摘要
我国正处于经济结构转型的关键时期,防范化解重大风险是我国当前经济工作的重要任务,因此有效预测企业债务违约风险具有十分重要的现实意义。本文系统性地运用机器学习方法,基于2014-2019年发生实质性债券违约的上市公司,采用7种机器学习算法构建企业债务违约风险预测模型。实证结果显示,基于随机森林所构建企业债务违约风险预测模型分类效果最佳,其中营业净利率、净利润、现金比率、财务费用和资产负债率5个财务指标可作为企业债务违约风险预警指标。研究表明,基于机器学习的企业债务违约风险预警系统能够有效地进行债务违约风险预测,不仅可以深化对企业债务违约风险影响因素的微观特征的认识,而且有助于监管部门对上市公司财务状况的监督更加有针对性。
China is in a critical period of economic structure transformation, so it is an important task for China’s current economic work to prevent and resolve major risks. Therefore, it is of great practical significance to effectively predict the risk of corporate debt default. This paper systematically uses machine learning method, based on the listed companies with substantial bond default in 2014-2019. This paper uses seven kinds of machine learning algorithms such as support vector machine, random forest, Lasso, naive Bayes, bagged trees, ordinary least squares regression and logistic regression to build the enterprise debt default risk prediction model. The empirical results show that the classification effect of debt default risk prediction model based on random forest is the best, and the five financial indicators of operating net interest rate, net profit, cash ratio, financial expenses and asset liability ratio can be used as early warning indicators of debt default risk. The research shows that the early warning system of enterprise debt default risk based on machine learning can effectively predict the debt default risk, which can not only deepen the understanding of the micro characteristics of the influencing factors of enterprise debt default risk, but also help the regulatory authorities to strengthen the supervision of the financial situation of listed companies, which is more targeted.
作者
王玉龙
周榴
张涤霏
Wang Yulong;Zhou Liu;Zhang Difei
出处
《财政科学》
CSSCI
2022年第6期62-74,共13页
Fiscal Science
关键词
企业债券违约风险
机器学习
预测模型
财务指标
Corporate Bond Default Risk
Machine Learning
Prediction Model
Financial Index
作者简介
通讯作者:王玉龙。
