The case for hybrid microfinance: Combining AI with social accountability

A recent AI-driven microfinance product in Indonesia, unveiled by a regional fintech, showcased the potential of blending artificial intelligence with social accountability. This new model leveraged transaction data from a payment app, mobile usage patterns, and social signals to generate credit scores for individual applicants within seconds, significantly improving approval and origination rates compared to traditional microfinance.

However, the AI model's approach raises concerns about the core principles of microfinance. A master's research from a decade ago highlighted the crucial role of the social mechanism in group lending, as the social bond was not merely a delivery channel for credit but the credit itself. The current generation of AI-driven products risks removing that essential social mechanism while retaining the label, termed the "solidarity break."

Historical microfinance models built on group solidarity, social pressure, and weekly meetings to achieve repayment rates of over 95 percent in low-income populations. The decline of group lending's effectiveness stems from three factors: the dense availability of individual data, the collapse in origination costs due to AI-driven processes, and the shift towards individual pricing.

The weakest borrowers, who historically benefited from cross-subsidization within the group, now face higher individual rates in AI-driven lending, despite lacking the social support that traditionally enabled repayment. Maintaining a hybrid model, combining group formation with individual scoring, might preserve the social architecture around loans while benefiting from AI's precision.

Some institutions, including Sharia-aligned microfinance providers, have begun experimenting with these hybrid models, aiming to achieve the best of both worlds.