以公司治理因素建構企業財務預警模型-運用模糊判別分析法

近年來，企業發生財務危機的例子層出不窮，如：美國的安隆(Enron)、美林證券(Merrill Lynch)、雷曼兄弟(Lehman Brothers Holdings Inc.)與台灣的博達、訊碟、力霸等公司。而傳統的分類方法僅將抽樣之企業以「非好即壊」的二分法作區分，但一個企業發生危機之徵兆通常是漸近地由好至壊或是被人為隱藏，要解讀這當中隱含之資訊，採用傳統的二分法判別表達似乎不甚合理。模糊理論(Fuzzy Logic Theorem)中的隸屬函數(Membership Function, MF)，除了可用來表達不確定的問題分析，且更能表達某些事件漸近性的變化程度與變化強弱。故較傳統的二分判別方法而言，模糊隸屬函數可取代非好即壊的不合理性，作出危機事件漸近變化程度的表示；因此，一個整合傳統判別分析與模糊隸屬函數之優點的模糊判別分析模型，作為本研究建構企業財務預警模型之依據，此模型不但保留傳統判別分析的區別特質，也更加合理表述企業發生危機的變化情況。如前所述，本研究整合常見之財務變數並加入公司治理變數，先以因素分析處理，再以模糊判別分析方法建構企業財務危機模型，實證對象為上市公司（不含金融業）發生變更交易（即全額交割），判斷其公司是否列入財務危機公司，提供投資人和相關利益人避免誤觸地雷公司而產生嚴重的損失。經由本研究之實證結果得知，可有效提供投資人和債權人判別公司是否即將發生財務危機之參考。

關鍵字

Fisher判別方程式；參數化模糊隸屬函數；危機預警模型；公司治理；模糊理論；財務危機；兩群判別分析；模糊判別分析

並列摘要

In recent years, financial crises example very many, such as: the United States Enron, Merrill Lynch, Lehman Brothers Holdings Inc. and Taiwan Bardon, Infodisc, Rebar etc. The traditional classification methods only sample of companies with “non-good or bad” dichotomy to distinguish, but the business crisis is usually asymptotically by the good to bad, or is artificially hidden, which implied to interpret this of the information, determine the expression of the traditional dichotomy seems not reasonable. The Membership Function, MF is fuzzy Logic Theorem, in addition to can be used to express the uncertainty analysis, and better able to express some of the asymptotic behavior of the degree of change level. Therefore, two points distinguish the more traditional method, the fuzzy membership function can replace the good that is non-bad unreasonable, to the degree of change crisis asymptotic representation; Thus, a discriminant analysis and integration of the traditional advantages of fuzzy membership function of fuzzy discriminant analysis model, constructed as the study of corporate financial early-warning model based on this model not only to retain the traditional characteristics of the difference between discriminant analysis, and more reasonable representation of changes in corporate crisis situations. This study integrated common financial variables and joined the corporate governance variables, the first order faction analysis and processing, and then to construct fuzzy discriminant analysis model of corporate financial crisis empirical objects of listed companies a change in trading to determine whether the company included in the financial crisis of its companies to provide investors and stakeholders who avoid inadvertently mine company and cause serious damage. By the empirical results of this study that can effectively provide investors and creditors determine the company’s impending financial crisis reference.

並列關鍵字

Fisher’s Discrimination Function ； Parametric Membership Functions ； Distress Model ； Corporate Governance ； Fuzzy Logic Theorem ； Financial Distress ； Two-group Discriminant ； Fuzzy Discriminant Analysis