Title: Analysing the stability of bankruptcy prediction models

Authors: Rohani Md-Rus; Kamarun Nisham Taufil Mohd; Rohaida Abdul Latif

Addresses: School of Economics, Finance and Banking, Universiti Utara Malaysia, Malaysia ' School of Economics, Finance and Banking, Universiti Utara Malaysia, Malaysia ' Tunku Intan Safinaz School of Accountancy, Universiti Utara Malaysia, Malaysia

Abstract: The aim of this study is to assess the predictive power of logit model and hazard model in predicting bankruptcy and to analyse the stability of the models. Using Malaysian listed companies and a sample span from 1998 to 2014, this study found that, for the hazard model, all variables were significant while for the logit model only five variables were significant. The results also show that the logistic and hazard models both had predictive accuracies of more than 90%. However, the hazard model had a predictive accuracy of 99.4% while logit model had a predictive accuracy of 91.8%. The hazard model was more stable than logit model as the predictive accuracy of the hazard only changed a little when a smaller sample was chosen. Lastly, the study showed that, even though both models were good in predicting distress, the hazard model is better than logit model.

Keywords: logit model; hazard model; bankruptcy prediction; stability; profitability; leverage; growth; cash flow; size; Malaysia.

DOI: 10.1504/AAJFA.2020.110493

Afro-Asian Journal of Finance and Accounting, 2020 Vol.10 No.4, pp.554 - 568

Accepted: 01 Oct 2018
Published online: 22 Oct 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article