This is holiday reading for nerds with an unnatural interest in electricity markets. Yet, perhaps perversely, it addresses a change of critical importance to how and what we pay for electricity in future. It illustrates the extent to which the changes flowing from the increase in intermittent renewable generation affect the costs of maintaining reliable electricity supplies and how those costs are passed on to final customers. I do not wish to disturb anyone’s holiday time-off but this is something to bookmark for reading when you return from the beach or any other holiday destination.
Much of what I write on CloudWisdom focuses on developments in the UK and Europe. In this article I will discuss the results of the recent capacity auction held by the PJM Interconnection, which covers the Mid-Atlantic region of the US East Coast, for the delivery year 2025-26.[1] Even this sentence may be sufficient to prompt some readers to switch off, but please bear with me.
The core issue is of fundamental importance to the evolution of electricity systems as intermittent renewable generation supplies account for an increasing share of total generation. If asked how power generators are paid to operate, probably 99% of people in Europe or the US would answer that they receive a price per kilowatt hour (kWh) of electricity that they supply to the grid or to customers. Surely that must be the natural – even the necessary – way of compensating generators?
Well, actually no! That has not been the exclusive case for three decades, since power markets began to be a key feature of electricity systems. Some generators have been paid for a variety of ancillary services required to ensure that the grid operates in a stable fashion with a very high level of reliability. In this respect, the recent PJM auction is no different from previous auctions and parallel arrangements in many other systems. However, it does signal a crucial qualitative change in the composition of revenues for many power generators.
The PJM auction points to a future in which many power generators will be paid primarily for providing backup and other services, while their revenues from generation contract as a share of total revenues. This shift should have been obvious to anyone who has observed the evolution of electricity markets and investment (or non-investment) decisions over the past decade. Yet, it has barely surfaced in political and public discussions of electricity markets. Even where the issue arises it is in disguised form because of the way in which subsidies for intermittent renewables are structured and financed. I will discuss that in a separate article.
We should start with some background. The PJM Interconnection LLC or PJM is the Regional Transmission Operator (RTO) which covers part of all of 13 states and DC from New Jersey to Kentucky.[2] It does not own transmission assets, but it acts as the electricity system operator responsible for managing and ensuring the reliability of the transmission network. Its equivalent in Great Britain (excluding Northern Ireland) is National Grid Electricity System Operator (NG-ESO) and there are similar Transmission System Operators in Europe. PJM has links to neighbouring transmission networks just as there are inter-connections from Britain to Ireland, France and other European countries.
The transmission assets controlled by PJM are owned by 51 companies including some of the largest US power companies (Dominion and Duke Energy) and a non-profit cooperative, East Kentucky Power Cooperative. PJM acts as the balancing authority for its network, which means that it is responsible for ensuring that the demand for and supply of electricity are matched continuously. It has interconnections to more than 20 neighbouring balancing authorities.
PJM is the largest RTO by population in the US. It serves about 65 million people, very similar to the GB network. However, peak demand and the capacity of power plants connected to the network are much higher than in Britain. The projected peak demand for 2025-26 is about 114 GW, which is more than twice the peak demand for the GB system in recent years.
The total capacity of generating plants connected to the PJM network is expected to be 186 GW in 2025. A substantial fraction of this generating capacity is heavily downrated because its availability cannot be guaranteed – e.g. intermittent wind and solar plants - or for other reasons. Thus, the capacity auction contracted about 135 GW, giving a reserve margin of 18.5%. This is adequate but significantly below the reserve margin of over 22% in 2019-20 and 2020-21 and an average of 20% since 2015. PJM has reduced its reserve margin to lower the cost of contracting guaranteed capacity. The auction prices for 2025-26 would have been much higher had it stuck to a reserve margin of over 20% as in previous years.
