The effect of environmental announcements on price performance in the European energy sector

This thesis investigates the impact of 450 environmental news events on the stock price behavior of 123 European companies. Using a diverse modeling framework Logistic Regression, Random Forest, XGBoost, and Quantile Regression—we explore both predictive accuracy and interpretive insight. XGBoost with cross-validation achieved the best predictive performance (AUC = 0.5723), making it the selected classification model. However, Quantile Regression revealed deeper economic patterns: while positive sentiment increased returns in the lower quantile (τ = 0.25), it had the opposite effect in the upper quantile (τ = 0.75), suggesting non-linear investor behavior. These results emphasize the value of combining machine learning with interpretive models to understand how markets react to ESG-related disclosures. This approach benefits analysts, investors, and policymakers aiming to decode market sentiment around sustainability.

The effect of environmental announcements on price performance in the European energy sector