Somewhere around year four of operation, a natural gas peaker plant runs its numbers and finds something unsettling. It ran for perhaps 400 hours over the past twelve months. The market paid it well during those hours, sometimes extraordinarily well. But 400 hours is not 8,760 hours, and the debt on the turbine doesn't know the difference.
This is the central problem of wholesale electricity market design, and it is not a bug anyone forgot to fix. It is, depending on whom you ask, either a feature or a slow-motion structural failure. The answer turns on a single architectural question: does the market pay generators only for the electrons they produce, or does it also pay them for being available to produce?
Energy-only markets and the missing revenue
Most wholesale electricity markets are built around what economists call an energy-only structure. Generators submit offers into a pool. A system operator ranks those offers from cheapest to most expensive and dispatches them in order until demand is met. The last unit called sets the price for everyone. This is the merit-order dispatch, and it is elegant in theory: the cheapest electrons always run first, and competitive pressure keeps prices honest.
The problem emerges when you think about what a peaker plant actually is. A gas turbine designed to run only during high-demand hours carries high variable costs (fuel isn't cheap at full throttle) but also significant fixed costs: financing the turbine itself, maintaining it through years of standby, keeping trained staff on site. To recover those fixed costs in an energy-only market, the plant must earn enough during its few hundred annual hours to cover costs spread across the entire year.
Mathematically, this requires occasional very high prices. If a peaker's fixed costs run to $80,000 per megawatt of capacity per year, and it runs 500 hours, it needs to net $160 per megawatt-hour above its variable costs just to break even. That kind of margin materializes only during scarcity events: a heat wave, a cold snap, an unexpected cluster of plant outages. Economists call the revenue from these moments the "scarcity rent," and the theory holds that it should, over a long enough period, be sufficient.
The theory has a vulnerability. Most energy-only markets also carry administrative price caps, set at levels intended to prevent market manipulation or protect consumers from shock bills. Texas's ERCOT market, the most prominent energy-only design in the United States, has wrestled with this tension for years. When the cap lands below the level at which scarcity rents would fully compensate peakers, the "missing money problem" is not theoretical. It is a line item on a balance sheet that keeps getting worse.
The capacity market answer, and its own complications
The response many grid operators chose was to layer a second market on top of energy-only dispatch: a capacity market. Generators are paid not just for the electricity they produce but for a forward commitment to be available. PJM Interconnection, which coordinates power across much of the northeastern United States, runs one of the most studied capacity markets in the world. Generators bid to provide a megawatt of guaranteed availability one to three years ahead, the market clears at a single price, and everyone who clears gets paid that price annually regardless of how many hours they actually run.
This solves the missing money problem for capacity. A generator that clears PJM's capacity auction can build its business model on two revenue streams: energy payments when it runs and capacity payments simply for existing and being ready. Fixed costs become genuinely recoverable.
But capacity markets introduce a different distortion. Consider two plant operators: Maria, who builds a highly efficient combined-cycle gas plant, and David, who builds a less efficient older-style unit. Both clear the capacity auction at the same price. Maria's plant, being cheaper to run, gets dispatched more often in the energy market and earns more there too. David's plant sits idle more, collects the same capacity payment, and earns less in energy. Over time, capacity markets can keep otherwise uneconomic plants alive on capacity revenues alone, like a patient on a drip that everyone agrees isn't getting better, slowing the retirement of assets that a pure energy market would have killed off years earlier.
There is also the question of what "available" actually means. A generator that has committed capacity but fails to deliver during a grid emergency is supposed to face financial penalties. How strictly those penalties are designed and enforced determines whether the capacity payment buys real reliability or just a contractual promise.
What people consistently get wrong about this
The common assumption is that high wholesale electricity prices mean generators are profiting. Sometimes they are. More often, a spike in the spot price is a peaker plant finally, briefly, earning back a sliver of the fixed costs it has been carrying for months of near-idleness. Strip out those peaks and the annual revenue picture looks very different.
People also tend to conflate the wholesale market price with what consumers pay. Retail electricity tariffs are typically regulated separately, smoothed over time, and bundled with transmission and distribution costs that have nothing to do with the generation market. A spot price of $500 per megawatt-hour on a hot afternoon does not mean your bill triples that month. The insulation runs in both directions: consumers are shielded from the worst spikes, but they are also shielded from the signals that would tell them to shift their dishwasher to midnight.
That asymmetry raises a question worth sitting with: if consumers never see scarcity prices, what exactly is the market disciplining?
The deepest misconception is that market design is a neutral, technical exercise. It isn't. Every structural choice, where to set the price cap, whether to run a capacity market, how far ahead to clear it, encodes a judgment about which risks generators should bear and which risks society should socialize. An energy-only market tells generators they bear the revenue risk, and if scarcity rents don't materialize, that is their problem. A capacity market tells ratepayers they are guaranteeing a return on investment for generators who may or may not be needed. Neither is neutral. Both are political.
The generator that can never win
The structure that is genuinely hostile to certain generators is an energy-only market with a binding price cap set below the theoretical scarcity rent, combined with increasing penetration of zero-marginal-cost renewables. Wind and solar, once built, bid at or near zero because their fuel is free. Their presence in the merit order suppresses the average price during hours when they run. They don't suppress peak prices during scarcity, but they reduce the frequency of near-scarcity events by covering more hours of moderate demand.
A mid-merit gas plant in this environment faces a narrowing band. The hours in which it was once profitably dispatched are colonized by cheap renewables. The scarcity hours that would have provided its financial lifeline come less often. The price cap prevents it from earning enough in the remaining scarcity events. Its fixed costs don't fall. The math stops working, and no amount of operational efficiency rescues it from the structure it is operating inside.
No market design solves this cleanly. Capacity markets can extend the plant's life, but they do so by asking consumers to pay for capacity that renewables have partly rendered redundant. Energy-only markets let the plant die, which is economically rational, but that raises the question of what provides firm power when the wind doesn't blow and the sun doesn't shine.
The grid that emerges from these choices is not an accident. It is a direct consequence of structural decisions made, often decades ago, by regulators who were solving a different problem entirely. The generators that cannot recover their fixed costs are not victims of bad luck. They are the predictable output of a system designed with other priorities in mind, and the next wave of capacity retirements will test whether those priorities still make sense.