Research On The Financial Crisis Warning Of Listed Companies In China Based On Bp Neural Network Model

The subprime mortgage crisis broken out in U.S.2007had graduated into a financial crisisgradually and had impacted its economy heavily, and it also influenced the other countries’economy in different way. Shortly thereafter, the European debt crisis took place, this madethe European economy which was not good even worse. The both had casted a shadow overthe global economy. As China’s economy continues to develop closer cooperate with theinternational economy, the U.S. subprime mortgage crisis and the European debt crisisinfluenced it more heavily. Meanwhile, the development of China’s economy has reached abottleneck period, the domestic market demand and the GDP growth rate are declining andthe pattern of economic development needs to change. The pessimistic situation of theinternational and domestic economy had caused the enterprises goods accumulated, moneyrecovery difficult, capital chain breaks, triggering the financial crisis, some even bankruptcy,led the enterprises lose a lot. Therefore, it will make a big difference of the company makingthe financial crisis early warning before it happens.According to the financial crisis forms, characteristics and causes，the paper has selected25alternative financial indicators of financial crisis to form an early warning index system.Choose the98A-share companies, listed in Shanghai Stock Exchange and Shenzhen StockExchange, in the financial crisis from2009to2012as the sample group, at the same time,choose98companies which financial is normal during the same period as the paired samplegroup to match the research. Collect the two groups of companies’ financial data of a yearbefore the sample group company broken financial crisis, do relevant test, and remove theunqualified data. Then use the first3years data of the remaining indicators to establish theBP neural network, apply the data of the4th year to do the simulation test. After repeatedlytraining and testing with MATLAB, the established BP network has a good and stableperformance and the simulation forecast accuracy is92%.The innovations of this paper are as follows: Firstly, the financial crisis early warningindexes are doing non-parametric test and multicollinearity test at the same time to improvethe effectiveness of these indicators; then narrowing the error of the BP network predictionerror and improve the accuracy of its determination; Finally, this paper has collected the lastest four years’ data and has got the satisfactory results, which is timeliness and has a highpractical value.