Evaluating the Effects of Green Credit Guidelines on Corporate ESG Performance Through a Double Machine Learning Approach

The double machine learning model effectively tackles challenges like model specification errors and regulation bias that conventional models may encounter. Evaluating the efficacy of policies is crucial for advancing the shift towards a low-carbon economy. This study explores the effects of Green Credit Guidelines on corporate ESG performance through the dual machine learning approach, aiming for a more robust estimation of policy effects. We analyze the data spanning from 2009 to 2021, focusing on companies listed on the A-share market. Our findings reveal that while the Green Credit Guidelines significantly improve overall corporate ESG performance, they have a negative impact on the social (S) component. Despite varying control variables, sample separation ratios, and machine learning models, the results remain robust. In contrast to the difference-in-differences model, the double machine learning approach produces smaller coefficient values. The effects of the policy impact varies with ownership and pollution levels, indicating that heterogeneity exists in the policy enactment.