A Research On Credit Risk Identification For Companies Of China's Commercial Bank

Credit risk has always been one of the main risks for commercial banks.In recent years,the non-performing loans and non-performing loan ratios of China's banks have continued to rise.By the end of 2017,the non-performing loans of banks has reached 1.71 trillion yuan.Commercial bank credit risk is widespread and can easily increase systemic risks with the negative effects of risk diffusion and amplification.Prerequisite for banks to manage credit risk is timely and effective risk identification.This study uses the financial data of listed companies,Lasso regression to select variable before building a logistic model and scientific methods to select evaluation index system.It is expected to construct an effective credit risk early identification model and provide new ideas for bank credit risk early identification,which will have positive significance for bank credit risk management.We describe the research background and significance,and the status of non-performing assets in China's banks.We Introduce the credit risk theory and source of risk,the research status and research methods on credit risk of banks.Then we propose a framework for credit risk analysis of banks and analyze the impact of various factors on credit risk.At the stage of empirical analysis,we construct four different credit risk early identification models for banks by the first-phase data and the two-phase data.We can choose the best model for credit risk identification by comparing and analyzing the accuracy and robustness of the four models.The study finds that,in the early identification model and the second identification model,Lasso-Logistic model is always the best model with very good identification effect.This model just needs a small number of variables,there are also advantages of high prediction accuracy and good model robustness to predict the default probability of next period and next two period effectively.Commercial banks can refer to the research results of this study to build a more effective credit risk early identification mechanism and explore a more complete credit risk management system.