Skip to content

Peak Electricity Pricing 2.0

It’s time to make critical peak pricing more flexible.

For meteorological reasons that I don’t pretend to understand, the same weather phenomenon that brought the devastating hurricane Harvey to Texas and the Gulf Coast played a role in last week’s record-breaking heat wave in California. It was so hot in Northern California that for the first time our local utility, PG&E, called a critical peak pricing (CPP) day on a Saturday, as well as last Thursday and Friday.CPPcallsFig1

Critical Peak Pricing programs (which go by different names at each utility…PG&E’s residential CPP is called Smart Days) are the most common way in which utilities in the US charge higher prices when demand is straining the system. In a May 2014 blog, I discussed why an alternative approach to reducing peak demand – paying customers to reduce their consumption – creates perverse incentives and is much less effective.

Traditional Design of Critical Peak Pricing

The structure of CPP programs is pretty standardized: over some set of summer months (June-September for PG&E), the utility has the option to call up to X days as critical peak days (15 for PG&E), typically making the announcement in the afternoon of the previous day.  On those peak days, the utility charges a much higher price for electricity during the high-demand period (2 PM-7 PM for PG&E).CPPcallsFig3

To date, residential CPP programs have been implemented as optional tariffs, as I blogged about in November 2015. The inducement to sign up for a program where you pay more for electricity on hot days is that you pay less for electricity on the other summer days. PG&E’s program, for example, charges an extra $0.60/kWh during those designated 75 CPP hours (15 days X 5 hours), but charges $0.024/kWh less during the other 2853 hours of those summer months.

Because the CPP price is set before the summer begins, and the total number of CPP days is limited, CPP programs are viewed as simpler and less risky for customers than full-blown real-time pricing, which changes the retail price hourly to reflect changes in the wholesale price of electricity. But the simplicity and risk mitigation of current CPP programs also reduces their effectiveness.

Guessing the Future to Call CPP Today

One of the big barriers to using CPP programs effectively is the limited number of days on which peak prices can be called. It’s easy to see one problem with a fixed number of CPP days: there will be some mild summers where the grid is never strained and no need for CPP days, and other harsh summers where the limit prevents the utility from sending appropriate price signals on all the scorching days, especially towards the end of the summer. CPPcallsFig2

A more subtle problem is that each time it considers calling a CPP day, the utility has to try to guess the weather for the rest of the summer in order to know the right threshold. If the summer turns out to be hotter than normal, the utility is going to regret some of the CPP calls it made in the beginning.

That’s the sort of summer we’re having this year. The California ISOs highest demand ever was 50,270 MW in summer 2006. Most annual peaks since then haven’t cleared 47,000 MW and none had cleared 49,000 MW, until last Friday. And then Saturday’s peak was over 47,000…on a Saturday!

With its CPP call on Saturday, PG&E used the 14th of its 15 days. So it has one bullet left to shoot with more than one-fifth of the program days left to go for this year, a period during which it has called 22% of all its CPP days over the last six years. (We have hot Septembers in California).

A More Flexible CPP Design

So is there a way to keep CPP programs relatively simple and low risk while allowing more flexibility in the number of days called? Yes! An approach that I studied in a paper back in 2013 (see section 7) would allow complete flexibility in the number of days called, but would still largely protect customers from the bill volatility that price spikes can cause.

The problem with the current structure is that it balances all of the additional revenue earned from a small number of CPP hours against a uniform price reduction during all other hours. That approach means that the utility has to stick to the planned number of CPP days — regardless of how stressed the grid actually is — or it will collect too much or too little revenue. To minimize customer bill volatility and utility profit volatility, typical CPP programs sacrifice pricing efficiency, which was the original point of the program.

But there is a way to reduce the risk to buyers and utilities and still call CPP events when and only when the system is stressed — as many or as few as is appropriate each year. Each time a CPP event is called, the utility also cuts prices by an approximately offsetting amount in the adjacent off-peak hours. Rather than balancing CPP revenues over the entire summer, this approach balances revenues within a day or two.

One question is whether a utility really could cut prices enough during surrounding hours to make each CPP call approximately revenue neutral. I examined that question in the 2013 paper and found that the answer is yes. The utility would have to cut prices to just a few cents per kWh in the non-CPP hours of the CPP day, and in some cases it would need to extend the “super discount” hours to off-peak times of adjoining days, but that would do it. And those super discount hours would help to further encourage timeshifting of some activities — such as running the dishwasher or washing machine — away from the CPP hours.

