基于大数据的互联网借贷平台信用风险传染及其监控研究

The Internet is bringing mankind into an era of big data, this promotes the rapid development of Internet finance, which effectively solve financing problems of small and micro-enterprises and vulnerable individuals in society. However, due to the specialties of Internet Lending, irresponsible runaway always happen, which causes considerable lending risk, so Internet Lending Platform needs to strengthen their credit risk management. It is an increasingly urgent issue to combine big data technology with platform risk management to achieve credit security of Internet lending platform. This project combines the complex network with contagion models based on reduced-form model to study the effects of credit contagion and the construction of platform risk monitoring system. Firstly, this project collects mass data which is closely associated with internet lending platform's risk. The data is collected by taking different approaches in accordance with various data characteristics. Secondly, we study the credit contagion of Internet lending platform from the perspective of complex networks and correlation mechanisms. Meanwhile, we use the clustering algorithm to gain multiple platform clusters. Thirdly, we propose different credit contagion models with general decay functions for corresponding platform clusters. Additionally, we derive joint distribution of default times by combing the change of probability measure method with the regression analysis method. Finally, we build the platform credit risk monitoring system to ensure the sustainable development of Internet Lending.

互联网正将人类带入大数据时代，互联网金融因此得到了快速发展，有效解决了小微企业和弱势社会个体贷款难的问题。但在具体运行过程中，由于互联网借贷的特殊性，存在诸多“跑路”事件，引发相当的借贷风险，互联网借贷平台的信用风险管理亟需加强。如何有效结合大数据技术和平台信用风险管理要素，实现互联网借贷平台的信用安全日益紧迫。本项目将复杂网络结构与基于约化模型的传染模型相结合，研究互联网借贷平台的信用风险传染效应及风险监控体系的构建。首先，本项目针对不同数据特征采取不同的方式收集与平台风险紧密相关的数据集合。其次，从复杂网络和关联机制的角度研究平台信用风险传染机制，并利用聚类分析法研究平台特征。然后，针对不同平台，构建带有一般衰减函数的信用风险传染模型，并通过结合测度变换和回归分析法求出多主体的联合违约概率分布。最后，构建互联网借贷平台信用风险监控体系，确保互联网借贷行业的稳健可持续发展。

互联网大数据时代的到来推动着互联网金融快速发展，互联网借贷平台由此形成。互联网借贷平台由于其借款手续简单、贷款周期短、服务范围广的特性，为小微企业及社会弱势个体贷款难问题提供了一定便利，但伴随而来的金融风险也不容忽视，尤其是信用风险。本项目基于收集的平台数据对我国互联网借贷平台基本特征进行了分析总结，并从关联机制及复杂网络结构视角深入研究了平台信用风险传染机制，指出网络借贷平台的信用风险可能形成螺旋式传染进而引发区域性风险。在此基础上通过构建集成策略模型及隐马尔科夫链扩展模型研究互联网借贷平台的微观风险评级机制、风险预测及宏观环境挖掘等应用，进而结合行业外部管理机制及内部监控机制提供相应的监管建议。