Skip to content

Are Demand Charges Fair?

Appeals to equity don’t salvage the argument for demand charges.

In the past, I have said some pretty unkind things about demand charges in electricity tariffs.

Demands charges are fees paid (mostly) by commercial and industrial customers based on their highest usage in any 15-minute (or sometimes, hour) interval of the billing period. They often constitute 30% or more of a customer’s bill, with most of the remainder based on total electricity consumption. Demand charges are going through something of a revival — even spreading to some residential rates. They have become particularly popular among utilities trying to get more revenues out of customers who install solar.

RethinkingDemandChargesFig2In a nutshell, my argument against the economic efficiency of demand charges has been that they accomplish nothing that couldn’t be better done with dynamic pricing. Demand charges are based on the customer’s highest 15 minutes of usage regardless of whether the cost of actually providing electricity in that period was particularly high. Charging prices that reflect the actual cost of supplying electricity all the time makes a whole lot more sense than a demand charge that whacks a customer for high usage in a single 15-minute period whether or not it is costly to provide electricity during that period. There are many branches to the argument about the efficiency of demand charges – such as whether they are a good way to reflect high energy prices or distribution-level capacity constraints, or to cover customer-specific fixed costs of service – but I addressed those in my previous blog.

I want to return to demand charges, because a number of people I respect in industry, government, and consumer groups still assert they should be part of modern rate design, and I recently realized that my arguments about economic efficiency may be missing their main point. Their case is often made on equity grounds, not efficiency. So, today I want to focus (mainly) on the equity aspects of demand charges.

RethinkingDemandChargesFig3The Revenue Shortfall Problem

Equity arguments arises in part because regardless of where you come out on efficient electricity pricing, those prices are not generally going to raise enough revenue to cover a utility’s total costs. To make up some of the difference from residential customers, most utilities levy a monthly fixed charge that is unrelated to consumption level. This doesn’t seem very fair, however, because heavy-use households pay the same fixed charge as light users. (For some very geeky economic thoughts about the ideal fixed charge, see the “bonus” section below.)

In fact, when I have surveyed people about this concern, they generally suggest that the fairest approach is to divide up any revenue shortfall proportionally to usage, that is, make it a volume-based charge. But that is when economists (including me) pipe up about the inefficiencies of charging a volumetric price that is far above marginal cost, because it discourages valuable electricity use. This is especially true when that discouraged usage would have substituted for more-polluting energy sources, like gasoline for transportation or natural gas for space or water heating.

This conundrum is worse when it comes to commercial and industrial customers. It seems even less fair to apply the same monthly fixed charge to a large factory that uses 3000 MWh per month as to a corner store that uses 3 MWh per month. Monthly fixed charges just don’t make a whole lot of sense when the scale of consumption varies drastically across customers.

That’s where demand charges come in. Demand charges are often seen as a way to charge something like a fixed fee that is higher for larger customers, without just making it part of the volumetric price. So, in practice do demand charges improve on the equity of tariff design compared to the other available pricing options? I don’t think so. RethinkingDemandChargesFig1

Evading Fairness

Once we agree on efficient dynamic pricing for electricity, demand charges seem to be a less equitable way to recover any remaining revenue shortfall than putting an adder on each kilowatt-hour. On efficiency grounds, I’m not a fan of such an adder, but it is pretty appealing on equity. To begin with, the equity concern most people have with fixed charges is that large-volume customers should pay more towards any revenue shortfall, not that customers with peakier demand should pay more.

Also, demand charges arbitrarily, almost randomly, impose more costs on some customers than others. Consider customers A and B that have the same total consumption and use the same quantity during times when the system is constrained. But customer A consumes its other energy in a more concentrated set of hours than customer B does. Customer A is likely to face a higher demand charge, and pay more towards the revenue shortfall, even though it is not doing anything more than customer B to constrain the system. With an adder to the volumetric price, that doesn’t happen.

Behind-the-meter (BTM) technologies have further undermined the fairness of demand charges. BTM storage can allow customers to substantially, and largely inefficiently, evade demand charges, which has become a beyond-cottage industry for consultants and battery producers catering to commercial and industrial customers. Installing batteries can be a bill-reducing strategy for the customer, but its often inefficient for society as a whole, because in the high-price states where it’s mostly taking place the individual savings are largely just cost shifts to other customers. And among residential customers, you can be sure that wealthy households are the first to install BTM storage to lower their demand charges. A per-kilowatt-hour adder does not create artificial savings for customers who install storage.

To be fair, a kilowatt-hour adder can be substantially evaded through installing BTM generation, a strategy that is also disproportionately pursued by wealthier households. In fact, this problem using BTM generation to evade cranked-up volumetric charges is what has led some utilities to advocate for demand charges instead. But that seems like a 2010 solution to a 2020 problem. As BTM generation is increasingly paired with storage, demand charges become a less effective mechanism for capturing additional revenue from such “prosumers.”

