Prediction of contractor default probability using structural models of credit risk: An empirical investigation

Yu-Lin Huang*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

23 Scopus citations


Structural models of credit risk can apply for quantitatively predicting contractor defaults and pricing performance guarantees. However, the application involves crucial empirical issues. Some of the empirical issues are investigated using market and accounting data of public construction firms in Taiwan. Statistical analyses are conducted using the Wilcoxon rank sum test, Shumway's discrete-time hazard model, and the receiver operating characteristic curve. Structural models are viable, and market value tends to dominate other measures of economic or financial distress in terms of prediction accuracy. However, when calibrated to minimize Type I and Type II errors, the default boundary of market value produces substantial residual errors. In addition, the calibrated boundary is at 151% of face debt, much higher than those suggested by previous empirical studies. This seems to reflect the idiosyncratic short-term debt structures of Taiwanese construction firms. Leland and Toft's model is recommended for further investigations, because their theory explains the higher than expected calibrated boundary.

Original languageEnglish
Pages (from-to)581-596
Number of pages16
JournalConstruction Management and Economics
Issue number6
StatePublished - 2009


  • Contractor default
  • Prediction
  • Probability
  • ROC curve
  • Regression analysis
  • Structural model


Dive into the research topics of 'Prediction of contractor default probability using structural models of credit risk: An empirical investigation'. Together they form a unique fingerprint.

Cite this