UPM Institutional Repository

Financial performance evaluation and failure prediction of Iranian food and agriculture corporation


Seyed Javadi, Mohammad Reza Haj (2015) Financial performance evaluation and failure prediction of Iranian food and agriculture corporation. Doctoral thesis, Universiti Putra Malaysia.


It is a common understanding that bankruptcy is not a sudden event for any community. The predicting bankruptcy by new methods could be taken as a shield of production to lower the risk and danger levels of company bankruptcy. Thus, these are much meaning to owners, creditors, investors and academics alike. With a simple plane to improve the availability literatures, The main goal of this study is to explore the predictive power of neural network model and Logistic regression model to bankruptcy prediction by measuring its percentage accuracy on the listed firms in Tehran Stock Exchange(TSE).Logistic regression model is also provided as performance benchmarks for neural network classifiers. Depending on the purposive sampling method, this study covered 94 listed companies in TSE, whereby 47 companies were healthy and 47 companies that have filed for bankruptcy during of 2005-2009.The sample analysis (85% of samples) was joint to train and validation the corporate bankruptcy prediction model. In addition, the sample validation (15% of samples) was employed to test accuracy of the corporate bankruptcy prediction models. ANNs were appraised as a tool to predict bankruptcy prediction model .They are often classified into two different training types : supervised or unsupervised .Prior researchers in business have mainly used Back Propagation(BP) networks . So this study examined BP model for their efficiency and profitability in bankruptcy prediction. Also the study uses the Receiver Operating Characteristic (ROC) curves and the accuracy rate approach to compare the predictive ability of all models that find which one model is the first. The result shows significant difference between the two models .The overall Ann's accuracy was 87% compared to 82% for Logistic regression. Besides, the Area under ROC values for ANN achieved above 88% compared to the Area under ROC values of above .82% for Logistic regression model. The study found significant difference between the two models. It means that nonlinear classifiers tend to outperform their linear model. Clearly, ANN is stands both efficient and valid preference in the terms of bankruptcy prediction for Iranian listed firms. A further result observes the majority of the selected factors belong to high leveraged and small liquidity and profitability groups cause the likelihood of bankruptcy. These findings agree strongly with the findings of as studies by (Zhang et al., 1999; Parker et al .,2002).

Download File

FP 2015 39IR.pdf

Download (1MB) | Preview

Additional Metadata

Item Type: Thesis (Doctoral)
Subject: Bankruptcy - Iran
Subject: Neural networks (Computer science)
Call Number: FP 2015 39
Chairman Supervisor: Zainal Abidin Mohamed, PhD
Divisions: Faculty of Agriculture
Depositing User: Haridan Mohd Jais
Date Deposited: 28 Feb 2018 02:58
Last Modified: 28 Feb 2018 02:58
URI: http://psasir.upm.edu.my/id/eprint/59115
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item