and percentage terms. The average peak period reductionacross the year is 0.16 kW or roughly 11%. In summer, the average is 0.21kW and 12.7%. Percentage impacts range from a low of roughly 6% to ahigh of approximately 14%. The 14% impact occurs in May, right after thehigher summer rates go into effect. 40
  43. 43. 41

×