50% contingency? Really?

After a long hunt, I received the missing link in the documents which led to the Corps' cost estimates for Options 1, 2 and 2a on the permanent pumping stations given to Congress in early 2009, months after their Congressionally imposed deadline.

The document is a cost estimate prepared by Corps personnel in the Walla Walla District. Here it is.

I will be posting in more detail about this soon, but I wanted to get this document out there as quicklly as possible. The findings are two-fold:

1) The costs the Corps gave to Congress in 2009 include a massive amount of slop. The Corps took base estimates prepared by Black and Veatch which did not include contingencies for each Option. The Corps figured out the contingencies in cost and schedule and added those to the base estimates to get the final numbers which were handed over to Congress. Those contingencies were all around 50% of the original base estimates. For example, the original Black and Veatch Option 2 estimate was $1.9 billion. The number reported to Congress was $3.4 billion. The difference - $1.5 billion - was made up of $850 million of contingency and $650 million of (I presume) escalation. That bit at the end - how much of the estimate reflected the massive uncertainty in the design - was never reported until now.

I wrote about this last fall, when I reached some improper conclusions because it was inconceivable to me to have an estimate that was nearly half contingency and escalation. But that is explicitly what the newest document reveals.

Imagine if that information had been conveyed to Congress and the public when the report was issued in 2009: "Our estimates include about 50% slop in the numbers." People would have had the reasonable reaction of, "Then your numbers are nothing more than guesses." That opportunity was foreclosed by the Corps.

2) This new Walla Walla report was issued just seven days (September 23, 2008) before the main cost report to Congress was due. The Walla Walla report included the stunning conclusion that the Corps' internal estimate for Option 1 (including contingency and escalation) ended up over $40 million more than the appropriated amount of $804 million. If this had been reported to Congress, all hell would have broken loose, because that would have meant that none of the options could get built for the money Congress had set aside.

So the Corps did what they do best - they made the Option 1 estimate fit the appropriation by just disappearing $40 million dollars and delayed the report five months, all without telling anyone. The problem is they didn't apply the same "methodology" (i.e. "just chop what you need to in order to make the numbers look good to Congress") to Options 2 and 2a - those estimates remained as high as possible, which was how they wanted it.

In effect, they lied to Congress. The internal Walla Walla estimates for Options 2 and 2a were based on a 90% confidence level (higher confidence translates into higher amounts of contingency), and those are the numbers reported to Congress. The original internal estimate for Option 1 - $845 million - was also created at the 90% confidence level, but it was more than the $804 million they had to spend. So alacazam! $845 million turned into $797 million with a footnote in the report to Congress that reads, "Within the authorized and appropriated programmatic cost of $804,000,000."

This way, the Corps got to keep the Option 1 train going without the uncomfortable question from Congress: "Why did your own cost estimate for your preferred option turn out to be more money than we gave you?"