Marginal pricing has been the backbone of every liberalised power market for 35 years. It nearly broke during the 2022 crisis. Here's why it exists, why it survived, and what reform actually changed.
Start with the misconception that gets repeated everywhere, including in serious newspapers and by serious politicians. The claim is that wholesale electricity prices are pegged to or linked to the most expensive form of generation, usually gas. This is wrong in a specific and important way.
There is no rule, no peg, no formula written in legislation that ties power prices to gas. What exists is something more elegant and a lot harder to argue with. The price clears at whatever it costs to produce the last megawatt-hour the grid actually needs in a given hour. If that last megawatt-hour comes from a gas peaker, gas sets the price. If it comes from a coal plant, coal sets the price. If demand is low and a wind farm is the marginal unit, the price is near zero. The price is not pegged to anything. It is whatever the most expensive necessary plant happened to bid that hour.
This is called marginal pricing or pay-as-cleared, and it is the design choice underneath every liberalised electricity market on earth. The UK has it. Germany has it. France, Spain, Italy, the rest of the EU, all clear at the margin. The US runs it across PJM, MISO, NYISO, ISO-NE, CAISO, SPP and ERCOT, with a locational twist. Australia's National Electricity Market runs it. New Zealand, Chile, parts of Latin America, all the same.
The mechanic itself is straightforward enough that you can sketch it on a napkin.
Three things to notice. First, the order is purely about marginal cost, the cost of producing one more megawatt-hour given that the plant exists. Solar and wind have near-zero marginal cost because the fuel is free. Nuclear is cheap because uranium is cheap and the plant is paid off. Gas is expensive because every megawatt-hour burns gas you have to buy at whatever the spot price is.
Second, the merit order is a description of what happens, not a regulation imposed on the system. Plants bid into a daily auction, the system operator stacks the bids from cheapest to most expensive, and dispatches them in order until demand is met. The last accepted bid is the clearing price.
Third, every plant that runs gets paid the clearing price, not their own bid. This is the part that drives non-specialists crazy. Why pay the wind farm €140 when its marginal cost is zero? The answer takes us back to the late 1940s.
If you let every plant get paid only what it bid, you create a distorted incentive. Every generator now has a reason to lie about its costs. The wind farm operator knows the gas peaker will be the marginal unit, so why bid €0? Why not bid €130 and hope the price clears at €140 anyway? Multiply this across hundreds of generators and the auction stops working. Bids no longer reflect costs. The system operator can no longer figure out the cheapest way to meet demand because the data they're getting from generators is strategic, not honest.
Marginal pricing solves this by making the dominant strategy tell the truth. If everyone gets paid the clearing price regardless of what they bid, the only thing your bid does is determine whether you get dispatched. If you bid above your marginal cost, you risk being skipped over and earning nothing. If you bid at your marginal cost, you run when you're needed and earn the clearing price. The optimal play is to bid your honest cost.
This is straight game theory, and it's why economists are unusually unanimous in defending marginal pricing as a design principle. It produces what the textbooks call a truth-revealing auction. The system operator gets accurate cost data, dispatches the cheapest combination of plants that can meet demand, and the market converges on the lowest possible cost of supplying electricity at any given moment.
The other defense is about investment signals. When the gas peaker sets the price at €140 and gas plants run for hundreds of hours a year at those prices, that's a signal that the market is short on cheap capacity. Investors see the spike, build wind, solar, batteries, or new CCGTs, and the next time demand peaks the cheaper new plants displace some of the gas. Spikes get rarer. Average prices fall. The signal worked.
Pull marginal pricing out of the system and you have to replace those signals with something else. Usually that something else is a planner, deciding what gets built where based on a forecast. The history of central planning in electricity is not a happy one. It produced the overbuilt, underused, vertically integrated monopolies that liberalisation was meant to fix.
The intellectual scaffolding of marginal pricing wasn't written for a market. It was written for a state-owned monopoly. Marcel Boiteux, a French engineer-economist working at Électricité de France in the late 1940s, published a series of papers between 1949 and 1960 working out how a single unified utility should set electricity tariffs to maximise welfare. The answer, in short, was: charge each customer the marginal cost of supplying them at the moment they consume.
