Digital Financial Inclusion and Poverty in Sub-Saharan Africa: Evidence from Panel Data and Methodological Lessons for Small Macro Panels

This paper examines the association between digital financial inclusion (DFI) and poverty across 30 Sub-Saharan African countries (2015–2022). Rather than relying on a single proxy such as registered mobile-money accounts, a two-dimension Digital Financial Inclusion Index (DFII) is constructed via Principal Component Analysis, separating account access from active usage — a distinction that matters given regional account dormancy ranges from near zero to over 90%. In the preferred Fixed Effects specification, selected via a Hausman test and estimated with cluster-robust standard errors, DFII is negatively and significantly associated with the poverty headcount ratio, a result that survives extensive robustness checks. Three more demanding extensions — dynamic System GMM, a Hansen (1999) panel threshold model, and panel quantile regression — each fail to detect a significant effect on the same sample. This pattern is interpreted as a single, coherent finding about sample size relative to estimator complexity in this literature, rather than three unrelated setbacks.