Proportional Cox Hazards Model to Quantify the Likelihood of Underestimation in Transportation Projects

Preparing accurate cost estimates for highway projects always has been challenging for transportation agencies. It is a common problem that the lowest submitted bid significantly deviates from owner's estimate. This might result in project delay or cancellation, budget pressure, and cost overrun, which are problematic for both owner organizations and highway contractors. There is a need to enhance understanding of transportation agencies about the likelihood of underestimation. This research assessed relations of several potential drivers to explain and forecast the likelihood of underestimation. This research for the first time used concepts and methods from survival analysis and applied them into construction bidding process. A Cox proportional hazards regression model was developed which is capable of examining significance of variables representing characteristics of project, bidder, and external (environmental) market, and using them to predict the likelihood of underestimation in transportation projects. The results showed that number of bidders, number of pay items, total number of projects awarded in the same month at state level, project types, producer price index for construction machinery manufacturing, value of construction put in place for commercial, unemployment, and highly active contractors are significant drivers of likelihood of underestimation. This research contributes to the state of knowledge in construction bidding analysis by identifying drivers of the likelihood of underestimation and creating a Cox model to explain and predict the likelihood of underestimation using information available from the identified drivers. It is anticipated that the results will help transportation agencies better understand the extent of risk of deviation between low bids and owner's estimates, prepare more-accurate cost estimates and budgets, and develop appropriate risk mitigation strategies for successful project delivery.