Download PDF (2097kb)

Non-performing loans (NPLs) represent a critical challenge to financial stability across nine emerging economies in Asia and Africa, where rapid credit growth, macroeconomic, and institutional volatility reinforce systemic risks. This study investigates the determinants of NPLs by employing a novel hybrid methodology integrating Panel ARDL estimation to identify long-run equilibrium, short-run relationships along with checking robustness through Fixed Effect and Random effect, and SHAP (SHapley Additive exPlanations) machine learning analysis for predictive feature importance and non-linear insights, moving beyond traditional approaches. Our findings reveal a complex interplay of factors, with distinct regional patterns. In Asia, NPLs are primarily driven by trend-following credit expansion and inflationary pressures, consistent with the Financial Accelerator mechanism. In Africa, macroeconomic instability, particularly high interest rates, and weak institutional frameworks are the dominant predictors, aligning with Institutional and Credit Rationing theories. The SHAP analysis corroborates these results, identifying bank credit and domestic credit as top global predictors, while highlighting regional asymmetries: inflation and regulatory quality are paramount in Asia, whereas interest rates and macroeconomic shocks generate higher predictive variance in Africa. Based on these insights, we propose distinct policy frameworks:Asia requires countercyclical credit regulations and sector-sensitive capital distribution, while Africa needs institutional reforms focused on collateral registries and interest rate stabilization, with simulations indicating potential NPL reductions. Our research emphasizes that NPLs are a macro-institutional challenge, necessitating integrated, region-specific strategies for financial stability.

Keywords: Non-performing loans, Emerging economies, Credit, Institutional Quality, Panel ARDL, SHAP analysis