財務危機預警模式之探討－以台灣上市上櫃電子公司為例

A Study on Prediction Model of Financial Distress-Taking Firms — Listed and OTC Electronic Companies in Taiwan as An Example

吳凌霄(Ling-hsiao Wu)

指導教授 ： 李賢哲

大同大學/經營設計學院/事業經營學系所/碩士(2005年)

摘要

本研究將研究範圍界定於台灣電子產業，藉以排除因產業特性的不同所造成的差異影響。在研究上，係使用國內上市上櫃企業之公開財務資料，以近期所發生上市上櫃電子地雷股作為財務危機企業研究之樣本企業，並挑選對照研究的正常企業樣本，研究期間涵蓋自民國90年至92年樣本企業在財務危機發生前三年的財務屬性。
研究流程區分為兩階段。首先，進行財務與非財務變數選擇，以無母數Mann-Whitney U檢定及多變量變異數分析進行導入變數選擇。第二階段利用主成份分析方式及逐步迴歸方式在處理時間系列之影響時，先以Durbin-Watson法檢驗自我迴歸，以Cochrane-Orcutt法進行變數轉換，然後進行Logistic迴歸分析。
本研究得到以下結論﹕
1.財務危機企業與正常企業兩組樣本群的實證結果顯示，在財務危機發生前一至三年期間，其財務與非財務變數的平均值已呈現顯著的差異。而且越接近危機發生點，正常與危機企業所呈現的差異越大。
2.以多變量變異數分析法中的多重比較法比較各群組間財務與非財務比率優劣之差異性。實證結果發現，現金流量比率、常續性利益成長率、稅前淨利成長率、營收成長率、淨值報酬率、營業利益率、稅前淨利率、資產報酬率、總負債/總淨值、借款依存度等10項財務變數與企業經營績效存有顯著的相關性。另TCRI評等是由台灣經濟新報所建立之TCRI〔Taiwan Corporate Credit Risk Index;台灣企業信用風險指標〕外部評等指標，屬於獨立客觀評等機構所建構出信用風險評等指標，亦具有一定之客觀性，故對於企業危機發生具有一定的參考價值。
3.發生危機通常不是一夕之間發生的，而是經過一段時間財務惡化的結果。財務比率本身往往也存在一定的趨勢，利用財務與非財務變數建立迴歸模式之前，應先將財務比率隨時間而變化的自我相關排除掉，再來建立迴歸模式較為合理。本研究以危機發生前三個年度個別年度進行迴歸模式診斷，藉由Durbin-Watson值檢定模型是否存在自我相關(Autocorrelation)問題，將自我相關干擾消除後，發現所建構的二個不同模式，無論是逐步迴歸或是主成分分析法後導入逐步迴歸所建立的Logistic預警模式整體區別能力皆達94.5％，型Ι誤差︰即實際為財務危機企業卻被歸類為財務正常企業皆為5.3﹪。

關鍵字

財務危機 ； 財務危機預警模式

並列摘要

This study will confine the scope to Taiwan’s electronics industry to eliminate any variations that may be caused by the characteristics of other industries. Financial information disclosed by companies listed in either the TSE or the Gretai Stock Exchange will be used in this study, and the samples are the companies affected by land mine stocks. Samples have also been taken from companies with normal operations as a control group for comparison. The sample period covers 2001, 2002, and 2003 in order to find out the financial attributes of these enterprises before the financial crisis.
　　The research process is divided into two phases. In phase I, financial and non-financial variables will be selected and the Mann-Whitney U test and multivariate analysis will be used in the selection process. In phase II, principle component analysis and step-wise regression will be used in handling the time-series effect. The auto-regression Durbin-Watson method will be used and the Cochrane-Orcutt method of the transformation of variables will also be adopted. Logistic regression analysis will be conducted afterwards.
The empirical findings described in the following:
1. The two groups of samples of enterprises in crisis and enterprises in good standing indicated that, in the period of three years before the outbreak of crisis, the mean values of the financial and non-financial variables indicated a significant variation. The closer in time to the outbreak of crisis, the wider the variation between enterprises in crisis and enterprises in good standing.
2. The method of MANOVA is applied for finding out the strength of the financial and non-financial ratios. The empirical result indicated that the 10 financial variables of cash flow ratio, continuous profit growth rate, EBT growth rate, revenue growth rate, return on net worth rate, operating income rate, EBT, ROI, total liabilities/total assets, and loan dependency are significantly correlated with the performance of enterprises. The TCRI index is an external indicator for independent and objective risk rating. Therefore, this indicator is also essential in the analysis.
3. Enterprises do not get into crisis overnight. Indeed, crisis is the result of prolonged financial malaise. Financial ratio in itself exhibits certain patterns. Before using the financial and non-financial variables for building the regression model, the autocorrelation of the financial ratios over time must be removed for constructing a reasonable regression model. In this study, a diagnosis has been conducted on the regression model covering each of the three years before the outbreak of crisis. The Durbin-Watson test is applied for testing autocorrelation. Two models were built up. The Logistic early warning model has the overall strength of 94.5% under both the principal components analysis and the partial regression method. The result also showed type I error: enterprises in crisis are misclassified as enterprises in good standing and is at 5.3%.

並列關鍵字

Financial Crisis ； Financial Crisis Early Warning Model

參考文獻

34. Ward, Terry J. and Benjamin P. Foster(1997), “Using Cash Flow Trend to Identify Risks of Bankruptcy”,The CPA Journal ,Sep.1997, pp.60~61.

1. Abrahamson, E. and Amir, E.(1996), “The Information Content of President’s Letter to Shareholders”, Journal of Business Finance and Accounting, Vol.23, No.8, pp.157.

2. Altman (1968), E.I. "Financial Ratios, Discriminant Analysis, and the Prediction of Corporate Bankruptcy," Journal of Finance 22, Vol.23, pp.589~609.

3. Beaver, W.H. (1966), "Financial Ratios as Predictors of Failure," Journal of Accounting Research Supplement, pp.71~111.

4. Bum. M (1974), “Failing Company Discriminant Analysis”, Journal of Accounting Research, Vol.12,pp.1~25.