“I can steal more money with a pencil than ten men with guns” – attributed to Al Capone’s bookkeeper
Whenever I did a press conference around a Space Shuttle event, there would always be one super hard question that made me stumble over the answer. Technical subjects I had down cold, the hard ones were always about the cost.
How much did the new safety gizmo cost? How much does ferrying the Shuttle across the country on the 747 cost? How much does the launch delay cost? What does each Shuttle launch cost?
If you look out on the internet there are ‘experts’ that will provide answers for these questions in plain and simple terms. I would honestly answer: ‘it depends’.
The simplest way to calculate the cost of each shuttle mission was to take the annual appropriation from Congress, adjust for inflation over the 40-year history of the program, and divide by 135 (the number of missions flown). Simple and totally inaccurate. Why? It’s complicated.
The Shuttle program manager was responsible for all the money spent but he could only control the portion called NOA – NASA Obligation Authority – which was a lot less than the money appropriated. Each NASA Center had a ‘tax’ on every program inside their gate. That is to say, if a program used a center, then the program contributed to the upkeep of the center: paying the guards at the front gate, mowing the grass, paying the light bill. Seem fair?
Well, if the Shuttle program was the ONLY program at a center – as it was for several of the largest NASA centers for a long time, the tax was eye wateringly high. Does the VAB need a new coat of paint? The Shuttle program gets to pay for that. Does the MCC need a new roof? The Shuttle program gets to pay for that. Does the A-2 Test Stand need a new flame bucket? The Shuttle program gets to pay for that. If the Shuttle program goes away, does the VAB still need paint and the MCC still need a roof? Yes. Paying for all those assets came to a head in about 2012 when the Shuttle program shut down and all the other programs had to scramble to find money to pay for infrastructure and center operating costs.
In that calculation of cost per launch, does one include those things that the agency still had to do whether or not the Shuttle flew? It’s a judgement call depending on what point you want to make.
And how about civil servant salaries? The Shuttle program was always in negotiation with the centers about how many civil servants to assign; but the program never got to vote on how many GS-13s vs GS-9s got assigned. Generally, there were more civil servants assigned to the Shuttle program than the program really wanted. Not negotiable, the agency needs to keep the ‘resource’ employed. So, CS salaries were extracted from the program which had very little input into that calculation. Carrying the salary costs so the agency had a ‘capability’ was not something that a Shuttle program manager really liked.
When one calculates the cost of ferrying the Shuttle from California to Florida, it makes sense to include the cost of the fuel used. And one should probably include some allowance for maintenance on the 747. But do you include an amount for the amortization of the capital expense of actually buying the plane in the first place? And the cost of the modifications made to change it from a passenger plane into a shuttle ferry, a one-time expense paid years ago? The pilots and the ground crews, do you include just the salary amount for the hours that they spend on the ferry flight, or include their entire salary for the year, maybe divided by the three-ish number of ferry flights performed? The generally stated cost of $1 million per ferry included all of that and more.
But if there were no ferry flights in a year, the shuttle program still paid for the pilots’ salaries and the maintenance of the airplanes – because the capability had to be there if needed.
Scrubbing a Shuttle launch was never fun, but they mostly saved the cryogenic fuel for the next launch attempt. Some hydrogen and oxygen were lost due to boiloff, etc., but the cost of lost fuel was in the single digit thousands. The work force was on payroll whether we launched or not. So, any calculation of cost for the scrub should really be based on the shift differential for the workers who were on the job overnight or on the weekend. The generally stated $750,000 cost of a scrub included all the salaries – even though folks would be back at their desks working on the next launch if we hadn’t scrubbed. So how accurate is that?
Is your head spinning? I’m the son of a CPA that grew up with dinner table discussions about depreciation schedules and amortization costs and I still find it confusing.
The standing joke around the Shuttle program office in the last years goes like this: “The first Shuttle launch of the year costs $3 billion; all the rest of the flights are free.”
In other words, if there is to be a program at all, a specialized skilled workforce dedicated to that program must be paid, specialized facilities dedicated to that program must be maintained, and all of those things must be paid for, never mind however many times a year they are used. A real space program is not a buy-it-by-the-yard kind of thing. The incremental cost of any additional shuttle flight was more realistically in the neighborhood of $200 million – not cheap – but a lot less than the $1.5 billion figure that comes from the ‘simple’ computation that throws in everything and divides by 135.
