Uncertainty analysis of procurement phase performance indicators using Extreme Bounds Analysis (EBA)

The procurement phase (PP) is one of the major phases of construction projects, which has a significant impact on the ultimate success of projects. Although some studies focused on identifying the PP cost and schedule performance leading indicators, however, the robustness/fragility of these variables have rarely been studied. An analysis of these indicators allow project managers to focus on the primary contributors, and the more robust indicators should receive higher priority when allocating scarce project resources as they are more likely to positively impact project performance. Therefore, the aim of this research is to differentiate between the robust and fragile PP cost overrun and schedule delay indicators. For this reason, this study used the two previously developed regression models, which predict the PP cost and schedule performances. Extreme Bounds Analysis (EBA) was used to study the robustness or fragility of the identified PP indicators. In this study, both Learner's and Sala-i-Martin EBA methods were used. Since Learner's method only focuses on the extreme bounds of the indicator's distribution while Sala-i-Martin considers the entire indicator's distribution, the final conclusions were made based on the Sala-i-Martin method. Findings which were presented in both numerical and graphical forms, indicate that "bulk material quality issues", "company's degree of familiarity with technologies to be utilized in the construction phase" and "number of design/engineering organizations" are the three robust PP cost performance indicators. Results of the analysis also reveal that "percentage of design completed prior to the start of construction", "number of execution locations", and "number of supplier organizations" are the robust schedule delay indicators in the PP. The findings of this research will guide project managers in allocating limited human and machinery resources more effectively and efficiently.