應用邏吉斯與資料包絡法建構財務預警模型－以台灣電子業為例

Using Logit and Data Envelopment Analysis to Construct the Financial Alert Model－A Case of Electronic Company in Taiwan

葉儒謙(Ru-Chien Yeh)

指導教授 ： 張文華

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

由於企業財務危機的消息時有所聞，造成投資者及資金往來的銀行龐大的損失，因此財務預警模型的建構是一個非常重要的課題。過去的學者多使用財務比率來建構財務預警模型，而忽略了經營績效對於企業財務危機發生的影響，因此本研究以財務比率變數加上資料包絡法所求得之相對效率值來建構財務預警模型，以期增加預警模型的正確率，以證明財務危機與經營效率之間的關係。
本研究以2002到2006年之台灣上市上櫃電子公司為樣本，根據台灣證券交易所對危機之定義，以一比一之配對方式總計64家公司，分為訓練以及測試樣本，接著再以Logit建構財務預警模型。一共建立四種模型分別為model 1：財務比率變數、model 2： 財務比率變數+技術效率、model 3：財務比率變數+純技術效率及model 4：財務比率變數+規模變數。
研究結果顯示不管危機前一年、前兩年或前三年的財務資料，負債比率及資產報酬率皆為重要的區別正常危機公司的指標，測試正確率分別為83.33%、77.08%及73.61%。離危機發生越近正確率越高。另外財務比率變數加入規模效率的model 4更能準確的區別為機公司與正常公司；加入技術效率的Model 3及加入純技術效率的Model 4皆能降低影響較嚴重的型二誤差。最後本研究發現，危機發生前三年開始，危機公司的固定資產有逐年降低的趨勢，因此固定資產應視為一項危機發生的參考指標。

關鍵字

財務危機 ； 財務比率 ； 經營效率 ； 邏吉斯 ； 資料包絡法

並列摘要

Owing to the frequent hearing of financial distress which may contribute to great loss of investors and banks, it is an important issue to construct a financial alert system. Most researchers used financial balance variables to construct financial alert model, but few of them considered the relation between operating efficiencies and financial distress. For that reason, this study both made use of financial balance variables and operating efficiencies to construct financial alert model and compared the accurate rates with the model constructed only by financial balance variables. If the accurate rates increased, it would tell that operating efficiency is also one of the important reasons which cause financial distress.
There were 64 companies selected as the samples which were separated into training and testing samples. The matching method is 1:1 and the period is from 2002 to 2006. Four models were constructed that are model 1: financial balance variables, model 2: financial balance variables combined with technical efficiency, model 3: financial balance variables combined with pure technical efficiency and model 4: financial balance variables combined with scale efficiency respectively.
The study showed no matter what years before distress, debt ratio and rate of return on assets are important variables discriminating normal and distressed companies. The overall accurate rates were 83.33%, 77.08% and 73.61% respectively. It also showed the closer to the time that distress happened, the higher accurate rates were. Comparing to model 1, the model 4 which added scale efficiency had the higher accurate rate discriminating normal and distressed companies in all years; the model 3 and model 4 diminished the type II error rate comparing to model 1 in all years. Furthermore, contrary to normal companies, fixed assets decreased from three years before distress and should be considered as an important sign discriminating the happening of distress.

並列關鍵字

Financial Distress ； Financial Balance ； Operating Efficiency ； Logit ； Data Envelopment ； DEA

參考文獻

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

[22] E. I. Altman and B. Loris, “A Financial Early Warning System for Over-the-counter Broker-dealers,” The Journal of finance, vol. 31, no. 4, 1976, pp. 1201-1218.

[23] R. D. Banker, A. Charnes, and W. W. Cooper, “Some Model for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis,” Management Science, vol. 30(9), 1984, pp. 1078-1092.

[24] R. D. Banker, and R. Morey, “The use of Categorical Variables in Data Envelopment Analysis,” Management Science, vol. 3(12), 1986, pp. 1613-1627.

[25] W. H. Beaver, “Financial Ratios as Prediction of Failure,” Journal of Accounting Research, spring, 1966, pp. 71-111.