At some point, the calculation depends on whether the calculator is selling or buying. Does the author of the calculation want the program to look horribly expensive or reasonably cheap?
The title of this piece is the punch line to an old bookkeeper’s joke:
When asked ‘how much is 2 plus 2?’ the wily old accountant responded, ‘what figure did you have in mind’.
Any cost estimate depends on how it is calculated and a good question to ask is what is the motivation of the person doing the calculation.
Wayne Hale's Blog 
During the past week, I read a single SLS launch (presumably Block 1) is expected to cost $2 billion. That likely assumes a flight rate of about one launch per year. I wonder what the incremental cost of an additional SLS launch in a year would be, or are the SLS supply chain and workforce so scaled-down that two launches in a year isn’t credible? That wouldn’t bode well for a launch-on-need capability or a rescue scenario.
I have been told that they can produce a core stage every 8 months with the tooling and workforce available. That works out to 3 cores every two years. With some additional tooling and supplementing the workforce they project 2 cores a year should be relatively easy to achieve. Orion spacecraft are similarly on a one a year production schedule and I don’t have any info on how hard it would be to increase production. Solid Rocket booster production will likely not be a constraining factor until they expend all the remaining shuttle steel case segments probably after Artemis-4. Then a replacement filament wound case will be required and preliminary work is underway to develop and certify that item. Similarly, after Artemis-4 all the remaining SSMEs will be expended and I think it will be a challenge for the AerojetRocketdyne folks to get a new expendable version of the engine into production.
Rescue mission? What rescue mission? No plans for that that I’m aware of.
So the cost of the James Webb ST is not $10B, like those knuckle-headed bloggers claim, but is actually ~$500M because that’s about how much it would cost to build a second one?
Marginal (i.e. incremental) cost is an interesting number after making a million widgets and you want to know how much the next widget costs. The fixed cost contribution vanishes. Marginal cost is an irrelevant number when only making, or launching, a 100 or so widgets. The fixed cost contribution doesn’t vanish – and no magical accounting or browbeating by a highly respected Flight Director can make it do so.
It’s definitely relevant to know who is doing the calculation but it’s also good to know if the calculation answers the question being asked. In this case, the question from taxpayers is simply how much did it cost to make those 134 Shuttle flights happen? If only $105B instead of $210B (in 2010 dollars) had been allocated, would 134 launches still have taken place? No, of course not. It is irrelevant if NASA used a substantial portion of the money for items like roofs and non-essential civil servant salaries. That’s what govt organizations do with their budgets. If half the total Shuttle expenditure had instead been allocated to NASA, half or fewer flights would have happened.
Yes, who calculates what number is a factor. We can be sure NASA in the next few years will calculate $500M as the cost of a SLS flight. And the $3B+ that it will cost to make each flight happened will be calculated by knuckle-headed bloggers.
I remember at MSFC in the early 90’s when there was a budget problem with Space Station Freedom that several station personnel got shifted over to our little tether project. The key was to find projects to move those people to small projects and keep the number on the project under 40, which was the threshold for the Center tax there at MSFC. Very familiar with what you are talking about.
There are those in FOD who think that the number of people on ISS operations is far too many, but since ISS supports the institutional overhead budget at JSC, that is what they do. So nothing has changed.
‘lies, damn lies, accounting’? urk
No more calls, folks—we have a winner!
Oh no! It just depends on how you make assumptions. Pay attention to the details when you get a cost estimate
As one of those on the “other side” this made me smile. To those of us lucky enough to get to stay close enough to the program to kinda’-understand-it the questions asked were sometimes frustrating because they took up a lot of time that could be otherwise used to get a new story to tell. That was especially true of something did not go as predicted. Somehow, research that led to surprises, maybe even delights were perceived as failures. That led to the simple, and your post demonstrates why, meaningless breathless reports that the drop-in reporters could understand. For many, the basic tenets and processes of science, engineering and related activities seem even more a mystery today.
My hat is off in admiration to the space media core folks who do a yeoman’s job of explaining the complexities of space flight to the taxpayers. It is not an easy job!