The bottom line is that neither demand charges nor kilowatt-hour price adders are a good solution to the revenue shortfall problem. The inefficiencies and inequities they create, particularly as BTM storage and generation expands, make the case even stronger for moving quickly to adopt efficient dynamic pricing. But efficient price variation alone generally isn’t going to raise all the revenue a utility needs. So the search continues for other revenue sources that fill the gap equitably without triggering evasive behavior and investments that primarily shift costs.

Bonus Section: Thinking about the ideal fixed fee (For those who really want to dive into economics.)

When considering pricing issues, I often find it useful to think about the ideal tariff design as a reference point, even if it is not feasible. When it comes to recovering a revenue shortfall after setting efficient prices, for efficiency sake we would like something that is not based on the level of consumption the customer chooses. But equity might suggest that customers who get more value out of the system should make a larger contribution.

A theoretical solution is straightforward, though impractical. If we knew each customer’s demand function, we could calculate the consumer surplus each customer obtains when the volumetric price is set equal to full social marginal cost (SMC, which includes the cost of pollution externalities, as I discussed in a blog last fall) in each hour. Then we could impose an individualized fixed charge equal to X% of their consumer surplus, where X (between 0 and 100) is set just high enough to cover the utility’s revenue shortfall. No customer would have an incentive to inefficiently reduce or increase their usage, because the consumer surplus from each unit they value more than SMC is still positive after the utility has taken X%, and the consumer surplus from any unit they value less than SMC it is still negative after the utility has taken X%.

The idea of a flat “tax” on consumer surplus is appealing on equity grounds, because customers who get a lot of value from their total electricity consumption – who would generally be high-usage customers – would pay more towards the revenue shortfall. (As we do now, we would probably want to carve out a different rate structure for customers who have high value and usage due to medical necessity.) But if one were considering equity from behind a Rawlsian “veil of ignorance”, one might opt for a progressive fixed fee where X is larger for consumers who get a lot of surplus, though probably only if such consumers were richer overall, not if their high usage were due to, for instance, having many people living in the same house.

This thought exercise illustrates the trade-off we face in the real world where we cannot observe a customer’s consumer surplus (though I suppose Facebook will have that information soon enough). If we try to use consumption quantity as a proxy for consumer surplus, and scale the “fixed fee” to quantity, it becomes a volumetric charge and creates an inefficient gap between price and SMC. If we ignore consumption quantity and ask everyone to contribute the same fixed fee, we are taking a much larger share of the consumer surplus of some customers than others, and likely a larger share from poorer customers, which most people would consider inequitable.

I’m still tweeting energy news stories/research/blogs most days @BorensteinS

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Borenstein, Severin. “Are Demand Charges Fair?”, Energy Institute Blog, UC Berkeley, July 8, 2019,

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.

70 thoughts on “Are Demand Charges Fair? Leave a comment

  1. Severin,

    One more comment.

    The portion of the fixed costs that are not recovered through scarcity rents can be allocated to customers based on their maximum physical capability of drawing energy, which is typically determined by the master fuse at their meter. This would roughly discriminate between small and large customers within each class as well as among the various classes. It’s not perfect but it is workable and appears to me to be “reasonably equitable.”

    • This violates the first principle of smart rate design: that customers should be allowed to connect to the grid for no more than the cost of connecting to the grid. Universal service has been a national goal since 1936. This approach would require an apartment dweller who has an electric water heater (4.4 kW), an electric range (4 kW), and electric clothes dryer (3 kW), an air conditioner (3 kW) and other appliances to pay a 15 kW share of the revenue shortfall, even though their annual usage may be only 4,380 kWh, or an average demand of 0.5 kW.

      And, of course, in the apartment sector, while each customer may be METERED separately, the only element of the grid that is sized based on that maximum demand is the meter itself. Ten or fifty apartment customers may be connected through a common transformer, and it will be sized based on something less than the maximum expected diversified demand of the group (transformers can handle an overload for a few hours here and there). That diversified demand may be more like 2 kW per customer.

      Dr. Borenstein’s suggestion, collecting it across average demand makes more sense. My suggested variation, an equiproportional surcharge on a time-varying rate is, in my opinion, an improvement on that.

      • Jim,

        Whose “first principle of smart grid rate design?” Adherence to dogma is what has delayed the adoption of economically efficient retail rates. Much has changed since 1936. Surely you are not suggesting that we continue rate making practices from the Depression days. We achieved universal service decades ago.

        One can argue that Allocating revenue shortfalls based on maximum potential demand is equitable because it would equally apply to all customers so the actual amount charged any one customer would be derated by the larger denominator (I.e., the sum of all maximum potential demands), which would apply to all.

        Your proposal to proportionately gross up the energy prices was raised in the 1970s. The problem is that any deviation from marginal costs would discourage efficient consumption decisions. In contrast, almost nobody is going to disconnect from the grid to avoid paying a reasonable monthly customer charge.

        • Relative usage isn’t the sole determinant of cost on the distribution system. As Jim Lazar points out, and is well recognized in utility ratemaking, multifamily and single family residences create different demands on the system. The California utilities measure these differential effects in their billing determinants.

