Seattle City Councilmember Mike O’Brien unearthed an interesting chart from the Washington Department of Transportation, showing that the state estimates that there’s a 40 percent chance that the tunnel project will go over budget…
That’s just the tunnel, mind you. It doesn’t include the other $1.15 billion slated for other parts of the project. That said, the tunnel—particularly the boring itself—carries the biggest technical and fianancial risks.
There are a couple of things worth noting here.
Find this article interesting? Support more research like this with a gift!
First: A 40 percent chance of a cost overrun is a pretty big risk. We’ll have to see how that plays out in the bidding process. If the state is looking for a fixed-price bid, perhaps that risk will be incorporated into the contract as a higher price. Regardless, I imagine that someone’s going to be awfully upset if costs start to escalate.
Second: While I have no real basis to critique the cost curve in the graph above, my understanding from people who are far more knowledgeable about the process is that it’s reasonable and technically defensible.
The curve itself was estimated using WashDOT’s Cost Estimate Validation Process ®, or CEVP ®. (Yes, CEVP ® is a registered trademark, so it needs a little ® thingy, also known as a Racol.) And by way of confession: I have a little crush on the CEVP ®. If I were a transportation cost estimate process, I’d want to date the CEVP ®, but I’d probably be too shy to ask it out. One of CEVP ®’s most attractive features is that it treats transportation costs as a probability curve, rather than a point. Big infrastructure projects just aren’t like appliances; they don’t come with a fixed price tag that’s known in advance. Typically, when you hear a cost quote for a transportation project, the number little more than an informed guess. Sometimes the engineers guess high, sometimes they guess low—but you never know in advance which direction they’re off, or by how much.
Because of that uncertainty, a CEVP ® probability curve is a much better way of thinking about costs than a point estimate. Not only is a cost curve better matched to reality, but it also helps project managers, politicians, and the public understand cost uncertainties—and prepare for cost overruns.
Third: Although CEVP ® seems to have a pretty good track record, and apparently is being adopted by other states, it’s hard to know how accurate any particular CEVP ® cost curve really is. And that may be especially true for tunneling megaprojects, where the track record for CEVP ® is limited—and the overall history of engineering estimates suggests a generous dollop of uncertainty mixed in with error, overoptimism, and strategic misrepresentation.
So perhaps the cost curve for the tunnel project is accurate, or accurate enough, in which case Seattle needs to wrestle with the implications of a 40 percent chance of cost overruns. Or perhaps—and don’t hate me for saying this, CEVP ®!—this particular cost curve is unduly rosy. Perhaps the chances are even greater than 40 percent that the cost of a tunnel will creep over $1.96 billion, and the odds of a serious overrun are far higher than WSDOT estimates.
Either way, it’s time for Seattle to stop arguing about whether there will actually be cost overruns—which is essentially unknowable before the work is well underway—and put its effort into figuring out how to pay for them if they do happen. To my mind, and given the stakes involved and the precarious finances of both the city and the state, the “who pays” question (along with many others) should be answered well before any shovel hits the dirt.
CEVP is better than most estimation systems. The problem with AWV is you need to estimate probability of risks and the additional cost of those problems. This then begs the question how do you evaluate risk and cost if the project is beyond precedence? Sent the probability of complications low or the cost of those complications low and it doesn’t affect the total cost as much as it maybe should be. Can you get your hands on the detailed inputs for the CEVP?