Bankruptcy Prediction Accuracy: Z-Score Vs Random Forest in Indonesia's Manufacturing Industry

Sari, Yulia Sindi
Arina, Natrabilla Cahya
Kurniawanti, Ika Atma

Jurnal Akuntansi dan Keuangan
Vol 15, No 1 (2026)

This study compares the predictive performance of the Altman Z-Score and Random Forest models in identifying financial distress among Indonesian manufacturing firms. Using unbalanced panel data from 1,476 firm-year observations over 2015 to 2024, the study evaluates both models through accuracy and the area under the receiver operating characteristic curve. The results indicate that Random Forest outperforms Altman Z-Score, achieving an accuracy of 88.68% compared with 78.66% and an AUC of 0.931. The evidence further shows that most observations remain in the non-distress category, while Random Forest is more effective in detecting financially vulnerable and borderline firms. These findings suggest that Random Forest offers a more robust early-warning mechanism than the conventional ratio-based approach for bankruptcy risk assessment in heterogeneous financial settings.