For the Virginia Department of Transportation (VDOT), preliminary engineering (PE) is a phase in the project development process whose expenditures are differentiated from the right of way (RW) and construction (CN) phases. PE funds support tasks such as planning studies, preliminary and final design, public involvement, and environmental processes. At the program level—that is, the aggregate funds from all projects that are to be allocated to PE as opposed to RW or CN—PE expenditures must be large enough to prepare future projects for construction yet small enough to build existing projects. The initial interest in this study resulted from the fact that, assuming a fixed program amount, higher PE spending will be associated with lower CN spending, and construction spending is logically of interest to VDOT’s stakeholders. At the project level—that is, the PE funds available for a specific project—forecast PE expenditures provide project managers with a guideline regarding what resources will be needed to prepare a specific project for construction. This report analyzes the extent and causes of this variability in PE expenditures at the program and project levels.
At the program level, the mean statewide PE pooled percentage was 14.7% based on data reflecting FY 2004-2012. The mean annual district PE percentage for the same period ranged from 11.3% to 21.3%, and much of this variation (89%) was explained by three statistically significant variables: the percent of expenditures spent on minimum-plan or no-plan jobs (as opposed to complete plan jobs); the percent of projects in the development phase (e.g., projects for which PE or RW but not CN funds were spent); and the percent of expenditures spent each year on large projects (defined as universal project codes [UPCs] in the 99th percentile for a given district and year). At the project level, variations in PE expenditures were explained by statistically significant variables such as the estimated construction cost, the project’s duration, whether it is administered by VDOT or some other entity, the length of the project in miles, and whether the project has a categorical exclusion. Based on a testing set of data that were not used to build the project-level forecasting models, the results suggested that the forecasting error—that is, the difference between forecast PE expenditures and actual PE expenditures for a given project—can be reduced from a mean value of about $195,000 (when one considers construction expenditures only as is currently the case) to about $109,000 (when one considers these other variables in a recalibrated model).
The study recommends that the project-specific model developed herein be further piloted to determine if it continues to offer improved accuracy relative to an existing approach that relies only on the construction cost estimate. The study also recommends that the experiences one district has had of providing historical PE percentages to stakeholders in the planning and programming process be shared with other VDOT districts. Because the study showed that interviews of district staff were essential for understanding how to interpret PE expenditures, the two ecommendations offered were applied at the district level.