判斷公司違約風險的大小是風險管理重要的一環，管理者利用各種模式評估公司的信用狀況。本研究以統計模式構建組合預測，期望能簡化管理者的決策過程。本文先分別以逐步迴歸法及因素分析法選取財務比率變數並應用Logit模型建構危機發生前一年及前二年之財務危機預警模式，除傳統財務比率變數外，另外再加入選擇權評價法的違約機率值，檢視模型之正確區別率是否能顯著提高。 根據實證結果，在危機發生前一年，以信用風險模型的正確區別率最低，而在財務變數為預測變數方面，不管是危機發生前一年或前兩年，利用逐步迴歸法選取財務變數的預測正確率優於採用因素分析法，但不論選取財務變數的使用方法為何，若與信用風險模型的違約機率合併做組合預測，皆能使正確歸類率改善，說明組合預測能改善單一指標對公司的違約預測。 It is an important part of risk management to assess company''s credit risk. The administrator use various kinds of ways to assess the credit condition of the company, but different ways may produce different results. This paper structures a portfolio by statistical method, expect to be able to simplify the administrator''s decision-making process. First separately with stepwise regression analysis and factor analysis choose financial rate parameter and use Logit model to construct the financial distress model in the previous year and the first two years take place in the crisis. Besides traditional financial rate parameter, accede to the default probability value of the option pricing method in addition, to observe whether these variables make the construction of financial distress model better accordingly. According to the reserch result , take place in the previous year in the finance distress, the correct difference rate of the credit risks model is the lowest. And the distress takes place in the previous year or the past two years, the correct rate of prediction of utilizing stepwise regression analysis to choose the financial parameter is superior to adopting the factor analysis. No matter with which kind of method chooses the financial parameter, if combine with the default probability value, we can get the better correct rate .
