系统性风险视角下的中国企业动态财务困境预警：基于AdaBoost CSSVM-TW模型

Dynamic Financial Distress Early Warning for China’s Enterprises from a Systemic Risk Perspective: Based on the AdaBoost CSSVM-TW Model

摘要: 为了解决由于数据不平衡、信息不充分和概念漂移等原因导致的中国外贸企业财务困境预测准确率低的问题。本文利用2010年至2022年中国A股上市外贸企业系统性风险指标和财务指标的年度数据，构建了时间加权结合AdaBoost成本敏感支持向量机模型，用于动态预测企业财务困境。实证研究结果表明，系统性风险指标具有独立于传统财务指标的预测潜力，因此可作为中国外贸企业的非财务动态预警指标，从而提高各种比较模型的预测准确性。结合系统性风险指标的时间加权与AdaBoost成本敏感支持向量机模型，有效解决了数据不平衡、信息稀缺和概念漂移带来的挑战，显著提高了在2015年中国股灾、中美贸易摩擦和COVID-19疫情下动态预测中国外贸企业财务困境的准确性，可作为中国外贸企业高效的风险动态预警工具。

Abstract: To tackle the problem of low accuracy in predicting financial distress in China’s foreign trade enterprises, attributable to data imbalance, insufficient information and concept drift, this paper utilizes annual data on systemic risk indicators and financial metrics of China’s foreign trade enterprises listed on the China’s A-share market between 2010 and 2022 to construct the time weighting combined with AdaBoost cost sensitive support vector machine model for dynamic corporate financial distress prediction. Empirical findings indicate that systemic risk indicators possess predictive potential independent of traditional financial metrics. This makes them valuable as non-financial, dynamic early warning indicator for China’s foreign trade enterprises, thereby enhancing the predictive accuracy of various comparative models. The time weighting combined with AdaBoost cost sensitive support vector machine model incorporating systemic risk indicators effectively addresses challenges arising from data imbalance, information scarcity and concept drift, significantly improving the accuracy of dynamic predicting financial distress in China’s foreign trade enterprises under the 2015 Chinese stock market crash, the Sino-US trade friction and the COVID-19 epidemic, which can be used as an efficient risk dynamic early warning tool for China’s foreign trade enterprises.