The Research On Small And Medium-sized Enterprises’ Credit Evaluation Model Based On Logistic Regression

In recent years, Chinese SMEs(small and medium-sized enterprises) have developedrapidly, and become an important part of the national economy.But its important statuscan’t weight on the status of the fnancing difculties of the SMEs. The main reason isthat the commercial banks lack specifc credit risk measurement method for the SMEs.According to the“new basel”, the bank should evaluate the credit rating of the cus-tomer, and needs to quantify the loan customer’s credit risk. This paper through theestablishment of credit risk probability model predicts customer’s credit risk state, andstrengthens the credit risk management of the commercial banks further.This paper introduces the logistic regression model, and on this basis, establishesthe logistic regression credit risk model of SME.120SMEs in Gansu Province are se-lected as the research sample in the empirical research.10fnancial indexes which havelittle conspicuouseness are removed from20fnancial indexes by K-S test, independentsamples t-test, Mann-Whitey U-test.Principal component analysis is done by the rest ofthe indexes, and5principal components which cumulative variance contribution rate is76.876%are extracted. Then the two-category logistic regression credit risk early warn-ing model is constructed by3principal components that are selected according to thewald test.Next,the accumulatively logistic regression credit rating model for the SME isconstructed.Firstly,8principal components which cumulative variance contribution rate is72.424%are extracted by principal component analysis.Secondly,the model is constructedby4principal components which is selected through reversely gradual selecting method.In the model test, this paper adopts diferent testing methods for the two models, andresults show that the prediction ability of models and the ftting results are relativelyideal. They can be used for internal risk early warning,and provide the enterprise creditrisk state and credit decision basis for the banking system.