Empirical Study On Financial Affairs Early Warning Model Of The Listed Company In Our Country By The Approach Of The Principal Component Analysis

After our country is entering into WTO, managing risks and financial risks with which companies in our country are confronted are increasingly improved. In this environment, how to avoid and disperse these risks has become the discussed hotspot in the financial scope. Many practitioners favor the financial affairs early warning as an important step of the financial risk management. This paper selects the listed company as the samples and tries to establish a financial affairs early warning system and make some innovations on the research ways by using the scientific statistics ways—the Principal Components Analysis. This analytical method uses a few key factors to represent overall change. The way is convinent to analysis the problem because it reduces the number of variables.In this paper, first, it shows the significance of conducting financial affairs early warning models in our country and emphasizes the necessities of this study. It summarizes the domestic and abord achievements empirical study on financial affairs early warning model. It gives the lessons conducting from this study on the basis of analyzing and concluding these results. Then, the writer simply introduces the Principal Component Analysis. At last, the study presents the conclusions and other problems, which are necessary to be pointed out.In this study, it selects 54 "ST" corporations and 54 normal corporations as samples in Shanghai and Shenzhen stock market in the year of 2003. The paper chooses some variables from all of the original variables by the test of significance. At last, it selects 9 finance indexes as discriminating variables. It applys the index of cash flow in order to find their discriminating effects. After that, the paper establishes financial affairs early warning model by using the approach of Principal Component Analysis. It also sets up an interval for identifing the extent of financial evaluations and presents amethodology to calculate the cut point that is rarely mentioned in most of previous studies for multivariate models. The paper tests and verifies the significance of the warning model by the test of hypothesis. It shows that the discrimination models can predict the finance distress to 84.25% correctly.