Making each CPP event revenue neutral would also make it easier to focus calls on the smaller number of days in which the system capacity is truly strained. In Josh Blonz’s paper from last year, which I blogged about in November, he showed that CPPs should only be called on average less than 10 days per year, more in some years, fewer in others. Josh also showed that on the CPP days that are called, the CPP price should be much higher than is used in most programs. That sort of implementation is a lot more difficult when a utility has little flexibility in the number of CPP days it can call each year.

The Growing Need for Demand Flexibility

Recent research — including a paper published by Max and co-authors just this week — shows that as the planet warms, demands on electricity systems will change, often in hard-to-predict ways.  Ramping up intermittent renewables will also increase the premium on being able to dynamically alter consumption of electricity. The emergence of a significant PV-driven “duck curve” years sooner than anticipated is just one demonstration of the value of being able to adjust demand at different times of the day.

Flexible pricing should be one of the primary instruments used to address changing stresses on the grid due to higher renewables penetration, increased A/C use, expansion of the electric vehicle fleet, and  other trends in electricity supply and demand. The current CPP programs were a fine first step in that direction, but we need to keep innovating in pricing structures, just as we need to keep innovating in technologies.  Revenue-neutral CPP days would be an excellent next step as we move towards a system that integrates new technologies, reduces carbon emissions, and controls costs.

I tweet energy articles/research/opinions most days @BorensteinS







Severin Borenstein View All

Severin Borenstein is E.T. Grether Professor of Business Administration and Public Policy at the Haas School of Business. He has published extensively on the oil and gasoline industries, electricity markets and pricing greenhouse gases. His current research projects include the economics of renewable energy, economic policies for reducing greenhouse gases, and alternative models of retail electricity pricing. In 2012-13, he served on the Emissions Market Assessment Committee that advised the California Air Resources Board on the operation of California’s Cap and Trade market for greenhouse gases. Currently, he chairs the California Energy Commission's Petroleum Market Advisory Committee and is a member of the Bay Area Air Quality Management District's Advisory Council.

19 thoughts on “Peak Electricity Pricing 2.0 Leave a comment

  1. Electricity 3.0?
    1. “The customer receives his or her electric power as a combination of real and reactive power (Watts and VARs).”
    2. “Watts, which deliver real energy, MUST come from a generator, but VARs can be created or used anywhere, using capacitors or inductors (Inverters and Wonderful Energy Resources). Both watts and VARs require current to deliver them.”
    3. “By managing loads (while balancing the current phases), distributed generation and distributed storage devices located near the grid edge, the efficiency of central generation can be maximized and maintained. At the same time, because loss varies with I2, the total delivery loss at peak load may be 4x the loss at the average (half) load.”
    Peak electricity pricing should not encourage higher losses and pricing should accurately reflect the accurate accounting of Watts, VARs, and Power Quality. I think James Prescott Joule would agree.

    • Only resistance losses are exponential. Core losses are constant.

      This underlying point is very valid: losses are much higher on-peak, and distributed resources, controlled water heating, ice-storage air-conditioning, and distributed batteries can avoid much more in cost than simply the kilowatt-hours of usage shifted in time.

      It works out to a kilowatt of load reduction at the end-load at the distribution system peak hour is worth about 1.4 kW at the generation level, once marginal line losses and avoided reserve requirements are taken into account.

      See our paper Valuing the Contribution of Energy Efficiency to Avoided Marginal Line Losses While this paper focuses on energy efficiency, the same principles apply to distributed generation and storage resources.

      • Jim, not sure to what “core losses” you’re referring, but your paper fails to consider inefficiencies inherent in ice storage air conditioning, controlled water heating, distributed batteries, and the lower-voltage transmission required to reach them. It fails to consider efficiency losses when end users are forced to curtail air conditioning or other appliance use at the behest of their utility or independent system operator. It treats “efficiency” as if it was an energy asset – as if it was infinitely scalable to meet projected future demand; as if the ratio of energy input/output might exceed 1.

        On a modern grid the most efficient route for energy from generator to its ultimate goal – the accomplishment of work – is one which matches supply to demand. Demand spread over millions of users, except in rare instances, results in a gentle load curve permitting CCGTs, nuclear power plants, hydroelectric, and geothermal energy sources to transmit energy efficiently using high-voltage AC or DC transmission, with a minimum of spinning reserves.

