Does capital adequacy moderate the credit-risk effect of profit-and-loss-sharing financing in Islamic banks? A contract-level risk framework and cross-country evidence

Islamic financing risk is frequently analyzed as a single undifferentiated category, despite substantial heterogeneity in how ownership, control, payment obligation, and loss are allocated across financing modes. This paper develops a contract-level risk-mitigation framework across ten Islamic financing instruments and empirically tests a hypothesis motivated by that framework: whether regulatory capital adequacy (CAR) dampens the relationship between profit-and-loss-sharing (PLS) financing share and non-performing financing (NPF) at the country-banking-system level, using a quarterly six-country panel drawn from the Islamic Financial Services Board’s Prudential and Structural Islamic Financial Indicators (PSIFI) database (256 country-quarter observations, 2014–2024). After confirming non-stationarity in the levels data via Augmented Dickey-Fuller tests and re-specifying the model in first differences with entity and time fixed effects and Driscoll-Kraay standard errors, we find no statistically significant CAR × PLS interaction effect on credit risk across three specifications and five robustness checks (including annual GDP-growth and inflation controls), the point estimate’s sign is inconsistent across specifications, and the models explain little of the within-country variation in credit risk (within R 2 between 0.02 and 0.06). The confidence intervals are wide and span both zero and economically meaningful effects of either sign, so the results reflect limited estimation precision at the country-banking-system level: they cannot distinguish a true effect close to zero from a modest effect estimated imprecisely. The paper’s contribution is therefore twofold: a reusable conceptual framework for contract-specific Islamic finance risk analysis, and a methodologically transparent demonstration of the limits of country-level macro panels for this class of question.