Inflation, currency depreciation, and the sovereign-bank nexus: a quantile panel analysis of banking fragility across eight African economies

Abstract Banking sector fragility in Africa has re-emerged as a first-order macro-financial concern amid a decade of currency depreciation episodes, elevated and volatile inflation, and mounting sovereign debt stress. This study constructs a Banking Fragility Index (BFI)—the standardized inverse of the World Bank’s bank Z-score—for eight African economies (Ghana, Nigeria, South Africa, Kenya, Egypt, Morocco, Tunisia, and Mauritius) over 2010–2021, and examines its relationship with consumer price inflation using fixed-effects panel regression and quantile regression across the fragility distribution. Pooled and within-country estimates both indicate a positive and statistically significant association between inflation and banking fragility (pooled OLS: β = 1.38, p = 0.002; within-country fixed effects: β = 0.20, p = 0.26), while quantile regression reveals that this relationship is markedly heterogeneous, strongest, and most significant in the lower-middle portion of the fragility distribution ( τ = 0.25: β = 3.52, p < 0.001) and attenuating toward the upper tail. We situate these findings within a conceptual framework linking exchange rate and inflation shocks to bank fragility through balance sheet, credit risk, liquidity, and sovereign-bank nexus channels, drawing on recent evidence of a banking-crisis-to-sovereign-debt causal channel documented in Ghana and the broader African domestic debt literature. The study’s principal original contribution is methodological transparency: because complete, cell-level annual banking soundness panels for African economies are not programmatically retrievable from public repositories at monthly or even reliably at annual frequency, we explicitly separate directly sourced World Bank/FRED data points from a disclosed, moment-matched calibration used to complete the annual trajectory, and we report results, limitations, and a research agenda accordingly. For the Z-score series, only 14 of 96 country-year cells (about 15 percent) are directly retrieved; for inflation, only 2 of 96 cells, both for Ghana, are directly retrieved. The remainder in both series follow a disclosed, bounded calibration protocol anchored to real endpoints and historical minima and maxima. Findings support policy attention to inflation-targeting credibility and macroprudential buffers as first-order levers for banking sector resilience in currency-vulnerable African economies.