Research On Financial Crisis Of Listed Company Based On Neural Networks

Now, with the continuous development of China’s securities market, the relevantpolicies and regulations on the listed companies become more perfect, which is bothan opportunity and a threat for listed companies. The opportunity is the continueimproving laws and regulations make the stock market more standardized; And thethreat is that this also greatly constrained listed companies, the company mightbecome "ST" or "*ST" because of this. Therefore, in the modern competitive market,it is necessary to control and prevent the financial crisis and risk if a company want tosurvive, develop and even make profit. So, it is extremely urgent to establish aneffective early warning system about financial crisis for listed companies.Firstly, this article simply reviewed the relevant literature in and abroad onrespects of the BP neural network and cash flow indexes. And then the articleintroduced the theoretical basis of financial warnings and elaborated the definition ofa financial crisis and the type of samples of financial crisis, and also introduced thetheory of BP neural network. On the basis of previous studies, this paper selected26indicators from six aspects of solvency, operational capacity, profitability, growth,financial flexibility and cash flow structure to establish a set of financial warningindicators of cash flows. This paper selected77ST companies between2007and2012of Shanghai and Shenzhen A-share listed companies as training samples in theempirical part, and also selected77non-ST companies which are in the same industry,the same period and the same principles of scale. This paper filtered indicators withdata of samples which were ST in a year, two years and three years and establishedthe mode with the screened indicators. The mode of this paper is established by neuralnetwork method. The empirical analysis shows that the accuracy rate of BP neuralnetwork financial warning model reached96.08%,88.24%and78.92%in the firstthree years of listed ST companies. This paper confirmed the superiority and accuracy of the neural network applying in financial early warnings.The conclusion of this paper is as follows:(1) The financial early warningindicator system based on cash flows established in this paper get very good financialwarning effect;(2)It predicts accuracy for the financial early warning modelestablished with BP neural network method in this paper, and it is also valuable inpractical applications;(3)We can get more stable results when the input of the neuralnetwork model constructed in this paper is between0and1.