Why negative electricity prices do not always switch off wind and solar

On a sunny or windy day, producing every available unit of renewable electricity is not necessarily useful. Curtailment—deliberately reducing output below what the weather permits—can be part of efficient power-system operation. Negative wholesale prices provide a financial signal to reduce generation: sellers effectively pay to supply electricity.

But that signal does not reach every generator unchanged. In an arXiv preprint, Lion Hirth examines two separate questions: which German wind and solar generators should find curtailment financially attractive, and how much production actually appears to stop? The supplied manuscript is dated 10 November 2026; its findings remain preprint evidence, not settled conclusions.

When producing still pays

The central mechanism is straightforward. A generator receiving a payment for each megawatt-hour produced loses that payment when it switches off. It may therefore keep generating at a negative market price if the support payment more than covers the loss.

A fixed feed-in tariff goes further: it pays a fixed amount for electricity supplied, shielding the generator from wholesale prices. Under that arrangement, a negative wholesale price alone does not make stopping profitable.

Germany's market premium works differently. It supplements market earnings when a reference price falls below a specified support level. The premium can make continued production worthwhile below zero, but the stopping threshold changes with the reference price. Rules suspending support during sufficiently long negative-price episodes change that calculation again.

Self-consumed rooftop electricity has another incentive: it avoids buying electricity at the household's retail tariff. If that tariff is fixed, negative wholesale prices do not change the immediate saving.

Reconstructing incentives and missing output

Hirth divides the fleet into 30 groups distinguished by technology, commissioning date, size and support arrangements. He reconstructs their financial incentives hour by hour, accounting for different payment levels and negative-price rules.

Measuring behavior is harder. Market-based curtailment is not directly recorded in a comprehensive statistic. The study estimates how much electricity the weather could have produced, then subtracts actual generation and recorded grid-related curtailment. The latter is output reduction associated with network constraints, rather than the market response being investigated.

The production model uses ERA5 reconstructed historical weather, including wind speeds at 17 locations and solar irradiance. To avoid teaching the model that curtailed production represents normal weather-driven output, Hirth calibrates it on hours selected to exclude likely price-driven curtailment.

This produces a fleet-level estimate, not a record of individual plants switching off.

Wind and solar respond differently

Across the 573 negative-price hours reported for 2025, the study estimates that 64% of potential wind generation had a financial incentive to curtail. About 27% of that exposed potential was actually curtailed, equivalent to 17% of total wind potential in those hours.

For solar, 28% of potential generation had an incentive to stop. The estimated response among that exposed portion was 9%, leaving total curtailment at roughly 3%.

These percentages describe potential electricity production during negative-price hours—not shares of annual production or counts of installations.

At prices below minus €50 per megawatt-hour, the contrast sharpened. Almost all wind potential had an incentive to curtail, but only about half was curtailed. For solar, just 39% had an incentive; around one in seven of that exposed portion responded, leaving approximately 6% of total solar potential curtailed.

Estimated market-based curtailment in 2025 totaled 2.1 terawatt-hours for wind and 0.6 for solar. Through October 2026, the corresponding estimates were 2.5 and 1.8 terawatt-hours. The latter figures cover an incomplete year.

What the estimates cannot explain

The gap between calculated incentives and output does not establish that owners simply ignored profitable opportunities. Private contracts with electricity aggregators can change who controls dispatch and who bears losses. Those terms are largely unobserved. Remote-control limitations and costs omitted from the calculation may also matter.

The incentive estimates assume generators correctly anticipate reference prices. They also simplify compensation delivered through extensions of the support period; the appendix acknowledges that this can overstate the incentive to curtail.

The output estimates have a separate vulnerability: any weather-model error specific to negative-price hours can look like curtailment. Reported generation itself is extrapolated from samples, and unrecorded network restrictions can contaminate the residual. Negative solar estimates in earlier years illustrate that measurement noise remains; Hirth does not claim measurable solar curtailment before 2025.

Two problems, not one

The policy distinction is between giving generators a reason to respond and enabling that response. Hirth discusses support independent of a plant's own production as one way to preserve price signals, while acknowledging a trade-off in investment-risk protection. His analysis does not establish the best replacement design or directly measure grid-stability incidents.

This article relies on the supplied PDF text extraction, which omits figures and may damage tables, equations or layout. Derived data and code are available from the author on request, rather than supplied here for replication. Hirth discloses that he runs an electricity-sector consultancy and reports no project funding or client commission.