Risk Assessment in Peer-to-Peer Lending Platforms: The Case of Lusaka, Zambia

Abstract

The rapid expansion of digital financial technologies has transformed credit access through Peer-to-Peer (P2P) lending platforms, which directly connect borrowers and investors. In Zambia, where 30.6% of adults remain financially excluded, platforms such as PremierCredit and Lupiya are expanding credit access for MSMEs and informal workers; however, this growth is associated with rising credit risk, operational challenges, information asymmetry, and regulatory uncertainty. This study addresses the lack of empirical evidence on risk assessment practices, credit scoring effectiveness, regulatory pressures, and investor protection in Lusaka’s P2P lending sector, and examines concerns surrounding AI-driven credit scoring models, particularly their limited transparency and contextual relevance. Using an explanatory sequential mixed-methods design, data were collected from 102 respondents and supplemented with 20 semi-structured interviews and three focus group discussions; quantitative analysis was conducted using SPSS, while qualitative insights were analysed thematically. Findings show that credit risk is the most significant concern (composite mean M = 3.957), driven by limited credit bureau coverage (M = 4.22) and the informal economy (M = 3.92); current risk assessment practices were rated below the scale midpoint (M = 2.943), with context-sensitive approaches scoring particularly low (M = 2.48); regulatory ambiguity (M = 3.75) was identified as a major barrier to platform scalability; and investor protection was perceived as inadequate, with borrowers reporting lower protection levels (M = 2.560) compared to investors (M = 3.297) and platform operators (M = 3.054), with statistically significant differences (F(2, 99) = 19.549, p < 0.001). Regression analysis revealed that credit risk assessment (β = 0.583) and methodological rigour (β = 0.446) significantly predict investor protection outcomes, explaining 76.8% of variance (R² = 0.768, F(3, 98) = 108.42, p < 0.001). The study concludes that stronger, context-specific risk assessment is essential and recommends regulatory harmonisation, expanded credit bureau coverage, improved disclosure standards, and locally adapted credit scoring models.