Assessing the determinants of corporate environmental investment: a machine learning approach

In recent years, experts and academics in the environmental management field have developed an interest in the factors and evaluation techniques that influence corporate environmental investment decisions. However, there are substantial differences between studies employing the most recent evaluation methodologies and those that use indicator systems. To explore the mechanisms that influence corporate environmental investment, this study investigated the determinants of environmental investment through the perspectives of firm, board, chair, and chief executive officer (CEO) characteristics using a machine learning approach. Based on a large-scale data sample from Chinese-listed companies, the results indicated that the extreme gradient boosting (XGBoost) model had an accuracy of up to 97.63%, thus performing the best. Additionally, the model that used SHapley Additive exPlanations (SHAP) to interpret XGBoost showed that a company's sales performance was the most important factor that influenced environmental investment, followed by CEO tenure, board independence, board gender diversity, chair academic experience, and the company's level of internationalization. Furthermore, when examining the sample of heavily polluting enterprises, sales, board gender diversity, CEO tenure, chair academic experience, board independence, and chair-CEO duality, all were found to play crucial roles in predicting environmental investment. Overall, this study aids in evaluating the factors that influence corporate environmental investment decisions and provides policymakers and practitioners with a machine learning approach for use when assessing these factors.