PJM’s contracted capacity as a proportion of its internal generating capacity has fallen from 88% in 2015-16 to 73% in 2025-26. There have also been large reductions in the amounts of demand response and imports offered. Changes in the way in which the amounts of capacity offered in the capacity auction are measured mean that detailed comparisons of offers and accepted bids over time are difficult. Still, there is two clear conclusions.
First, the capacity offered in the annual auction as a fraction of internal generating capacity has fallen by more than 10 percentage points. This is largely a consequence of the growth in intermittent generating capacity, though PJM is a relative laggard in this respect. Since additions of new dispatchable generating capacity are low, while some existing capacity is planned for retirement, the ratio of capacity offered to internal generating capacity is likely to fall further up to 2030.
Second, the margin between accepted and offered capacity has fallen from over 22 GW for 2022-23 to barely 1 GW for 2025-26. With retirements and continued growth in forecast peak demand it will be difficult to contract sufficient capacity in 2026-27 and later years without some combination of (i) higher capacity auction prices, and (ii) lower system reserve margins.
At some point, again quite soon, PJM will have to find ways of paying generators to build more dispatchable capacity. If it does not do that, the system reserve margin will fall to a level that jeopardises the reliable operation of the whole network. A widespread outage in the PJM network would probably be worse than the North-East blackout of 2003 because the PJM network is so closely linked to transmission systems in the North-East, Mid-West and South of the US.
This leads us to the auction prices in the 2025-26 auction. The system price was just under $270 per MW of capacity per day or just over $98,500 per MW per year. The auction prices were much higher for two constrained zones – Virginia (Dominion) and Maryland (Baltimore Gas and Electric) with capacity prices of $444 and $466 respectively.
The system price was over 9 times the system prices for 2024-25 of $29 per MW per day, even though the contracted volume fell by 8% from 147 GW to 136 GW. This illustrates just how tight the auction for 2025-26 was. Over the 10 years from 2015, the average system price was $40,400 per MW per year at expected 2025 prices, so in real terms the 2025-26 auction prices was over 2.4 times the 10-year average.
To put the 2025-26 prices in context, the EIA’s 2022 estimates of levelized costs for plants entering service in 2024 suggested that the average annual cost of capital + O&M expenses for a gas turbine operating with a load (capacity) factor of 10% would be $57,000 per year at expected 2025 prices.[3] In the US, gas turbines are the standard option for providing dispatchable peaking capacity that is expected to be used for less than 1,000 hours per year. In the UK, reciprocating engines have become popular, especially for plants that are expected to be used for less than 200 hours per year. Usually, they have lower fixed costs and higher operating costs than gas turbines. They can be built more quickly gas turbines, but they tend to be small and less suited to meet US system requirements.
Whether it is gas turbines or reciprocating engines, the average PJM system price for capacity has been less than the cost of building and operating new peaking capacity over the last 10 years. It is not a surprise that power generators have had little interest in building new peaking capacity, though they have been able to cover O&M costs for existing plants.
The system price for 2025-26 would be sufficient to warrant investment in new peaking plant, but only if high prices for system capacity are expected to persist. The standard deviation of the system price has been $22,700 per MW per year at 2025 prices. Depending upon the investor’s tolerance for risk and financial resources, the system price might need to stay above $200 per MW per day at 2025 prices to stimulate substantial new investment.
There is an alternative approach, adopted in the UK, which is to offer 15-year contracts rather than relying on year-by-year auctions. This has not been common in the US because it was thought that there was sufficient capacity available to meet short term capacity requirements. Clearly that is no longer the case with the squeeze on reserve margins discussed earlier.
The increase in the system price should not be treated as an isolated blip. It is a warning that the average level of capacity prices will be much higher in future than they have been over the last decade. Indeed, if the current administration’s target of decarbonising the power system by 2035 is taken seriously, then the capacity prices must rise even further. Other than gas turbines, the primary option for providing peaking capacity is battery storage.
