Analyzing peer-to-peer lending secondary market: What determines the successful trade of a loan note?

Abstract

Predicting loan default in peer-to-peer (P2P) lending has been a widely researched topic in recent years. While one can identify a large number of contributions predicting loan default on primary market of P2P platforms, there is a lack of research regarding the assessment of analytical methods on secondary market transactions. Reselling investments offers a valuable alternative to investors in P2P market to increase their profit and to diversify. In this article, we apply machine learning algorithms to build classification models that can predict the success of secondary market offers. Using data from a leading European P2P platform, we found that random forests offer the best classification performance. The empirical analysis revealed that in particular two variables have significant impact on success in the secondary market: (i) discount rate and (ii) the number of days the loan had been in debt when it was put on the secondary market.