The enhanced benefits of ESG in portfolios: A multi-factor model perspective based on LightGBM

This paper explores the enhanced benefits of ESG in portfolio models from the perspective of a multi-factor prediction model. Firstly, using the traditional Barra multi-factor model as a basis, the Fama-Macbeth test is employed to evaluate the explanatory power of the multi-factor model after incorporating the ESG factor. Considering the non-linear nature of ESG factors, the Light Gradient Boosting Machine (LightGBM) model is utilized for feature identification and optimization, thereby enhancing the accuracy of predicted returns in the multi-factor model. Secondly, two portfolio models, namely Barra+ESG+MV and Barra+ESG+BL models, are constructed by introducing the predicted returns based on Barra and ESG factors into Mean-Variance (MV) and Black-Litterman (BL) models, respectively. This facilitates a further analysis of the benefits of ESG in asset portfolios. Finally, the advanced BPZ model is used to examine the robustness of the gain effect attributed to the environmental (E), social (S), and corporate governance (G) factors in ESG from a non-linear perspective. The findings reveal that ESG significantly improves portfolio performance, particularly in terms of reducing systematic risk. Furthermore, the E, S, and G factors bring universal enhancement effect to the multi-factor model. This study provides new avenues for research on ESG in portfolio management.