Why capture rates matter
A wind or solar plant does not earn the average power price. It earns the price that happens to prevail when it is generating — and because weather drives every similar plant in a region at once, those are disproportionately the hours when output is high and prices are low. The capture rate measures this directly: it is the volume-weighted average price a technology actually earns, expressed as a share of the simple average (baseload) price over the same period. A capture rate of 100% means the plant earns exactly the average price; below 100% means it earns less — the revenue erosion commonly called cannibalization.
The mechanism is the merit order. Generators are dispatched cheapest-first, and the last unit needed sets the price everyone receives. When wind and solar — whose marginal cost is near zero — supply a large share of demand, the clearing price falls into the cheap part of the stack. The more of a technology there is, the more its own output coincides with these depressed prices, so its capture rate declines as its market share rises.
This matters for project finance because revenue, not output, repays debt. Two plants with identical generation can have very different earnings if one consistently produces into low-price hours. You can see this trade-off across all 14 zones — capture rate against renewable penetration — and by zone and technology over time.
How we compute it
We use the simplest well-posed definition — the naive day-ahead capture rate:
capture rate = [ Σ( DAM price × output ) / Σ( output ) ] ÷ mean( DAM price )
The numerator is the volume-weighted average day-ahead price the technology earns; the denominator is the period’s baseload price. We compute it for the Nordic price zones (NO1–NO5, SE1–SE4), Denmark (DK1, DK2), Finland, Germany (DE-LU) and the Netherlands, for 2019–2025, from hourly ENTSO-E Transparency data — day-ahead prices and actual generation per production type.
Two deliberate choices. First, the data is resolution-agnostic: 2019–2024 clear hourly, while 2025 spans the European shift to 15-minute settlement. We weight every interval by its energy (MW × interval length) and the baseload by time, so a single formula is exact across the transition. Second, we keep negative-price hours rather than discarding them; they are part of what a generator actually faces, and removing them would flatter the result.
The forecast-error question
Most published capture rates — including ours here — multiply actual output by the day-ahead price. That implicitly assumes the whole volume is sold a day ahead at that price. In reality a wind or solar plant sells its forecast in the day-ahead auction; the difference between forecast and actual is settled afterwards in the imbalance market, usually on worse terms, and precisely when the whole system is long. So the price a producer actually banks differs from the naive number by a forecast-error penalty.
It is tempting to add that penalty back at the market level — to take a market imbalance price and apply it to “all wind’s” deviation. We do not, because it is not well defined. The imbalance market is reflexive: its prices and volumes are the aggregate forecast error. If the market were perfectly balanced at the day-ahead auction, imbalance volume would be zero and the imbalance price would mean nothing. You cannot price the aggregate error with a number that exists only because of that error, and you cannot attribute the system’s net imbalance to one technology’s gross deviation.
The consequence is clean rather than limiting. The naive day-ahead capture rate is the only well-posed quantity at the market-aggregate level — it needs no forecast and nothing endogenous. The realized-versus-banked gap is real, but it is asset-specific: it depends on a given plant’s own forecast quality and balancing position, and is unobservable from public market data. It belongs to a per-asset analysis, not a market-wide headline. Most analyses quietly present actual × day-ahead as “what producers earn”; we state plainly where that number stops.
Interconnectors
A cross-border cable earns nothing from generation — it earns from the price difference between the two zones it connects. We report each Norwegian DC link (NorNed to the Netherlands, NordLink to Germany, Skagerrak to Denmark) two ways. The first is congestion rent: flow × (price on the expensive side − price on the cheap side). This rent is collected by the two transmission system operators jointly and split 50/50, so it is a system quantity, not a generator’s revenue.
The second is the one that matters for Norway’s energy balance: the timing premium on the flow itself. We compare the Norwegian price during export hours, and during import hours, against the Norwegian baseload. This exposes an asymmetry that a single rent figure hides — exporting drains flexible reservoir hydro, the most valuable resource in the system, while imports arrive as intermittent surplus that cannot refill a reservoir. Only rain and snowmelt do that. See the cables.
What this is — and isn’t
- Naive, not settlement-realistic. Actual output × day-ahead price; the forecast-error penalty (above) is not priced in.
- Market-aggregate, not per-asset. These are zone-and-technology averages, not any single plant. A specific project’s location, profile and forecasting change its numbers.
- Day-ahead only. No intraday or balancing revenue; no ancillary services.
- No storage modelling. We do not model batteries shifting output between hours.
- Onshore and offshore wind are merged, as the source data does not reliably separate them.
These are revenue-side metrics. They are necessary for judging a project, but not sufficient — cost structure (CAPEX, LCOE) and any contracted revenue (PPAs, CfDs) sit alongside them.
Data & sources
Hourly day-ahead prices, generation per production type, load and physical cross-border flows from the ENTSO-E Transparency Platform, 2019–2025. Capture rates are computed by MeginLeid’s capture-rate engine and recomputed as new data arrives. Back to capture rates.