| 摘要: | 本研究主旨在探討國內股票上市、上櫃和興櫃公司之財務危機預警模式在國際金融海嘯期間之預測能力的有效性實證。研究的重點在以金融海嘯期間前的企業財務危機資料所建立之預警系統，是否也能夠應付金融海嘯期間之樣本外的危機預測能力。本研究將採用財務指標、及財務指標結合公司信用評等與總體經濟因素等兩種類型的解釋變數，研究期間從2004年到2008年底為止以年度公司財務資訊或總體經濟數據為主，同時並將研究期間區分為非金融海嘯期間 (估計期間：2004年到2006年) 與金融海嘯 (樣本外預測的期間：2007年到2008年) 兩個期間為主。其次，將採用2006年下半度到2007年度之間發生財務危機之非金融類股及配對樣本公司共計有64家估計期間的樣本公司，以做為建構企業財務危機預警模型所使用。另外，運用Kolmogorov-Smirnov檢定、及Wilcoxon 檢定等無母數統計方法，以及Logit 迴歸分析等方法來進行實證分析。最後，再經由金融風暴期間以發生財務危機及非財務危機共計52家的上市櫃及興櫃公司，以所建立的四種財務危機預警模型進行樣本外預測能力的評估後，發現建議採用負債佔資產比率、長期資金佔固定資產比率、股東權益報酬率和每股盈餘等四項財務指標變數再加入股價(異常報酬)資訊後，並配合較長期間的資料所建構的預警模型來做為金融風暴期間樣本外的預測模型是較為適宜，因為在公司發生財務危機的前一年(2007年)時，其預警模型的危機鑑別率高達79.17%；誤判率僅30.43%，具有不錯的危機鑑別率和誤判率。另外，本研究也發現僅有股價此項經濟因素會與公司發生財務危機事件有關；但經濟成長率、失業率、利率、通貨膨脹率和景氣指標等經濟數據與公司發生財務危機事件則無關。關鍵字：財務危機；Logit 迴歸分析；財務指標；總體經濟變數；金融海嘯；樣本外預測能力 The purpose of this study to investigate the effective evidence of the domestic listed companies, OTC and Emerging Stock Company financial crisis early warning mode Evidence of the effectiveness of the predictive ability of the international financial tsunami period. The study focused on the early warning system established during the financial crisis of enterprise financial crisis information, whether able to cope with the crisis outside the predictive ability of the samples during the financial crisis. Financial indicators and financial indicators of the company's credit rating and overall economic factors and two-line content of the explanatory variables were be used in this study, during the study period from 2005 to the end of 2008, the annual financial information or the overall economic data was the main, and during the study divided into the two main periods, non-financial tsunami period (estimation period: 2004 to 2006) and the financial tsunami (of-sample forecast period: 2008). Secondly, the second half of 2006 degrees to the financial crisis occurred between the 2007 non-financial stocks and the matching sample companies totaling 64 estimates during the sample companies, used as the construction of corporate financial distress prediction model. In addition, Kolmogorov-Smirnov test, and Wilcoxon test, nonparametric statistical methods, and Logit regression analysis method were used to an empirical analysis.Finally, through the financial turmoil to financial crises and non-financial crisis, a totaling of 52 listed counters and emerging companies, established four financial crisis early-warning model for the evaluation of sample forecasting ability, it is found that suggested using the liability to asset ratio, long-term funds to fixed assets ratio, return on equity and earnings per share four financial indicators variables and then add the stock (abnormal returns) information, and with a longer period of data to construct the early warning model to do the financial turmoil of samples outside the prediction model is more appropriate, because of the financial crisis occurred the year before (2007), the crisis of its early warning model to identify up to 79.17%; misjudgment rate of 30.43%, with a good crisis identification rate and false positive rate. In addition, the study also found that the only stock the economic factors will relate to the incident of the company financial crises; economic data and the rate of economic growth, unemployment, interest rates, inflation rates and the economy indicators of financial crises has nothing to do.Keywords: Financial Crises; Logit Regression Model; Financial Ratios; Macroeconomic Variables; Predictability |
