An Empirical Study On Default Risk Of Corporate Bonds In China

With the release of the policy documents such as"Pilot Measures for Cor-porate Bond Issuance",the number of corporate bonds issued has increased significantly in China since 2007.The scale of corporate bonds has also increased from 10 billion yuan in 2007 to 1102.474 billion in 2017 yuan.Corporate bonds have become a major method of low-cost financing for companies.The feature of low risk and stable returns has made investors and institutions choosing it posi-tively.With the rapid growth of the size of corporate bonds,the number of default events has increased year by year.In 2019,due to corporate bond defaults,a total of 37.721 billion yuan of defaults have been caused,causing unpredictable losses to investors.Therefore,measuring corporate bond default risk has important practical significance.The mainstream method for measuring default probability of bonds is the KMV model.Because the model's data is relatively easy to obtain and the calcu-lation steps are few,it has a wide range of applicability.However,the KMV model was derived from the analysis of US bond's data in the last century,and it is not,fully applicable to the Chinese corporate bond market.Non-listed companies cannot obtain the two necessary indicators—equity value and stock price volatil-ity,so the second chapter selects the listed corporate bonds that have defaults in China.Through their equity value,stock price volatility,liabilities,and the default point calculate the default distance,and then use the default distance to mea-sure the default risk.We divided china's corporate bonds into two groups named default samples and non-default bonds.After empirical research,it is found that the default probability of the two groups is significantly different,which shows that the KMV model can explain the magnitude of the default risk of China's corporate bonds.It is informative for bond investors.KMV model cannot estimate the default risk of non-listed companies.In the third chapter,we use the method named industry substitution to estimate the asset value and asset value volatility of the non-listed company.There are few data of China's listed corporate bond issuers,and the industry substitution method is difficult to implement.Therefore,it is proposed to use the EBITDA,book value,sales,and asset-liability ratio to fit asset valne and its fluctuation rate.Chapter four uses the classicneural network method-error backpropagation algorithm(BP algorithm)to optimize the regression estimation method.The BP algorithm is an algorithm that can reflect the non-linear mapping.It is good at obtaining the optimal output in the premise of unknown input variables and output variables,through the method of self-learning and self-adaptation and constantly adjusting parameters,so it is suitable for the situation of this article.Select indicators such as EBITDA,book value,sales,and asset-liability ratio as input variables,and select corporate bonds with substantial defaults and non-defaults in 2018-2019 as the training set.Then analysis the applicable method to estimate the value of non-listed corporate bond assets and their volatility in China.By analyzing the default distance of corporate bonds in 2018-2019,the results show that the default distance calculated by the model will make a significant difference between the default group and the non-default group,which has practical significance to prevent the default risk of corporate bonds in China.