傳統財務危機預警模型的發展基礎係建構於財務數字之上,但在進行危機預警模型建構前,必須注意財務報表資訊是否經人為操縱及控制與操縱的影響性。由於不實財務報表可能無法充分反應該公司的實質經濟體質,也很可能會對危機之預測產生偏誤結果,故本文研究目的即在於將盈餘操縱程度變數納入財務危機預警模型建構之中。本文採取Jones模式與Modified Jones模式進行不可裁決性應計項目的迴歸估計,並以pannel data並分產業別方式進行裁決性應計項目之迴歸估計;除此之外,亦計算各樣本歷年總應計項目以及裁決性應計項目之標準差作為盈餘管理程度變數加入危機預警模型中,並衡量各模型之效度。實證結果發現,以裁決性應計項目來看,不論是在Jones模式或是Modified Jones模式之下,危機公司之盈餘操縱程度皆明顯大於正常公司,證實了危機公司與正常公司盈餘操縱之程度有顯著差異;而由模型效度結果顯示,不論是KS值或是ROC值,加入本文所設計之盈餘操縱程度變數,其對純財務績效模型之效度確實有提升的效果。
In traditional studies, the financial warning models were constructed on the basis of financial variables, but the windows dressing effects on financial statement always cause prediction accuracy bias on financial warning model. The purpose of this study is to establish the financial warning model considering the variables about earnings management. We use the Jones Model and the Modified Jones Model to estimate the non-discretionary accruals of each sample first and then calculate the standard error of the total accruals and the discretionary accruals for each sample respectively to evaluate the degree of earnings manipulation. We find the failure firms indeed have larger total accruals and discretionary accruals than non-failure firms. In conclusion, the financial warning models of adding earnings management variables are better than pure financial variable's model in the prediction of financial distress.