The Study On The Financial Crisis Early Prediction Model For Chinese Listed Companies

Financial crisis prediction has been a topic of broadly concern for a long time. In the rapid developing but yet immature Chinese securities market, financial security of listed companies is vital to its stakeholders, so studying the listed companies' financial situation is important for keeping the securities market robust, moreover, one can expend the research findings to guide other forms of enterprises on its operation and risk management.However, many researches are found impracticable due to unsolved technical problems and unsatisfactory long- term prediction effects. Among the problems of existing models, the problem of selecting of prediction parameters and the problem of multicollinearity have never being solved effectively. To solve these problems needs a break-through of the current prediction methods.Under this background, the paper reviews the empirical researches done by others, then selects 33 listed firms whom have been special treated (ST) in the year of 2004 as research samples, and according to the same industry and same business size matching principle, an equal amount of financially healthy firms are chosen as the control samples. Base on sample firms' available financial data, research time span is decided and needful sequential data is gathered. Finally, the author establishes a new financial crisis early prediction model for listed companies using principle component analysis method, discriminate analysis method and the SPSS software.The main contributions of this paper are:(1) The financial crisis prediction model established by the author is based on sequential data;hence its sample data contains more information of financial trend of the firms.(2) Using principle component analysis to solve the problem of multicollinearity among parameters, which greatly improves the quality of prediction.(3) The research result is practicable;it provides a prediction signal two years before a company incurs continuous losses with an accurate rate of 80.3%. The model this thesis developed will serve as a useful management tool and save management's time on the decision-making.