Comparative Study On The Listed Company Financial Fraud Identification Model In Our Country

Financial fraud of the listed company is a universal phenomenon, which hinders the open, equity and justice of the development of securities business, and becomes an obstacle of the capital market in each country. Under this circumstance, how to detect the financial fraud so as to clean the honesty environment of capital market becomes a serious problem. Therefore, how to build a valid detecting model of financial fraud is the emphasis of this thesis.Above all, through the domestic and foreign literature review, the thesis starts from detecting models of financial fraud, this step provides the foundation and reference for the design of detecting model; Second, the thesis identifies the definition of financial fraud and analyzes economic theories of it; Third, the intention, means and indications of financial fraud are analyzed, and the reasons for which technological methods should be selected are explained; Fourth,120sets companies are selected as samples, which composed of the financial fraud and related non-financial fraud companies from the year1998to2009. Meanwhile, this research chooses related detecting ratios and constructs four types of detecting models, including Logistic model, BP, PNN and Elman neutral network model. Finally, the thesis employs the Expected Misclassification Cost Method to compare these models, and gets the result that the accuracy of Elman neutral network model is the highest. On the basis of the above research, the thesis summarizes the conclusion and shortage.That a new method-Elman neutral network is adopted to detect the financial fraud is the innovation of the thesis. Elman neutral network has strong retrospect of historical information, dynamic modeling, this character adds great help to enhance the accuracy of detecting model of financial fraud and to provide prewarning for the regulator.