以財務比率衡量企業財務危機預警模型之實證研究

A Study on Measuring Corporate Financial Crisis Prediction Models by Financial Ratios

蔡長裕(Chang-Yu Tsai)

指導教授 ： 梁世安 教授

國立臺北大學/商學院/國際財務金融碩士在職專班/碩士(2010年)

摘要

有鑒於近年國內企業、集團因受經濟不景氣及金融風暴衝擊，接二連三爆發無預警之財務危機、破產倒閉，除導致員工、投資人、金融機構、游廠商蒙受衝擊與損失外，甚有可能引發本土型金融風暴。
故為了減降或消除前述信用違約風險之發生機率;擬採用國內上市、櫃公司2003至2007年發生財務危機之企業36家，依據1:2之隨機、不區分產業，但考量總資產規模之方式進行配對，共篩選出108家企業作為研究樣本，並挑選出國內外相關文獻及銀行授信實務上常用之16項財務比率相關數據及TCRI評等作為研究變數；在運用多項變數區別能力分析方式篩選出可能不具區別力變數及採用多元迴歸、逐步迴歸過程將顯著變數予以保留;最後再納入現金流量變數、TCRI評等虛擬變數，分別建立四個邏輯斯迴歸模型;進行企業財務危機預警模型之實證研究。
經實證結果本研究之最佳模型，包括未分配盈餘/總資產、營收淨額/總資產、負債/總資產、每股盈餘、(現金&約當現金+短投)/流動資產、資產總額等六項變數;其中(現金&約當現金+短投)/流動資產、資產總額等兩變數雖在過去文獻中顯少納入探討或結果為不顯著，惟經本研究實證納入前兩項變數後對邏輯斯迴歸模型之預測正確率具顯著提升效果;另值得一提的是銀行授信實務上相當重視之現金流量變數(息前稅前折攤前淨利/總負債)與經濟新報TCRI評等虛擬變數,在此模型中均呈現不顯著結果。
其次，就模型預期準確率而言，在Cutoff Point 0.5下；本研究所建構之四個邏輯斯迴歸模型(即納入現金流量變數、TCRI評等虛擬變數與否)中係以模型一【即包括未分配盈餘/總資產、營收淨額/總資產、負債/總資產、每股盈餘、(現金&約當現金+短投)/流動資產、資產總額等六項變數】之預期準確率達92.59%為最高；且銀行業所關心之型ㄧ誤差(Type I Error)亦為最低，僅5.56%。
最後，為驗證最佳模型之效度與實用性，進一步利用2008年發生財務危機之上市櫃企業做為Outsample，進行樣本回測；結果顯示，前一至三年之模型正確預測率均達93.33%，預測能力不錯，且其中金融機構最為關心之型ㄧ誤差(10%)，尚在可接受範圍;故整體而言，本研究本所建立之『企業財務危機預測模型』應可供國內金融業做為授信前決策或貸後管理之參考。

關鍵字

企業財務危機預警 ； 邏輯斯迴歸模型

並列摘要

In this study, 36 domestic publicly listed & OTC’s listed companies which suffered from financial crisis in 2003 to 2007 are selected. We do the matching comparison by the total assets of the companies with these randomized, 1:2, non-discrimination industries. A total of 108 companies are selected as the research sample, and 16 financial ratio variables, TCRI related literatures which bank’s credit practice common-used as research variables.
With applying several variables discrimination analyses, variables are screened out for non-discriminated variables. We also apply Multiple Regression Model & Stepwise Procedure which retains significant variables, add up the cash flow variables and TCRI dummy variables to build up four Logistic Regression Models; and do the empirical analysis on corporate financial crisis prediction models.
From the empirical study, the best fit model includes the following six variables: Inappropriate Retained Earnings / Total Assets , Net Sales / Total Assets, Total Liabilities / Total Assets,Earnings Per Share, (Cash & Cash Equivalent + Short–term Investment ) / Total Current Assets and Total Assets.Although variables ( Cash&Cash Equivalent + Short – term Investment ) / Total Current Assets and Total Assets are seldom included in previous literature, the accuracy of predicting result improved significantly by adding these two variables in Logistic Regression Model. A worth noted is that the bank’s credit practice commonly used which often emphasized more in cash flow variables ( EBITDA / Total Debt ) and the TCRI rating of Taiwan Economic Journal shows insignificant in this model.
Secondly, the accuracy of this model is expected to run under the cutoff point 0.5. The 92.59% accuracy rate of Logistic Regression Model, which contains previous 6 variables, is obviously superior to the other three models (models that including EBITDA / Total Debt and TCRI dummy variables) of 91.67% accuracy rate. As for the Type I Error, which most banks concerned seriously, it’s the lowest-- 5.56% to this model.
Finally, we use publicly listed & OTC’s listed companies that suffered financial crisis in 2008 in order to re-verify the validity and practicality of the model. The results show that the model prediction rate risen up to 93.33% by means of appling previous three years data. It is a good estimation, and the Type I Error (10%), which most banks seriously concern, is still in an acceptable range.
In conclusion, this proposed model of predicting financial crisis to an industry could be a good reference to the domestic financial sector in decision for loan granting or post-loan management.

並列關鍵字

Logistic Regression Model

參考文獻

16.黃振豐、呂紹強（2000），「企業財務危機預警模式之研究-以財務及非財務因素構建」，當代會計，第一卷，第一期，19-40頁。

21.鄭文英、李勝榮、葉憲弘（2006），「台灣上市上櫃電子公司經營財務階段判別模式之建立」，風險管理學報，第八卷，第一期，71-96頁。

24.盧俊安（2005），「我國上市櫃公司財務危機預警機率模型之建構與驗證」，崑山科技大學企業管理研究所碩士論文。

25.謝劍平（2006），「財務報表分析Financial Statement Analysis」，智勝出版社。

1. Altman, Edward I.（1968）, ”Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy”, The Journal of Finance, Vol.XXIII, No.4, pp.589-609.