Sound Transit 3 is not going to match the stellar project delivery record of Sound Transit 2 barring a bailout from another government or the economy. In attempt to understand the cost estimation failures that got us to this point, on June 24th ST accepted the report of consultants asked to investigate why (report, slides, video).
The report points out parts of ST’s own cost estimation methodology that it did not follow during initial project estimation in 2015 and 2016. In particular, ST did not seek a second opinion on costs and did not sufficiently invovle its own Real Estate division in determining acquisition costs.
It’s a long report and hard to summarize. It identified eight key drivers of the increases between “phase 1” and “phase 2” estimates, and was able to assign a subjective importance to five of them:
1. Inaccurate original assumptions for major design elements such as stations, foundations, storm water requirements and sitework were contributing factors to cost increases from ST3 to Phase 2. The use of robust analysis from ST’s programmatic design team appeared absent based on various interviews with ST staff. [40%]
2. Utilization of a ROW methodology that employs a formulaic approach that is repeatable and can be applied consistently to a wide variety of projects and alternatives. This formula contains many inputs including contingency and was updated to reflect current market conditions throughout the estimate phases. [20%]
3. The use of a “buffer method” for defining preliminary ROW impacts instead of a properly defined project footprint that evaluated ROW requirements in light of construction needs. [20%]
4. A dynamic real estate market with rapidly rising property values and the redevelopment of lower density uses to higher density uses. [10%]
5. Updates to ST’s Unit Cost Library (UCL) to more reasonable units. Areas contributing as large cost drivers were bridge and tunnelling cost per route foot and vertical conveyance (i.e., elevators). It appears ST updated these units based on similar projects in lieu of escalating older information that was previously utilized. [10%]
6. The rigid nature of the estimate practices. For example, environmental costs were originally captured using high, medium, and low-cost allocation per route foot of the alignment. While having a system in place to ensure these items are not overlooked is important, it also creates an environment where an outside opinion from the design consultant performing the estimate appears unnecessary. [N/A]
7. ROW Ownership was not utilized to the extent it should have been due to the primary focus being on ST2 projects that were going on at the same time. [N/A]
8. Second opinions were not utilized to the extent they should have been. ERP approved methodology but did not perform an in-depth assessment on the accuracy of the cost estimates that could have identified inaccuracies earlier than what were presented. [N/A]
So what’s to be done? There are 38 recommendations. One answer is simply to do more engineering earlier, which is expensive but would reduce surprises. Short of that, involving more agency experts earlier would help a lot, as well as moving away from convenient formulas into real assessments of the cost to acquire the property at hand. The last leads to a less formalized method; any opening for judgment also creates one for wishful thinking. In the hands of well-intentioned staff, though, flexibility would be powerful.