Research On Early Warning Of Credit Debt Default Risk Of Listed Companies

Since the first credit bond default occurred in the bond market in 2014,bond defaults have occurred frequently.Since 2018,the scale of default has shown an explosive growth trend,reaching a record 120.6 billion yuan,causing investors to panic.Against the backdrop of the "three major battles",the risk of China's credit bond defaults has received continuous attention from investors and regulators.Based on the introduction of China's credit debt default situation in recent years,this article has carried out theoretical and practical analysis from two aspects of direct and indirect factors that trigger corporate credit debt default.For how to accurately and effectively predict corporate credit debt defaults,a large number of documents use the KMV model.This study found that the default distance and default probability measured by the KMV model have certain limitations in the Chinese market.First,due to the immature domestic secondary market and the huge fluctuations in stock prices,the equity value and volatility may be There are distortions at certain times;secondly,the capital structure of different industries will affect the company 's debt value,which may affect the calculation of default distance and default probability,resulting in large differences in default probability of different industries.This article introduces the KMV-SMOTE random forest model,selects 32 default listed companies and 160 non-default companies as samples,and selects the listing from macroeconomic factors and micro-enterprise financial factors,market transaction factors and structural factors.The company's data of 23 variables in the year before the first default occurred,and the SMOTE algorithm was used to correct the data to improve the problem of unbalanced data,reduce the possibility of overfitting,and improve the accuracy of model prediction.Empirical results show that net profit / total operating income,equity pledge ratio,net sales rate,EBITDA / interest-bearing debt,default distance and net profit(year-on-year growth rate),return on net assets(ROE)and other variables are affecting the default of listed company bonds Important factor.At the same time,the prediction accuracy of the modified random forest model can be increased from 92.7% to 96.9%,an increase of 4.2%.It shows that the KMV-SMOTE random forest model can effectively predict the default of credit bonds and has important practical significance.Finally,based on the results of the model in this article,we provide relevant recommendations for investors,listed companies and regulators.