應用邏吉斯迴歸技術探討財務危機預警變數與資料長度之適用性研究--以台灣上市電子產業為例

Applicable Study on Taiwan Electronic Industry Financial Crisis Predictive Model Based on Logistic Regression

陳義方(Yi-Fang Chen)

指導教授 ： 羅淑娟

國立臺北科技大學/管理學院/工業工程與管理研究所/碩士(2009年)

摘要

近幾年間發生多起地雷股事件，讓社會大眾措手不及遭受重挫，其中頻傳財務危機的公司多為一般投資大眾所注意的電子產業，為保障廣大投資大眾及債權人，電子業之企業財務預警系統之機制確實有研究之必要性。財務預警模型多以年資料為主，但年資訊通常要等年末或次年初才可獲得，對於預警時程可能緩不濟急，本研究探討導入季資料的可行性，經實證結果得知，季資料表現優於年資料，尤其在危機發生前一年度；對配對比率而言，國內研究多以假設帶過，並未進行實際實驗，本研究將樣本分成對（1：1）與非成對樣本（1：3），經實證分析結果，很明顯的成對樣本優於非成對樣本；在變數方面，財務危機預警模型皆以財務變數為主，但是仍無法有效預警公司營運問題，本研究除財務變數外，將囊括非財務變數（公司治理變數與信用評等）於預警模式中，試著尋找出最適合危機預警模式之變數，經實證分析發現在財務變數模式中再導入公司治理與信用評等所建構之財務預警模型危機前一年的分類正確率可提升至90.9%，從模式中得知，負債比率、流量比率、總資產週轉率、每股盈餘、董監事質押比率與TCRI信用評等為危機公司評估的先期指標，而總資產成長率為後期指標。本研究最佳模型為變數包含財務變數、治理變數、信用評等並以成對的季資料建構之財務預警模型，提供給投資大眾與企業經理人作為檢視投資標的物與本身企業之工具，以減少投資的風險。

關鍵字

財務與非財務指標 ； 資料長度 ； 配對比率 ； 預警模型 ； 邏吉斯迴歸

並列摘要

Recently, the occurrence of the landmine share contributed to great loss of public investors and most of them were the electronic industries which were the focus of investing. To protect the rights and interests of investors and creditors, it is necessary to construct a financial alert system. Most researches used the data of year, but it was too late to get the data in the end of the year or the beginning of next year. In this study, we introduced the data of quarters to the alert model. The result showed that the data of quarter was better than data of year, especially one year before crisis. To matching principle, most domestic researches only assumed the matching principle without experimenting it. This study categorized the data to pair (1:1) and non-pair (1:3). The result showed that obviously the pair data were better than non-pair data. To variables selection, most researches used financial balances to construct alert model but the result did not satisfying. This study added non-financial variables(governance variables and credit rating) to derive the most suitable model. The result showed that the accurate rate of the model adding governance variables and credit rating increased to 90.9%. The result showed that debt ratios、EPS and Total asset turnover ratio are all prior index. Among governance variables and Credit rating, Debt ratio、Flow rate、Total asset turnover ratio、EPS、Pledge ratio of directors and credit rating of TCRI were prior index discriminating normal and crisis. The posterior index were Growth rate of total assets. The best model in this study used financial balances, governance variables and credit rating of quarter. Hopefully, we could provide the public of investors and managers a tool to examine enterprises and lower the investing risk.

並列關鍵字

financial and non-financial index ； data span ； matching method ； crisis alert model ； Logistic regression

參考文獻

[6] 許溪南、歐陽豪、陳慶芳，「盈餘管理、公司治理與財務預警模型之建構」，金融風險管理季刊，第3卷，第3期，第1-40頁，2007。

[2] A. Agresti, An Introduction to Categorical Data Analysis. New York:Wiley Series in Probability and Statistics, 1996.

[3] J. Argenti, Corporate Collapse: The Causes and Symptoms. London: McGraw- Hill, 1976.

[7] E.I. Altman, “Financial Ratios, Discriminant Analysis, and the Prediction of Corporate Bankrupty.” Journal of Finance, Vol. 23, no. 4, 1968, pp. 589-609.

[8] F.A. Amir, “Bankruptcy Prediction for Credit Risk Using Neural Networks :A Survey and New Results,” IEEE Transaction JNL on Neural Networks, Vol. 12, no.4, 2001, pp. 929-935.