考慮偏誤修正後之財務預警模式：以台灣上市電子業爲例

Financial Warning Model under Bias Correction: The Taiwanese Listed Electrical Companies as Example

周賢榮(Shyan-Rong Chou) ； 鄭文英(Wen-Ying Cheng) ； 菅瑞昌(Andy Chien) ； 吳千慧(Chien-Hui Wu)

《美和技術學院學報》 29卷1期 (2010/03) Pp. 209-225

https://doi.org/10.7067/JMHIT.201003.0209

摘要

從Beaver(1966)與Altman(1965)提出企業財務預警模式後，便逐漸引起後續學者對此議題廣泛討論與研究。在歷經40年的研究中，雖然財務預警的分析能力有長足進步，但仍然存在些許偏誤問題尚未解決，例如在樣本選取、變數分類及財務預警模型的選擇，因此本文提出解決財務預警偏誤的方法。此外，在修正偏誤的基礎下，本文以台灣上市電子業公司爲研究對象，估算發生財務危機的風險機率，並提供投資者在選股時用來判別公司財務經營狀況的參考指標。
實證結果發現，在修正偏誤下偵測台灣上市電子業公司財務經營階段之整體判別率達93%，且負債佔資產比率及每股盈餘具有最適判別能力。在估算相對風險方面，當負債佔資產比率減少1%時，發生財務危機的風險可減少33%，而每股盈餘減少1%時，則其發生財務危機的風險會增加6.98倍。此外，將負債佔資產比率及每股盈餘財務比率值代入羅吉斯迴歸模式中，若大於－0.539，則判定該公司爲經營正常公司；若計算之值介於－0.539至－12.059問，則判定該公司爲輕度危機公司；若計算後之值小於－12.059，則判定該公司爲重度危機公司。

關鍵字

財務預警模型 ； 偏誤修正 ； 次序羅吉斯迴歸模型

並列摘要

After the pioneering study by Beaver (1966) and Altman (1968) proposed corporate financial warning models, many scholars have completed empirical research on this topic over the last four decades. Even though the discrimination of financial warning is more accurate, existing biases, such as sample selection, variable classification and traditional model selection, still present problems. This study proposes an approach to overcome the problems listed above. In addition, we examine the financial administration states of Taiwanese listed electrical companies under the corrected biases. Further, we calculate the risk ratio and threshold value of financial administration states.
The empirical results show that on the trichotomous classification test, the indices of debt ratio and earnings per share (EPS) have significant differences between financial administration stages. The correct classified rate of all companies is 93%. This paper measures the relative risk in the financial stages and financial ratios. The debt ratio decrease to 1% lead to financial distress risk decreasing 33%, the EPS decrease of 1% lead to financial distress risk increasing 6.98 times. In addition, this paper substituted debt ratio and EPS into the ordered logistic regression model, we classify that the companies are in a normal state of operation when the threshold value is greater than -0.539. The classification of a slight level of crisis is given when the threshold value is between -0.539 and -12.059. The classification that companies are in a heavy degree of crises is given when the threshold value is less than -12.059.

並列關鍵字

Financial warning model ； Bias correction ； Ordered logistic regression model

參考文獻

王濟川、郭志剛(1995)。Logistic迴歸模型─方法及應用。台北:五南。

陳肇榮(1983)。運用財務比率預測企業財務危機之實證研究(博士論文)。國立政治大學企業管理研究所。

郭晉源(2002)。整合灰色預測與鑑別分析於企業財務危機預警模式建構之研究。中國統計通訊。13(2)，20-31。

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。8(1)，71-96。

Alexander, C. O.,Leigh, C. T.(1997).On the Covariance Metrices Used in Value at Risk Models.The Journal of Derivatives.4,50-62.