Capacity Markets: Paying for Power You Hope Never to Use
Electricity systems are built for their worst hour, not their average one. That hour might be a windless evening in February with the temperature below freezing, and it might occur once in a decade. The plants that exist to cover it face an awkward commercial fact: their entire purpose is to sit idle.
The Missing Money Problem
In an energy-only market, a generator is paid for each megawatt-hour it delivers and nothing else. In principle this should be sufficient. During the tightest hours of the year prices ought to rise steeply as supply approaches demand, and those brief spikes ought to cover the annual cost of plants that run rarely.
In theory the arithmetic works. A peaking plant might run 100 hours a year and earn a price high enough in those hours to recover its costs. The trouble is that the prices required are extraordinary - thousands of euros per megawatt-hour - and no political system tolerates them for long.
So markets impose price caps, or regulators intervene during crises, or operators take administrative actions that suppress the spike before it forms. Each intervention is individually defensible and collectively fatal to the business case for peaking plant. The revenue the theory relies on never arrives. This is what economists call the missing money problem.
The consequence is that plants needed for reliability close, because staying open loses money, and the system discovers its shortfall during the first severe event afterwards. Since reliability is a public good and nobody experiences a partial blackout, the market does not correct itself in time.
How the Auction Works
A capacity market separates two products: energy, which is paid for when delivered, and capacity, which is paid for being available. The system operator forecasts peak demand for a future year, adds a reserve margin, and auctions contracts for that quantity of firm capacity.
The auction is descending. The operator starts at a high price and lowers it while participants withdraw, until the remaining bids exactly match the required volume. Everyone who clears is paid the clearing price per kilowatt of capacity per year, regardless of how much electricity they actually generate.
The contract carries obligations. A plant holding capacity must be available when the operator declares a stress event, and pays penalties if it is not - which is what prevents the payment from becoming a subsidy for merely existing. Britain's mechanism, running since 2014, auctions four years ahead so that new plants have time to be built, alongside a shorter one-year-ahead auction for fine adjustment.
What counts as one kilowatt of capacity varies by resource, which is where the design gets interesting. A gas plant is derated slightly for the chance of an outage. A wind farm is derated heavily, since its contribution during a still evening is small. A four-hour battery is derated according to whether stress events typically last longer than four hours. These derating factors are technical judgements with large financial consequences, and they are argued over accordingly.
The Case Against
The most common objection is that capacity markets pay fossil plants to stay open. In several British auctions a substantial share of contracts went to existing gas and, early on, coal - assets already built, already depreciated, receiving a payment to postpone closure. Whether that is a scandal or exactly the intended function depends on whether one believes those plants are still needed, which is the actual disagreement.
A second objection is that the problem is self-inflicted. If regulators allowed scarcity prices to reach their true level, the revenue would appear and no separate market would be necessary. Texas ran on this principle for years with a very high price cap and no capacity mechanism. The February 2021 winter storm, which killed hundreds of people during an extended blackout, became the central exhibit for both sides: critics said it showed energy-only markets fail to procure reliability, defenders said the failures were of winterisation and gas supply rather than market design.
A third objection concerns forecasting. The operator must estimate demand and derating factors for a year several years ahead, and errors in either direction are costly - over-procurement charges consumers for capacity never needed, under-procurement leaves the system short.
The middle position, now common in European system operators, is that some explicit mechanism is needed because the political tolerance for extreme prices does not exist, but that it should be technology-neutral and short-duration so it does not lock in assets for decades. Whether real auctions achieve that is an empirical question that varies by jurisdiction.
What It Means for New Technologies
For anything new connecting to a grid, the capacity market is where reliability is converted into revenue, and it rewards a specific property: being available at the system's tightest hour, with a penalty for failing.
This is measured rather than asserted. Derating factors are calculated from historical output during past stress events, which means a resource earns its capacity value through demonstrated performance when the system was under strain. A source that genuinely produces during a still, dark, freezing evening is worth far more per kilowatt than one that produces a great deal in total but little then.
Batteries illustrate how quickly this can change. Early auctions derated them severely because of their limited duration. As stress-event data accumulated and durations lengthened, their derating improved and they began clearing large volumes. The route was empirical, not rhetorical: the asset performed, the data showed it, the factor moved.
That is the same route open to any new generation technology. Continuous output is valuable here in proportion to how reliably it appears in the hours the operator worries about, and the mechanism for proving it already exists. It measures delivery at the worst hour, which is the most demanding and most honest test a grid applies.
Frequently asked questions
What is a capacity market?
A mechanism that pays generators and other resources for being available at a future date rather than for electricity produced. The system operator forecasts peak demand, adds a reserve margin, and auctions contracts for that volume of firm capacity, usually several years ahead so new plant can be built in time.
What is the missing money problem?
In an energy-only market, plants needed only during the tightest hours should earn their annual costs from extreme scarcity prices. Price caps and regulatory intervention suppress those prices, so the revenue never arrives, and plants the system needs for reliability close because staying open loses money.
Why are wind farms derated in capacity auctions?
Because capacity value is measured by contribution during system stress events, which often occur on still, cold evenings. A wind farm producing a great deal over a year may contribute little in those particular hours, so it is credited with a fraction of its nameplate rating.
Do capacity markets subsidise fossil fuels?
They have often paid existing gas and, early on, coal plants to stay open, which critics call a subsidy and supporters call the intended function of paying for reliability. The disagreement is really about whether those plants are still needed, not about the mechanism itself.
Can storage and demand response participate?
Yes, and increasingly they clear large volumes. Both are derated according to duration and demonstrated performance during past stress events. Batteries were derated severely at first and improved substantially as operational data accumulated and typical durations lengthened.