Based on the EIA’s levelized cost estimates for new plants entering service in 2024, the average capacity price would have to be at least $120,000 per MW per year at 2025 prices to warrant investment in new battery storage to provide an adequate reserve margin for PJM. The constrained capacity prices for 2025-26 for Virginia and Maryland exceed that level by a large margin. This may indicate that the actual costs of building new battery storage within a short period are very much higher than the EIA’s levelized cost estimates. Alternatively, investors may not be convinced that the constraint premia will continue to apply beyond 1-2 years ahead.
The total cost of the capacity auction contracts for 2025-26 will be $14.7 billion. PJM’s average actual load in 2023 was 84.1 GW which translates to 737 TWh for the whole year. Spread uniformly over all load the capacity payments translate for 2025-26 to $19.9 per MWh of load. For the current year the equivalent cost is $3.0 per MWh of load. The unweighted average of day-ahead locational marginal price across all PJM hubs in 2023 (this is equivalent to the day-ahead wholesale price for the GB market) was $29.0 per MWh.[4]
Hence, in the current year capacity payments amount to just over 10% of the wholesale cost of electricity in 2023. In 2025-26 that share will increase to nearly 70%. Of course, we cannot know what wholesale electricity prices will be in 2025-26, but the likelihood that wholesale electricity prices in 2025-26 will be more than 6 times their level in 2023 is practically zero.
This is the fundamental change in the role of capacity payments that I flagged at the beginning of this article. As I have argued, the large increase in total capacity payments reflects a squeeze on the margin between retirements of dispatchable generation and an increasing level of peak load. Retirements of older plants may be delayed by the higher capacity payments which they can earn. Even then, operators have a limited incentive to invest in upgrades to improve operating efficiency and environmental performance as the number of operating hours per year falls due to displacement by subsidised renewable generation. For gas plants, such investments are often mandatory when turbines are replaced or refurbished, because environmental regulations have been tightened since the plants were first built.
While the future rate of plant retirements is uncertain, the overall direction of change is clear. The system reserve margin will only be maintained or improved by offering capacity payments which encourage investors to provide new backup capacity. This means that the level of capacity payments must match the marginal cost of building and maintaining either new gas turbines or battery storage.
It is, therefore, inevitable that the average level of PJM capacity payments in the next decade will be between 60% and 90% of the average PJM wholesale power price in 2023. In turn this means that the general level of consumer electricity prices paid by both businesses and households will rise significantly as energy suppliers pass on the higher capacity payments that they will incur.
The impact on the cost of living may be felt widely. It is unlikely to be accepted easily, since average electricity consumption per head is so much higher than in Europe, where levies of various kinds, including capacity payments, have pushed consumer electricity prices well above their typical level in the US. For example, in the UK the wholesale price of electricity accounts for about 22% of the average price paid by households in mid-2024. That reflects both (i) the decreasing share of revenues earned by generators that come from sales of power at wholesale prices, and (ii) the increasing burden of policy costs as well as transmission and distribution charges on final electricity users.
State regulators may attempt to claw back some of the increase in generator revenues from the increase in capacity payments. However, this is very difficult to do in any electricity system that has been even partially liberalised because of the impact on merchant generators and competing energy suppliers. The outcome is likely to involve a lot of noise and heat but rather limited rollbacks of the impact of the increase in capacity payments.
In summary, what may appear to non-specialists as a large but technical change in the way in which the PJM capacity market works should be seen as foreshadowing a basic shift in the way in which electricity markets operate in the US. This is, of course, ultimately driven by the push to replace dispatchable fossil fuel generation by intermittent renewable generation. However, it is far from clear how many of the advocates of the shift have any real appreciation of its wider impact on electricity markets and consumers. In the UK we are still regularly treated to claims about how wind and solar generation are so much “cheaper” than thermal generation. As the case of the PJM capacity market illustrates, such claims rest on simply ignoring the wider costs of ensuring that our electricity systems continue to operate at the level of reliability that we have come to expect.
