<![CDATA[This study aims to develop an interpretable early-warning framework to predict financial distress among non-life insurers in Bangladesh. The research addresses the question of which financial and operational indicators most accurately signal early signs of distress in the insurance sector. Using a firm-year panel covering 2014–2024, the study applies penalized logistic regression, random forests, and gradient-boosted trees, combined with class-balancing remedies and SHAP-based interpretability techniques, to identify the key determinants of insurer distress. The results show that gradient-boosted trees achieve the highest out-of-time recall performance. At the same time, SHAP analysis consistently identifies the management expense ratio, lagged underwriting performance, and reinsurance intensity as the most influential predictors. Robust tests across alternative sampling and feature reduction methods confirm the stability of these findings. The study concludes that monitoring expense efficiency, underwriting results, and reinsurance practices provides the most reliable early warning signals of financial distress in thin-premium markets. The originality of this research lies in integrating explainable machine learning with operational financial indicators in a developing-market insurance context, producing a transparent and policy-ready predictive model that balances accuracy and interpretability.]]>
Copyrights © 2025
