Study On The Borrower's Credit Evaluation Of P2P Lending

P2P lending market is an important complement to the traditional financial markets,however its development is restricted due to asymmetric information,therefore there is need to establish effective credit evaluation models and methods to evaluate P2P borrowers.This thesis proposes two main methods to solve the information asymmetry problem of P2P lending,one is by examining personal credit systems and the other involves analyzing credit rating models.The relevant literature and theory include national and international sources as well as in depth studies of market practices.The chapter 3 studies the development of domestic and international P2P lending markets and their elaborate and multi-faceted networks.This chapter details the differences between P2P lending networks and their features in relation to traditional financial loans.This chapter also examines the business model of foreign P2P lending networks in order to trace the trend of foreign P2P networks and to situate the domestic development accordingly.This chapter also examines risk identification and risk sources of P2P lending network systems,and the status of Chinese P2P lending market regulations and its challenges.Three game theory models of personal credit system are discussed in detail in this chapter.This chapter recommends that in order to have a rigorous credit information system,a sound,adaptable legal system is required,and various credit products and supports need to be developed.Problems of P2P networks regarding loans and credit information systems are imperfect for various reasons:the central bank's credit system is not linked to the P2P lending market and private businesses' credit system have startup problems such as large data processing constraints that force these enterprises to undertake costly and time consuming tasks.This chapter notes that big data models and in depth integration of online credit systems mark the future direction of the P2P lending platforms.The chapter 4 examines P2P lending market interest rates and repayment issues from the information asymmetry perspective with the use of an artificial neural network model.This model is self-adapting,self-organizing,can perform with real-time data and can manage nonlinear information.With the help of the database obtained from world leading P2P lending platforms like Prosper and the Lending Club,this chapter analyzes the significance of the borrower's credit rating,this former reflects the lending interest rate and other corresponding characteristics of other credit indexes.This research discovered that only a few indexes determined the borrower's interest rate,however borrower's information distortion borrower's or credit rating misalignment will inevitably result in pricing distortions in the P2P lending market.Therefore a comprehensive multi-index pricing mechanism is suggested in order to avoid such pricing distortions.The artificial neural network model analyzed several borrower credit index effectively,while the payment status response curve shows that certain indexes have application direction between them.Due to the absence of credit rating system and big data platform,Chinese P2P lending platforms are involved in every other form of transactions that change the nature of its service provider platform.This chapter concludes that P2P credit indexes provide theoretical support for P2P lending platforms whether they are intermediaries or customized platforms.From the perspective of big data inquiry,an appraisal model for P2P credit ratings was created.This chapter 5 introduces a multi-angle,multi-level model that places credit information in a spatial dimension to cross retrieve borrowers' data that would reflect the credit history continuously.Base on this concept,the Bayesian network model was used to test the effectiveness of the P2P borrower credit evaluation model and resolve the information uncertainty issues.To put the model into practice,this model was performed on the data obtained from Prosper and the Lending Club.Some results of the Bayesian network model include an unified expression for the complicated relation between P2P borrowers;the inference credit evaluation accuracy reached 87%within the samples based on the Bayesian network;the critical probability of credit evaluation value increased;external sampling of other credit evaluation models was quite effective and stable;increasing the data set could significantly improve the effectiveness of the credit evaluation model;the greater the dimension of borrower's information,the higher accuracy and efficiency could be achieved to evaluate the borrower's credit status.Another statistical test performed on this chapter checked the validity of static and non-static values by using a regression model.A logistic model,a decision tree model and a support vector machine(SVM)model comparison were compared and tested with the mentioned data from Prosper and the Lending Club.This analysis found that the Bayesian network model has good stability and predictability since it could explained the interdependent relationship between credit indexes and payment status in an acyclic graph.The overall performance of the Bayesian network model is the best of the three tested models.Choosing the Bayesian network model as the evaluation model for P2P borrows' credit condition gave the most efficient results.Chapter 6 choose domestic representative "ren ren dai" as the object of study,empirical evaluate the credit issues of the domestic P2P lending.On the basis of full analysis of sample data,construct the artificial neural network model to evaluate the relative importance for credit variables and research the borrower's influence on the payment status.The empirical results show that "the underlying information + credit files" indicator set contains the reimbursement status information,most of the second is the user information,the last is the state audit;Found that the difference for relative importance of every index is small,the most important is credit rating,serious overdue,successful loan,repayment period and interest rate,but the credit reports,certification of work and the borrowing amount are relative unimportance.Secondly,this chapter using Bayesian network model to predict "ren ren dai" borrower's loan status,whether it is inside or outside the sample prediction samples,the constructed model can accurately predict the status.Finally,according to the specific environment of domestic P2P lending,this chapter from the legal,regulatory and challenges discuss the particularity of P2P lending.Chapter 7 summarizes the main research conclusions in the previous chapters.Based on these conclusions,a series of policies were suggested for evaluating P2P credit conditions in China.Some policy recommendations,for instance,include strengthening the construction of a credit system,expanding the scope of credit lending to P2P networks,developing an effective credit evaluation model,improve the legal system for P2P lending,and strengthening the market supervision of P2P lending markets.