Bankruptcy Prediction Using Cerebellar Model Neural Networks

Abstract

In Taiwan, more and more enterprises face the problems of financial distress in recent years. Besides, it is also noted that the volume of outstanding debt to corporations increases in Taiwan. An improvement in distress prediction accuracy can lead to save tens of billions of dollars. The firms which face financial distress will reveal many signs on financial data. Therefore, this study hopes to provide to managers and investors as a reference for decisions making through systematic approach as it can look for firms facing financial distress. In this paper, a novel prediction system is proposed which is based on intelligent classification to distinguish bankruptcy prediction. This method is referred to as a cerebellar model neural network (CMNN). A CMNN can be thought of as a learning mechanism imitating the cerebellum of a human being. Through training, this CMNN can be viewed like an expert of financial analyzer and then it can be applied to bankruptcy prediction. This study uses an artificial neural network, a genetic programming, and the proposed CMNN to construct financial distress prediction models and compare the performance of above three models using some Taiwanese company data, and it confirms CMNN is better than the others. By doing this, it can help understand and predict financial condition of firms and prevent the firms from insolvency. The CMNN yields the best prediction through the efficient infer reasoning of CMNN. Thus, the result is feasible to construct the financial distress prediction model.