          The method of grossing up the price from marginal cost relative to marginal cost shares is Ramsey pricing, which is shown to be efficient. It’s the heart of equal percentage marginal cost (EPMC) pricing used for California utilities.

          • Ramsey pricing is not efficient pricing; it is a second-best pricing scheme that minimizes the societal cost (i.e., deadweight loss)of deviating from marginal cost pricing. Ramsey pricing prescribes that deviations from marginal cost should be inversely proportional to the elasticity of demand, e.g., that the largest price markup is applied to the load that with the lowest price elasticity. This only produces equal percentage markups (EPMC) if all loads have the same price elasticity (which is like – never). EPMC is done because it is easy to implement – not because it is correct in any theoretical sense.

          • Correct that Ramsey pricing is only second best pricing, as is every other single option actually feasible. (Equal lump sum payments without accounting for differences in wealth or income is not politically feasible in any possible sense so we are left only with 2nd best solutions.) Do you have a valid iron-clad study with no ranges of uncertainty that can distinguish the differences in elasticity of demand among every individual customers (going to rate classes is a rough kluge)? I didn’t think so. So you’ve proven my point that there is no available first-best solution, and what is agreed as the best second-best solution is really impossible to implement too. So we’re left with no option that implements marginal cost pricing in a manner that acceptably achieves efficiency.

  2. Severin,

    I understand that this article addresses the revenue requirement issue, rather than that of how to set the dynamic rates; however they are interrelated.

    Economic theory informs us that economic efficiency is maximized when all prices are set equal to their respective SHORT-RUN marginal costs (not LONG-RUN marginal costs as some people believe). But for this to be true we need to accept an expanded definition of SRMC that includes a scarcity adder when the system is capacity-constrained and the constraint is binding, i.e., the system is incapable of delivering enough energy to satisfy demand what would exist without the surcharge. The function of the scarcity adder is to toll the lowest value loads off the system in an efficient manner so that the remaining demand can be served. Fred Kahn discussed this in Volume 1 of his landmark word, “The Economics of Regulation.” Of course, to make this work the retail customers need to see the dynamic price signals and develop predictable responses to price increases (i.e., price-responsive demand, PRD).

    Now back to the revenue requirement. The scarcity adds produces rents for the utility because there are no offsetting costs. These surplus rents can then contribute to fixed cost recovery. While they are unlikely to fully cover those costs, they do reduce the the shortfalls and make the problem more manageable.

    Today utilities oversize their systems so that they can serve demand without tolling but that actually imposes excess costs on customers who are forced to pay the resulting higher fixed costs. The tolling scheme I described provides a means for balancing the cost of adding capacity with the value that the added capacity produces for consumers, as measured by total consumers’ surplus.

    Overall, I fully support using dynamic pricing that is directly tied to the short-run marginal costs (LMPs) of the relevant wholesale market. This is what I argued for in FERC Docket RM10-17, which erroneously produced Order 745.

    • “Economic theory informs us that economic efficiency is maximized when all prices are set equal to their respective SHORT-RUN marginal costs”… Not true, as the “putty/clay” model of cost definition demonstrates. And your definition of “costs” is violated when you add the scarcity premium, which you define as having no cost. The scarcity premium is set at a probability weighted value of the barrier to entry to the market–which is the long-run marginal cost.

      Regardless, the conditions for relying on a scarcity premium require optimality that is impossible to achieve, and it ignores the political conditions in which markets always operate. Market failures will always exist which means that governments will always have to intervene. No regularly traded commodity, good or service is relying on price spikes to recover large capital investments. We need to dismiss the idea for electricity too.

      • MCUBED,

        OK, I’ll bite. Explain the putty/model? LOL!

        The scarcity surcharge I described does represent a cost – the opportunity cost of the last (I.e., marginal) kWh not served. But note that this is a societal cost borne by customers, not the utility, which is why it produces rents that offset some of the utility’s I recovered fixed costs.

        The probability-weighted definition you presented is not the scarcity premium I described.

        And you are wrong; price spikes are used to recover capital investments in a number of industries, including the power sector, e.g., ERCOT in the US, Alberta in Canada, the NEM in Australia, etc.

        • As I stated before the scarcity premium comes from the barriers to entry for new suppliers. Consumers then bid up the price to point of the alternative cost for another supplier or to their budget constraint. You are not describing a marginal cost concept–you are describing a value based pricing approach which is not related to marginal costs. Suppliers in that market must judge how high they are willing to let a price go before the scarcity premium is high enough to attract another entrant which will drive down the price again.

          Note that scarcity pricing is not consistent with an efficient competitive market in which entry is costless. You are describing a market such as a competitive monopoly or oligopoly. The efficient pricing for that type of market is either a Cournot equilibrium in which the price is set by price response function of the other market participants or one with perfect first degree price discrimination that takes all consumer surplus for the oligopolist suppliers. The efficient competitive market implies a Bertrand equilibrium in which suppliers cannot gain sufficient profits to recover their fixed investments. None of these are particularly desirable from an overall societal standpoint unless you are hard core believer in chasing unachievable “efficiency.”

