會計師意見與總體經濟在財務危機的應用與分析

Abstract

過去研究以會計師意見單期模型預測公司發生財務危機的機率，單期模型存在抽樣偏差與忽略信用風險的問題，且過去研究並無得到會計師意見模型優於傳統模型的一致結論；本研究針對單期模型問題加以修正，採用Shumway(2001)多期的離散型危機模型建構財務危機預警模型，並以總體經濟變數控制總體經濟風險(亦有討論以Belkin、Forest與Suchower（1998）之研究(AR(1)-GARCH(1,1))去修正有納入總體經濟的財務預警模型)，探討會計師出具意見、市場變數與總體經濟變數是否有助於財務危機預測。 本研究以1986年至2006年台灣(曾經)上市上櫃公司為研究對象，以Shumway(2001)離散時間涉險模型、納入市場變數的離散型模型以及納入市場變數及總體經濟變數的離散型模型，分別針對財務比率、會計師意見、市場變數與總體經濟訊息，建立30個財務危機預警模型，並比較30個模型的精確度。 本研究結果發現：1.會計師出具「繼續經營假設有疑慮」、「長期投資前期未經查核或由其它會計師簽証」意見能提升模型的解釋能力。2.納入市場變數及總體經濟的離散型較未納入總體經濟訊息的模型解釋能力為佳，財務危機預測較為精準。3.本研究所建構的財務比率加市場變數、總體經濟、產業因素、會計師出具「對繼續經營假設有疑慮」意見(模型26)，或是財務比率加市場變數、總體經濟、產業因素、會計師出具「繼續經營假設有疑慮違反一致性及或有事項」三項意見(模型29)（考慮所有的衡量方法）兩個離散時間涉險模型財務危機預測的準確率較佳，研究結果顯示會計師意見、市場變數、總體經濟變數以及產業因素能提供財務危機預測增額的資訊。

Previous studies have not found out the consistent results about the relative performance of financial ratio models and auditors’ model in predicting financial distress since prior research ignore the credit risk and sampling bias in static models. This study reevaluates the relative performance of auditors’ opinions versus traditional financial models. We modify Shumway (2001) discrete-time hazard model by employing macroeconomics variables to control the macroeconomics risk. This paper explores the auditors’ opinions, market variables and macroeconomics variables in financial distress prediction. The study constructs 30 financial distress prediction models by employing discrete-time hazard model and by using financial rations, auditors’ opinions, market variables and macroeconomics factors. We further compare the prediction ability of the 30 models. The results exhibit that the auditors’ going concern opinion increases explanatory power of the predicting model and the accuracy of auditors’ going concern opinion model is superior to financial ratio model or other auditors’ opinion models. Besides,the prediction ability of the market and macroeconomics discrete-time hazard model is superior to that of other model without macroeconomic variables. We further analyses that the prediction ability of discrete-time hazard model, which employs financial ratios,auditors’ going concern opinion ,market variables and macroeconomics factor and employs financial ratios,auditors’ going concern opinion,auditors’ against consistency ,auditors’ contingency event ,market variables and macroeconomics factors, is better than other models. The accuracy of financial distress prediction can be improved by properly using auditors’ opinions, market variables and macroeconomics factors.