Predicting Financial Distress of Companies in Malaysia: A Comparison of Adaptive Neuro-Fuzzy Inference System and Discriminant Analysis

Abstract

Financial distress prediction has been a topic of great interest over decades, not only to managers but also to the external stakeholders of a company. The aim of this study is to compare the ability of the adaptive neuro-fuzzy inference system (ANFIS) and multiple discriminant analysis (MDA) in predicting the financial distress of public listed companies in Malaysia. The sample consists of 42 financially distress and 42 non-financially distress companies for the period 2005–2015. The financial data of the companies were collected for 3 years prior to classification as PN17 companies by Bursa Malaysia. Five financial ratios existing in the Altman model were used as the input variables. The results of this study indicate that the ANFIS model could accurately predict 95.98 and 84.62% of the respective training and holdout sample. On the other hand, the MDA model achieves 83.91 and 76.92% overall accuracy prediction rate for the training and the holdout sample respectively. This study will be useful to financial institutions, investors, creditors and auditors to identify companies that are likely to fall into financial distress.