          • I think we are talking past one another. The scarcity surcharge I described earlier was indeed a value-based metric set by customers’ desire to buy energy, as determined by the demand curve for electric energy. It is not set by the supply side.

            The scarcity surcharge performs two functions:

            In the short-run it rations the available limited capacity so that it can serve the loads that the customers value the most. This is important because new capacity cannot be added instantly.

            In the longer-run the scarcity surcharge incentives new capacity additions, either by attracting new entry (in a competitive market) or by informing a monopoly utility that it is cost-beneficial to expand capacity.

            What you described is the behavior of potential suppliers in assessing when to enter the market, i.e., at what price point and indeed, there is a game-theoretic aspect to this because there often is a “first-mover” advantage involved. But a Bertrand equilibrium does not apply to the electric power industry because one of its preconditions is that a single supplier has enough capacity to supply the entire market. That just doesn’t happen. Also, there is tacit collusion among generation project developers, although this gets us back to the “firs mover” game you described earlier.

            I think we have wandered too deep into the weeds. LOL!

          • “The scarcity surcharge I described earlier was indeed a value-based metric set by customers’ desire to buy energy, as determined by the demand curve for electric energy. It is not set by the supply side.”

            Then by definition, this is not a marginal “cost”. You are describing the ability to price discriminate as a monopolist–exactly the reason why we have regulation.

            Andy Ford at WSU has a good set of papers that describe how infrastructure industries that rely on scarcity pricing, e.g., real estate, go through boom-bust investment cycles. The difference there is that we allow developers to go bankrupt (but even then, as evidenced by our current President, that system can be easily exploited). As we’ve seen twice with PG&E, we are unwilling to let our utilities actually suffer the consequences of bankruptcy, instead shifting the risk to ratepayers.

  3. Any legal reason why it wouldn’t be possible to set the fixed fee proportional to property values times a residential/commercial/industry deflator? Seems like that would take care of equity concerns pretty well while staying efficient, no?

  4. In Australia some regulators have pushed for demand charges in residential offers. However we have contestable retail markets and given the choice customers have rejected offers with demand charges – after three years almost no customers have chosen such offers and only a small handful of retailers offer them. The reason is rather obvious: customers have no way of knowing how they fare on these complex structures than they do on the simpler structures. None of the price comparison websites bother to include demand offers in the suite of offers they care. All of this should have been quite obvious from the outset and one wonders what time and expense might have been spared had rate designers, regulators and policy makers paid more mind to how retail markets actually work and what customers actually value.

  5. Nice post.
    To me, the tricky question is how to get the full energy costs from consumers to private firms under fully dynamic pricing. But I’m not worried about the grid costs.
    I totally agree that we want non-volumetric charges for the grid, and we want grid charges to be fair. This has income tax written all over it. So what is stopping us from reimbursing the the TSO for its cost using income tax revenue, possibly complemented with a capital gains tax? I mean other than politics…
    I understand that it is difficult to distribute public revenue to private firms. But giving tax revenue to the TSO is fine from a public finance point of view, as this is a regulated natural monopoly anyway. As we know, natural monopolies are supposed to run a deficit, so reimbursing the TSO is nothing but standard textbook economics.

    • Why should we charge for the electric grid differently from any other commodity? We don’t provide large tax subsidies to other public utilities like water and sewer (and I sat on a city utility rates commission so I’ve seen the balance sheets). As another example of a network system, cell phones and Internet service charge full cost prices. Instead of continuing down the line of the fantasy about dynamic pricing that will affect a minority of the overall cost, maybe we should be rethinking the proposed pricing model.

      • I find it more interesting to look at how grid costs are recovered for non-monopoly grids. After all, a principal function of regulation is to impose on monopolies the pricing discipline that competition imposes on competitive industries.

        When we go to Home Depot, we connect to a global hardware grid, that brings us products from Ohio, Germany, Mexico and China. We pick out the items we want, and the cost of “the grid” including trucks, ships, and trains, are all built into the unit costs we pay at the cash register.

        The same is true at the supermarket. We bear the costs of connecting to the grid, by traveling to the store. All of the costs of producing and distributing the products are built into the unit prices. And they even provide us an express lane, so we don’t have to wait behind people with baskets full of stuff.

        And also at Macy’s or at the gasoline station: all of the costs of the product, and of the grid that transports it, is built into the prices we pay for each unit we buy.

        In each of these examples, we bear the cost of connecting to the grid. On the electrical grid, this is the final service drop to the house, the meter, and the bill. Everything upstream is a “shared grid” and belongs in some sort of time-variant unit price. I prefer critical peak energy pricing as that form; others may prefer full dynamic pricing.

        Only a monopolist can impose an admission charge on customers wanting to buy their products. In competitive markets, sellers embrace customers, and welcome their business. They are happy to have us, whether for a $1.99 bag of screws or $4,000 for a bunch of new appliances at Home Depot, a can of gas for our lawnmower or a tankful for our motor home at Shell, a single pair of socks, or a whole new wardrobe at Macy’s, or a single bunch of radishes or a whole basket full at Thanksgiving at Kroger.

