In this scholarly investigation, we meticulously assess the effectiveness of Environmental, Social, and Governance (ESG) metrics in predicting financial hardship across a cohort of 3,111 publicly traded companies on the Chinese stock exchange from 2012-2022. This study employs Python software for comprehensive data analysis to process and interpret large datasets efficiently. Our empirical findings robustly validate that incorporating ESG metrics significantly enhances the predictive prowess of our model, thereby elevating precision in discerning instances of financial distress. A striking feature in the process is that the chance of incorrectly identifying the distressed or defaulting firms as sound business enterprises due to the implementation of ESG is impossible. The foundation of our predictive model, bonded by a strong methodology, includes integrating various tools such as classical statistics methods and state-of-the-art new machine learning models. For a comparative analysis seven machine learning models have been employed, such as Logistic Regression, Decision Trees, Support Vector Machines, Random Forest, Naïve Bayes, AdaBoost, and Gradient Boosting. To interpret the results, three performance parameters have been used which are sensitivity (Se), Area Under the Curve (AUC), and F1 score. Among all the models Random Forest Model stands out as the most stable model, which shows 100% accuracy in all the parameters, with and without the inclusion of ESG Scores. The implications of our work also affect the market scene, making an impact through prospective investors, policymakers, and financial parties affiliated with the forefront companies. Additionally, the present contribution develops the existing literature on distress prediction, as it helps to understand which sustainability factors work best for a comprehensive analysis of the company's problems.
