Advancing Corporate Credit Risk Assessment in Emerging Markets: A Comparative Analysis of Machine Learning Classifiers in South Africa

This study examines the predictive power of machine learning techniques in corporate credit rating assessment using firm-level financial data from 208 companies across three key sectors in South Africa. By employing statistical models alongside advanced classifiers, including logistic regression, support vector machines, random forest, decision trees, k-nearest neighbors, and XGBoost, the analysis evaluates model performance using accuracy, sensitivity, specificity, precision, and the Matthews correlation coefficient. The empirical design incorporates financial ratios capturing liquidity, solvency, profitability, and efficiency, thereby aligning predictive analytics with established financial theory. Results demonstrate that while traditional models provide a baseline framework, ensemble and kernel-based methods deliver superior classification accuracy, particularly when sectoral heterogeneity is considered. These findings underscore the growing role of artificial intelligence in improving credit risk assessments, enhancing financial inclusion, and supporting regulatory oversight in emerging markets. The study offers theoretical contributions to credit risk modeling and provides policy recommendations for integrating explainable machine learning into financial supervision and lending practices.