    • Beat,

      Interesting idea! Yes, covering monopoly utility deficits with income tax revenues would solve the equity problem without sacrificing economic efficiency. However, I doubt that it would be politically feasible.

      • I think the deal would be if deficits are covered with tax revenues, then the utilities must become publicly owned. That could solve other problems related to privately owned monopolies of the network grid.

  6. I’ll start with current monthly demand charges are little better than monthly customer charges. Once a customer hits a demand high early in the month, they have no incentive to reduce their demand below that level. The solution to this problem using demand charges is to go to DAILY demand charges since we generally use energy on a diurnal cycle.

    But the real problem here to start is that the current definition of “marginal costs” used too often in discussing electricity pricing is inadequate. As Jim Lazar points out, the appropriate metric is the long run MC which for electricity is much more than just the capital cost of a CT plus the energy cost of a CCGT. And it isn’t the market clearing price of the hourly exchanges for a number reasons, including that regulatory entities always intervene to prevent shortages that might cause price spikes. Notably, the CAISO MCPs have not been sufficient for a new entrant to recover even one year’s worth of investment costs. We first need to dispel policy makers of the notion that they can see the electricity marginal cost in regularly transacting markets.

    Beyond that issue, the generation long-run marginal costs (which seems to be the only element that gets academic attention) are perhaps one-third California’s electricity rates. At least another third is distribution and transmission costs (with no discussion of how those might be incorporated into dynamic rates), and the remainder is a mix of stranded costs and public policy initiatives. How do we recover those costs? That’s the real conundrum–the generation costs are almost trivial in comparison.

    BTW, I don’t know of a utility where customers drawing 3 MWH and 3,000 MWH pay the same customer charge…

    • MCubed,

      We have had this conversation before. You keep harping on the fact that the CAISO MCPs do not support new entry. Of course they don’t! Why? Because they do not include any scarcity adders – even when the system is capacity-constrained.* Unlike ERCOT (and other energy-only markets in other countries) the California market relies on the individual utilities to assure resource adequacy thorough their resource planning.

      Read my earlier comment to Severin in which I define SRMC to include a scarcity adder. For a more detailed explanation read volume one of Fred Kahn’s two-volume set.

      BTW, since you advocate setting rates based on LRMC, tell us how you would calculate LRMC. What is the time horizon and the technologies that would you use to determine the appropriate LRMC?

      I submit that the concept of LRMC is a convenient fiction used as an academic teaching tool. In the real world it doesn’t really exist. As Keynes said, “In the long-run we are all dead.”

      * A caveat. I think (but am not sure) that the CAISO does Operating Reserve Pricing that does add a scarcity surcharge to the energy price when the ISO needs to dispatch into the desired Operating Reserves to avoid load curtailments. That is certainly true in other ISOs, such as ERCOT and PJM, so I suspect that CAISO has adopted a similar vehicle. But all of these pricing schemes place a cap on how high the energy price can go and, except for ERCOT the caps are too low to fully support new entry.

      • Scarcity priciing isn’t happening in California. Most importantly the CPUC has and will always require planning reserve margins that suppress any scarcity pricing of note. And now in Texas they are running out of capacity because they’re realizing the same problem that California had initially–financiers are no longer going to be fooled by the scarcity pricing myth. You can talk about a theoretical world that really doesn’t function anywhere, and I’ll point out what is really happening and what are the actual constraints on public policy. Wishing that politicians would be willing to allow rolling blackouts just so we could have scarcity pricing isn’t a policy proposal.

        You miss my point–where does the scarcity premium come from? Thin air? You allude to relying on economic theory, but then you leave your most important point to magic. I’ve explained where scarcity pricing comes from–it’s the cost of the barrier to entry which is a long-run marginal cost (an investment cost) weighted by the probability of incurring the supply constraint that creates the scarcity. Essentially we’re back to Jim Lazar’s definition.

        BTW, Kahn laid out useful general principles, but many of his particulars have not held up over the nearly five decades, this being one of them.

        CAISO does not add a scarcity premium.

        I want to add one more point: All of this discussion relates ONLY to the generation portion of the electricity price. For the T&D and other charges, a “short run” MC doesn’t really exist–we only have the cost of incremental investment, or even of rebuilding the entire system. This leads to a substantial temporal discontinuity between generation and T&D if we attempt to apply marginal cost pricing using different principles.

        There’s a general consensus in ratemaking that generation charges should be in cents/kWh, and temporal variation is desirable. In California at least, demand charges are much more related to T&D. So the question should be focused on how to price T&D, not generation. Scarcity pricing isn’t really relevant to distribution pricing because the costs are so localized and there are not yet any functioning markets. So we’re left again with LRMC measures.

