Research On Credit Risk Assessment Method Of Unlisted Small And Medium-sized Enterprises

With the growing number and size,small and medium-sized enterprises(SMEs)have become an important force in promoting China's economic.However,they are faced with the development dilemma of financing difficulties.Especially,for the majority of unlisted SMEs,limited credit from financial institutions has become their main constraints for sustainable development.High risk of SME loans for banks caused by information asymmetry between banks and enterprises should be an important factor behind it.At the same time,under the background of financial disintermediation,the interest rate liberalization and the introduction of new regulatory policy,when avoiding risks and responding to regulatory,banks are bound to set their sights on the SME credit market,which is regarded as a new profit growth point.So,the accurate assessment of SME credit risk has become the key issue that should be solved first.In recent years,theoretical and practical circles have conducted positive exploration and research about it.Through literature review we can find that,the existing research on SME credit risk focus on the theoretical study of risk assessment index and method and the empirical research based on public data of listed companies,lacking of research on the actual credit data of unlisted SMEs.Moreover,scholars prefer financial factors and contempt non-financial factors when conducting empirical research.Therefore,it is necessary to expand the research field.By using the enterprise information and loan data of 1297 unlisted SMEs obtained from a commercial bank in 2012-2013,this paper constructed an index pool which contains 12 dimensions,24financial factors and 21 non-financial factors,and then established two risk assessment models with logistic regression and BP neural network,which were tested later.The empirical results shows that some financial factors have a negative correlation with the probability of default,such as the net cash flow,liquidity ratio,operating margin,product sales rate,growth rate of total assets and so on.For the non-financial factors,the duration of the establishment of enterprise and the cooperation time between enterprise and bank have a negative correlation with the probability of default.Besides,enterprise maybe has a low probability of default when its annual report is audited.The empirical test shows that both models have good predictive power and can be applied to the credit risk assessment for unlisted SMEs.Compared with the logistic regression model,BP neural network model has a stronger predictive power.At last,in order to deal with the limitation of single risk assessment method in actual credit business of bank,this paper constructed a multi-angle assessment method for SME credit risk based on the theory of information asymmetry from the perspective of identify characteristics of the "hard information","information" and the third party credit information of enterprise.This multi-angle assessment method can aid banks in choosing suitable assessment methods to evaluate SME credit risk according to different situation of the relationship between banks and enterprises,the characteristics of the credit information of enterprises and the cost for obtaining information.