Bankruptcy Prediction Using Multilayer Perceptron Neural Networks In Jordan

Abstract

This study attempts to develop bankruptcy prediction model for the Jordanian industrial sector with a recent approach—neural networks. The multilayer perceptron neural network (MPNN) approach was used to develop the bankruptcy prediction model for the Jordanian industrial companies for the period from 2000 to 2015. The samples have been divided into two subsets: the first set for developing or building the model, made up of 14 companies, of which 7 are bankrupt and 7 are non-bankrupt; while the second is a hold-out sample for testing the model, made up of 18 companies, of which 9 are bankrupt and 9 are non-bankrupt. The main variables in predicting bankruptcy were ten financial ratios. The results show that the accuracy rate of final prediction model is found to be 100 percent. While the hold-out sample testing provides that the model correctly predicted all 18 test cases.