The Comparative Research On Management Model Of Credit Risk

This paper starts with the essence and characteristics of risk and credit risk and the evolvement of the credit risk management methods, which includes the periods of expert assessment, financial ratio analysis and structural models. Then, in brief, the paper introduces the typical traditional credit risk management methods and the modern ones. The traditional methods are technically dropped behind, but they still fit the modern credit risk management in many aspects, while on the other hand, each modern method has its own advantage and area of application. The writer creates a simple model with 13 standards to judge each modern method's advantage and disadvantage. We reviewed the practical terms in Chinese market with the information approaches to estimate each model's efficiency. Pay attention to the high requirement of the data collection, the problems in Chinese stock market and the interest rate marketization, and the critical tentative preconditions, we come to this conclusion that CreditRisk+ and Credit Portfolio View cannot be used in China at present. KMV requires historical default data and mapping between DD (default distance) and EDF (expected default frequency). Although we do not have much historical DD and EDF data in hand, they can be quite good theory reference in credit risk estimate. As a result, there's only KMV can be practically used in China.We used KMV and Logistic models separately to do diagnostic analysis on Chinese listed companies. In the static KMV study, 20 companies, 10 with outstanding achievement and 10 the opposite, are chosen to be our sample. Then we tested KMV with the each company's EDF on March 13 2007 to see whether it can really forecast the default frequency of Chinese listed companies. The result shows that, in a certain degree, KMV can reflect credit condition of those samples well. We supposed that the stock prices obey log-normal distribution, so there may be some windage in practical use. KMV model can catch up with credit condition changes in listed companies not only through data from stock market but also financial report forms from the companies within this market. KMV has a quite sensitive presentiment so that it can feel tiny influence bring by financial risk and management risk in those non-ST listed companies. In dynamic study, 4 outstanding achievement companies and 4 unwell performance companies are chosen out from the sample we had during static study to build a new sample. We collected information about daily closing prices of those companies from 2001 to 2006. After we calculated DD and EDF of each company on December 31 from 2002 to 2006, we gave our prediction of this year's DD and EDF on December 31. The result is: the 4 non-performed companies improve their credit condition by some certain degree each. It may be real improvement in running status and credit condition, which may be caused by fearing to be kicked out of the market, or may be forecasting error, as a result of substituting data of December 31 2006 with that of September 30. And on the other hand, the 4 well performed all trend in decline in DD and raise in EDF after share structure reform. This shows that after the share structure reform, fluctuation in company's asset value will lead to a decline in credit quality as a result of the entire circulation, but ignoring the entire circulation, just the opposite, will lead to a overrate in the company's credit quality. Besides, another important reason for EDF decline is: under the Bull Market effect, a company's equity capital and assets value increase and the distance to the default point goes longer. The influence of the Bull Market effect on both of the two groups can be seen obviously. Under the diagnostic analysis with the Logistic model, 45 ST companies coupled with 45 well-running companies constitute a parameter estimation sample space. We compared the Logistic model with KMV model after the model parameter was estimated, and found that the accuracy of the Logistic model is 88.89% when it was used to estimate the ST sample, and 90% when to do with the non-ST sample. The average accuracy counts up to 89.45%, which turns out to be a very good result. Meanwhile, two abnormal data showed up in KMV analyzing to ST sample. Ignoring these two abnormity, the accuracy reaches to 77.78%, considering the rate 99.99% in non-ST analysis without abnormity, the average accuracy of KMV is 88.89%. Both of the models do well in analyzing credit risk of Chinese listed companies. Logistic model does better when analyze credit risk in ST companies, while KMV does better in non-ST. Comparing the two models, we can see that KMV which runs on stock price forecasts future better, while Logistic model concerning nothing with the price estimates the past better.