The Studies On The Model Of Warning For Corporate Financial Distress

There are many different opinions among domestic and overseas scholars about"financial distress"conception. In this thesis,"financial distress corporation"is defined as ST corporation.Because of restricting corporation development, financial distress has been the linchpin question that can't be avoided. In order to enhance competitive power, it is necessary to strengthen corporate management about financial distress, so as to offer a better economic environment, and avoid or reduce financial distress.The financial condition of corporation is the integrative reflection of the corporation's management. At the same time the financial condition determine whether the corporation can go on running. So the research of the warning for corporate financial distress is a key to the management of corporation. The thesis focuses on the point. I hope it will be helpful for enhancing the corporation's management level.This paper begins with fundamental theories about financial distress and lays a theoretical foundation for the following demonstractive research. Then it makes a retrospect about domestic and foreign current research products. Two multivariate statistical analysis and two artificial intelligence methods are employed to make demonstractive research, which is the focus of this paper. Firstly I construct a principal component analysis warning model. Secondly, I construct a discriminating analysis model. Because linear model have some defects, I also construct BP neural network model and support vector machine. The correct results of their prediction are 70%, 80%, 85% and 82.5%, respectively. The results of the discriminating show that all these four methods can be used in forecasting financial distress, in which the effect of BP neural network and support vector machine are the better. By comparing the four models, their merits and demerits are shown.Based on the comparison of merits and demerits of the four models, we advance a new model---mixed system model, and its prediction correct ration is 85%. Because this model combines the former four models, as to every corporation, mixed system model has more warning reliability than any single model. So mixed system model has preferable application prospect.