Cox Proportional Hazards Model Based Research On Credit Risk Measurement Of The Commercial Bank

For a long time,the credit risk has been one of the important factors that restrict the normal operation of commercial banks.The 2010 release of the Basel agreement again related to commercial banks are facing credit risk problems and put forward new requirements.In order to accurately measure the credit risk faced by commercial banks,the bank in China must establish the credit risk model.As a dynamic point prediction of econometric model,now Survival analysis is widely used in the field of related finance,compared with traditional commercial bank credit risk measurement model,survival analysis method has certain superiority,so this paper will try to use based on the Cox proportional hazards model survival analysis to measure the commercial bank faces the credit risk,and the dynamic display of the model point prediction ability.Firstly,it is important to introduce basic concepts of survival analysis and the applicability and superiority in the field of credit risk measurement,then describe the introduction of Cox proportional hazards model.Make a simple analysis of the estimation method and the test method of the basic principle.Secondly,from four aspects of the solvency,operating capacity,profitability and development capacity of the sample index to select,and select the sample data for the missing values and outliers.Then the selected indicators are Kruskal-Wallis non parametric test,multiple linear test and proportional hazard hypothesis test to screen the appropriate indicators for modeling.According to the screening index and estimating the baseline survival function value Cox proportional risk model final,and draw out the listed company survival function curve and risk function curve,through empirical testing,the results of the likelihood ratio test and Wald test.In addition,in order to eliminate the cut-off point as well as the different effects of different sample proportion selection model discrimination ability to predict,the samples were divided into modeling samples and testing samples,respectively for ST and non ST listed companies in accordance with the 1:1,1:2 and 1:3 of the proportion of the sample distribution,and choose whether the different cut-off point judge ST of listed company.The empirical results show that the model can be used to determine the proportion of the total sample as the dividing point,and the accuracy of prediction is higher than that of the normal samples.The final choice of the traditional credit risk measurement model Logistic regression model,modeling the use of existing data,and compared the risk and the proportion of Cox model's discrimination ability,through the comparison of the two models while the prediction accuracy is higher,but the Cox proportional hazard model compared with the Logistic regression model is better at predicting ability analysis is a dynamic prediction model,therefore in the commercial bank credit risk measurement has great practical value.