Performance measurement and the prediction of capital project failure

This paper examines how changes in project-management performance in the execution phase affect project outcomes at completion. While identifying the key determinants of project-management performance is critical, few studies examine the discriminatory power of performance variables for predicting capital project failure at completion. Using 130 capital projects and a longitudinal design, this study develops a performance-measurement model based on changes in project-management performance during the execution phase. Subsequent hierarchical logistic-regression analysis reveals a good explanation of the variation in the failure of capital projects and high classification accuracy. Validating out-of-sample data demonstrates that the optimal model provides a reasonably good overall classification rate of 81.54%. Ultimately, our findings suggest that performance changes in the execution phase explain an important part of project outcomes and, more importantly, are useful predictors for project failure.