Study On The Evaluation And Prediction Of Financial Situations For Listed Manufacturing Companies In China

With a rapid development and intense competition in capital markets in China, the quality of financial information of listed companies is getting more attention. Listed companies, shareholders, creditors and other related stakeholders need to know not only the financial situation of a company but also accurate financial status of the company to make right decisions. The emphases of this thesis are placed on studying a multi-classification on the financial situation of listed companies and on developing a financial situation prediction model.In this study, financial situations of listed manufacturing companies are classified into three categories: distressed, normal, and healthy financial situations. The classification variables are selected by using R-Type Hierarchical Clustering Method. The study indicates that current ratio, rate of return on assets, and ratio between cash and retained earnings are three distinguished classification indices. These three indices are able to reveal a company's financial situation completely. Listed manufacturing companies in 2005 and 2006 are classified into three categories using the Q-Type cluster analysis proposed in this thesis. All samples used in the classification can be used for the new prediction model developed in this thesis. Compared with previous classification systems, cluster analysis appears to be more objective and is able to provide more useful information.Financial situations of a company can be evaluated from financial indices, which include quality of assets, capital structure, profitability and cash flow of the company, and corporate governance of the company, which can be undertaken by optimizing the structure of shareholder, the structure of board of directors, and performance and motivation. 534 listed manufacturing companies in 2005 are used as the unpaired whole sample. Distinguished variables for financial characteristics and corporate governance characteristics of two classifications and three classifications in the years t-1 and t-2 are determined by using the nonparametric testing approach and are used in the financial situation prediction model.534 listed manufacturing companies in 2005 are used as the modeling samples and 560 listed manufacturing companies in 2006 are used as the testing samples. They are both unpaired. Financial prediction models are formulated to analyze the samples by using Logistic Regression Analysis and SVM method respectively and the results are more universal and reliable. In case of two classifications, the result by using Logistic Regression Model is good and the prediction correct rate in the pre-crisis year (t-1) is up to 80%. In case of three classifications by using SVM model, the prediction accuracy of three classifications in the pre-crisis year (t-1) is up to 71% and indicates that the model has an excellent predictability and provides more information than previous studies.The SVM model is able to predict not only the financial situation of a company but also accurate financial status of the company.The major contributions of this study include applying data minig technology to financial situation classification,using a unpaired whole sample of three-classification and formulating a new prediction model based on SVM. More accurate information on the financial situation of a company can be obtained by using the new classification system and prediction model.