Expanding on that issue of expenses and salaries, it disgusts me that Representatives and Senators view NASA as a jobs program for their state (Florida, Texas, Alabama, Louisiana, and a few others). NASA is America’s space program! But that viewpoint of NASA being a cash-cow for their state started with LBJ.
Oh no, well before Lyndon.
All the way back to the start. NASA’s formation from NACA + various labs took a lot of horse-trading among the various branches working on their own rocket programs, and an even more monumental effort with the setup of Apollo to get that program through the Senate & Congress. Releasing the funding required meant everyone had to have their pound of flesh in order to agree to the project as a whole, and decisions made there have had knock-on effects on every subsequent program (down to vehicle component dimensions limited by the height of the MAF). For all that people like to poke fun at the ESA’s inefficiency for distributing various tasks among member states, they were working from a proven model.
Good points regarding what counts, and what doesn’t count, but as a technical person, not an accounting person—isn’t that the point of modern, computerized accounting? Ie: it should be somewhat straightforward given the right practices to determine the pro rata use of facility overhead by given programs. At the very least within some margin. It would be nice to see that actual SLS/Orion (who are we kidding, this is the discussion) costs are, both fixed program costs, and marginal launch costs.
In the case of the former, I realize that it gets pretty messy.
I’m with Robert Zubrin: engineers at NASA are smart and knew how to save propellants after scrub, but the human spaceflight program was always designed to spend as much money as possible because the purpose of the program was to spend money. You will never be able to convince anyone that those “paint the VAB” costs explain the development costs of SLS (which could have been done in under a billion dollars by someone who paid out of their own pocket) and you would not be able to explain the marginal cost of an SLS flight or of the EUS. BTW, Commercial Cargo demonstrated how to buy by the meter, and it did not require a Nobel prize level scientific breakthrough.
From my perspective we were always challenged to do more for less. The idea that Spaceflight exists to spend as much money as possible is foreign to the universe I inhabit
Ever come in under budget?
I’m not sure his observations are directly opposed to your experiences. One thing I remember from my military experience is being challenged to do more with less-yet being limited by constraints imposed by decisions being made far above my pay grade. My outsiders view of this space program makes me suspect the same dynamic exists. Of course everyone’s boss tells him to stretch resources, but that only works so far when higher echelons or Congress have already created inefficient and wasteful structures.
I still don’t see why all of this stuff shouldn’t be counted. I’m sure similar infrastructure costs are all counted in the cost to taxpayers of a Falcon 9 or Heavy launch. If it wasn’t, we’d have to cut that $50-150 millionF9/Heavy price maybe in half—so ~$25-75 million? Would that be an honest assessment of the cost to taxpayers for those rockets?
Seems to me that the first cost range for F9/Heavy was much more accurate. And honest.
And the fact that many of these Shuttle-related costs continued to be spent even after the Shuttle program was retired seems irrelevant to me.
I dunno. Does SpaceX contribute to security at CCAFS? Is that included in their price or do they get a free ride because the government is trying to promote commercial Spaceflight? As a taxpayer I’m interested in whether or not commercial users of the facilities are paying their fair share or whether we take the infrastructure support out of the government programs
This point about security makes me wonder: will that be handled in future by the Space Force? It seems logical that a whole bunch of tasks would be transferred from NASA budget line items to the Space Force budget. Whether that means a consequent reduction in NASA’s budget, or a freeing up of funds within NASA is a mess for Congress to work out.
SpaceX contracts with the 45th for a number of things, and though I can’t state it with certainty, I imagine that security is one of those things. Your other points, though, regarding paying (or repaying) the cost of the launch pad are very good ones. In aviation, the cost of maintaining FAA-licensed airports is taken out of the aviation trust fund, which in turn is funded by taxes on aviation fuel. It’s analogous to how road and highway maintenance is paid for through gasoline and diesel fuel taxes. FAA/AST floated the idea a long time ago, and it was sunk rather quickly. Since then, the idea has been floated by Congress a number of times – and sunk just as quickly. It has a lot of ballast, though really the cost of propellant is such a small portion of space launch that it shouldn’t. Even if aircraft-like reusability is achieved with space launch vehicles, propellant cost (which would then be the bulk of the operating cost) would be very small. As long as the government didn’t get greedy, it would not adversely affect anyone’s business if such taxes were imposed.
