Corporate governance and financial distress in China a multi-dimensional nonlinear study based on machine learning

Promoting governance efficiency is crucial for preventing corporate financial distress. However, previous research has been constrained by limited dimensions of governance predictors and insufficient linear estimations to predict financial distress. To address this issue, this study gathers 37 corporate governance indicators of Chinese publicly listed firms from 2009 to 2022 in the dimensions of ownership structure, board features and executive traits. The LightGBM machine learning approach is then used to compare the predicting power of these individual indicators, as well as the predicting power of the dimensions. The SHAP (SHapley Additive exPlanations) method is further adopted to conduct an in-depth interpretability analysis upon the established prediction. Our approach identifies the nonlinear effects of important corporate governance indicators on financial distress and prioritizes those indicators based on their impact levels. Our results show that the most influential indicator is institutional ownership, followed by managerial ownership and executive compensation disparity. In addition, the dimension of ownership structure has the highest predicting power among the three. Overall, our study provides new insights into how firms can optimize their corporate governance mechanisms to prevent financial distress.