27 April 2011

NASA on Shuttle Costs

At USA Today Dan Vergano has an article taking a look back at the Space Shuttle program, which is scheduled to fly its last flight this year.  I chatted with Dan yesterday about the program and in particular the difference between the estimates that NASA provided to him of total program costs, $113.7 billion, and those that Rad Byerly and I recently published, $192 billion.

My first impression was that NASA did not adjust for inflation in their tabulation (what the students in my quantitative methods seminar this past term learned was a methodological no-no).  Dan did a follow up and this was indeed the case. Here is an excerpt from the email that Dan received from NASA Public Affairs, shared with Dan's permission (emphasis added):
The number I gave you is actual dollars (that is we added the dollar amount for each fiscal year and came up with a total). The author [Pielke] seems to have adjusted his numbers to 2010 dollars then added them up. Because we don't know how he computed his adjustment, we can't comment on how he arrived at his number.

We estimate the total cost of the program in 2010 dollars from FY1971 to FY2010 (which does not include STS-133, STS-134 or STS-135) would be about $209.1 billion.
I'm not sure what is in NASA's numbers that is not in ours, or how they computed the inflation adjustment, but we did our calculations conservatively, so I'm not surprised at the higher tabulation. However, this is the first time in the 20 years that I've been looking at this issue that NASA reports a higher number than we do -- the per-flight difference is small -- $1.6 billion per flight from NASA and $1.5 billion from us, so not a big difference, but it is nice to see a convergence of views.


  1. Maybe I'm not smart enough to attend your quantitative methods class, but if the goal is to assess the average cost per launch of a program that is ending, and the costs were actually incurred in the years 1971 to 2010, then why is it wrong to compute the average of the non-inflation-adjusted dollars? After all, the true cost of the program to the taxpayer was $113.7 billion, not $192 billion (or $209.1 billion). This is a fundamentally different question than e.g. trends in disaster losses, where using the non-inflation-adjusted costs introduces a spurious trend not related to disasters.

  2. -1-Eric

    It is a good question. The reason is very much like the disaster cost issue -- we want to compare apples and apples over time.

    If I ask how much you spend on milk as compared to your grandfather, using unadjusted dollar data I might find that you spend 3 times as much. But if I adjust for inflation I might find that you spend about the same. That difference matters in making sense of question about spending over time. Factoring out inflation gives a common baseline to looking at dollars over time. There are other ways to make such comparisons as well, but for government budgets, inflation-adjusted dollars are a standard metric.

    For more than you probably want see:


  3. Another way to look at it - when they made cost predictions in 1971, they did so in 1971 dollars. When we judge their accuracy, we need to do so in the same units.

    I'm more concerned about the massive failure of productivity than dollar estimates. We were told that the space shuttle would by flying damn near every week. It was going to be no different than scheduling a passenger jet. The problem was less that they underestimated the cost - although that was certainly important - but that they overestimated the usefulness.

    And of course, there's the ultimate problem of manned space flight - there's really not much to do up there. The analogy of exploration on earth was a non-starter. When the Spanish found the Western Hemisphere, they could set up colonies and go about their business no differently than they did in Spain. Space? Not so much.

    Rant off.

  4. -3-Mark B.

    To convert the numbers in our Nature piece and blog posts from 2010 to 1971 dollars just divide by 4.43 (Source: OMB GDP price deflators)