Credit Risk Model and Bayesian Improvement for Companies in China
- 1k Downloads
The paper focuses on credit risk measuring methods and tries to find suitable model for China’s listed companies’ credit risk measurement. According to the default condition in Chinese listed companies, we apply the factor analysis to the correlative data, and give the default discrimination model by Logistic regression. As the precision of the credit risk default model affect the bank’s risk status and profit directly, the paper uses the Bayesian estimate to improve the predictive power of credit risk default models. Comparing the precision of two models by AUC value and Brier Score, the result shows that the value of AUC of Standard estimator is 0.834 while the same value of Bayesian estimator is 0.870. It shows that the Bayesian estimate has a higher predictive power of precision and stability.
KeywordsCredit Risk Empirical Bayesian Estimate Factor Analysis Logistic Regression
- Adkins L, Hill RC (1996) Using prior information in the probit model: empirical risks of Bayes, empirical Bayes, and Stein estimators. In: Berry DA, Chaloner KM, Geweke JK (eds) Bayesian analysis in statistics and econometrics. John Wiley and Sons, New York, pp 79–90Google Scholar
- Altman E, Haldeman R, Narayanan P (1977) ZETA analysis: a new model to identify bankruptcy risk of corporations. J Banking Finance (l):29–54Google Scholar
- Stein RM, Jordao F (2003) What is a more powerful model worth? Moody s KMV Inc, New YorkGoogle Scholar