        • MCUBED,

          Why is California relevant to this discussion of optimal retail rate design. The CPUC has repeatedly demonstrated its ineptitude in setting rates, e.g., the four – tiered rate structure that incented affluent customers living in McMansions to bypass the top tiers by installing solar panel. And back in 2000 the CPUC exacerbated the energy crisis by refusing to raise retail rate when the IOUs could not serve existing demand; resulting in rolling blackouts. Most recently its adoption of static TOD rates, TOD rates were innovative in 1980 when metering capability was limited. Smart meters make them obsolete today.

          So let’s leave California out of this discussion.

          You ask where the scarcity premium comes from. Apparently you don’t understand the concept, which may be my fault for not describing it in greater detail. Fred Kahn described it in volume one of his landmark work. He defines the Short Run Marginal Cost (SRMC) curve as extending vertically upward at the point where the system reaches its maximum capability to serve load. The scarcity premium (or surcharge) is the price increment between the starting point on the vertical extension to the point where the customers’ composite demand curve intersects the extended SRMC.

          For this concept to be actionable the system operator needs to know what the price elasticity of demand is at each point on the demand curve that is likely to intersect the vertical portion of the SRMC curve. This can be empirically determined through trial and error (like just about everything else in life).

          Theoretically, this concept can produce premiums equal to your idea of Long Run Marginal Cost under hypothetical conditions of static equilibrium. But those conditions will never be achieved in the real world of evolving technologies and random shocks to the economy. As in other markets, power markets are never in equilibrium. Bill Hogan explored this idea with respect to transmission planning and expansion.

          You are wrong in claiming that the scarcity surcharge concept doesn’t apply to T&D, both of which include capacity constraints. Transmission constraints are managed by ISOs partly through LMPs that include scarcity surcharges.

          Distribution systems also include capacity constraints that are inefficiently managed through overbuilding and load interruptions, rather than through price signals and demand response. This is perhaps the greatest untapped potential for efficiency gains through scarcity pricing. The New York REV proceeding funded two studies that demonstrated this potential by revealing the role of a Distribution System Operator in setting locational dynamic retail prices (DLMPs),

          Lastly, I suggest you take care in criticizing Fred Khan’s work. Granted, he was human and not God, but you don’t have the credentials to credibly do so.

          • Of course California is centrally relevant here. This blog is hosted by University of California. Almost all of the state-specific topics are about California. Most of the readers and commentors are from California. If you want to talk about other states, I suggest you go to a different blog site that focuses on those other states. I will continue to bring up the relevance to California, in large part because what happens in California drives what happens in other states as well.

            (BTW, your explanation of what went wrong in the California energy crisis is off the mark, but I won’t go into that.)

            I know exactly where the scarcity premium comes from. That’s why I explained that the cost of the barrier to entry is the determinative factor. No barrier to entry, no scarcity. Marginal costs are about SUPPLIERS costs, not consumers VALUES (which you are confusing with costs.) That these markets are out of equilibrium implies that a certain points, the market clearing prices are above or below the long run marginal costs. What we’ve seen in electricity markets is that MCPs are never above the LRMC for a sufficient period to justify capital investments. And this is true for the reasons that I’ve spelled out previously.

            While transmission may have scarcity premiums (although those are again priced artificially through FERC tariffs), that is not true for distribution systems, which are the larger share of utility costs. A DSO might have dynamic pricing for DERs, but there won’t be a basis for consumption. You are basically proposed value-based, rather than cost-based, pricing. That is wide open to manipulation by monopoly providers. It also will lead to transfer of consumer surplus and wealth to producers. That’s exactly what we’re trying to avoid through monopoly regulation.

            I will criticize Kahn’s work as I choose. That he was first to lay out these concepts does not make him infallible. I’m not the only one here critiquing certain parts of his concepts. He doesn’t have rigorous theoretical foundations for some of his findings. His strength is that he made may fundamental principles accessible to practitioners, but I do not put great weight on the deeper, finer conclusions.

          • @mcubedecon

            “You are basically proposed value-based, rather than cost-based, pricing. That is wide open to manipulation by monopoly providers. It also will lead to transfer of consumer surplus and wealth to producers. That’s exactly what we’re trying to avoid through monopoly regulation.”

            Cost-based pricing is just as open to manipulation as value-based – it’s a two-way street. That vertically-integrated utility holding companies are permitted to buy fuel from their own subsidiaries then recoup its cost from customers as an expense; that regulators allow CCAs to tell customers they’re being served 100% renewable energy at night when the wind isn’t blowing, only perpetuates the myth a “market” or “consumer surplus” in electricity actually exists (when a consumer sells surplus electricity, he/she becomes a producer).

            There is no free market when the end user is denied freedom of choice, and that’s always been the case with grid electricity. Sure, checking a box for a 100%-renewable-energy CCA might give customers the feeling they’re doing something good for the environment, but without verification it’s of psychological value only – snake oil. If we’re trying to avoid a transfer of wealth to producers through monopoly regulation, it includes natural gas producers, wind farm producers, and homeowner-producers with their roofs covered in $80,000 worth of Tesla solar tiles. And we’re not trying hard enough.

