Credit Risk Assessment of Selected Banks

Abstract

The thesis is devote to analysis the credit risk of China commercial banks, we will focus on describe probability of default of banks by credit scoring models. First, in the theoretical part, we will briefly introduce the financial analysis for banks and two main categories of credit scoring models which are linear discrimination analysis and regression models (linear and logit). Then, in the analysis parts, we through analysis the financial indicators about the largest five banks in China to realize the situation about credit risk in China. The most important part is assessment credit risk, we will work with the 36 sample with complete public information in hundred China commercial banks which will be separated into two groups (non-default and default) by problem indicators compare with mean values based on historical data. Subsequently, we will create model functions and estimate the probability of default through the banks sample, through comparing the result from ROC curve, we will find the model which will get the best result and then, testing the efficient predict period for three models by another group data during past four years since modeling. Last, we will give some suggestion about credit risk and list the unsatisfied points about the models.

The thesis is devote to analysis the credit risk of China commercial banks, we will focus on describe probability of default of banks by credit scoring models. First, in the theoretical part, we will briefly introduce the financial analysis for banks and two main categories of credit scoring models which are linear discrimination analysis and regression models (linear and logit). Then, in the analysis parts, we through analysis the financial indicators about the largest five banks in China to realize the situation about credit risk in China. The most important part is assessment credit risk, we will work with the 36 sample with complete public information in hundred China commercial banks which will be separated into two groups (non-default and default) by problem indicators compare with mean values based on historical data. Subsequently, we will create model functions and estimate the probability of default through the banks sample, through comparing the result from ROC curve, we will find the model which will get the best result and then, testing the efficient predict period for three models by another group data during past four years since modeling. Last, we will give some suggestion about credit risk and list the unsatisfied points about the models.