Research On The Assessment Of Borrowers’ Credit Risk In Online P2P Lending

Online P2P lending is a kind of credit lending pattern which is based on Internet technology,achieving peer to peer credit lending directly and solving the problem of financing of Mid-small size clients and companies effectively.With the vigorous development of the online P2P lending market, however, some risks come into emerge,of which credit risk is the most serious one.With the climb of borrowers’ default rates in online P2P lending in recent years,the credit risk has done serious damage to the interests of investors.Therefore,it is a hot topic to assess the credit risk of the borrowers in the online P2P lending market. In view of the development of online P2P Lending market in our country is still not mature,so this thesis select Lending Club company as the example to assess the borrower’s credit risk.Firstly,some common risks in online P2P lending market would be introduced,and the credit risk will be analyzed emphatically.Secondly,several well-developed methods to assess the credit risk would be introduced and analyzed, and we find that the logistic regression model is more suitable to assess the credit risk of borrowers in Online P2P lending.Finally,based on the trading data of Lending Club Company,a logistic regression model is constructed to assess the borrowers’ credit risk.In the process of model builing,we first select 18 indicator variables that maybe explain the credit risk of borrowers and a k-means clustering analysis is carried out on the samples,then those representative default and normal individuals are selected to form the training samples of the Logistic regression analysis model.Secondly,all 18 indicator variables get discretized and classified,and then information value and weight of evidence value are introduced,and all indicator variables get a preliminary screening by the IV.Thirdly,the original values of the remaining indicators after the preliminary screening are replaced by the interval WOE values of the indicators,and the Logistic regression model would be built that the selection is stepwise regression to quantitate the credit risk of borrowers.What’s more, the ROC curve and discriminate matrix would be applied to test and verify the validity and the ability to predict of the model.Results show that installment,interest rate,purpose,annual income,home ownership,loan amount to annual income,credit history length,inquiries last 6 months and revolving utilization are correlated with the the borrowers’ credit risk significantly,and the Logistic regression equation that consists of these 9 indicator variables can quantify the default rates of borrowers and offer support for investors to make rational bidding decisions. In view of the fact that most of online P2P lending companies in China have not established the credit risk assessment system,the study in this thesis may provide guidance for those P2P companies in our country to establish a assessment system of credit risk of borrowers.