Impact of Corporate Social Responsibility on the Financial Performance of Tourism Enterprises in Provinces Hosting China's Mixed World Heritage Sites: A Data‐Driven Machine Learning Approach

This study examined the relationship between corporate social responsibility (CSR) and financial performance in the context of tourism firms operating in China's Mixed World Heritage regions. Using data from 2012 to 2019, four machine learning (ML) algorithms were evaluated for their ability to predict financial performance, with eXtreme Gradient Boosting (XGBoost) demonstrating the highest accuracy. The SHapley Additive exPlanations (SHAP) method was then applied to quantify the contribution of each CSR dimension. The findings revealed that shareholder responsibility had the strongest yet negative impact on financial performance, followed by social, employee, and supplier/customer/consumer responsibilities, all showing positive effects. Environmental responsibility exhibited mixed effects across financial indicators. Overall, CSR showed a negative influence on financial performance. By integrating interpretable ML techniques, this study offers methodological contributions to CSR research and provides practical insights for tourism firms and policymakers seeking to optimize CSR strategies in Heritage tourism contexts.