Abstract. Previous studies on bankruptcy prediction focused on identifying significant indicators to predict bankruptcy. Few studies analyzed the impact of the change in space and time but there are limited studies which attempted to investigate the sensitivity of these models to the change in econometric methods. The current study analyses the impact of the change in econometric methods on the predictive performance of Singh and Mishra (2016a) bankruptcy prediction model. A matched pair of 208 companies comprising distressed and non-distressed firms for the period 2006 to 2014 were selected randomly. The study utilises Multivariate Discriminant Analysis (MDA), logit and probit econometric techniques to model bankruptcy. Secondary sample, long range accuracy and Receiver Operating Characteristic (ROC) tests were used for the validation of bankruptcy prediction models. The major findings of the study suggest that accounting information’s, namely, leverage, profitability and turnover ratio remained significant indicators to predict bankruptcy for Indian manufacturing firms. The study further concludes, if significant indicators of bankruptcy are identified, then there is no significant impact of the change in econometric methods on the predictive performance of the default prediction models.

Keywords: bankruptcy prediction, Indian manufacturing companies, logit, probit, MDA.