A1174
Title: Deepening the relationship between ESG score and firms' performance via machine learning
Authors: Susanna Levantesi - Sapienza University of Rome (Italy) [presenting]
Rita DEcclesia - Sapienza University of Rome (Italy)
Valeria D Amato - University of Salerno (Italy)
Abstract: Several firms are already adopting Environmental, Social, and Governance (ESG) into their governance, investment strategies, and risk management. Existing literature provides limited evidence of the relationship between the ESG score and the firm's profitability. We explore this matter by analyzing a sample of the companies constituting the EuroStoxx-600 index using different machine learning models. We aim to assess whether the ESG score significantly influences the firms' profitability measured by the earnings before interest and taxes (EBIT). We further deepen the relationship between ESG score and EBIT through machine learning interpretability toolboxes such as partial dependence plots and individual conditional expectation, which help to visualize the functional relationship between the predicted response and one or more features, and the Shapley value allowing us to examine the contribution of the feature to the prediction. Our findings show that the model can reach high levels of accuracy in detecting EBIT and that the ESG score is a promising predictor, compared to other traditional accounting variables.