Credit losses in peer-to-peer lending: a quantile regression approach Author

Abstract

Peer-to-peer (P2P) lending is a form of direct lending that connects borrowers with investors through online platforms. This study analyses factors that determine credit losses in peer-to-peer lending with a dataset consisting of 253 948 charged-off loans from the Lending Club. Most of the previous quantitative studies have focused on understanding and predicting defaults which has left a gap in the literature in understanding default amounts. This study aims to deepen the understanding of the relatively unexplored area of credit losses in P2P lending. Through quantile regression, we have examined the relationship between borrower characteristics and credit losses to answer the question “What are the determinants of credit losses in P2P lending”. The results show that loan terms, application type, and interest rate are among the most important determinants of credit losses. Employment length is an important determinant in the lower quantiles, but the effect becomes less significant compared to other covariates in higher quantiles. Lastly, loan purpose is shown to be among the top determinants and becomes more important toward the higher quantiles. By identifying the determinants of credit losses, we offer valuable insights for investors, P2P platforms, as well as policymakers in assessing credit risk. This study contributes to a deeper understanding of credit losses in P2P lending and we encourage future research to use this paper as a theoretical framework to go even deeper into credit losses. For example, a study that uses extremal quantile regression to analyze the tails of the distribution.