When Elon says he expects the marginal cost of a Starship launch to be $2M, ya know he’s doing his “that’s aspirational” trick. When Eric reports on SLS being $2B/flight, ya know what he’s doing too.
To be fair I think Erik Berger reported exactly what the director of the OMB said about SLS costs. How the OMB got to that estimate is not explained. Similarly, if SpaceX achieves a fully reusable ‘gas and go’ Starship, I think a $2 million per flight estimate is high.
But remember there is a lot wrapped up in that ‘if’
It’s not entirely fair to say that dividing (operating) costs of the Shuttle program by 135 is simple and inaccurate. It would be fair to say it lacks lots of context. It’s accurate to the extent it starts the conversation at a whole cost whole system context. It says something of value that for a result, a number of launches if that’s even a valid measure of results (some would say a means to a result, transportation not being an end in itself), a certain total amount of resources came to bear. Alternately, in other paths, what might have been done with the same resources?
The questions persist today for current programs. Compare options A vs. B. It’s not valid to say Option A carries all these fixed costs we’d have anyway, as if arbitrary yet true. That assumes we can’t decide what we carry, ending the discussion more often than not for for political reasons.
Keeping people around (mostly contractors) or a building or painting these (like the VAB), or a test stand, are choices we make with resources. This is a valid question to ask of future programs as options – how much resources will you require, as nothing is written in stone unless someone wants it to be. If we are to be results focused against certain resources then context is important, but dividing vast resources, say for a program today, against a flight rate (say 1 a year) vs. applying whatever portion of these resources can be freed up with other choices is a valid question. But we can’t get honest comparisons across options can we? Not when one side says oh by the way, I’m still keeping all these other costs and I need not justify that to anyone. Want an honest answer, keep the context, understand it, and compare across future paths. Don’t just throw up your arms in resignation at having to carry the VAB and a thousand other things – forever.
Excellent points, Eric. You point out the inherent assumptional flaws in Waynes argument, rather than just helplessly accept them.
I agree. The shuttles were not meant to be ends in themselves, but a means to an end. So the cost per launch per se isn’t really a good question. Better is, what ends were accomplished for the amount spent, and can we do better? Could we have done better?
Agree that you have to understand the context.
I wonder if we should have included some amortized cost for building the launch pads in the first place in the 1960’s. After all, the shuttle program never carried that as part of the cost. Nor does SpaceX even though they use that same structure.
Edgar!
Here’s some opinions from the peanut gallery (no experience in the space industry):
It seems to me fixed cost is something all launch providers will have to face, this is not limited to the Shuttle (or other government owned launch vehicle). I assume ULA’s pricing for Delta-IV Heavy takes into account the money needed to maintain SLC-37B and SLC-6, and Orbital Sciences (now Northrop Grumman)’s pricing for Pegasus takes into account the money needed to maintain the L-1011. So when talking about the cost of a government launch, taking into account the fixed costs is not about motivation, it has important finance purposes, as in a launch company it is how you set the price of a launch to customers.
It is true that sometimes journalists mixed up fixed cost and marginal cost, and they should be called out and corrected on these. But that doesn’t mean launch price based on fixed cost has no value, both fixed cost and marginal cost are important. If your launch system has high fixed cost and no way to amortize the high fixed cost over many launches, this tells you the launch rate is too low, either because the design is not adequate to support high launch rate, or there is not enough payloads, both are concerns that need to be addressed.
Sometimes when an asset has been fully depreciated, the old sunk costs don’t show up on the bookkeeper’s ledger any more.
Accounting is probably more complicated than rocket science.
I would argue that all those fixed costs you mentioned SHOULD be included. Counting only marginal costs is misleading. The question then becomes, why was the Shuttle launch rate so low as to allow fixed costs to dominate? This was a mistake.
Of course the early, overly optimistic, projections for shuttle flight rate were much higher. 64 flights per year would have reduced the cost by a lot but it wasn’t achievable. In 1986 the goal was to achieve 15 flights building toward 24 a year. That wasn’t achievable either
Reasons for the low flight rate of the shuttle are numerous and a long discussion of them is something I would like to take on at some point but not today.
