Predicting Prepayment and Default Risks of Unsecured Consumer Loans in Online Lending

Online lending provides a means of fast financing for borrowers based on their creditworthiness. However, borrowers may undermine this agreement due to early repayment or default, which are two major concerns for the platform and lenders, since both affect the profitability of a loan. While default risk is frequently focused on credit scoring literature, prepayment has received much less attention, despite a higher prepayment rate being observed in online lending when compared with default. This article uses multivariate logistic regression to predict the probability of both the underlying prepayment and default risks. Real consumer lending data of 140,605 unsecured loans provides evidence that these two events have their own distinct patterns. We consider systemic risk by incorporating macroeconomic factors in modeling and address the influence of economic conditions, which are lessons learnt from the last financial crisis. The out-of-sample validation has shown that both prepayment and default can be accurately predicted. This article highlights the necessity of regulations on prepayment given the fast growing online lending market.