The Risk Measurement And Audit Countermeasures Research Of Credit Debt Default Of Listed Companies Based On KMV-BP Neural Network Combination Model

In recent years,the domestic bond market has developed rapidly,with a scale of nearly 100 billion yuan.It has become the second largest bond market in the world,and its market position is gradually improving.However,with the booming of the bond market,the credit risk of the bond market has gradually been exposed.From 2018 to 2019,more than 300 bonds have fall to a real debt defaults.However,the credit ratings of credit bond entities remain high.In the past three years,more than 90% of the defaulted bonds were high-rated bonds which the credit ratings were all above the "AA" level.The rating inflation situation of bond market is serious.Distorted credit ratings not only mislead investment behavior of the bond investors,but also lead to financial regulators' erroneous estimates of the level of credit risk in the credit bond market.Facing the severe situation of frequent defaults on credit bonds,audit department should play an active role in the regulatory system.Under the macro underground of preventing systemic financial risks,the objectivity and accuracy of risk assessment and risk measurement would influence the quality of audit work.Considering the advantages of BP neural network in solving the risk distribution of credit assets,this paper builds a KMV-BP combination model for the empirical study of the credit risk of listed bond issuers and gives an innovative Suggestions for audit countermeasures according to the empirical analysis results.The first part of this article introduces the research background,purpose,significance,method and other contents of this article,including the reviews the previous research results.The second part introduces the overview of the entire credit bond market in 2019 and the risk of credit bond defaults,and gives a discussion of the reasons for the occurrence of credit bond defaults.The third part introduces the basic theory of model application.The fourth part is the empirical research part.First,we construct the training set including the 2018 A-share and ST-share listed companies and the test set including the empirical research of the 2018 Shanghai and Shenzhen listed corporate bonds.Then,the training set data is calculated using factor analysis theory and KMV model theory as the input signal and output signal of the BP neural network training model,and the corresponding relationship between the factor signal and the KMV output signal is established.Finally,the factor signal of the financial data of the test set is used as input,and the trained neural network model is used to calculate the real asset value and asset value volatility of the listed bond issuer.Empirical research results show that the credit risk of listed bond issuers is between A-share listed companies and ST-share listed companies;the KMV-BP combination model measurement is consistent with the measurement results of traditional credit ratings,and the accuracy of measurement is compared Credit ratings are more accurate,and a brief analysis of the causes of credit rating inflation is due to the current rating payment model.And from the perspective of the audit supervision department,from the three perspectives of national audit,certified public accountant audit and internal audit,put forward corresponding audit countermeasures.