The Curves Behind the Price of Power

The current argument is not obscure at all. If wholesale electricity prices become politically intolerable, should governments soften the signal inside the market itself, or leave the market-clearing logic in place and deal with the consequences elsewhere?

That question returned with force when Italy moved to reimburse certain thermoelectric producers for carbon costs. Whatever view one takes of the proposal, its meaning is clear enough. Carbon costs would matter less at the moment bids are compared. The curve would still exist, but it would not be the same curve.

That is why Lion Hirth has been trying to explain the merit order all over again. His point is not nostalgic and not doctrinal. The merit order, he argues, is simply the short-run supply curve. Offers are ranked from cheaper to dearer. Demand cuts through that stack. The marginal accepted unit helps determine price. Only after that comes the next question: how are accepted bids paid?

The confusion begins when these layers are folded into one another. "Merit order" starts to sound like a philosophy of power markets, as if someone had chosen it out of taste. In practice it names something more basic. If scarce supply is being matched against demand, some ranking has to happen somewhere. The dispute is not over whether the ordering exists. The dispute is over how much political intervention should take place before the market is allowed to clear.

In the current public debate, the merit order is best understood as the short-run supply curve behind wholesale power prices, not as the rulebook of one particular auction. Particular markets and products express that logic in different ways.

Most producers and consumers do not live directly on the spot market. They hedge. The political argument is about the role of short-run prices in system coordination and forward-price formation, not about every household instantly paying the hourly auction price.

Three different things are often collapsed into one.

First comes ranking. Bids are ordered from cheaper to dearer.

Second comes selection. Enough capacity is awarded, or enough energy is activated, to meet the requirement.

Third comes payment. Successful bids are then paid either a common clearing price or their own bid price.

That distinction matters because the current political argument begins in the wholesale market, where the European Commission describes the system as one of marginal pricing, also known as pay-as-clear. Bids are compared, the market clears, and all generators selling in that moment receive the same market price. German balancing markets show that the same ranking problem can sit alongside different payment rules. FCR capacity keeps the ranking logic and pays the highest awarded price to all successful bids. aFRR and mFRR capacity also rank bids, but successful providers are paid their own bid prices. On the energy side, aFRR and mFRR bids are transferred into merit-order lists and activated in ascending price order.

These three layers do not define the merit order by themselves. They are ways specific products express parts of the broader short-run supply-curve logic.

The reserve examples that follow do not replace the wholesale concept. They make one part of it visible.

Before it became a slogan in policy arguments, the merit order had a quieter life.

Its first home was not the exchange screen but the control room. Long before today's coupled auctions, power systems faced a simpler question: which available plant should run next if demand has to be met as cheaply as possible? That instinct later became formalized as economic dispatch.

Its second life was mathematical. In 1962, Jacques Carpentier extended the dispatch problem into what became known as optimal power flow, adding the network itself to the optimization. The cheapest stack was no longer enough. Transmission constraints and the physics of the grid had entered the calculation.

Its third life was economic and political. By the late 1980s, spot-pricing theory had made the connection between short-run marginal cost and visible electricity prices much more explicit. In Europe's liberalized markets, that logic later became embedded in wholesale auctions and market coupling. In Germany, the phrase took on a wider public life again once the renewables debate turned toward the "merit-order effect": the way low-marginal-cost renewable generation could push prices down along the conventional stack.

The current dispute is younger than the idea itself. The merit order is not a recent ideological invention. It is an old dispatch instinct, later formalized as an optimization problem, and only then turned into a visible market curve and a political keyword.

The clearest public window into ranking and award is not the day-ahead auction. It is reserve.

When most readers say "the power price," they mean the wholesale market for the next day. But if the aim is to make the logic of ranking visible, German balancing procurement is the better place to start. Regelleistung publishes auction outcomes and anonymized awarded bids. That is not full transparency. It is enough to show the shape of a real auction at the point where it matters most: the awarded edge.

Take one product: negative aFRR, 08:00 - 12:00. Price rises up the vertical axis. Awarded megawatts accumulate along the horizontal axis. On the left sit cheaper accepted bids. On the right, the expensive awarded edge. Offers were ranked. A requirement cut through them. The margin landed in one place on one day and in another place on another day.

The public reserve file does not show the whole tender book. What appears is the anonymized awarded side of the auction. The rejected tail beyond the award is missing. What appears on the page is therefore not the full depth of the market, but the visible frontier where the requirement was met.

This aFRR capacity chart shows the ranking logic unusually clearly: bids ordered by price until the procurement volume is met. The awarded edge is visible even though remuneration on the capacity side is pay-as-bid rather than pay-as-cleared.

That partial view is still enough to teach something decisive. On 29 June 2025, the same aFRR- 08:00 - 12:00 product cleared materially higher than on the calmer comparison day shown here. The product did not change. The auction format did not change. The system behind the auction changed, and the awarded edge moved with it.

What most people have in mind is the day-ahead auction. That is where the benchmark price for the next day is formed, and it matters directly for storage, flexible demand, pass-through tariffs, industrial hedging, and decentralized portfolios.

