Research On The Influence Of Narratives On The Overdue Rate Of Borrower

A new mode of borrowing develop rapidly in our country since PPDAI, the First P2P lending platform was introduced in 2007. At the same time of rapid development, all kinds of problems arise, such as the net credit platform running, the low overall borrowing rate, and the widespread default. In these problems, the phenomenon of borrower overdue is the biggest risk faced with investors. It's inevitable that the borrower's credit risk is which caused by asymmetric information exist in the credit process.The only way to ease the information asymmetry problem is information. The disadvantage-side investors can easily identify borrower's credit risk as long as they have enough complete, real information. But in the net credit markets, both debit and credit side is a two-way anonymous. The information obtained by investors mainly from the borrower's credit rating and parts of identity information provided on the network platform, and they can't get the borrower all the information.Investors will take more attention to the unproven subjective information, and take it as the basis of investment decisions when the borrower will be unable to provide complete objective information in the process of borrowing. Existing studies have shown that narratives of economic agents has certain influence on the result of the economic behavior. In the P2P lending, narratives refers to the borrower describe own condition (including personal quality, personality, borrowing, etc.) with the narrative language. So the purpose of this paper is to study that narratives of borrowers can influence the rate of overdue or not,and how much it influence. Base on that, the study want to establish a prediction auxiliary model of the borrower's overdue rate, and help investors and net credit platform to better identify the credit risk of the borrowers, which can help the P2P platform play its role in public welfare for poverty alleviation and the financing for small and medium enterprises as much as possible.This article grab the 4490 sample data directly from the PPDAI website platform by the way of programming data. It extract eight characteristic variables from narratives, which is refer the quantitative methods of narratives at home and abroad.The eight characteristic variables respectively is integrity, stability, aggressive, family,moral,brush credit,experience,improving the quality of life.The two characteristics variables,family and moral,are delete in the model when they can't through the consistency check. Base on that, the study use the Logistic regression model to do the research. The study found that the objective information-inadequacy (the lower the credit rating) borrowers is more inclined to create favorable image by providing more features in narratives in order to obtain loans. It also found that more information provided in narratives, the higher the probability of default. Finally, the study found that the different characteristic variable narratives have different influence on the overdue rate of borrower. The three characteristics variables, honesty, credit, improving the quality of life are positively correlated with the overdue rate of borrower. It means the investors have a negative appraise to borrowers showed these three characteristics and think the overdue rate is higher. And other three characteristic variables, steady, enterprising, experience are negatively related with the overdue rate of borrower.The investors have a positive appraise to borrowers and believe the overdue rate is low. But only the "experience" characteristics have significant correlation with the overdue rate of borrower while others not. But no statistical significant relationship doesn't means no economic significant relationship.This is a exploratory study about the influence of narratives on the overdue rate of borrower, and further discussion about the role of narratives. According to the results of Logistic regression, it build the prediction auxiliary model of borrower's overdue rate. The study find that, no matter from the view of the model goodness of fit or model prediction accuracy, narratives can obviously help to predict the borrower default and help investors to identify high-risky borrowers. Finally, according to the conclusions, we give relevant recommendations to borrowers, investors and P2P online lending platform, and it's helpful to the credit risk management on network platform.