Boiteux was solving a planning problem, not a market problem. EDF was a vertically integrated monopoly nationalised in 1946. There were no competing generators, no auctions, no bids. But the math turned out to be portable. The same equations that told a state-owned utility how to price its output could tell a competitive market how to clear. The marginal cost of the last unit dispatched is the right price whether you're a planner trying to allocate scarce capacity or an auctioneer trying to make the auction efficient.
This is the deep reason marginal pricing is so resilient to political attack. It isn't a free-market construction. It's a welfare-maximising planning principle that happens to also be the equilibrium of a competitive auction. When critics argue marginal pricing is a Thatcherite imposition, they're missing the history. The math came out of a French state utility before Thatcher was a chemist.
Boiteux's later work on peak-load pricing in 1960 extended the framework to handle capacity constraints and time-varying demand. This is the direct ancestor of today's scarcity pricing mechanisms in ERCOT, capacity markets in PJM, and the cone of debate around how to price reliability under high renewables penetration. The man basically wrote the field's foundational documents and most people working in power markets today have never heard his name.
For forty years Boiteux's framework stayed mostly academic. Most countries ran electricity as a state monopoly. The shift came in 1990 with the privatisation of the Central Electricity Generating Board in England and Wales. The reformers needed a market design, and they reached for marginal pricing because it was the only design that had been worked out to a serious degree of mathematical rigour.
The England & Wales Electricity Pool began trading on 1 April 1990. Generators submitted bids for each half-hour of the next day. The system operator stacked the bids, dispatched cheapest first, and paid every dispatched generator the System Marginal Price set by the most expensive unit needed to meet demand. It was the first competitive wholesale electricity market anywhere in the world, and the design was copied, sometimes word for word, by countries setting up their own markets in the years that followed.
One subtle thing about the UK Pool worth understanding. The original design wasn't pure pay-as-cleared in the way most people think. It included capacity payments calculated separately, and it had quirks in how strategic bidding played out, especially with only two large generators dominating the market early on. But the core principle, that the marginal unit sets the price for everyone, was there from the start and has survived every restructuring since.
The American story is messier because the US never had one electricity market. It has seven regional markets plus large parts of the country (the Southeast and most of the West outside California) that still operate as vertically integrated utilities with no wholesale competition at all. But where competition exists, marginal pricing is the design.
The intellectual home of US market design is the work done by William Hogan and others at Harvard in the 1980s and 1990s, refining marginal pricing into Locational Marginal Pricing (LMP). The innovation: instead of a single clearing price for the whole market, each node on the transmission grid clears at its own price, reflecting the cost of delivering one more megawatt-hour to that specific location given transmission constraints.
This matters because transmission lines have capacity limits. If a wind farm in west Texas is generating cheap power but the line to Houston is congested, you can't actually deliver that cheap power to Houston. The wind generator is stuck in west Texas with too much supply, while Houston has to dispatch a more expensive local plant. LMP captures this by giving west Texas a low price and Houston a high price simultaneously. The gap, called the congestion component, signals where the grid needs more transmission.
PJM was first to adopt LMP at scale in the late 1990s. The other US ISOs followed. ERCOT was the holdout, running a zonal market until December 2010, when it switched to nodal. Today every competitive US wholesale market runs locational marginal pricing, with some variations in the details.
ERCOT deserves special mention because it took the marginal pricing logic further than anywhere else. Most markets pair an energy market (where marginal pricing applies) with a separate capacity market that pays plants just for being available, even if they don't run. ERCOT skipped the capacity market entirely. Generators in Texas earn money only from selling energy and ancillary services. There is no payment for showing up.
The implication: scarcity pricing has to do all the work. When demand surges and reserves fall, prices in ERCOT can spike to the system cap, currently $5,000/MWh, and on rare nodal events have hit numbers like $28,000/MWh in February 2025 due to local congestion. Those spikes are the only mechanism for plants to earn back their fixed costs and for new capacity to get a return on investment.
This design has fans and critics. Fans like Hogan argue ERCOT is the cleanest expression of efficient market design in the world. Critics, especially after the February 2021 winter storm, argue energy-only markets don't deliver enough reliability when demand is correlated with extreme weather. The debate continues. ERCOT itself has been adding mechanisms (the operating reserve demand curve, the performance credit mechanism currently in design) that look increasingly like soft capacity payments without admitting it.
