Summary: | 碩士 === 國立中正大學 === 會計所 === 94 === This research is trying to propose a diagnostic model to help the accountants discern whether a particular company will be sued or not in the future based on the financial information and corporate governance mechanism. Data set consists of the 62 sued companies filed by the Securities and Futures Bureau as well as Investors Protection Center, and the other 124 non-sued companies of similar scale were collected in the same industry. Based on the financial and corporate governance factors, the litigation warning model is constructed by using logit and neural network respectively. The empirical results of logit show that profitability is negatively correlated with the sued probability significantly, and debt positively correlated with the sued probability significantly. On the other hand, for corporate governance factors, shareholding,and the number of directors and supervisors are negatively correlated with the sued probability significantly。 The proposed neural network litigation warning model can really improve the forecasting accuracy of the logit model. Besides, the 2-stage neural network can improve the original one. Furthermore, the warning signal is increasing stronger when the time approaches the breaking out point no matter for logit or neural network. The proposed model can help the related people take some actions before the litigation happened, and help the accountants evaluate the customer’s litigation risk。
|