Peer-to-peer (P2P) lending: default, default dependency and industry potential

The thesis presents three empirical chapters on the credit risk and industry potential of the Peer-to-Peer (P2P) market. The first empirical chapter explores default risk which considers macroeconomic factors in addition to traditional determinants such as credit scores, loan-to-income level, employability, and the credit history of the borrower. The importance of the macroeconomic environment on the lending market is highlighted by Gremi (2013). Logistic regression has been applied to one of the biggest P2P platforms operating in the United States (US), Lending Club, on borrowers from 2007 to 2016. The results show the grading system and FICO score are the significant factors associated with the default of loans. This thesis also finds a significant negative relationship between the macroeconomic environment and the performance of the P2P loans.The second empirical chapter explores default dependency in P2P lending with a specific focus on asymmetric tail dependency. Five different copula families are used to capture possible non-linearity and asymmetric properties of the loan portfolios. Using the monthly loan dataset of one of the biggest P2P Small and Medium-sized Enterprise (SME) lenders, Funding Circle, from 2010 to 2018, the study covers six different risky loan segments and, therefore, fifteen loan portfolio pairs. From the estimations of the entire distribution and upper tails of fifteen portfolio pairs, the study finds that the Gumbel copula, which represents right-tail dependence, fits the empirical data best in eight out of fifteen portfolios. In three cases, the joint default distribution shows symmetric tail dependence represented by Student’s t copula. The results highlight the importance of the joint default distributions of portfolios, in addition to individual defaults of each borrower.The third empirical chapter investigates the industry potential of P2P loans in the UK. Quarterly growth rates of P2P consumer and business loans are fitted to the Bass, Logistic, and Gompertz models from 2005 to 2019. The goodness of fit measures strongly supports the soundness of the Gompertz model for P2P Business and Consumer Lending. Selected models are used to forecast the diffusion pattern for the next 5 years and predict the industry potential of P2P lending. According to the estimations, total cumulative loan allocation through P2P platforms in the UK will reach nearly £24 billion by the end of 2024. The market potential is expected to be triple the current size and reach £32 billion, while the P2P business lending keeping its dominant position, compared to P2P consumer lending.