Systemic Risk and Crisis Warning of Stock Price Collapse in Enterprises—A New Perspective Based on Machine Learning

建设“金融强国”的过程中，系统性风险加剧，企业发展环境险峻，企业股价崩盘已成为动摇一国金融市场稳定的重要影响因素。本文结合Logit回归模型与LightGBM模型、XGboost模型等前沿的机器学习方法，探究系统性风险指标在预测我国企业股价崩盘危机中的作用。结果显示，系统性风险在预测各行业企业的股价崩盘危机时具有显著的效果，同时，总资产收益率较低的中游、下游及服务与支撑企业更易受到系统性风险的影响，进而面临股价崩盘的危机，总资产增长率较低的下游企业更易在市场波动中出现股价崩盘危机。此外，本文进一步发现基于主成分分析构建的系统性风险指标与LightGBM模型或XGboost模型相结合可以取得较好的预测效果。在此基础上，本文强调微观企业股价崩盘预警机制中宏观审慎视角的重要性，建议差异化监管各行业企业，引入机器学习方法，为维护金融系统稳定提供一定参考。In the process of building a “financial powerhouse”, the collapse of corporate stock prices has become an important influencing factor that shakes the stability of a country’s financial market. This article combines Logit regression model with cutting-edge machine learning methods such as LightGBM model and XGboost model to explore the role of systematic risk indicators in predicting the stock price collapse crisis of Chinese enterprises. The results show that systemic risk has a significant effect on predicting stock price collapse crises in various industries. At the same time, midstream, downstream, and service and support enterprises with lower total asset returns are more susceptible to the impact of systemic risk and thus face stock price collapse crises. Downstream enterprises with lower total asset growth rates are more likely to experience stock price collapse crises in market fluctuations. In addition, this article further found that the systematic risk indicators constructed based on principal component analysis combined with LightGBM model or XGboost model can achieve good predictive performance. On this basis, this article emphasizes the importance of macro prudential perspective in the micro enterprise stock price collapse warning mechanism, suggests differentiated regulation of enterprises in various industries, introduces machine learning methods, and provides certain reference for maintaining financial system stability.