Electricity Rate Design for the Real World
For decades economists have bemoaned the fact that retail electricity prices don’t adjust to reflect the volatile cost of providing energy. Because electricity is not storable, the wholesale cost can change by a factor of five or more within a single day, but the price to most end-use customers remains constant. It’s the equivalent of the price at the gas pump being held fixed while the world oil price ranges between $20 and $140 a barrel…only compressed in time.
Time-varying electricity pricing offers benefits both now and for the future. The immediate benefit is that raising prices at peak times (when producing each extra kilowatt-hour is most expensive) and lowering them at off-peak times would move some consumption off the peak and reduce the need to build additional “peaker” power plants. In the longer run, sending such time-varying price signals would allow us to better synchronize consumption with electricity production from intermittent resources, such as solar and wind.
A lot of research has attempted to estimate how much time-varying pricing causes customers to reduce or shift their demand, while some theoretical and simulation work has focused on what the “best” pricing structure would be. A new Energy Institute working paper contains elements of both of these approaches, but it also takes a third tack that is likely to be more helpful to policymakers.
In “Making the Best of the Second-Best: Welfare Consequences of Time-Varying Electricity Pricing,” Josh Blonz (a researcher at the Energy Institute who is finishing his PhD this spring) first estimates the impact of a critical peak pricing (CPP) program, under which the utility more than triples the retail price of electricity on up to 15 hot weekday afternoons of the summer, and lowers the price slightly at all other times.
Next, he compares the CPP pricing to the “gold standard,” real-time retail pricing (RTP), under which the retail price would change every hour of the year to reflect changing wholesale prices. Josh shows that while the CPP program is cost effective, it is capturing less than half the benefits of RTP. But the paper doesn’t stop there. Josh shows that some simple and intuitive adjustments to the CPP program could greatly increase the benefits it yields, thus offering policymakers not just analysis of what has been done, but also practical steps for future improvements.
The study examines Pacific Gas & Electric’s (PG&E’s) CPP program for small commercial and industrial (C&I) customers, which is itself a bit of an innovation. The vast majority of CPP studies look at residential programs, despite the fact that C&I customers consume two-thirds of the electricity and nearly all of the demand that faces time-varying prices in the U.S.
How Much Does Price Variation Change Consumption?
If you read this blog regularly, you already know that empirical economists are obsessed with sorting out causality from mere correlation. In this case, we want to know how much raising price on hot summer afternoons causes customers to consume less by comparing the customers on CPP to another group who are otherwise virtually identical, but are not on the program. The gold standard is a randomized control trial (RCT) as is done in many medical studies (and as Catherine and Meredith, along with four other co-authors, have done in a study of CPP implementation by Sacramento Municipal Utilities District).
If you can’t do an RCT, a clever alternative takes advantage of thresholds or cutoffs in eligibility criteria for the program, such that the subjects who just barely meet the criteria for participation are virtually identical (as a group) to the subjects who just barely miss the criteria. Lucas and Judd Boomhower, for instance, used this approach in a study of the energy savings from a refrigerator and air-conditioner replacement program in Mexico. Josh takes this path, comparing those who just barely met with those who just barely missed eligibility for the CPP program based on the date their smart meters were installed.
He finds that when PG&E calls a CPP day on hot summer afternoons, C&I customers in the inland (hot) part of their territory cut their consumption by an average of 13%. Customers on the coast may reduce their usage a bit, but it is likely not much and statistically indiscernible. The study points out that this is probably because temperatures on the coast are still very moderate on CPP days (averaging around 70 degrees during the CPP call hours in the two sample years), so the most effective price response action — raising air-conditioning setpoints — is less available to coastal customers. In fact, in the inland areas the study finds that CPP-driven savings rise as the temperature increases.
The inland businesses that save the most on CPP days, Josh shows, are the “non-customer-facing”: offices, manufacturing, warehousing, etc. The “customer facing” businesses — retail, restaurants, movie theaters, etc. — don’t seem to respond, which makes sense since part of what they are selling is a pleasant, air-conditioned atmosphere. Plus, it is much more practical for a business to warn its employees to dress for a warm day at work than it is for a retail store to give the same warning to potential customers.