My big regret is that we never built a shuttle mark 2 that would have evolved with the lessons learned and perhaps approached those high flight rates which would drive the per flight costs down.
The cost of the Space Shuttle program?
14 Astronauts
at least 3 ground crew
2 search personnel
People die when doing great things in service of their country. Many more will perish as we venture out from our little blue planet. In 2019 I still see brand new TV advertisements that feature the Space Shuttle and it hasn’t flown since July 2011.
As has been observed on multiple occasions in this blog, risk is ever-present in human spaceflight. The question is: what justifies taking a risk with fatal consequences? I, for one, have a hard time with arguments such as, “they died fighting NASA complacency.” Such an argument will not play well with taxpayers or in Congressional testimony. Weak rationale will play even more poorly in association with a commercial enterprise, where profit is a primary motivation.
I’m not sure this article succeeds at telling me what it is trying to tell me. The majority of examples listed seem like absolute no brainers to include in the cost of a launch.
If you use a building, you get to pay its electricity bill. This is the most natural thing in the world. If you’re the only user of a facility, you are responsible for its full cost.
Granted, a program managed by a company wouldn’t pay for anything directly, but it would have to produce enough profits to set off its cost and then some in the book of whoever is doing the accounting. Pretty similar concept at the end of the day.
Commercial entities usually don’t want the roof to fall on their heads either, so they pay to get it fixed – either directly or through the rent they pay. If you want to own a plane, you pay for its amortization. And yeah, that’s expensive, that’s wy you generally don’t buy a plane if you can avoid it at all. Renting it is usually a much more efficient utilization of the resource. Much less waste. Commercial launch providers have scrubs too and they carry their cost. They can’t just magically remove that cost to make their launches look better.
OF COURSE all these costs have to be accounted for if you want to know what a single launch costs. It’s a different story if you want to know the cost of an additional launch over what you’re already doing, but doing that kind of accounting in any other context is disingenuous. If you were to create a non profit to manage your program and your non-profit is bankrupt almost immediately due to your cost estimate, something is very wrong. What would you have to pay that non profit to keep the program going? That’s the interesting question.
Is my head spinning? Maybe a bit, but I don’t see a single argument against dividing the program cost by the number of flights, only arguments FOR it. Precisely because accounting can be so difficult, it’s such a powerful tool to just watch where the money is going, akin to how it’s much easier to just run a piece of software and empirically watch what it does instead of trying to perform static analysis and formally prove its behavior. Money goes in at one and and disappears in various directions at the other. That difference is the cost of your money eating apparatus. It’s not the cost of making your apparatus bigger, but it is, very fundamentally and tautologically, the cost of keeping that apparatus running.
I get the general spirit of “this isn’t our fault” especially from someone responsible for managing it all, but all it comes across as is “this is not wasteful because it is mandatory waste.”
At some point, the calculation depends on what you want to use your number for. Do you want to buy an additional launch or do you want to compare the entire program to less wasteful commercial alternatives? For the former, you want to separate out fixed cost as accurately as possible. For the latter, total cost divided by products is the most accurate number you’re ever going to get. If you think this is done to make the program look bad and more importantly that looking bad in comparison is the only possible outcome, then it seems that you don’t think very highly of the program in question either.
Great piece, Wayne! I still recall discussing the costs and risks associated with hurricane preparation at KFC and JSC with you. Another shuttle expense… You were the daddy warbucks of 3 sites :-).
This is why SLS will cost a fortune to operate as well – as Chris Craft warned. And while flying additional missions will bring down the unit cost, there is a limit to how many missions we can fly that’s driven by some fairly immovable constraints such as: launch site facilities, production capacity, and manpower requirements that are driven by applying heritage complexity to ground and flight planning, integration, and operations. I wonder if there has been a study to examine how many flights are actually possible on an annual basis, or for mission / campaign sprints.
As an S&MA type I worry that reliability growth will be slow as well. This is another issue raised by Chris Kraft.
And, finally, I am sympathetic to the points raised by the humorous “I Want to do Apollo Again” video. https://youtu.be/A4J9uvhJQM0
I really wonder if it isn’t time that we rethink this big booster model and start looking at smaller incremental launches that build capabilities in LEO and cislunar that we can leverage for longer range missions.
Flight-Guidance …. signing off … 🙂