Research On The Credit Risk Measurement Of Non-Listed Companies In China Based On KMV-SVM Model

With the rapid development of the economy,the development of Chinese commercial banks and corporate credit bond market has gradually diversified,especially in recent years,credit default events have occurred frequently,and the issue of credit risk has attracted more and more attention.In addition,compared to listed companies,non-listed companies account for a large proportion of the defaulting parties.Therefore,it is of great significance to find reasonable methods to measure and study the credit risk of non-listed companies in China,so as to achieve the effect of preventing default risk.Currently,the credit risk measurement model for listed companies has achieved initial results,but there is currently no recognized model for measuring the credit risk of non-listed companies.In the modern credit risk evaluation system,the KMV model has a solid theoretical foundation,effectively combining historical book information and market transaction information,and has become a significant achievement in academia.However,due to the lack of two key indicators required by the KMV model for non-listed companies,this article chooses to analyze the credit risk of non-listed companies in China by introducing a support vector machine(SVM)regression model based on the KMV model.Due to the lack of stock market information for non-listed companies,there is no way to obtain their equity value and volatility of equity value.Therefore,this article first uses the stock market data and financial data of listed companies and adopts the KMV model to obtain the asset value and volatility of listed companies by Matlab software.Then,using traditional regression methods to construct regression equations based on the idea of the KMV model,and combining the index data of non-listed companies to obtain the default distance and default probability of non-listed companies in China.Then,this article uses SVM regression model to train and predict the data of listed companies,calculates the default distance and expected default probability of non-listed companies by KMV-SVM combination model,and compares it with the default probability calculated under the KMV single model.After that,by drawing ROC curves and calculating the AUC values under the two models,the accuracy of model judgment was compared and verified.The result shows that the KMV model has a certain ability to identify the credit risk of enterprises.Based on this model,the evaluation results of measuring the credit risk level of enterprises by constructing KMV-SVM combination model are closer to the actual default situation of enterprises and have higher reliability,which can better measure the credit risk of non-listed companies in China.At the same time,the application of SVM regression model effectively overcomes the problem of poor accuracy of traditional econometric regression for nonlinear data.Finally,based on the current situation of non-listed companies’ default risk in China,combined with theoretical analysis and empirical research,this article provides relevant suggestions and prospects for the credit risk assessment system of non-listed companies in China.