營建公司違約機率預測模型之比較研究

Abstract

自1997年發生了亞洲金融風暴及1998年本土型金融風暴至今，國內接連發生不少知名企業因財務危機而倒閉，其中營建業之相關危機事件竟高達十數餘起，有鑑於此有不少學者投入營建公司違約預警模型之研究，探討各種可能影響營建公司發生違約的因素，例如財務因素、景氣循環波動因素等，並利用各種不同的違約模型建構方法來預測公司發生違約之可能性機率。 本研究應用巴塞爾銀行監理委員會(2005)所建議之驗證方法，使用貝氏失誤率、ROC曲線與AUC值來評分並衡量模型預測之準確度，探討近年來針對營建公司所建構之違約預警模型，何者具有較高的預測能力與辨識能力，以提供對於營建公司違約風險評估之參考。

Since the Asia financial meltdown in 1997 and the local financial meltdown in 1998, several famous enterprices had gone bankrupt because of the financial crises. The number of the financial crises in construction industry is even higher than ten; therefore many scholars threw themselves into the research of forecasting models of default probability of construction firms and treated kinds of factors that would cause the the construction firms to default, for instance, finance, inflation or deflation. And they use different methods of building default models to forecast the default probability. This study applied the method which was adviced by Basel Committee on Banking Supervision (BSCS), and exercised Bayesian error rate, ROC Curve and AUC Value to rate and estimate the forecasting accurancy of different models. Treating the forecasting models of default probability of construction firms which were developed in recent year- which one qualified higher ability of forecasting and identifying- as a reference to estimating the default probability of the construction firms.