Predicting Financial Distress in Select Indian Listed Companies Through Machine Learning Algorithms

Abstract

In the era of digital revolution, identifying and predicting the signs of financial distress in advance have become a strategic consideration. It enables in preserving the confidence of investors, while making informed decision to avoid distress. The financial distress concerns of listed companies become a matter of debate in academic and professional realms, as early detection enables companies, investors and regulators to implement corrective measures even before it strikes. In developing Indian markets, the vulnerabilities of businesses and structural imbalance have added to the pressure of wanting to have a sound predictive framework. This study aims to examine the effectiveness of machine learning algorithms in forecasting financial distress in select Indian listed companies. This research is based on 7,705 observations derived from 1,938 Indian listed companies across 11 sectors. The financial distress was analysed through three well-established traditional models namely Altman Z-score, Fich and Slezak, as well as Zmijewski index. In addition, key financial ratios representing profitability, leverage, liquidity, and operational efficiency are incorporated into it. The outcomes from these conventional Models were examined with the machine Learning algorithms encompassing of logistic regression, decision tree, K nearest neighbours and random forest. The predictive performance of these models was then evaluated on metrics such as precision, recall, F1 score and area under curve measure. The results indicate that the random algorithm shows strong predictive performance and delivered the highest accuracy of 93% for Model 1- Altman Z-score, 99% for Model 2- Fich and Slezak and Model 3 - Zmijewski. The results evidenced that asset turnover, return on asset, interest coverage ratio, and debt ratio have the significant impact on financial distress under different models.