Research On Rural Loan Default Model And Its Prediction Accuracy

In rural areas of China,there are serious information asymmetry between rural financial institutions and farmers,it is difficult to predict the farmers' overdue behavior and default behavior.In such circumstances,in order to reduce the loss of famers' overdue behavior and default behavior,the rural financial institutions will raise lending rates and reducing the money supply which makes credit rationing phenomenon appeared in the the rural credit market.A large proportion of farmers' credit demand is difficult to satisfy.Therefore,it is necessary to study the overdue behavior and default behavior of famers,to alleviate the information asymmetry between financial institutions and farmers.At the same time,a transformation happens during China's rural and agricultural development,there have been some changes on household characteristics,the overdue behavior and default behavior between different types of farmers need to be studied separately.In this paper,based on the data of Z microfinance institutions,we study the overdue behavior and default behavior of farmers,trying to find out the influence factors of famers' overdue behavior and default behavior,and compare the prediction accuracy of Logistic model,the rare event Logistic model and Skew-Logistic model are compared.In this paper,the main content is as follows:The first chapter is an introduction to this paper,mainly introduced the background and significance of this paper.Carding domestic and foreign relevant research,put forward in this paper,the research contents and research methods,and the possible innovation and deficiency existing in this paper.The second chapter discusses the theoretical basis,analyzed the influence factors of peasant household credit risk in our country,and classified peasant household according to the current situation of the farmers in our country and the data of Z microfinance institutions.The third chapter introduced the source of data and the indicator system.Introduced three models:Logistic model,Rare Event Logistic model and Skew-Logistic model.At the same time,this chapter introduced the model prediction ability evaluation index:ROC curve and AUC values.The fourth chapter does an empirical research on the farmers' overdue behavior and default behavior.First use the Logistic model to analyze the overdue behavior for each sample.Second,using Logistic model,Rare Event Logistic model and Skew-Logistic model analyzes the default behavior for each sample,and compares the three kinds of model prediction ability.The empirical results found that the overdue behavior of agricultural management famers and non-agricultural farmers is easily influenced by famers'debt levels,the overdue behavior of hired famers is easily influenced by the guarantor and the co-borrower;in terms of the impact factors,Default behavior is more complex and the model constructed is not strong enough to explain the default behavior of farmers.The fifth chapter is the conclusion and prospect.On the basis of the above empirical analysis,this paper present a statement of the research results,and the prospect of credit evaluation is also discussed.