Analysis Of Credit Risk About The Bond Portfolio Based On Revised Creditmetrics Model

Since the occurrence of "11 Chaori bond",the first bond default in China in 2014,the number of bond defaults has continued to increase in recent years,and the number of defaults and the amount of breaches have gradually increased.From the perspective of China's policy,the environment of policies which is positive is difficult to sustain,it has been replaced by a situation named "narrow balance".China's macroeconomic operation has also presented the "new situation" with a little lower speed of develop-went.Under such circumstances,many industries have experienced serious overca-pacity.Driven by the task of de-capacity and de-leveraging,some corporate's situation in financing is not optimistic and there are some corporates whose chain of capital have broken down.Thus all of these lead to an increase in credit risk.Although the credit risk becomes more obvious,the measurement of credit risk in China is only at the initial stage.Many investment management companies only analyze the bonds for investment purposes,and lack quantitative measures.This arti-cle uses CreditMetrics model to measure the credit risk of bond portfolios based on the perspective of investors.Since the CreditMetrics model was originally developed by J.P.Morgan,its input parameters are not very suitable for China's actual situation.Therefore,we modify and optimize the model's parameters and calculation methods.The main work of this article is as follows.1.For the credit rating transformation matrix,we compare the transformation matrix from Standard&Poor's,Dagong and the transformation matrix based on im-plicit rating.The results show that due to the high ratings of China's bonds,there lack low-grade rating data.At the same time,the number of defaulted bonds is small,and there is a zero margin on the rating transformation matrix of domestic data.Therefore,the rating transformation matrix provided by Standard&Poor's is used in the Credit-Metrics model.For the credit rating of the underlying bonds,this article uses the im-plicit rating instead of the traditional debt rating.2.For the discount rate,this paper adopts the stripped securities method to grad-ually calculate the forward interest rate together with credit spread based on maturity yield of bonds of various grades,and finally get the term structure of long-term risk discount.3.For the correlation between bonds,this paper abandons the linear correlation,but uses the properties of the copula function to calculate the correlation between the corresponding stocks.In this process,this paper uses the method of kernel density es-timation to estimate the marginal distribution of the corresponding stocks returns data.4.For the bond portfolio,this paper adopts the Monte Carlo simulation method,and uses the Cholesky decomposition method to decompose the correlation coeffi-cient matrix,and then generates the datas which obey the given correlation and nor-mal distribution.Through relating the threshold value to the given credit rating change,the relevant credit rating based on the simulated data is obtained.5.For the measurement of credit risk,this paper adopts the CVaR method.And based on the characteristics of the credit risk which has peak and thick tail,finally we proves that the CVaR calculation result is better than the VaR calculation result,and the CVaR method is more suitable for the credit risk that has a thicker tail.Although this paper improves CreditMetrics model from parameter modification and calculation method optimization,this does not obscure the deficiency of China's rating data.Therefore,based on the research process and the results,this paper pro-poses to improve the rating system and guide the rating agencies to actively give the small businesses a rating.And we should establish the probability databases of rating transformation and default rate databases.