中文摘要：

准确预测违约风险是减少P2P网络借贷平台＂跑路＂现象的重要对策。该文使用WDW上海直营店的运营数据考察了借款人信息与其违约行为之间的关系,找到了借款人违约的关键因素并确认了其可信度。Logistic回归模型和Cox回归模型的结果表明：外地户籍、已婚和历史违约记录与借款违约率和违约速率均正相关,揭示借款人特征信息具有一定个性趋势;债务收入比仅与借款违约率负相关而对违约速率不敏感,表明债务收入比无法准确预测违约动态;家人知晓借款和借款目的真实性与借款违约率和违约速率均负相关,说明软信息更能体现借款动机及偿还意愿;平台的信用评级指标与借款违约率和违约速率均存在非常显著的负相关关系,表明平台对借款人信用评价较为准确。该文的研究结论为构建借款人违约风险的量化指标提供了支撑。

英文摘要：

Accurately predicting the default risk of borrowers from online Peer-to-Peer （P2P） has become one of the most effective ways in reducing P2P bankruptcies. Using the internal data from Shanghai branch of WDW, this paper discovers the key and reliable factors that may influence loan performances. By means of Logistic model and Cox Proportional Hazard test, the paper reveals that personal characteristics of borrowers such as household register, marriage, historical default record affect both the default rate and hazard rate; debt-income ratio only influences the default rate, but it doesn＇t decrease with the hazard rate of a borrower＇s default; family awareness of the loan and the truth of borrowing purpose are negatively related with default rate and hazard rate, meaning soft information is representative of borrowers＇ willingness to pay; credit grade is significantly negative-associated with both the default rate and hazard rate, proving that it is accurate for P2P to predict default rate of borrowers. These results provide support for establishing quantified indicators of default risk.