Bankruptcy prediction models for NSE-Listed firms: A comparative analysis
Uma Maheshwari B, Kavitha D. Sujatha R., Abinaya R and Abhirami B
Bankruptcy prediction models for NSE-Listed firms: A comparative analysis - Hyderabad IUP Publications December 2023 - 5-22
According to the Insolvency and Bankruptcy Board of India, the number of companies filing for insolvency witnessed a 30.29% jump to 3,312 in the fourth quarter of 2019. The increasing rate of company failures has prompted efforts to provide better measures to predict bankruptcy well in advance. The objective of the current study is to predict the bankruptcy of firms listed on the National Stock Exchange (NSE) using the data for three years prior to bankruptcy. The data consists of financial variables (categorized into profitability, solvency, liquidity, and activity ratios) and non-financial macroeconomic variables. To achieve this objective, the study uses predictive models such as Altman Z-score, logistic regression, support vector machine (SVM), ensemble methods, artificial neural networks, etc. It also compares the accuracy with performance measures to find the best model for the prediction of financial distress in Indian firms. The findings suggest that logistic regression model has relatively higher (96%) bankruptcy prediction ability, and SVM has the highest model accuracy of 88.54% and demonstrates great ability in predicting healthy companies.
Bankruptcy
Artificial neural networks
Prediction models
Support vector machines
Comparative studies
Bankruptcy prediction models for NSE-Listed firms: A comparative analysis - Hyderabad IUP Publications December 2023 - 5-22
According to the Insolvency and Bankruptcy Board of India, the number of companies filing for insolvency witnessed a 30.29% jump to 3,312 in the fourth quarter of 2019. The increasing rate of company failures has prompted efforts to provide better measures to predict bankruptcy well in advance. The objective of the current study is to predict the bankruptcy of firms listed on the National Stock Exchange (NSE) using the data for three years prior to bankruptcy. The data consists of financial variables (categorized into profitability, solvency, liquidity, and activity ratios) and non-financial macroeconomic variables. To achieve this objective, the study uses predictive models such as Altman Z-score, logistic regression, support vector machine (SVM), ensemble methods, artificial neural networks, etc. It also compares the accuracy with performance measures to find the best model for the prediction of financial distress in Indian firms. The findings suggest that logistic regression model has relatively higher (96%) bankruptcy prediction ability, and SVM has the highest model accuracy of 88.54% and demonstrates great ability in predicting healthy companies.
Bankruptcy
Artificial neural networks
Prediction models
Support vector machines
Comparative studies