Two Ways Electricity Prices Change Tell Different Stories About Renewables

When people say wind and solar power make electricity prices more volatile, what kind of price change do they mean? A new preprint by economist Lion Hirth argues that the answer matters: two movements commonly discussed together can reflect different market conditions and carry different risks.

The paper combines an economic model with German electricity-market data spanning up to 25 years. It finds that prices for different delivery periods have become more dispersed since the 2010s, while a measure of price changes for the same delivery period declined over much of the available record. The analysis is descriptive, however, so it cannot establish how much of either trend was caused by renewable energy.

Two clocks, two kinds of change

Electricity markets work with both a delivery time and a trading time. In a day-ahead auction, buyers and sellers agree on prices for electricity to be delivered during separate periods the following day. Those prices can differ substantially: electricity for a sunny midday period is not the same product as electricity for an evening period.

Hirth calls differences across delivery periods variability. They can exist even if everyone knows tomorrow's supply and demand perfectly. Electricity can be stored, but storage remains costly and limited, so a surplus in one period cannot always be shifted cheaply to another.

After the day-ahead auction, trading continues closer to delivery. A forecast may change, or a generator may go offline, prompting traders to revise what they will pay for electricity in the same delivery period. Hirth calls the resulting change across trading time volatility.

The terminology is the paper's proposed distinction, not a universal rule: Hirth notes that many studies use volatility for differences between delivery-period prices. Under his definitions, the two measures describe different things. One can be high without the other being high.

What the German data show

Hirth examines day-ahead prices from 2001 to 2026, alongside shorter records of intraday prices and imbalance settlement prices. The latter are used to settle remaining mismatches between market participants' positions and metered electricity; they are set by a formula rather than emerging directly from voluntary trades.

To compare years with very different overall price levels, the main measures divide price dispersion by the annual average day-ahead price. The analysis also trims observations outside the first and 99th percentiles to reduce the influence of extremes.

By this measure, day-ahead variability fell from the 2000s to the 2010s and rose again after 2020. In 2024–2026 it was roughly 50–51% of the annual mean price, near the upper end of the range observed in the 2000s.

The measure of volatility based on the spread between imbalance and day-ahead prices fell from 76–147% in 2010–2015 to 42–53% in 2025–2026. Intraday volatility measures generally showed no trend. These are relative measures: in absolute monetary terms, both variability and volatility rose with the price level after 2021.

The paper also groups days by actual wind and solar generation. Since 2024, within-day price variability was highest on days with high solar generation and lowest on windy days. A measure of intraday volatility had risen with wind generation until 2020, but since 2021 showed little dependence on wind or solar generation.

Why the distinction matters

Hirth's simplified model offers one way to understand the solar pattern. If solar production initially coincides with high daytime demand, it can flatten differences in the remaining demand that other generators must meet. With more solar capacity, the model suggests, additional generation can instead deepen the midday trough and widen differences across the day. The observed pattern is consistent with that mechanism, but does not prove it caused the change.

The distinction also matters for market participants. Differences between delivery periods create price spreads that storage operators may try to exploit. Changes in the expected price of the same period affect positions held as delivery approaches and can increase the value of being able to adjust those positions.

What remains unresolved

Germany's energy system changed in many ways during the period studied. Fuel prices, market integration, trading liquidity, flexible capacity and settlement rules all changed alongside renewable generation. The study does not separate their individual effects. In particular, the imbalance-price formula changed repeatedly, complicating comparisons over time.

The intraday measures use averages of trades rather than complete transaction-by-transaction price paths, and hourly averaging understates the measured volatility of individual quarter-hours. Market participants' forecast errors were not publicly available for the analysis. Normalizing by average prices also affects the results; the paper reports alternative measures and normalization checks.

The finding is therefore not that wind and solar cannot affect uncertainty or trading risk. It is that Germany's observed price movements depend on which clock is being measured—and that the long-run data alone cannot assign the trends to a single cause.