Credit Risk Assessment Models in Peer to Peer Lending Platforms

Abstract

This study provides a systematic literature review of credit risk assessment models in peer to peer lending platforms. The review is motivated by growing concerns about platform failures, loan underperformance, and investor protection in markets characterised by high information asymmetry and thin borrower credit files. Using structured database searches and explicit inclusion criteria, the study synthesises evidence on statistical, machine learning, and ensemble based models that are applied to large scale lending datasets. The findings show that advanced modelling techniques consistently outperform traditional scorecards in predicting default and, when designed to target risk adjusted returns, can support more sustainable portfolio performance for investors. The review also highlights the rising importance of soft and unstructured information, including narrative loan descriptions, personality related indicators, and other non traditional signals, which enhance discrimination power when combined with numerical features. Finally, the study identifies an emerging shift toward explainable artificial intelligence and profit oriented scoring within governance frameworks that emphasise transparency, fairness, and regulatory compliance. These developments highlight the need for integrated and transparent credit frameworks.