Research On Credit Risk Evaluation And Risk Control Model Improvement Strategy For Internet Small Loan

With the rapid development of internet finance,how to reduce the default risk through timely judgement of high-risk clients is always the critical problem for the internet loan following their birth.Quantitative analysis method such as Regress Analysis Model,can deduce conclusion and make prediction for credit-default probability from vast information of borrowers.It can not only decrease the risk-control cost,but also improve the loan efficiency and that inspires confidence rather than fear.However,the new internet financial product called Cash-Loan,which is evolved from the payday-loan financial product initiated in the west,featuring with much smaller loan amount,much shorter loan-cycle time.A new challenge to its credit fault analysis,prediction and risk control came out because the cash-loan clients in china have much less credit-related information than in the west.The related researches,nevertheless,focused on much more macro-level institution,mechanism and policies than micro-level quantitative analysis.Some western studies on payday-loan are rather different with cash-loan in china in many aspects like internet market,economical environment and clients' portrait.Thus risk control of cash-loan is unique in china and less studied than normal internet-loan products.The current risk control approach and modeling method are still uninspected under the specific financial scenario.The effectiveness and applicability of Regress Analysis need to be inspected with actual data.New studies are still necessary to find out how to focus on less variables to predict the probability of high-risk clients.This article tried to make further research about the international and domestic development of internet finance and related risk control theories by using the method of literature research and comparative study.the recent market evolution and risk control practice about internet small loan and P2P(peer to peer)loan were described as well.Then we integrated and compared various models and analytical tools in credit-default domain,and companies and products in mega-data risk control domain.We took cash-bus as a benchmarking platform for cash-loan further study and investigated the internal operations including its clients,products,processes and risk-control-system of a newly established platform called KH to give a comprehensive understanding about an actual organization in Chinese cash-loan industry.The following part of this article operated the actual data of KH platform from Q1 to Q 3 in 2017 to make the related analysis.Firstly,we classified the original explaining-variables,made random sampling and divided the data into 2 groups for later comparison.Secondly,we made qualitative client-portrait analysis from different dimensions.Thirdly,we made quantitative analysis to get the credit-fault probability model for cash-loan domain by using methods like factor analysis,factor rotation,logit transformation,and etc.The results indicate the good reliability of forecast and goodness of fit.The prediction accuracy achieved at 84.8%for outside-sample data examination.The conclusions of these analysis indicated:(1)Comparing with normal loan,cash-loan domain variables related with repayment willingness and stability are statistically and significantly co-related with and credit-fault probability.(2)Variables related with loan repayment ability are related with the credit-fault probability at a lower middle level.(3)Variables of lending rate and lending purpose are weakly related because the borrowers mostly cared rather the liquidity than the lending rate.(4)The continuous improvement of risk-control model will be closely linked with multiply coordinated operations such as data,models,systems,people and mechanism.The healthy and normalized development of cash-loan industry will depend on not only the marketing supervision and industrial self-regulation,but also the sharing of data and risk-control capabilities,especially for the sharing of multi-point-loan which will dramatically power the industrial companies and decrease their bad-debt rate.On the other hand,the mode of "supernormal profit cover higher bad-debt"cannot last for long time.Companies have to quickly make transformation and optimize the design of financial products with more accurate client-portrait studies,more professional risk controls.Thus,the balance of regulation-compliance and maximum profit could be achieved.This research provided a reference and lessons learnt for construction and improvement of internal risk control.In addition,the factor-analysis attached regression methods and data sets were derived from actual cash-loan practice and helped us to find out a detail explanation about the key factors on credit-fault probability.A primary variable-set framework for cash-loan risk control and guidance to risk control strategy were built up as a micro case for further researches on internet finance.