台灣企業集團財務預警模式之研究

Abstract

隨著產業結構轉型，企業集團經濟影響力也日益增加。近年來數起集團財務失敗案例，造成社會重大衝擊，如何於事先發現集團財務惡化，亡羊補牢，是對經營者、投資者與管理當局都是相當重要的。為此，本研究針對企業集團財務預警模型進行研究，希望達成下列三目的：1.建立集團財務預警模型，並分析其預測能力。2.企業集團預警模式與一般預警模式分類上優劣。3.了解企業集團財務危機過程，以及預測能力最佳的變數。 本研究選取台灣地區51個集團（包括：38個財務正常集團、13個財務危機集團）作為研究樣本，以羅吉斯迴歸建立財務預警模型，而模型的變數是以：集團的14項財務比率進行主成份分析後所翠取的主成份。本研究建構兩個預警模型，第一個模型以16個集團（8個財務正常、8個財務危機集團）財務比率資料，建構發生危機前一年至前三年之財務危機預警模式，並在建立後進行：模式內樣本分類準確測試、新樣本（30個財務正常集團、5個財務危機集團）預測準確測試、分類錯誤樣本後續分析、三年度最具判別能力之主成份比較；第二個模式以26個集團（13個財務正常集團、13個財務危機集團）建立財務預警模式，並在建構後進行交叉效度研究，以補模式一新樣本預測準確測試中財務危機樣本過少的缺點。 本研究經過實證模型建立後發現：1.模式分類能力顯著，且具有績效評估與財務預測效果。2. 對集團企業而言，只以單一企業預測該公司發生財務危機與否，並不恰當，必須考慮集團企業相互支援的因素，同時對個別企業與集團作財務危機預測後再作判斷。3.企業集團財務危機發生過程，與賴世權所提出的「企業失敗三段論」頗為一致。

As the structure of industry changes, the influence of business conglomerates is increasing. Nevertheless, some financial failure cases of business conglomerates shocked the community in the last few years. Therefore, it is an important issue for managers, investors, and supervisors to detect whether the business conglomerate’s financial status is deteriorated. This study focuses to business conglomerate’s financial alert model and hopes to achieve the three goals: 1.to construct the business conglomerate’s financial alert model, and to analyze it’s forecast ability. 2.to know the difference between business conglomerate’s financial alert model and business financial alert model. 3.to know the procedure of business conglomerates’ financial distress and the variables that have the best forecast ability. The study selects 51 business conglomerates as studied samples, included 13 financial distress conglomerates and 38 normal conglomerates. The study adopts logistic regression to construct financial alert model. Before constructing model, the study performs principal component analysis to 14 financial ratios of business conglomerates. The extracted principal components are the variables of logistic regression model. The study constructs two financial alert models. The first model was constructed by financial ratios of 16 studied samples, 8 of which are financial distress. The second model was constructed by financial ratios of 26 business conglomerates, 13 of which are financial distress. Each of the two models has three sub-models, which predict the business conglomerates’ financial status on 1-3 years prior to the conglomerates’ financial distress. After the two models were constructed, the study performed validation test, investigated why some samples was usually classified wrong, and compared the three sub-models to choose the principal components that have better forecast ability. There are 3 conclusions the study made: 1. the classification ability of model is significant, and the model help people to evaluate performance. 2. For an enterprise of business conglomerate, it’s financial status is not all depend on itself, but is influenced by all other enterprises of the business conglomerate. Therefore, when someone wants to forecast business conglomerates’ financial status, he must forecast single enterprise and conglomerates’ financial status. 3. the procedure of conglomerate’s financial distress conforms to “the three steps of business failure”, which was delivered by Lai.