Cross-National Benchmarking of Bankruptcy Prediction Models Across V4 Economies

In recent years, the prediction of corporate bankruptcy has become an increasingly important topic in financial and economic research, particularly in the manufacturing sector of Central and Eastern Europe. Accurate early-warning models are essential for mitigating financial risks and ensuring business sustainability. This study investigates the comparative performance of classical statistical and machine learning (ML) models for predicting corporate bankruptcy across manufacturing firms in the Visegrad Group (V4) countries, addressing the problem of financial distress forecasting in transitional economies. The purpose of the research is to evaluate whether pooled regional models perform as effectively as country-specific models and to examine the influence of national data characteristics, such as sample size and heterogeneity, on predictive accuracy. A balanced dataset of firm-level financial indicators from Slovakia, the Czech Republic, Hungary, and Poland was employed, and three classification techniques, namely logistic regression (LR), artificial neural networks (ANN), and decision trees (DT), were applied to develop predictive models for individual countries as well as for the combined V4 region. Model performance was assessed using multiple classification metrics including accuracy, F1 score, AUC (area under the receiver operating characteristic curve), precision, and recall, with careful attention to handling class imbalance. The results indicate consistently high discriminatory power across all models, with AUC values ranging from 0.929 to 0.991, classification accuracy between 94.9% and 98.3%, and F1 scores from 0.972 to 0.991. Artificial neural networks slightly outperformed logistic regression and decision trees, particularly in countries with larger samples, while pooled models demonstrated performance comparable to country-specific models, highlighting the generalizability of predictive models across V4 economies. The findings carry practical implications for policymakers, creditors, and business managers, supporting the development of scalable early-warning systems, enhancing risk assessment practices, and informing strategic decision-making in dynamic economic environments. Overall, the study contributes both to the theoretical understanding of model performance in bankruptcy prediction and to applied knowledge for regional economic foresight and business intelligence.