MSc dissertationIndividual project

Electricity Price and Renewable Energy Stock Returns

Is wholesale electricity-price volatility associated with the daily share returns of European renewable energy generators?

Context: MSc Finance and Investment Management, University of Liverpool · submitted April 2026
Role: Individual research project
Data and tools: Daily data, Jan 2018 – Jan 2026 · Python

Python
GARCH
Panel regression
Energy economics

Problem

As European renewable generators move from fixed feed-in tariffs to market prices, their revenue becomes more exposed to wholesale price swings. The research question was whether that volatility is associated with their daily share returns, and whether expected and unexpected volatility behave differently.

My contribution

An individual dissertation: I collected the data, built the price proxy, volatility model and regressions in Python, and wrote the thesis, Electricity Price Volatility and Renewable Energy Stock Returns: Evidence from European Liberalised Electricity Markets.

Approach

Price proxy. The daily mean of German, French and Swedish wholesale prices, January 2018 to January 2026.
Volatility. A mean equation with seven daily lags, weekday effects and a crisis dummy, followed by a GARCH(1,2) model with Student-t errors. This splits each day's price move into expected volatility (σₜ) and an unexpected shock (zₜ).
Returns. Pooled regressions of the daily returns of five listed generators (EDP, Verbund, Fortum, ERG and PNE) on zₜ and σₜ, controlling for the STOXX Europe 600, Brent crude and the euro–dollar rate. I also tested lagged shocks and pre- and post-crisis periods.

Two stacked time series, 2018 to 2026: the daily price proxy, which spikes to nearly 600 euros per megawatt-hour during 2021–22, and its GARCH conditional volatility, which rises from under 10 to nearly 70 euros per megawatt-hour in the same period — **Figure 1. Price and expected volatility.** Daily price proxy (top) and GARCH conditional volatility σₜ (bottom). The shaded band is the 2021–22 crisis window used in the analysis.

Two stacked time series: raw daily price shocks swinging up to about 150 euros per megawatt-hour during the crisis, and the same shocks divided by GARCH volatility, which stay within about plus or minus 4 throughout — **Figure 2. From price shocks to volatility innovations.** Raw unexpected price changes (top) grow sharply during the crisis. Scaled by the model's expected volatility (bottom), they become an even series with a standard deviation of 1.0. This scaled series zₜ is the shock variable tested against returns.

Results

Unexpected shocks (zₜ) showed no statistically significant association with same-day returns in any model.
Expected volatility (σₜ) was only weakly associated over the full sample (p ≈ 0.09), but significantly associated after the 2021–22 crisis (p = 0.034).
Shocks lagged by two days were significantly associated with returns (p = 0.043).
The equity market dominated: a 1% move in the STOXX Europe 600 was associated with a 0.62% move in the generators' returns (p < 0.001).

Taken together, the association between electricity prices and these stocks appears delayed and regime-dependent rather than immediate. These are statistical associations; the study does not establish cause and effect.

Limitations

The price proxy averages three markets, so it may not match the prices individual firms actually received.
Five firms is a small sample, and firm-level data was only available quarterly or annually.
Some serial dependence remained after the GARCH fit, regression residuals were heteroskedastic and non-normal, and the models explain under 10% of return variation.
The models are linear; non-linear methods may capture effects these miss.