DIGITAL CREDIT ALGORITHMS AND AGRIPRENEURS’ ACCESS TO FINANCE IN SUB-SAHARAN AFRICA: EVALUATING MOBILE MONEY TRANSACTION HISTORIES AS COLLATERAL SUBSTITUTES

Abstract

In Sub-Saharan Africa, many smallholder farmers struggle to obtain loans because traditional lenders require collateral, which these farmers often lack. However, with the rise of mobile money, there is a new opportunity to assess creditworthiness. This study explores whether mobile money transaction histories can substitute traditional collateral in digital credit algorithms, potentially helping agripreneurs access finance. The research examines data from four countries: Kenya, Ghana, Nigeria, and Cameroon, involving 2,400 agripreneurs and over 847,000 mobile money transactions. Interviews with 60 agripreneurs, 24 lenders, and 16 regulators were conducted for deeper insights. The quantitative data were analyzed using random forest machine learning models, logistic regression, and propensity score matching, while qualitative data were analyzed using thematic analysis. The results showed that mobile money transaction history-based credit scores predicted loan repayment with 78.3% accuracy, comparable to traditional collateral-based assessments but with higher inclusion rates. Borrowers approved through the algorithm had similar default rates to those approved through traditional assessments. Despite positive findings, challenges such as algorithm distrust and regulatory gaps remain. Recommendations include establishing regulatory sandboxes, mandating transparency disclosures, developing value chain-specific scoring models, implementing consumer education programs, and creating portable credit reputation systems across FinTech providers. Overall, mobile money transaction histories could significantly benefit agripreneurs in Sub-Saharan Africa by improving access to finance, provided appropriate regulations and safeguards are implemented.