Research On Loan Credit Risk Evaluation Model For Small And Micro Enterprises In China

According to the characteristics of operating and credit risk of small and micro enterprises in China with a large historical credit data set from a number of commercial banks in a region of northern Jiangsu province,this thesis studies and constructs a probability of default evaluation model and a probability of loss given default evaluation model for the commercial banks in China to evaluate the credit risk of small and micro enterprises by the theory and practice research method.This thesis starts from the existing theory of loan default,summarizes the credit risk characteristics of small and micro enterprises and the main difficulties for commercial banks to evaluate the credit risk of small and micro enterprises based on the current development status of Chinese small and micro enterprises,then explores the possible sources of credit risk of Chinese small and micro enterprises,considering that most of the current credit evaluation models for small and micro enterprises only focus on the heterogeneity factors of different borrowers,but ignore the other possible influence sources that might affect the credit risk of small and micro enterprises during different periods,like economic situation,business environment,etc.,and put forward the concept of credit environmental factors.Then based on the theoretical analysis the proxy indicators of the credit environmental factors are preselected,further according to the large sample data set of historical credit loan records of several commercial banks in a region of northern Jiangsu province,the correlation between these proxy indicators and the credit risk of small and micro enterprises are confirmed by experiments and finally picks out the key indicators according to the result.All these above establish the theoretical and factual basis for the construction of the credit evaluation model for small and micro enterprises in China.According to the indicators that have significant correlation with the probability of default of small and micro enterprises and the sample characteristics of the historical credit loan data set,this thesis design a three-stage hybrid credit evaluation(TSHCE)model to evaluate the probability of default for small and micro enterprises in China.Logically,the TSHCE model could be divided into three stages:first stage,get several evaluating results about importance of each attribute according to multiple attribute subset selection strategies,then generate the attribute importance ordinal vector that defined by this thesis by ranking them,which measures the average ranking importance of each attribute in multiple attribute subset selection strategies;second stage,according to the attribute importance ordinal vector,use the roulette method to extract subsets with different attributes,and then train several basic predictors;third stage,combining the prediction results of each basic predictor trained in the second stage to construct the re-training sample set,and further train the connection predictor to learn the feature of re-training sample set to get the final result of the model.Compared with the existing mainstream credit evaluation models,the TSHCE model can dynamically select the superior combination of different attribute subset and basic model according to the inherent characteristics of the data set to build the basic predictor,and then dig out the predictable features of the specific data set,thus could effectively improving the flexibility of the model itself and the applicability on different data sets.Empirical study shows that the TSHCE model has a better prediction accuracy.According to the indicators that have significant correlation with the probability of loss given default of small and micro enterprises and the sample characteristics of the historical credit loan data set for the small and micro enterprises,this thesis design a multiple predictors hybrid credit evaluation(MPHCE)model for the evaluation of probability of loss given default for small and micro enterprises in China.The key idea of MPHCE model is that in large sample data sets,the contribution of the same attribute in different local value areas are different to prediction performance.So firstly propose the concept and definition of the positive related attribute set,negative related attribute set,positive boundary vector,negative boundary vector and so on to describe the correction between different attributes and the distribution characteristics of default samples in small micro enterprise data set,and prove their main property.Then according to the similarity between the samples and the positive boundary vector,the negative boundary vector,MPHCE model divides the whole dataset into several subsets,and then training the basic predictors respectively on each subsets.When need to predict a new sample,MPHCE model firstly divides the sample into the appropriate subset according to the similarity between the sample and the positive boundary vectors,negative boundary vectors,then uses the basic prediction model which built on the subset to predict the probability of loss given default of this sample.Compared with the existing mainstream credit evaluation models,the mechanism setting of MPHCE model that divides the samples into different subsets based on the similarity between the sample and the boundary vector and trains the base predictor respectively,.could effectively dig out the predictable pattern in different local areas on the training sample set,and selecting the appropriate basic predictor according to the similarity of the new sample and the boundary vector makes the prediction process more sensitive to the characteristics of the new sample.The empirical study shows that the MPHCE model has a better prediction accuracy.This research work has certain implications to the establish of the theory and method of credit risk evaluation model for Chinese small and micro enterprises,while also have some realistic meanings like could help to prevent the credit risk of Chinese bank industry evolves into systematic financial risk,improve the competitiveness of Chinese commercial banks,alleviating the problem of financing for small and micro enterprises in China.