本文旨在彙整探討以Kothari et al(2005)模型進行盈餘管理行為、公司治理機及財務資訊對於財務危機發生之預警研究。以臺灣上市(櫃)公司2005年至2009年因發生財務危機而打入全額交割股之公司做為危機樣本,其餘為正常樣本,並以盈餘管理、公司治理及財務資訊透過Logistic迴歸分析建構出4種財務危機預警模型。 實證結果發現,公司應用裁決性應計項目操縱盈餘管理的程度愈大時,公司發生財務危機的機率愈大;公司治理機制愈弱,並應用裁決性應計項目操縱盈餘管理的程度愈大時,公司發生財務危機的機率愈大;當公司營運不佳、並應用裁決性應計項目操縱盈餘管理的程度愈大時,公司發生財務危機的機率愈大;公司應用裁決性應計項目操縱盈餘管理的程度愈大,公司治理機制愈弱、公司營運不佳時,公司發生財務危機的機會率愈大。而從財務危機預警模型之預測來看,以盈餘管理、財務資訊及公司治理所架構之財務預警模型最佳。另外透過敏感性分析發現,不管是使用Kothari et al模型或Modified Jones模型、有無刪除無財務危機產業,甚至2008年與2009年的全球性金融風暴,皆對本研究所架構之財務危機預警模型之敏感性影響不大。由此可知,本研究架構之模型對於公司發生財務危機可做為參考之依據。
This paper aimed to apply the model developed by Kothari et al., (2005) and study the predictability of financial distresses by monitoring companies’ earnings management decisions, corporate governance practices, and changes in financial information. The subjects of this research consisted of the TSE and GTSM listed companies between 2005 and 2009 whose shares were traded on a gross settlement basis due to financial distress. We further applied a logistic regression analysis on earnings management, corporate governance, and financial information to construct 4 financial alert models. Our empirical study showed that the more extensively a company manipulates earnings using discretionary accruals, the higher probability that the company will encounter financial distress; the more extensively a company with weak corporate governance manipulates earnings using discretionary accruals, the higher the probability that the company will encounter financial distress; the more extensively a company with poor business performance manipulates earnings using discretionary accruals, the higher probability that the company will encounter financial distress; the more extensively a company with weak corporate governance and poor business performance manipulates earnings using discretionary accruals, the higher the probability that the company will encounter financial distress. As far as predictability is concerned, the financial alert model that takes into consideration a company’s earnings management, financial information, and corporate governance produces the best results. Furthermore, we have discovered using sensitivity analyses that the financial alert models constructed in this research were less susceptible to the impacts of whether the Kothari et al or the Modified Jones model was used, whether the industries in financial distress were eliminated from the sample, or even the worldwide financial crisis that took place between 2008 and 2009. We conclude as a result that the models constructed in this research provide a useful reference in predicting a company’s financial distress.