Kmv Model And Neural Network-based Listed Companies' Financial Crisis Early Warning

Competitive market economy environment has produced great opportunities for corporations, as well as great risk and crisis. Once the financial crisis happened among listed companies, it is no doubt that all shareholders, lenders and other interest-relating owners will suffer huge losses. Not only to survive the fierce competition and create the well development, but also to maximize the interests of shareholders and lenders to ensure the financial distress of listed companies has attracted much attention.Throughout all existing research about the financial distress of corporations, it is mainly based on the typical financial indicators to forecast financial situation. However, an analysis based solely on financial indicators of enterprise is difficult to cover all the reasons which cause the financial crisis. Also the lack of reference information on the capital market is another aspect. In view of that, the financial crisis forewarning indicator system in this paper includes the typical financial indicators, the default distance indicator which come from the stock market data, and the input-output efficiency indicator which come from the pure financial data. Then, an empirical analysis has been done by the criterion that "ST" is similar to the financial crisis, based on the public data of listed companies to build the training and test samples. Finally, the financial distress forewarning model was constructed through BP neural network method. The research results of this paper contain:First, considered the full circulation of stock market, we first modified the default point of KMV model parameter by adjusting the proportion between the medium-long term and short term liabilities. Then we calculated the default distances of all samples by using the default point modified KMV model.Second, based on the default distances of all companies, it's turned out that the default distance of the ST is less than non-ST companies which is in accordance with the actual situation. The default distance indicator has reflected to some extent true of stock market.Third, DEA method was chosen to calculate the input-output efficiency indicators of all selected samples. The pure financial data turned into the input-output relationship. In this paper, the inputs including: the operating cost, expense, asset and liability, while the outputs including: income and net profit. The results showed that compared with non-ST, ST companies had lower input-output efficiency and greater fluctuation.Fourth, we established the financial distress forewarning model of BP neural network based on the default distance, input-output and basic financial indicators. The empirical results showed that BP neural network has better forecasting effects. By comparing the forecasting results, the model after including the default distance and input-output efficiency indicators performed better.