        In my city, each block does not have its own police and fire departments. We don’t pump our own water from the ground nor generate our own electricity, for good reason – it’s more efficient, and thus both cheaper and more environmentally-friendly, to do it that way.

  2. Severin, in 2015 your Haas colleague Lucas Davis and Catherine Hausman at the University of Michigan published an analysis of the 2013 closure of San Onofre Nuclear Generating Station (SONGS), showing

    “…the lost generation from SONGS was met largely by increased in-state natural gas generation. In the twelve months following the closure, natural gas generation costs increased by $350 million. The closure also created binding transmission constraints, causing short-run inefficiencies and potentially making it more profitable for certain plants to act non-competitively.”

    Binding transmission constraints are transmission bottlenecks which can’t be overridden in CAISO’s Energy Management System. That they’re now a problem highlights the fact California’s grid was largely designed around baseload sources like SONGS, and that changes in priorities favoring renewable energy are responsible for transmission problems facing the CAISO grid, undiminished emissions, and higher electricity prices.

    So, sure – we can redirect the burden to consumers (under the patronizing label of “Flexible Demand”); we can build out new gas generation in areas more geographically advantageous for transmission; we can facilitate imports of electricity of dubious provenance from other Western states.

    Or, we can power California’s grid using the carbon-free baseload power for which it was designed. If that’s not a no-brainer, the current direction of California energy certainly is.

  3. Responding to Sev asking for evidence PTR works.

    BGE and other PTR results. Slide 10 70+% participation, measurable results. As a result, BG&E expanded the program to include all consumers.

    Also: results of 17 PTR pilots, averaging 10% – 20% savings on peak for enrolled participants. Page 27.

    Better with advanced “hands free” technology.

    • Hi Jim

      Thanks, but that wasn’t exactly what I’m looking for. The second paper is just a survey of studies and claims these reductions, but doesn’t explain how they were estimated. All it says is that some of the studies are more scientifically valid than others. I suspect that many are the ones I have already looked at.

      The presentation by Harbaugh looks interesting, but it is just a slide deck. Is there a paper that actually explained in detail how the study was done? I couldn’t find a link or reference to one in the slide deck.


  4. Baltimore Gas and Electric does it the other way around. The pay a credit for reductions compared to a baseline. There is no limit to the number of events they can call. Every customer is enrolled, as there is no penalty for non-performance, only rewards. They are getting almost as good response with Peak Time Rewards as other utilities are with CPP, but without the limited number of events or the consumer resistance. We are gradually learning the intersection of economic theory and consumer psychology. It does not always lead where we economists think it should.

    • Hi Jim
      I’d like to see the study that shows that peak time rebates reduce consumption compared to a real control group. As I point out in my blog on peak time rebates (, the studies that I have seen suggests that the real savings are extremely small. The studies that claim large savings generally are only counting the customers who get the rebates. There are many other problems with peak time rebates, including rewarding customers who have more volatile consumption. Please see my blog and the paper it (and this blog) refers to.
      But I am open to new evidence, please post the link to the study that suggests PTRs actually save substantial electricity.

  5. Pricing of electricity should also be graded for Power Quality (see link that follows). I would add Current Phase Balancing to the pricing structure as well. The potential amount of electrical grid capacity that could opened up with new technologies (3DFS) that address the loads collectively that you mentioned, like increased AC use and electric vehicles along with anything connected in the system that requires reactive power is huge. Additionally, we should also be measuring the quality of active and reactive energy since both quantities are measured in Joules and this reduction of heat should be reflected as discounted pricing.
    “A Three-Part Electricity Price Mechanism for Photovoltaic-Battery Energy Storage Power Plants Considering the Power Quality and Ancillary Service” –

  6. Since the system is increasingly stressed, perhaps we need to consider what is done in many other parts of the world. On peak demand periods implement rolling power outages for ALL. Those who wish to and can afford get local backup systems; emergency systems [hospitals etc] either would be on separate circuits, or also get backup systems.

    This is a 3rd world solution for the 3rd world problem we are facing. It makes absolutely no sense to set up generation capacity to meet PEAK loads, when the slight outages can trim the peaks. We cannot depend on demand management [by the consumer] so we manage supply.
    It also will likely lead to greater use of solar perhaps even without storage to address only those really hot days when the sun is out there in force.