[1] The results of the PJM capacity auction for the year from June 2025 to May 2026 were announced in a press release on 30th July 2024 with an accompanying report on the detail results of the auction. As is the way with such organisations, the report is full of almost incomprehensible acronyms and jargon, but the key feature of a huge jump in the market clearing price from the current year of 2024-25 to next year cannot be missed.
[2] The acronym PJM was based on the primary coverage of the RTO when it was formed – Pennsylvania, New Jersey and Maryland. The RTO has expanded well beyond that original scope.
[3] The EIA’s levelized costs should be regarded, at best, as a general indication of the costs of building new plants. Their capital cost estimates tend to be too low, and they are certainly over-optimistic about the period over which investors expect to recover their capital costs. As a guide to actual investment decisions, a general rule of thumb is to increase the EIA cost estimates by 20%-25%. This would suggest a range of $70,000-$75,000 per MW as the minimum capacity price required to induce investors to build new gas turbine as peaking capacity for the PJM market.
[4] Data on PJM hourly actual loads and day-ahead locational marginal prices is available on the EIA at https://www.eia.gov/electricity/wholesalemarkets/data.php?rto=pjm .
My translation, to enable me to work out what was going on, is as follows.
A typical power station, like the new Keadby B, has a maximum power output of 500 MW or about 5 billion units per year (5 BU/y)
In the old days, gas fired power stations like the new Keadby B which could provide 5 BU/y if run flat out, would on average run for about half the year because they were on full power in winter and low output in summer, delivering 2.5 BU/y.
With very rough figures, at today's prices, they would be charging the grid about 10 p/U. With distribution, metering, billing costs and profit, this ends up costing the consumer about 25 p/U.
The annual income of Keadby B is easy to calculate - 2.5 BU/y x 10 p/U = £250 million/y, ie £250M/y
The power station needs to pay for its fuel, around £150M/y, leaving a gross profit of £100M/y. It has to staff and maintain the power station. But the biggest cost is building the power station itself - £350M. The investors need an income of £35M/y just to repay that investment. All in all, this means that gas power stations running as much as they can are financially viable.
In these new greener times, a power station like Keadby B is not run for half the year. Indeed the whole purpose of government energy policy is to use as much wind and solar as possible and run the gas power stations as little as possible, perhaps for only 40% of their normal total output.
So instead of delivering 2.5 BU/y, it delivers 1 BU/y and, at 10 p/U, charges the grid £100M/y.
Proportionally its gas costs actually go up because the power station is being used intermittently according to how hard the wind blows. But even if we ignore that extra loss, gas costs are about £60M, leaving a gross profit of only £40M/y. That simply isn't enough to cover the costs of running and making a £350M power station. So investors stop making new gas power stations and governments panic because, without gas, we get blackouts when the wind stops blowing.
That's where the capacity prices come in, which are fees paid to gas/oil/coal power stations just to be there to save us from dunkelflaute (dark windlessness).
The passage quotes a capacity price of $120,000 per MW per year, roughly £100,000 per MW per year. I have to explain this to myself slowly.
This means the government pays fossil fuel suppliers £100,000 per year for each MW of generating capacity they keep available.
Since Keadby B has a capacity of 500 MW, it gets a subsidy of 500 MW x £100,000/MW.year = £50M/year. It is that subsidy which makes it worthwhile to make gas power stations which end up being used well below their potential.
Remember that our under-used power station delivers 1 BU/y or 1000 MU/y. For that it gets a subsidy of £50M/y.
The subsidy per unit is (£50M/y)/(1000 MU/y) = 5 p/U. So our underused and subsidised power station, normally delivering power to the grid at 10p/U costs an extra 5p/U in subsidy.
The above is worth an article on its own, I think. Feel free to use it, or I'll place it elsewhere.
Good wishes, Mark Ellse
Hi
You might find my own take on refuting the idea that Weather-Dependent "Renewables" are nine times cheaper that fossil fuels.
Please reuse or reference these posts at will.
https://edmhdotme.wpcomstaging.com/the-myth-of-cheap-renewable-power-in-the-uk/