Australia's National Electricity Market launched on 13 December 1998, covering Queensland, New South Wales, Victoria, South Australia and (after a 2005 connection) Tasmania. The Western Australia and Northern Territory grids are separate. The NEM is structurally similar to ERCOT in that it's energy-only with no separate capacity market. It's also a gross pool, meaning all electricity must clear through the wholesale market. There's no pay-as-bid trading happening on the side.
The five regions each clear at their own price, with limited interconnection capacity producing different prices in different states. South Australia, with very high wind and solar penetration, regularly clears at near zero or even negative prices when the sun is shining and the wind is blowing. Queensland and New South Wales, more dependent on coal, tend to clear higher.
The Australian experience is interesting because it's a stress test of the marginal pricing model under heavy renewables penetration. South Australia is essentially a live experiment in what happens to a marginal pricing system when more than half of generation has zero marginal cost. The answer so far: prices become extremely volatile, with frequent zero or negative prints during the day and spikes in the evening peak. Investment has shifted toward batteries and demand response that can profit from the spread.
The NEM also introduced a Run-of-Plant PPA standard for renewables: generators sell their output to a counterparty at a fixed price regardless of when it's produced. The Australian government adopted a 32 GW Capacity Investment Scheme in late 2023 that uses two-way Contracts for Difference, which is essentially the same instrument the EU eventually settled on after the 2022 crisis.
The EU is where the politics gets loudest. Twenty-seven member states, wildly different generation mixes (France 70% nuclear, Poland heavy coal, Germany post-nuclear gas plus renewables, Spain heavy renewables), all coupled into a single day-ahead market via an algorithm called EUPHEMIA that clears prices simultaneously across most of the continent.
The architecture, sometimes called the target model, was built up through a series of directives in 1996, 2003, 2009, and 2019. By 2014 day-ahead market coupling was operational across most of the EU. By 2018 intraday markets were coupled. The cross-border mechanism only works because every member state runs marginal pricing. You can't couple a marginal-price market with a pay-as-bid market: the math doesn't compose.
This matters for the reform debate. When France or Spain proposes structurally decoupling electricity prices from gas, they're proposing something that would force the EU to abandon market coupling, or build a parallel system, or fragment the single market. None of those are easy. Marginal pricing in the EU isn't just a domestic design choice. It's the protocol that makes the cross-border market exist.
Pre-2022 the EU merit order looked roughly like this: nuclear and renewables at the bottom, then lignite, then hard coal, then combined-cycle gas, then peaking gas and oil at the top. Gas set the price perhaps 40% of the hours. The rest of the time it was coal, or in mild seasons, renewables and nuclear. The system worked. Wholesale prices in 2019 averaged around €40/MWh across the major markets.
Then came 2022.
Russia's invasion of Ukraine and the subsequent weaponisation of gas exports created a textbook stress test of marginal pricing. Russian gas exports to the EU dropped 60% in 2022. Gas prices on the Dutch TTF benchmark peaked above €300/MWh in August, roughly nine times the pre-crisis level. Several other things went wrong simultaneously: more than half of France's nuclear fleet was offline at points due to corrosion problems, southern Europe had a severe drought that wrecked hydropower, and CO2 prices in the EU ETS were running 70% above 2019 levels.
The result was wholesale electricity prices that averaged around €220/MWh across the EU in 2022 versus €40/MWh in 2019, with spot peaks in the summer of €600/MWh. A Ramboll analysis put the additional cost to EU consumers at €336 billion in 2022 alone.
And here's where the politics turned sharp. Because of marginal pricing, gas plants set the clearing price for nearly every hour of 2022, even when most of the generation in any given hour was renewable, nuclear, or hydro. Operators of those low-cost plants, who hadn't seen their costs change, were earning the gas-set clearing price. The gap, the infra-marginal rent we mentioned earlier, ballooned from a normal year's reasonable margin into what was widely called a windfall. Spanish and German renewables operators, French nuclear, Norwegian hydro: all of them earning revenues unrelated to their actual costs.
Two narratives competed. Narrative one, mostly from southern Europe (Spain, Portugal, Greece, France): the system is broken. Why are consumers paying gas prices for non-gas electricity? Decouple the prices, cap the rents, fix this structurally. Narrative two, mostly from northern Europe (Germany, Netherlands, Nordics): the system is doing exactly what it's supposed to do. The high prices are sending the right investment signal. Build more renewables, fewer gas plants. Don't break the mechanism.
