Journal of Accounting and Management Information Systems (JAMIS)

The generalizability of financial distress prediction models: Evidence from Turkey

Vol. 14, Nr. 4/2015 ,   p685..703

Ibrahim Onur Oz
Tezer Yelkenci

Keywords:   Financial distress prediction, emerging markets, model comparison, financial crisis, multiple discriminant analysis, logit, probit, hazard model, financial ratios

Abstract:   This study analyzes five of the well-known and most cited distress prediction models in the literature. The models are implemented to continuous publicly listed industrial firms in Turkey through their original and re-estimated coefficients in a comparative way to examine their generalizability in different time periods and samples. The effect of 2008 financial crisis is also assessed to conduct a fuller analysis of the models’ prediction accuracies. The results emphasize that Ohlson (1980), Taffler (1983), Zmijewski (1984), and Shumway (2001) provide highly accurate distress classification results through their original coefficients for Turkish industrial market. On the other hand, the re-estimation of the models (other than Ohlson’s [1980]) fails to improve the prediction accuracies which are also found insignificant by considering the pre and post crisis periods.