A Ressearch On The Credit Risk Measurement Of China's Listed Companies

Credit risk measurement is the base of credit risk management, which is the kernel of bank management. With the progress of science and technology and the rapid change of international economic environment, credit risk has become more and more complex, attracting more and more focus from countries around the world. When much research effort was paid and much fund was invested into the research of credit risk, many new models and methods have been developed and put into practice. It is, therefore, an important task for China's banking to take the advanced technology of credit risk management from other countries for reference and set up models and methods suitable for China. With the above background, this paper decided to choose the credit risk measurement of China's listed companies as its research subject.Firstly, the paper reviewed the literature about methods and articles relative to credit risk measurement. Then the author chose samples from listed companies, and adopted an empirical approach by using the KMV model and LOGISTIC model. It is found that the discriminate ratio of the KMV model, which is based on the option pricing theory, was low, comparing with the relatively high effectiveness of the multivariate regression model mostly built on financial index during the one or two years before default happened. It is found from the predicted models that during the one-year period before default happened, the default probability has high relevance with asset solvency and receivable turnover, but little relevance with profit earning capability. Whereas, during the two-year period before default happened, the default probability has high relevance with profit earning capability, cash flow and receivable turnover. High discriminate ratio was achieved in the models' actual testing. The discriminate ratio reached 94% and 85% respectively for the one-year model when tested using the previous sample and new added sample. The discriminate ratio also reached 94% for the tow-year model when tested using the previous sample.This paper took an approach combining theoretical and empirical analysis. A few creative points were made in such areas as the selection of defecting sample, the application of KMV model, and the building of the index system for LOGISTIC model, etc.