Spain and Portugal got an EU derogation in June 2022 to cap the price of gas used for power generation, effectively breaking the link between gas and electricity prices on the Iberian peninsula. It was popular politically and reduced consumer bills in the short term. It also increased gas consumption (because subsidising gas for power makes it cheaper to burn), required substantial public spending, and was discontinued at the end of 2023 once gas prices normalised. A useful cautionary tale for anyone proposing similar interventions elsewhere.
What didn't happen, despite the political pressure, is anyone actually scrapping marginal pricing. Spain came closest to proposing it. The Commission reviewed alternatives. Bruegel, Cambridge EPRG, and most academic energy economists pointed out that pay-as-bid auctions produce worse outcomes (you've now incentivised generators to lie about their costs), average-cost pricing breaks dispatch efficiency (you'd run expensive plants when cheap ones were available), and tech-specific pricing creates a regulatory mess. Every alternative had bigger problems than the current system.
The European Commission published its reform proposal on 14 March 2023. After eighteen months of Council and Parliament negotiation, the final text was adopted in May 2024 as Regulation (EU) 2024/1747. The core compromise: keep marginal pricing in spot markets, push everyone toward long-term contracts.
The structural insight here is worth pausing on. The reform doesn't fix the gas-windfall problem by changing how spot prices form. It fixes it by changing what generators actually earn. A wind farm under a two-way CfD doesn't pocket €600/MWh during a gas crisis. The spot price might be €600/MWh, but the wind farm is earning its strike price (say €60/MWh) and the difference flows back to the state, which can rebate consumers. The signal still gets sent (high prices say build more capacity), but the windfall is intercepted.
The UK has been running CfDs for renewables since 2014, predating the EU reform. The Contracts for Difference scheme has been the main support mechanism for offshore wind, onshore wind, solar, and tidal generation. The UK's 2023 Review of Electricity Market Arrangements (REMA) consultation explored more radical changes including locational marginal pricing in Britain, splitting the market into nodal zones similar to the US. As of 2026 the government has decided to retain the national pricing model but pursue zonal reforms in transmission constraint management. Marginal pricing stays.
In the US, the conversation has been quieter because the 2022 crisis hit Europe much harder than North America. ERCOT's design debate post-Winter Storm Uri (February 2021) is more relevant: should Texas add capacity payments? The answer being slowly worked out is "yes, but call them something else." The Performance Credit Mechanism is being designed as a way to compensate generators for being available without admitting it's a capacity market. Locational marginal pricing as a mechanism is not under debate anywhere in the US.
Australia's Capacity Investment Scheme, announced November 2023, mirrors the EU CfD approach: government underwrites long-term Run-of-Plant PPAs for new wind and solar at strike prices set by competitive auction. The intent is to add 32 GW of renewables by 2030 with revenue certainty, while keeping the spot market's marginal pricing intact. Same architectural choice as the EU: layer contracts on top, leave the mechanism alone.
Three things are worth holding in your head simultaneously if you work in or around power markets.
First, marginal pricing is not going away. The 2022 crisis was the most serious political pressure the design has ever faced and it survived. Every major reform package, EU, UK, US, Australia, has chosen to layer contracts on top rather than replace the underlying mechanism. The mechanism is too useful for dispatch efficiency, too embedded in cross-border market coupling, and too well-defended by the economics profession to be torn out.
Second, the share of revenue actually settled at the spot marginal price is shrinking. With CfDs becoming mandatory for new EU capacity, with PPAs growing fast in Spain and the US, with capacity-investment-scheme contracts in Australia, and with the UK's contract regime maturing, more and more generation is earning a fixed price under contract, not a variable price from the spot market. The spot market still clears at marginal cost, but fewer megawatt-hours are actually settling at that price.
Third, the implication for anyone selling into the energy industry is that the buyer landscape is shifting. Off-takers signing long-term PPAs, structured product desks designing CfDs, asset managers underwriting contracted renewable portfolios, industrial consumers procuring electricity directly from generators rather than through utility tariffs: these are all bigger categories than they were five years ago and will be bigger still in five years' time.
The plumbing of marginal pricing isn't disappearing. It's just becoming the substrate that everyone hedges against, rather than the price most participants are actually exposed to.
EnergyLead Partners runs B2B outbound for companies selling into US energy markets. Sales Navigator, enrichment, multichannel sequencing, fully managed.
Get in touch