Abstract
Purpose – Clean energy stocks have recently received significant attention from both investors and researchers, reflecting their growing importance in financial markets. This paper forecasts clean energy stock (CES) returns using many predictors, including technical, macroeconomic, climate risk and financial predictors. The goal is to reveal how different predictor groups work and their time-varying patterns.
Design/methodology/approach – This study establishes a robust forecasting framework using monthly data from the WilderHill Clean Energy Index, spanning January 2009 to December 2023, and integrates 56 predictors across four categories. To address multicollinearity and identify key drivers, the framework applies advanced shrinkage methods, regularization, quantile regression and model combination. This offers a dynamic solution for forecasting CES returns.
Findings – The study identifies macroeconomic predictors as the most stable and powerful drivers of CES returns; the Chicago Fed National Activity Index (CFNAI) is a particularly important indicator. Climate predictors show temporal variability, while technical and financial predictors are more important during market volatility. A group-level analysis highlights macroeconomic variables as key to forecasting accuracy. Climate predictors play critical roles in specific periods. Medium-term dynamics (2–4 months) associated with macroeconomic predictors have the strongest impact on performance.
Originality/value – This paper introduces a novel approach to forecasting CES returns by integrating 56 diverse predictors. This addresses research gaps, given the previous focus on traditional predictors or single-model frameworks. The study further examines the roles of predictor grouping, component selection, rolling windows and forecasting horizons in increasing prediction accuracy and in describing the dynamic interactions driving CES returns.
