運用市場基礎模型預測營造公司財務危機之研究-以美國營造公司及台灣營建業為例

Abstract

營造公司的違約預警問題一向為營建項目業主及其他利害關係人所關注。為提升對營造公司財務危機的預測能力，過去的營建管理學者將與管理能力及經濟因素相關的變數，加入一般會計基礎預測模型之建構中。然而，管理變數必須依賴專家主觀的判斷，會計數據容易受到管理階層的操縱，而且和經濟數據同屬過時的資訊，用它們來做為模型的輸入變數，引起一些學者的質疑。 近年來，以公司股價為主要輸入變數的市場基礎信用風險預測模型，引起學者們廣泛的研究，由於營建業特殊的產業特性和會計處理原則，前述研究大部分將營建業排除在其驗證樣本之外。本文針對營建業，採用美國營造公司為驗證樣本，研究市場基礎模型對營造公司發生財務危機的預測能力。不同於以往營建管理相關文獻以配對方式選取少數樣本，本研究採取大橫斷面樣本，以避免選樣偏差。同時，本研究採用接受者操作特徵曲線 (ROC Curve) 衡量不同模型依據營造公司可能發生財務危機之風險高低排列的能力，用以挑選出真正對營建項目業主及其他利害關係人有用的模型。 本文研究結果顯示：市場基礎模型區分財務危機與正常營造公司的能力優於Severson et al.(1994),及Russell and Zhai (1996)所建構加入管理能力或經濟相關變數的會計基礎加強模型；此外，市場基礎模型在依據營造公司可能發生財務風險機率高低排列的能力上，明顯優於傳統會計基礎模型，其預測效力也高於Reisz and Perlich (2007)對整體產業但排除營建業樣本的驗証。本研究另以台灣營建公司為樣本驗證模型的效力，發現市場基礎模型的預測能力仍不低於會計基礎模型。本文因此建議可採用市場基礎模型為評估營建業發生財務危機可能性的替代方法，本研究最主要的貢獻是為營建業之違約風險預測問題開拓一新的研究方法。

The prediction of construction contractor default has always been an important issue for construction owners and other stakeholders. Previous construction contractor default prediction models incorporated managerial or economic variables into traditional accounting-based models to enhance predicting power. However managerial variables are qualitative and depend on human judgment, while accounting numbers are subject to manipulation by management. Furthermore, both economic variables and accounting ratios are only available periodically and may not provide necessary information in time. Using these variables as model inputs has caused doubt among scholars. The market-based default prediction models which use stock market information in predicting company default risk have appealed to scholars in recent years. Perhaps due to the unique industrial characteristics and accounting rules in the construction industry, the construction industry is usually excluded in their empirical validation. This is the first study applying market-based models to predict the default of American construction contractors and assert that the option-pricing framework is very suitable to describe the behavior of contractor default. Different from existing literature of contractor default prediction models, this research builds and validates models using a large cross-section of contractors, and uses all available firm-years data during sample selection period in empirical analyses to alleviate sample selection biases. The Receiver Operating Characteristics (ROC) curve is employed to assess the model performance in ranking contractors from riskiest to safest, as to choose the optimal model for construction owners and other stakeholders. The empirical results of this study exhibit that the market-based models have a smaller misclassification rate in classifying defaulted and non-defaulted contractors than the enhanced accounting-based models, which, as proposed by Severson et al. (1994), and Russell and Zhai (1996), additionally incorporate managerial or economic variables into accounting-based models. Besides, the market-based models obviously outperforms traditional accounting-based model in ranking contractors from riskiest to safest. They also have markedly better discriminatory power than that of Reisz and Perlich (2007) based on the data set of all industries except the construction industry. The overall results conclude that the market-based models, which use stock market information in predicting company default risk, has significant advantage for the construction industry, and it provides an alternative to measure construction contractor default. The contribution of current research is that it proposes the possibility to explore the default risk of the construction industry using a more powerful new tool.