ADAPTING THE ALTMAN Z-SCORE WITH ARTIFICIAL INTELLIGENCE FOR HOUSEHOLD FINANCIAL DISTRESS PREDICTION

Resumen

This study aims to modify the corporate Altman Z-Score model for application in household finance and enhance its predictive accuracy by integrating an artificial intelligence approach. Employing an exploratory-predictive quantitative design, the research utilized a sample of 467 Indonesian households with debt obligations. The original Z-Score variables were successfully adapted to relevant household financial components. The modified model classified 82.4% of the sample into the safe zone, 11.8% into the grey zone, and 5.6% into the distress zone. A notable discrepancy was observed between the model's objective assessment and households' subjective perceptions of financial distress. However, robust statistical tests confirmed a significant relationship between the Z-Score categories and self-reported distress status (Chi-Square p < 0.001, Cramér's V = 0.676), with substantial agreement (Cohen’s Kappa = 0.643). The implementation of artificial intelligence via ordinal logistic regression validated the adapted Z-Score as a highly significant predictor (p < 0.001), indicating that a one-unit increase in the Z-Score reduces the odds of being in a higher distress category by 4.4%. The study concludes that the AI-enhanced, adapted Altman Z-Score model serves as a valid and robust predictive tool for identifying household financial vulnerability. This instrument is recommended for financial institutions and regulators to monitor systemic risk and contribute to sustainable development goals, particularly SDG 1 and SDG 8.