Financial Distress Prediction Models Of Listed Companies Based On Longitudinal Data

As we know a company's financial situation directly influence its operation, and listed companies are the foundation of the stock market. So it's of great importance to keep these listed companies running well. But with the real economy keeping sliding in recent years, especially more and more manufacturing factories were in danger of collapsing, an efficient financial distressed prediction model is of great importance.We can take measures in advance if we have an efficient model.These will make great sense to investors and business managers.In this paper, we took the companies as financial distressed if they were “special treated”,and 50 manufacturing factories which were firstly special treated in the period of 2011 and 2014 were selected as samples. At the same time, we chose 50 healthy companies randomly as matched samples. This study has considered factors that were frequently used in other people's research.Firstly, we constructed a logistic regression model and fisher discriminant analysis model to predict the financial distress probability based on cross-section data. The result showed that logistic regression model performed slightly better than discriminant analysis method. However, review the former research most of the models were based on cross-section data which containing the data in only one year. So, we constructed two prediction models based on longitudinal data. One was Generalized Estimation Equations(GEE) which represented marginal model, and the other was random effect model which on behalf of conditional models. The result of these two models were almost the same. The forecasting accuracy of ST companies and healthy companies were 92% and 98% by GEE respectively, and 94%, 98% by random effect model.In conclusion, no matter the model base on cross-section data or longitudinal data, we can get high forecasting accuracy of financial distress. That means the financial ratio were efficient. Financial distress is not only influenced by profit but also determined by Solvency and operational capacity. The model based on longitudinal data performed slightly better than the model based on cross-section data, and the model gave a broad prediction range of financial distress.