新经济业态P2P网络借贷的风险甄别研究

发布时间：2018-08-04 16:00

【摘要】：国内P2P网贷平台存在的问题及风险令人触目惊心,本研究旨在探索P2P网络借贷这一重要新经济业态的风险甄别问题。本文在我国444家P2P平台基本信息与交易信息基础上,通过网络爬虫获取了30万余条网民评论和新闻数据,利用文本挖掘技术提取了重要的外部信息,综合使用11种统计模型评估变量的重要性并讨论其经济意义。研究表明:1平台风险与基本信息、交易信息和外部信息间存在复杂的非线性关系,SVM、树类模型等非线性模型对预判P2P平台风险更加有效;2网民评论、关注度等基于互联网的非结构化信息,以及平均利率波动、资金净流入波动、未来待还金额波动等信息对甄别风险具有重要的作用;3根据拍拍贷的案例,小额、超短期、男性借款人具有更高的违约风险;4基于线性模式的传统风险指数构建方法并不适用于甄别P2P平台风险。有关部门应从完善法律法规、加大信息披露力度、发布风险预警综合指数等对P2P行业进行监管。
[Abstract]:The problems and risks of domestic P2P network lending platform are startling. This study aims to explore the risk discrimination of P2P network lending, an important new economic business. Based on the basic information and transaction information of 444 P2P platforms in China, this paper obtains more than 300000 comments and news data from Internet users through web crawlers, and extracts important external information by using text mining technology. Using 11 statistical models to evaluate the importance of variables and discuss their economic significance. The research shows that there are complex nonlinear relationships between the risk and basic information, transaction information and external information. The nonlinear models such as SVM, tree model and so on are more effective in predicting the risk of P2P platform. Attention and other unstructured information based on the Internet, as well as average interest rate fluctuations, net capital inflow fluctuations, future outstanding amount fluctuations and other information play an important role in identifying risks. 3 according to the case of PPDAI, small, ultra-short-term, Male borrowers have higher default risk. The traditional risk index construction method based on linear model is not suitable for peer-to-peer platform risk identification. Relevant departments should supervise P2P industry from perfecting laws and regulations, increasing information disclosure, issuing comprehensive index of risk warning and so on.
【作者单位】：中国人民大学统计学院;国家统计局国际统计信息中心;中国国际经济交流中心;
【基金】：中国人民大学科学研究基金(中央高校基本科研业务费专项资金资助)项目“中国跨越中等收入陷阱研究”(16XNH103)资助
【分类号】：F724.6;F832.4

【参考文献】