An Empirical Study Of Financial Crisis Early-warning Model For Listed Companies Of Our Country

The empirical study of financial crisis early-warning for listed companies is a research subject widely concerned. Since Beaver' single variable research, this study has been a durable subject in the fields of accounting, stocking and financing for nearly 40 years. While in China it's a fresh field just initiated. Being a weatherglass for economics performance as well as an indicator for enterprises' operation, financial crisis warning not only has high academic value but also great significance applied value.In view of the limitation of the statistical method which exists in the corporate financial crisis warning, this thesis proposes the superiority and potential of Artificial Neural Network (ANN). LVQ neural network has the advantage of the pattern-identification. Therefore, this thesis sets up the financial crisis warning model based on LVQ neural network.The thesis selects 148 listed companies as the study's samples and picks up 5 financial ratios for the warning model after studying the domestic and international financial crisis warning models. The samples are divided into training samples and testing samples. Based on training samples, the Matrix P of the model is set up. After fixing the model's parameters, the financial crisis warning model is set up based on Matlab7.0.In order to verify the accuracy of the model, the thesis uses the testing samples to test the model. The testing result shows the warning model which is set up has 88.3% accuracy rate, which shows that the model has the excellent function of classifying and LVQ neural network could be employed in the pattern-identification, and the model based on LVQ neural network could predict financial crisis of enterprises efficiently. We can forecast the development of LVQ neural network in financial crisis warning of listed companies.