世界各國企業破產事件頻傳,財務危機公司預警模型的相關研究,發現愈來愈強調公司治理的重要性,雖然已有學者將公司治理變數納入危機預警模型,但卻忽略公司治理中的「資訊透明度之揭露」。本研究主要的研究目的在於驗證資訊揭露透明度對危機預警模型的重要性,即探討公司治理變數在考量資訊揭露透明度後,是否顯著提升預警模型的準確性;以台灣上市櫃公司財務比率、資訊透明度之揭露、公司治理、總體經濟等變數,分別以羅吉斯法(Logistic regression model, Logit )、多層感知器類神經網路法(Multilayer Perceptron Neural-Network, MLPNN)、支向機(Support Vector Machine, SVM)及基因支向機(Genetic Algorithms-Support Vector Machine , GA-SVM)四種方法來建構財務危機公司預警模型,期望找出最適合台灣上市櫃公司的財務危機預警模型。 本研究將樣本區分以1:2的配對方式,為訓練與測試樣本。以89年至93年之樣本作為訓練樣本,建立預測模型; 以94年至96年之樣本作為測試樣本,來檢驗測試樣本預測的準確度與模型的穩定度。 研究結果發現(1)納入資訊透明度變數之揭露,對危機預警模型的重要性可獲得驗證;(2)比較四種研究方法所建立的財務危機預警模型中,基因支向機(GA-SVM)的預測準確性優於支向機(SVM)、多層感知器類神經網路法(MLPNN)及羅吉斯模型(Logit)。
A lot of bankruptcy incidents have happened. The related research that financial distress company model emphasized more and more company governs importance. Although the scholar add govern variables to the distress company model, but actually neglects the company to govern “disclosure information”. This main research goal is to lies in the importance disclosure information model to the failure company. The discussion that added governs variable after the consideration disclosure information, whether remarkable promotion early warning model accuracy. The aim of this study is try to use Logistic regression model (Logit), Multilayer Perceptron Neural-Network (MLPNN), Support Service Machine (SVM) and Genetic Algorithm-Support Service Machine (GA-SVM) to establish financial distress predictable model. For building up Taiwan bankruptcy prediction model, the data set is arbitrarily split into two subsets: about 2/3 of the data is used for a training set (2000 to 2004) and 1/3 for a validation set (from 2005 to 2007), comparing the prediction accuracy with the firms from 2000 to 2006 dataset. The empirical findings from the research shows that (1) added the variable of disclosure information, financial distress predictable model is importance to be possible to obtain the confirmation. (2) The predictive accuracy of the GA-SVM model is better than SVM, MLPNN and Logit models.