Default or Profit Scoring Credit Systems? Evidence from an Emerging High-Risk P2P Loan Market.
42 Pages Posted: 10 Sep 2020
Date Written: July 31, 2020
For the emerging peer-to-peer (P2P) lending markets to survive, they need to employ credit risk management practices that ensure an investor base that is profitable in the long-term. In this paper, we propose a profit scoring decision support system that is dynamically updated and based on modeling the annualized adjusted internal rate of return of a loan. Our statistical approach is based on logistic and linear regularization methods complemented with Bayesian model averaging and selection techniques. Using data on loans from an emerging European P2P market, we document that in an out-of-sample framework, our approach overwhelmingly dominates standard credit scoring models that are based on labeling loans as either defaulted or not. In fact, even if we take data snooping bias into account, we find that realized returns tend to be significantly--almost 2.5 times--higher when using our profit scoring approach compared to the standard credit-scoring model based on regularized logistic regressions. Finally, as our results are robust across different modeling choices, we conclude that the management of credit risk can be significantly increased by designing systems that model profitability instead of loan (non)failure. Our results thus suggest a paradigm shift in modeling credit risk in the P2P market as modeling loan returns leads to much more accurate decisions than that achieved by more elaborate models that model loan's probability of a default.
Keywords: profit scoring; credit scoring; financial intermediation; P2P; fintech;
JEL Classification: D12, E41, G20
Suggested Citation: Suggested Citation