Research On The Credit Risk Assessment Of Peer-to-peer Lending Borrower On Logistic Regression Model

As the enforcer of Internet finance, Peer-to-peer lending began to rise in China since2007. Through a short period of promotion and development, this new pattern of small private lending began to be widely accepted by the public because of its advantages of high yield, low threshold and convenient operation. By the end of March31,2015, there were1,728P2P platforms in operation with a monthly increase rate of around2%.49.26billion Yuan transaction occurred in March the single month, showing an extreme prosperous situation. But at the same time, lack of thresholds, standards and regulators also sets the industry into chaos. By the end of March2015, accumulative550platforms existed problems. Regardless weak legal regulation, this article aims to study the credit risk assessment of P2P borrower in a healthy market. RenRenDai was chosen to be the empirical analysis example because of its integrity, justice, transparency, innovation and efficiency. It is one of the earliest P2P platforms and now has become the industry leader with good reputation. On its website, the borrowers’information is fully disclosed which greatly improves the efficiency and accuracy of data collection.The history of P2P websites in foreign developed countries was just ten years, but owning to comprehensive regulation and healthy market environment, they maintained the benign cycle of development. P2P companies are powerless than the traditional commercial banks in risk control because of small scale. The basic risks are credit risk, legal risk and regulatory risk, as well as the specific risks of asymmetric information risk, investment risk, self-discipline risk, settlement risk and information security risk. The biggest risk is credit risk, and most of the P2P platforms are using traditional credit risk assessment models without upgrades towards new characteristics of Internet lending.From the perspective of P2P platform credit risk, the main body of this paper is the borrower. The first step is referring to the commercial bank’s personal risk evaluation system to select indexes and collect data from Renrendai. The second step is through data processing, classification and quantitative methods to ensure the final modeling indexes with the aid of information gain technique. The third step is using Logistic regression methods to get a new credit risk assessment model. The last step is using another volume of testing data to test the accuracy and affiance of the new model and compare it with the existed P2P credit risk level system. The result shows good and convictive. In the end of this paper, the author makes a conclusion and proposes a few of improvements of constructing a comprehensive P2P platform credit risk assessment system.The main innovation of this paper is using the traditional Logistic regression model for the emerging P2P platform to evaluate the borrower’s credit risk, as well as with the aid of information gain technique and the concept of WOE. It shows a good result and also performs better than the existed system on credit risk assessment.