    • California has so much excess capacity for the foreseeable future, that rolling outages is not a reasonable suggestion. In fact, the premise of CPP in the current environment is about fighting the last war. The utilities go through the motions on hot days of reaching for capacity resources, but the generation system is never really under stress. (It’s a different story for the distribution network which we have consistently ignored when discussing real resource adequacy.) The new “war” is about providing flexibility through customer interactions. The utilities have been slow (reluctant?) to offer real rate incentives to offer up these services.

      • mcubedecon, the excess capacity you describe, whatever it is, is not available in the areas which need it. That’s why there are FlexAlerts – electricity does not magically fly to wherever it’s needed. It must travel through wires, which is a very, very complicated process. One with constraints which, if ignored, could cause those wires to literally melt from their poles.

        CAISO and utilities really don’t want that to happen. So during California’s last CPP day, to prevent a Level 3 Emergency and rolling outages, CAISO instituted “Auto DR” – a term they use to describe remotely cycling, or even shutting down, large air conditioning systems of some customers.

        Would allowing CAISO to shut off your air conditioning fall under the heading of “flexibility”, or “dependency”?

  7. I just read Stephen Kaffka’s comment.

    This is the typical reaction of consumer advocates. Electric rates are the wrong tool for addressing poverty and inequality. That’s the role of tax policy and social safety nets like Medicaid.

  8. “The problem with the current structure is that it balances all of the additional revenue earned from a small number of CPP hours against a uniform price reduction during all other hours.”

    The revenue neutrality issue is hardly the most important problem with CPP programs. A far worse shortcoming is that these programs are economically inefficient and potentially ineffective. Why? Because they impose a peak price that is predetermined, thus is not responsive to the actual system marginal cost during the peak period. Thus, the peak price will be either too low or too high. Furthermore, during periods of capacity scarcity the peak price may not be high enough to suppress demand by enough to avoid involuntary load shedding.

    The way to ameliorate this shortcoming is to vary the peak price based on how severe the capacity shortfall is forecasted to be. Oklahoma Gas & Electric has implemented a program (SmartHours) that charges four different peak prices based of expected next-day system conditions. It has been well received and has been successful in deferring the need for adding new generating capacity.

    Still, the ideal retail rate design is a real time price that varies in each hour (or even each 5-minute interval) throughout the day and night, 24/7. This is how wholesale market prices vary and eventually retail prices will reflect the wholesale prices (plus include surcharges for distribution losses plus a scarcity component when generation or distribution capacity is in short supply) as distributed generation and local storage gets ever cheaper.

    We should be getting consumers acclimated to real time pricing, rather than confusing them with these inefficient workarounds. Protecting consumers from paying the true cost of serving them is not doing them any favors.

    • Us electricity wonks may be willing to pay attention to 5 minute pricing info, but the vast majority of consumers want consistent, predictable pricing. Almost everyone has something more important in their life than managing their electricity loads minute to minute. (And relying on technology to do so is only attainable by the wealthy few, and not likely to be personally cost-effective.) Notably, the single biggest driver of rooftop solar PV is the assurance of a constant, known price for decades, even if the panel cost more initially. (Home ownership runs on the same incentive.) Consumers should be offered a menu of pricing options, with differentials based on the risk they are willing to take on (let’s end the “revenue neutrality” requirement that makes TOU customers pay the same overall as flat rate/hedged customers.) And we should be rewarding consumers who make long-term energy-saving investments with pricing grandfathered to moment when they make the investment, just as we do now with generators.

  9. Missing from this blog is any reflection on the experience of people in very hot areas like the Central Valley or low dessert regions when temperatures reach extremes. People were advised last week to remain indoors due the combination or extreme heat and smoke from numerous wildfires (a different regulatory failure). Cutting down on energy use or increasing prices greatly during peak demand would penalize people in such areas disproportionally, especially people with marginal incomes and less than comfortable housing. A writer at the Hoover Institution who lives in the San Joaquin Valley reports seeing many people in Walmarts and other large stores in his part of the valley during peak heat events, who are trying to find cool spaces. He infers that they cannot afford the current costs of power or have poor domestic conditions to deal with heat, so have to leave home to manage their discomfort. He further notes that people in favor of higher energy prices often can afford such prices and live in cooler coastal regions where energy bills are much lower. This blog would seen to support those charges.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: