A Comparative Study On Factors Influencing The Default Risk Of P2P Lending In China And The United States

P2P loan refers to a small amount of credit loan transaction between individuals that requires the help of a professional online platform to establish a lending relationship between the two parties and complete relevant transaction procedures.Since 2007,the first network platform in China found,P2P industry fast development quickly.The concept of Internet+and publicity of the borrowing of P2P network are quickly known by people.Every coin has two sides.Beginning in 2013,part of the P2P network lenders frequent borrowers malicious enforcement of negative news,then in 2015,the large-scale platform closed.P2P lending sparked an unprecedented crisis of confidence.The core of P2P online lending is risk control.It is of great practical significance to find out the factors that influence that default behavior through scientific methods which reasonably avoid risks and guide the benign development of the industry.P2P online lending started in the UK and grew stronger in the us,which has a relatively complete P2P online lending risk control system.The comparison between Chinese and American online lending platforms is of certain guiding significance for the development and construction of P2P platforms in China.This paper takes the overall development of P2P industry and the specific factors that affect P2P default risk.Selecting China and the United States as the research objects,and using the methods of literature analysis,quantitative and qualitative analysis to carry out the research is meaningful.First,the article carries on the qualitative analysis.Lending to contrast China and United States Internet platform overall development situation,and from the P2P network industry loan market scale,platform,operation mode and market regulation mechanism three aspects in detail.We found that market demand is big,industry of our country P2P platform operation mode is not sound,national regulation mechanism is imperfect and so on.Then,the article carries on the quantitative research.In this paper,specific transaction data of Lending Club and Renrendai,two representative online Lending platforms in China and the United States are selected as targets.Building a logistic regression model based on LightGBM algorithm as well as analyze the influencing factors of default risk.The results show that for Lending Club platform,credit status is the key to risk of default.Indicators such as the total amount of loans on the platform and loan interest rate have a positive impact on the default risk;while annual income of the borrower and the maximum current balance of all revolving accounts have a negative impact on the default rate.The total amount of moneythe borrower applies for on the platform has the greatest impact.For Renrendai platform,personal information and loan information are the key factors that affecting the default risk.Age and the number of successful loans have a positive impact on the default risk,while the number of loans and the number of completed loans have a negative impact on the default risk.The number of pens returned has the greatest impact.Finally,based on the results of qualitative and quantitative analysis,the following Suggestions are put forward for China's P2P industry.Secondly,the privacy of borrowers should be fully protected to reduce the negative impact caused by information disclosure in an era when data is as expensive as oil.Thirdly,risk diversification mechanism should be set up so as to spread risks and reduce credit that costs by seeking other lenders to share a loan line.As a linear model,logistic regression model has limited learning ability,while lightGBM method based on integrated learning can independently find distinguishing features and directly go for the path of decision tree into logistic regression.In the empirical part of this paper,lightGBM model and logistic regression model are combined and applied,which eliminates the need to manually search for features and makes the model have the advantages of shortening cycle and improving efficiency.This paper chooses Lending Club,a representative platform in the United States,and Renrendai,a representative platform in China,for comparativing analysis,which is of certain reference significance for the development of China's P2P industry.However,due to the imperfect information disclosure rules of the Renrendai website,the integrity of the data obtained by the crawler software cannot be guaranteed.In addition,the process of model establishment is faced with the contradiction between "precision" and "accuracy" and an appropriate solution has not been found.It is believed that in the future,with the perfection of website information disclosure rules and scholars' further in-depth study of the algorithm,there will be more optimized methods to solve this problem,so that the model prediction results will be closer to the real value.