Yet the market that dominates headlines is not the market that can be shown most cleanly in public. In DE-LU, orders go into the coupled European day-ahead process. EPEX closes the day-ahead order book at noon on the day before delivery. The outcome sits inside Single Day-Ahead Coupling and is solved through EUPHEMIA, the common price-coupling algorithm that handles cross-border constraints and complex order logic across bidding zones.

The official exchange view is visible as an aggregated supply-demand crossing on EPEX. The hidden book is not.

EPEX publishes results and aggregated curves. That is already useful. It does not amount to an open reusable auction book in the way readers might imagine when they hear "the curve." The richer market data sits inside licensed products and usage conditions. A public notebook can show the official exchange outcome and build a transparent benchmark beside it. That is the boundary.

On EPEX, the day-ahead auction is visible as an aggregated supply-demand crossing. That crossing is the market result for a given delivery period. It is the closest public view of the actual auction.

The benchmark on this page is a different object: a parsimonious open-data explanation built from load, wind, and solar.

Public load, wind, and solar are turned into an hourly residual-demand proxy. A stylized inflexible floor is removed. The remaining demand is then cleared against a transparent benchmark stack of lignite, coal, gas, imports or flexible demand, peakers, and scarcity. That gives a reconstructed hourly benchmark price.

The daily chart compares actual clearing prices with that benchmark across the day. The selected-hour view then turns one delivery hour back into a supply story: given observed load, wind, and solar, which benchmark layer would be marginal, and how far away does the actual auction clear?

The gap is where omitted costs, constraints, and bidding structure start to matter. It reflects both the simplifications of the benchmark - a fixed floor, fixed benchmark costs, and a stylized oversupply rule - and the structure the model cannot see directly: block bids, CHP heat constraints, hydro opportunity cost, outages, tighter cross-border conditions, portfolio bidding, and the non-convex logic of the coupled auction.

Each control changes a different assumption. The offset changes the overall level. The technology sliders change the order of the stack. Strategic-renewables mode changes how part of renewable output enters the stack at all. Below the stylized inflexible floor, the benchmark switches into a simple oversupply pricing rule.

FCR is a useful point of comparison.

It too is a ranked auction. Offers are ordered. A requirement cuts through them. But the remuneration rule is different. In Germany, FCR is a symmetric product, procured in six four-hour slices, and all successful bids receive the highest awarded price. That is pay-as-cleared.

Day-ahead wholesale, FCR, and aFRR each combine ranking, selection, and payment in different ways. Day-ahead and FCR both end with a common price for successful bids. FCR and aFRR are both reserve-capacity products procured against a system requirement. aFRR capacity then changes the remuneration rule while keeping the ranking logic. Seen together, the examples clarify more than any one of them alone.

Changing the payment rule in one product does not by itself abolish the underlying short-run supply-curve logic. FCR belongs here as a point of comparison. Like day-ahead, it ends with a common price for successful bids. Like aFRR, it is procured against a system requirement rather than wholesale demand.

The deeper lesson appears in aFRR.

Some of the changes in German and European balancing design have moved the system closer to a clearer merit-order logic. Germany's common merit-order list reduced fragmentation across control areas. The balancing energy market introduced in November 2020 allowed balancing energy to be delivered without prior participation in the capacity auction, so cheaper energy bids could compete more directly. Later integration into PICASSO pushed that logic further by combining local merit-order lists into a common European activation process aimed at cost-efficient selection within system-security limits.

Other choices still preserve a different structure. aFRR and mFRR capacity remain pay-as-bid. Capacity procurement still has the character of an insurance product. Operational deviations from the Germany-wide merit order still occur in exceptional cases, including grid bottlenecks and process-control issues.

That is not a departure from the merit order so much as a reminder that merit-order activation lives inside a system that still has to survive real operational constraints.

The closer the market moves to real time, the less persuasive it becomes to talk as if the choice were simply "market" versus "intervention." The ranking logic survives, but it survives inside a denser shell of capacity rules, platform design, settlement rules, fallback arrangements, and security constraints.

Once that structure is visible, the policy question sharpens.

If governments dislike the price, they have several options, and they are not the same option with different slogans attached. They can let the market clear and compensate households or firms afterwards. They can reimburse or suppress costs before bids are compared, which changes the visible supply curve itself. Or they can move more volume outside the short-term auction altogether, into contracts, regulated procurement, or other administrative allocation methods.

This is why the current fight matters. Complaining about the price is easy. Replacing the mechanism is not. Germany's day-ahead market is embedded in SDAC and EUPHEMIA, with coupled clearing across bidding zones and a common algorithmic design. Changing that world would mean changing how costs enter bids, how coupling works, or what share of volume bypasses the auction entirely.

The deeper point is the one Hirth has been trying to drag back into view. Germany's power price does not come from one villain, one technology, or one theory. It emerges at a moving edge where demand meets the next unit of supply under a particular market design. The fight begins when politicians decide that edge has become too painful to leave alone.