From day-ahead to mid and long-term horizons with econometric electricity price forecasting models

Abstract

Robust econometric models for mid to long-term electricity price forecasting have become increasingly critical due to evolving market dynamics and price volatility. The European energy crisis in 2022 led to unprecedented fluctuations in energy prices, underscoring challenges for operational and risk management. After a comprehensive literature analysis, we address key challenges: (1) Constraining coefficients with bounds derived from fundamental models for interpretability; (2) Incorporating seasonal expectations of regressors such as load and renewables to stabilize long-term forecasts; (3) Managing unit root behaviour of power prices by estimating same-day relationships and projecting them forward. We develop interpretable models for forecasting horizons from one day to one year and provide guidelines on modelling frameworks and key variables. A practical application with scenario analysis demonstrates the framework. We conduct forecasting studies on Germany’s hourly electricity prices, by applying regularized regression and generalized additive models. • CO 2 , gas, coal and oil futures explain short-run unit root behaviour of power prices. • Including fuels drastically decreases the forecasting accuracy beyond 3 months ahead. • Long-run forecasts of renewables and load are crucial for model stabilization. • Spurious effects handled by estimating current relationships, then projecting. • Econometric models rival black-box models, adding interpretability and scenarios.

Topics & Concepts