          • I agree in general with your characterization that our current regulated electricity market has led to large transfers of wealth (including large central nuclear generating stations with huge cost overruns), but I was talking about a technical definition of “market manipulation” that leads to regulation. Just about anytime a market relies on “scarcity pricing” as the primary means of recovering capital investments, that market is subject to overt market manipulation by the suppliers due to the high barriers to entry from other suppliers.That’s precisely what happened in 2000-01 and the reason for the price caps in the CAISO markets. So Borlick’s proposal to rely on “value-based” pricing exposes consumers to even more wealth transfers.

          • “Sure, checking a box for a 100%-renewable-energy CCA might give customers the feeling they’re doing something good for the environment, but without verification it’s of psychological value only – snake oil.”

            Are you accusing the CCAs of failing to verify that they have the RECs to support their 100% green product? And that the electricity market does not have verifiable tags to track generation sources? That has an even bigger implication that these LSEs are failing to comply with the GHG caps in AB32. That’s a very serious accusation of breaking state law that requires substantial evidence.

          • Well, this all feeds into the fiction that electricity is a commodity and not a service. Electricity is generated and used at the moment of demand. It is a service. The wires to your house allow this service to be delivered to your door.

            Solar panels mean you want the service to be available full time, all the time, but you only intend to use it some of the time. Sorry but the cost of the service is fixed – you have to pay more per unit if you use it less. Just like any other service.

            “That regulators allow CCAs to tell customers they’re being served 100% renewable energy at night when the wind isn’t blowing, only perpetuates the myth a “market” or “consumer surplus” in electricity actually exists (when a consumer sells surplus electricity, he/she becomes a producer).” Exactly. It’s a scam. The electricity is not stored in a big warehouse. Use it or lose it.

            The hours the waiter spent standing around waiting for a customer are unrecoverable. They can’t be bought, sold, or traded. Wait, I will produce a certificate that will be a waiting “credit” to show that the waiter was available when I didn’t need him, and trade that for times when I do need him! The whole idea is preposterous.

            The only possible fair solution is the simplest one – every user is charged a refundable connection fee to the utility. If you buy enough electricity, the fee is refunded. If you don’t, you pay the fee. Simple. Don’t like it? Add some power walls to your solar panels and cut the wire to the utility. No one is stopping you. Want to sell back your excess electricity? Sure – but the utility gets to buy it for whatever they think it is worth. And it likely ain’t worth very much.

            Wait, this screws up the economics of your solar panels? Well your solar panels screw up the economics of the utility, so fair is fair I guess.

          • “I agree in general with your characterization that our current regulated electricity market has led to large transfers of wealth (including large central nuclear generating stations with huge cost overruns)…”

            Curious whether your characterization of nuclear plant cost overruns as “huge” amortizes these costs over the plants’ potential lifetime (80-100 years), vs. the lifetime of plants abandoned as soon as capital costs are recovered (typically, the initial 40-year licensing period)?

          • Nuclear plants require large supplemental investments to continue operating beyond 40-50 years, costs which are not included in the initial investment estimates. That’s one reason why the cost of Diablo Canyon jumped so much for its relicensing in 2024.

          • “Nuclear plants require large supplemental investments to continue operating beyond 40-50 years, costs which are not included in the initial investment estimates. That’s one reason why the cost of Diablo Canyon jumped so much for its relicensing in 2024.”

            Nonsense – the cost of generating electricity at nuke plants decreases over time. Why? They’re uprated, for pennies on the dollar.
            Last week Browns Ferry Nuclear Plant, the second-largest nuclear plant in the country, finished adding 465 megawatts for a total price of $475 million, or $1.02 million/megawatt. With a capacity factor above 99%, the added capacity is now providing clean electricity reliably and predictably fifteen times cheaper than California solar (Topaz).

            The cost of Diablo Canyon “jumped for its relicensing” because PG&E pulled a big number out of a hat.

            Keep thinking up bizarro talking points, I’ll keep smacking ’em down!

          • Speculation about whether PG&E lied about Diablo costs is far from snack down. You have no evidence. Updated costs from 2010 is not lying.

            Meanwhile nukes are demanding subsides to keep running in the East.

          • “Meanwhile nukes are demanding subsides to keep running in the East.”

            Another in your unending list of specious talking points. After eleven years of being excluded from Ohio’s Solar Renewable Energy Credit (SREC), nukes are only demanding equal recognition for their clean-energy contribution. Long overdue, Ohio’s zero emission credit (ZEC) will level the playing field by awarding the same credit to all clean sources of electricity. That’s right – solar will get the same credit as nuclear, based on the EPA social cost of carbon.

            Why wouldn’t solar interests welcome such a program? Because they recognize clean, reliable nuclear energy represents an existential threat to their industry. Since Ohio’s SREC was repealed, plans for a 500MW gas plant expansion have been cancelled in the state, one that would have relied on providing backup for intermittent solar and wind to be profitable. After eleven years of bailing out wind and solar, legislators realized they were only locking in gas emissions for decades into the future – and like four other states with ZECs, Ohio finally said “enough is enough”. More to come.