The primary value when these businesses do reduce their demand on hot summer afternoons is that less generation capacity needs to be built. The study unpacks the regulatory process that determines these capacity requirements in order to evaluate how much money the CPP program saves due to reduced capacity building. To evaluate the program fully, however, one has to also take into account the loss of value customers suffer from not consuming as much electricity, such as setting the A/C temp a bit higher than they normally would. Josh does this and finds that the benefits of the CPP program well outweigh the costs.
Sharpening a Blunt Tariff Design
Of course, CPP is a blunt instrument: The number of days that can be called each summer is in a fixed range, the hours it is in effect are the same for all such days, and likewise the price increase is the same for all such days. Economists have long argued for more flexible pricing that changes hourly – real-time pricing — but have met with stiff opposition from regulators and some consumer groups, due to perceived complexity and fairness concerns. Josh shows that even though this CPP program has to be considered a success, it is still producing only about 43% of the benefits of full-on RTP.
So, the study asks, if RTP is not feasible, can the CPP program still be improved? The answer is an emphatic yes. Since the benefits are primarily from reducing usage on just the few hottest days of the year, the paper looks at lowering the total number of CPP days called, while increasing the price more on the days where it is called. The paper finds that cutting the number of CPP days nearly in half, but charging a price that is more than 50% higher on those fewer days, greatly increases the net benefits of the program.
Even with weather uncertainty and having to call the CPP event a day in advance, cutting back to 8 CPP days per summer still makes it very likely the utility can hit the 2 or 3 summer days that actually strain capacity. And raising the price by a greater amount on those days means that more capacity gets saved. At the same time, eliminating up to 7 CPP days per summer when capacity wasn’t going to be strained means that customers aren’t losing the benefits of the extra electricity consumption on those days. Josh’s proposal doesn’t get to the perfectly efficient outcome, but working within the real-world constraints of regulation and political feasibility, it turns out to nearly double the net benefits of the program.
It’s unusual for an academic research paper to have a punchline that can be so easily appreciated and implemented by regulators. This isn’t pie-in-the-sky thinking about fixing all the flaws in electricity rate design, but first-rate research on a good pricing policy and insight about how to make it much better.
Severin Borenstein View All
Severin Borenstein is Professor of the Graduate School in the Economic Analysis and Policy Group at the Haas School of Business and Faculty Director of the Energy Institute at Haas. He received his A.B. from U.C. Berkeley and Ph.D. in Economics from M.I.T. His research focuses on the economics of renewable energy, economic policies for reducing greenhouse gases, and alternative models of retail electricity pricing. Borenstein is also a research associate of the National Bureau of Economic Research in Cambridge, MA. He served on the Board of Governors of the California Power Exchange from 1997 to 2003. During 1999-2000, he was a member of the California Attorney General's Gasoline Price Task Force. In 2012-13, he served on the Emissions Market Assessment Committee, which advised the California Air Resources Board on the operation of California’s Cap and Trade market for greenhouse gases. In 2014, he was appointed to the California Energy Commission’s Petroleum Market Advisory Committee, which he chaired from 2015 until the Committee was dissolved in 2017. From 2015-2020, he served on the Advisory Council of the Bay Area Air Quality Management District. Since 2019, he has been a member of the Governing Board of the California Independent System Operator.
Rather than using this quasi-experimental approach, did you consider using within-subjects regressions on non-event days to estimate counterfactual usage on event days? I work at a firm that uses this method to evaluate event-based programs (including CPP programs) and have had success with this approach, particularly when there are non-event days that are similar to event days in terms of temperature.
Your characterization of CPP in the fourth to last paragraph makes an unnecessary limiting assumption. CPP rates were designed, implemented, and have been operated successfully since year 2000 that include the following: (1) no limit on how many days per year the rate can be dispatched and (2) hours of dispatch on a CPP day based on system need, no fixed dispatch term. While RTP is a theoretically more desireable rate from an economics perspective there are in fact two other features that can improve CPP, specifically:
(1) There is no reason CPP can’t include two or three levels of critical peak pricing based on pre-defined system conditions. Automated control signals to customers and customer response can easily be adapted to respond accordingly. Smart meter/utility billing systems can also be adapted.
(2) There is clear regulatory advantage to targeting selected customers for participation, either geographically or based on known usage characteristics (smart meter data). Targeting allows technical support to be focused on those who might need automated or other assistance. Targeting also allows utilities and other organizations to focus on those customers not suitable for CPP to provide educational, investment, and other resources to prepare them for transition to CPP. These options were actually presented in a NARUC webinar on November 30, 2011 that you and I participated in.