Predicting corporate failure using accounting information : the Malaysian experience

Financial ratios have long been used as predictor of important events in the financial markets. Researchers have formulated business failure prediction models utilising financial ratios. However, relatively few failure prediction studies on Malaysian firms have been documented. The objective of t...

Full description

Saved in:
Bibliographic Details
Main Author: Muhamad Sori, Zulkarnain
Format: Thesis
Language:English
English
Published: 2000
Subjects:
Online Access:http://psasir.upm.edu.my/id/eprint/8824/1/FEP_2000_6%20IR.pdf
http://psasir.upm.edu.my/id/eprint/8824/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Financial ratios have long been used as predictor of important events in the financial markets. Researchers have formulated business failure prediction models utilising financial ratios. However, relatively few failure prediction studies on Malaysian firms have been documented. The objective of this study is to develop a model that can discriminate between Malaysian failed and nonfailed firm. Also, this study investigates the distributional properties of the financial ratios of failed and non-failed listed firms. One-to-one sampling technique was utilised, where 33 failed and non-failed mixed industry sector firms, and 24 failed and non-failed industrial sector firms for the period from 1980 to 1996 were sampled. Using Kolgomorov-Smirnov test adjusted to Lillifors test, it was found that, only one financial ratio was normally distributed. Nine financial ratios were found to be lognormal in mixed industry sector and the number increased to 18 in the industrial sector In addition, 3 financial ratios were square root normal in mixed industry sector and 6 in industrial sector It is found that the log transformation technique was the most effective procedure and the square transformation technique was the least effective to transform non-normally distribution data to the family of lognormal distribution Finally, industry sector played an Important role in determining the normality level, where focused into specific industry sector gave better results than mixed industry sector However, it is found that the equality of variance covariance of the failed and non-failed firms was not observed However, the impact of this inconsistency was minimal on the classification accuracy After the assumptions of discriminant analysis were satisfied, stepwise multiple discriminant analysis was utilised to develop failure prediction models The mixed industry model correctly classified 86 2% and 91% of the original sample and holdout sample respectively The model was further validated using leaveone- out classification or U-method (86 2% correct classification) The results remain robust and the failed and non-failed classification accuracy was found to be significantly better than chance An alternative prediction model was developed based on accounting information, which outperformed the original model and correctly classified 88 1% of the original sample and 86 7% in U method The models for industrial sector were equally accurate for the mixed industry, which correctly classified more than 80% of the failed and non-failed firms and the original model outperformed the alternative model. The selected variables in the final model were a good proxy for the profit, cash flow, working capital and net worth variables.