          • Maybe Michael Moore, in his new documentary Planet of the Humans, has been misreading too:

            “It turned out the wakeup call was about our own side,” [Producer Jeff] Gibbs said in a phone interview. “It was kind of crushing to discover that the things I believed in weren’t real, first of all, and then to discover not only are the solar panels and wind turbines not going to save us … but (also) that there is this whole dark side of the corporate money … It dawned on me that these technologies were just another profit center.”


  7. Those of us that have spent years working on rate designs probably will not agree with your entire assessment. Dynamic pricing is the path forward for all utilities, but it doesn’t take care of the full rate recovery requirements. Nor is it always fair and equitable. There is the size of the “pipe” and there is the “volume” that flows through it. I have many examples of large industrial customers that fuel switched based on the volume pricing between electricity and natural gas. Without larger demand charges, it is difficult to recover the costs for infrastructure (size of the pipe) that might sit idle much of the time while another fuel (including storage) supports the load.

  8. This is an excellent piece. Indeed, 15-minute integrated demand was never the “right way” to recover system capacity costs.

    A century ago there was great controversy over whether TOU rates or demand charges were the preferable way to recover these costs, and demand charges won out due to much lower administrative costs. At that time, the only way to measure TOU was with chart recorders which had to be hand-read. There is an excellent paper on this history entitled: “Electric charges: The social construction of rate systems” that was published in Theory and Society (2005) 34: 579–612.

    Indeed, an office building, which operates from 7 AM to 7 PM has a 50% load factor, and a street lighting system, which operates from 7 PM to 7 AM also has a 50% load factor. Assuming that they rigidly held to exactly those hours of usage, however, they could share the same generation, and transmission capacity. (I’ll assume they are in different locations, so would require separate distribution capacity). By contrast, a 7-11 or Denny’s, which operates 24/7, would pre-empt capacity at all hours, and could not share capacity. But typical non-coincident peak demand charges would charge each of the three “customers” exactly the same for generation and transmission capacity. And the newer evolution of demand charges, peak-period demand charges, applied based on the highest 15-minute demand from 5 – 8 PM only, would do the same thing. The office building would still be operating from 5 – 7, and would pay a full load, and the street lighting system that started at 7 PM would do so as well, double-charging for the same capacity while the Denny’s would pay only once for the capacity it pre-empts.

    Suppose instead those capacity costs were spread across ALL usage during the 5-8 period. Then the office building would pay for two hours, the street lighting system for one hour, and the Denny’s for three hours. That’s how a TOU energy rate works — apportioning capacity costs in proportion to capacity utilization.

    I’ll turn now to Sev’s suggestion of a uniform per-kWh surcharge to recover the residual revenue requirement, above long-run marginal costs.. Two questions arise. First, how should we measure long-run marginal costs? Second, assuming we find an acceptable method, and indeed there is a residual marginal cost, should it be uniform, or should it be equiproportional? I’ll address these sequentially.

    I’ll suggest that the “right” measure of long-run marginal cost is what the telecom industry came to call “Total Service Long-Run Incremental Cost (TSLRIC). That is the cost of building a new system, using today’s technology, to optimally meet today’s requirement. That is, it’s a “replacement cost” analysis, not a “start with what we already have and build from there” which is how most utilities do long-run marginal cost analyses today (begat by NERA in the PURPA era of 1980 or so). The telecom industry came to this as a measure of the costs a new entrant would incur, and then allowed incumbent local exchange carriers to price competitively at or close to that level if doing so would preserve some sales margin while covering short-run and intermediate marginal costs. This is likely to produce a higher measure of long-run marginal cost than we see in the NERA-methodology.

    Having reached agreement (or not) on how to measure marginal costs, and agreeing on a form of dynamic pricing where prices are different at different hours of the month, I’ll then ask if a UNIFORM per-kWh adder is equitable, or whether the adder should be a PROPORTIONATE adder. I lean toward the latter, but hope to generate some discourse on this issue in the comments on this blog post.

    For example, assume that our TSLRIC analysis shows us that the long-run marginal costs are $.05/kWh in off-peak periods, $.10/kWh in mid-peak periods, and $.25/kWh in on-peak periods, with an average of $.12/kWh, but our revenue requirement is $.15/kWh. Should the adder be $.03/kWh at all hours, or should it be 25% at all hours. The first would produce retail rates of $.08, $.13, and $.28. The latter would produce retail rates of $.0625, $.125, and $.3125 for the three TOU periods.

    If one goal is to attract flexible load (water heating and EV charging, for example), the lower off-peak rate of an equi-proportional approach would be more effective. If the goal is to discourage increases during peak periods, the equi-proportional approach would be more effective.

    We have a long history in most existing marginal cost states (California, Oregon, and a few others) of using EPMC adjustments to reconcile marginal costs to the revenue requirement. Should that principle be preserved as we transition from an archaic demand/energy framework to a more dynamic energy-driven marginal cost framework?

    I look forward to the discourse.

%d bloggers like this: