Construction Of Risk-warning Model For Commercial Bank Based On The Logistic Regression

Bank risk identification and disposal is one of the important issues in bank supervision.How to establish an effective risk-warning model of bank, thus to help bank regulators to identify and warn risk,allocate the scarce supervision resources more efficiently, and take timely measures to prevent and resolve risks, prevent bank failures or minimize the losses caused by bank failures,this is an important research topic in the field of banking supervision,and have great theoretical and practical significance in improving banking supervision.However,up to now,researchers in this field are based on the empirical studies on American commercial banks.Because different countries have different economic and financial development level, different manner of banking development, different bank supervision mode,so there are some differences in the construction of bank risk-warning model between China and foreign countries.The most obvious point is that,U.S.A have a long history in banking development and have accumulated many supervision data,so they are able to analyze bank samples with different risk degree.But in China, due to lack of banking samples, we can not do that like America.Building a feasible risk-warning model for Chinese banks is the purpose of this article.This article learns relevant domestic and foreign studies.Firstly,it analyzes types of bank risk and risk-warning methods, and their functions,objectives.Then it constructs a indicator system for bank risk-warning, applies factor analysis and AHP method to evaluate the risk condition of sample banks. This article builds up a high-risk bank sample and a low-risk bank sample.And then it applies K-S test and Q-Q Probability Plot method to test whether financial ratios meet normal distribution.And it applies M-W-W tests and student's t test method to analyze the financial characteristics of these two groups, the test results show that there are differences in the means of each of the various financial ratios between these two samples. As a consequence,it is feasible to apply appropriate statistical methods in building a risk-warning model.By logistic regression it finally develops a risk-warning model with three explain variables,and it tests model's predictive ability finally.