P2P网络借贷中羊群效应与信用评价研究

发布时间：2018-04-19 04:24

本文选题：P2P网络借贷 + 自动投标机制　；参考：《合肥工业大学》2017年硕士论文

【摘要】：近年来,互联网金融迅速发展,P2P网络借贷作为互联网金融的典型模式之一,其对于资金借贷双方具有重要意义,能够有效促进普惠金融的发展。然而,由于高度的信息不对称,可能导致其中存在非理性羊群效应、信用评价困难等问题。国内P2P平台较多,为了满足客户的需求,自动投标机制极大地提高了交易效率,节约了交易时间,但是自动投标机制对羊群效应的影响,尚无文献研究。此外,虽然P2P网络借贷能够缓解资金供求双方的矛盾,但是交易双方高度的信息不对称加大了贷款违约风险,如何有效地进行信用评价已经成为研究的热点。故本文围绕上述两点展开研究。第一,研究P2P网络借贷中自动投标机制对羊群效应的影响。首先,收集拍拍贷平台的交易信息,对P2P网络借贷市场行为模式进行分析。其次,依据信息不对称、羊群效应等相关理论,对P2P网络借贷中的羊群效应及其影响机制开展研究,并提出研究假设。最后,通过面板数据回归,实证检验了自动投标机制环境下的羊群效应问题。实证结果不仅表明P2P网络借贷中存在羊群效应,而且发现自动投标机制可以减弱羊群效应。更进一步,自动投标机制与借款金额、借款期限等变量类似,都表现为理性羊群效应。第二,研究P2P网络借贷中的信用评价。首先,阐述了一个能够在复杂数据环境下擅于处理分类问题的XGBoost模型,并给出重要变量判断方法。其次,在一个公开的大规模信用卡违约数据集上,检验XGBoost模型的性能并将其与Logistic回归、Lasso-logistic回归、决策树、随机森林等模型与方法进行比较。最后,将其引入到P2P网络借贷信用评价中,对拍拍贷网络借贷平台真实借贷数据进行实证分析。实证结果表明:在对P2P网络借贷大规模、非均衡数据进行信用评价时,XGBoost模型不仅能够得到比已有模型更高的AUC值,而且能够识别出关键变量。这些研究结果,为改善交易机制以促进P2P网络借贷健康发展提供了决策参考,对于P2P网络借贷违约的预测与控制等都具有重要意义。
[Abstract]:In recent years, the rapid development of Internet banking, P2P networks borrowing as one of the typical model of Internet banking, the borrowing of funds both sides has important significance, can effectively promote the development of Inclusive Finance. However, due to high degree of information asymmetry, which may result in irrational herd effect, credit evaluation problem of domestic P2P platform more. And in order to meet the needs of customers, the automatic bid mechanism greatly improves the transaction efficiency, saving transaction time, but the effect of automatic bidding mechanism for the herd effect, there is no literature research. In addition, although the P2P network lending can alleviate the contradiction between supply and demand of funds, but the information asymmetry in both transactions increased loan default risk, how to credit evaluation effectively has become a hot research topic. This paper focuses on the above two research. First, research by P2P network The loan in automatic bidding mechanism influence on herding. Firstly, the transaction information collected a pat on the loan platform, the P2P network lending market behavior model is analyzed. Secondly, on the basis of information asymmetry, herding and other related theories of herding P2P network lending and the impact of development mechanism, and put forward the research hypothesis finally. And through the panel data regression, empirical test the herding problem of environmental automatic bidding mechanism. The empirical results not only show that the herding effect exists in the P2P network lending, but also found that automatic bidding mechanism can reduce the herd effect. Further, automatic bidding mechanism and the loan amount, loan period and other variables are similar, performance for rational herding second, the credit evaluation of P2P network lending. Firstly, elaborated a complex data environment in dealing with classification problems and XGBoost model. Are important variables to determine method. Secondly, the default data set a large-scale public credit card, the performance test of the XGBoost model and the Logistic regression, Lasso-logistic regression, decision tree, random forest model and comparison method. Finally, it will be introduced to the P2P network lending credit evaluation, empirical analysis the real lending data pat loan lending platform network. The empirical results show that: in a large network of P2P lending, credit evaluation of non equilibrium data, XGBoost model can not only get higher than the AUC value of the model, and to identify the key variables. The results of these studies, in order to improve the trading mechanism to promote the healthy development of the P2P network borrowing provides reference, is of great significance to the P2P network lending default prediction and control.

【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F724.6;F832.4

【相似文献】