This article discusses the sources and methods behind the "How much did Apollo cost?" resource.
Buried within one of the half-dozen boxes of Apollo budget documentation at NASA's historical reference collection in Washington, D.C. is a piece of paper outlining a "loose agenda"—sadly undated—for an "Apollo Cost Consensus Meeting." Among the goals stated on the agenda are for "the cost estimating community to reach consensus on Apollo costs."
Though the convener is anonymous, you can feel their resignation in the written notes summarizing the meeting: "[it] had the predictable results...no one was fully prepared to discuss the problem...data extremely sketchy...discrepancies typically 15% - 20%."
Such is the lament of the Apollo cost curious. Somehow, the United States' greatest triumph in human exploration—a triumph of engineering, cooperation, and organization—evades answers to seemingly simple questions: how much did it cost and when was the money spent?
The most authoritative account of Apollo's cost was released in early 1973, submitted as written testimony by NASA to Congress. According to this document, the total cost of Apollo was $25.4 billion. NASA summarized major program costs within Apollo but did not detail amounts spent per year and rounded all numbers to 1 significant digit. Other data, first published in the 1970s, claim Apollo cost $19.4 billion—a significant discrepancy that demands an explanation.
The lack of detailed, reliable cost data on Apollo—easily the most important human spaceflight effort of all time—still baffles me. Many of the details have been lost to time, as accounting practices changed over the past half-century and undigitized records were lost, misfiled, or forgotten in historical vaults throughout various NASA centers.
The participants of the cost consensus meeting apparently shared my desire to understand the cost of Apollo, particularly as it applied to effective cost estimation and financial planning for future NASA projects. So despite the disappointing outcome of the initial meeting, the participants agreed to revisit cost data contained in congressional budget submissions from the 1960s. Whether this actually happened there is no indication.
Taking the lead from the anonymous notetaker in NASA's financial office, I did the first part, which was to revisit all NASA budget submissions to Congress between the fiscal years 1961 and 1974. Though these documents were digitized by NASA headquarters' history office, they were not readily available online (until now). Every annual budget submission contains a wealth of data, including self-reported obligations (contracted expenditures) for NASA programs 2 fiscal years' previous. For example, the 1965 NASA budget submission lists "actual" spending amounts contracted by the agency in the 1963 fiscal year.
Using this information, I reconstructed Apollo's publicly-reported cost history year by year, from 1960 to 1973. I also found additional Apollo-era cost documents in NASA headquarters' historical reference collection, including rare data reporting annual construction and operations numbers for Apollo-related ground facilities.
According to my reconstruction, direct costs for Apollo amounted to $20.6 billion between fiscal years 1960 and 1973. Additional indirect costs, such as the construction of ground facilities, civil servant salaries, and deployment of the tracking and data network, increases the total to $25.8 billion for the Apollo program. This agrees with the congressionally reported number within 1.6% and represents a significant improvement over detailed cost data presented elsewhere.
Adjusting for Inflation
The relative value of the U.S. dollar changes over time. In 1961, the cost of a loaf of bread, an hour of a construction worker's time, or pound of aluminum was generally lower, than the same items today. This doesn't necessarily mean that those were cheaper in 1961. Paying 50 cents for a loaf of bread sounds cheap, but it occupied a larger share of household expenditures (0.009%) in 1961 when the average household salary was $5,700 than a $3 loaf of bread does today (0.006%) for average household incomes of $50,300. Good inflation adjustments attempt to account for such subtleties.
Inflation adjustment is far from an exact science. The makeup of the U.S. economy was significantly different in the past, as were labor laws, regulatory systems, productivity, technology, and the relative strength and structure of the global economy. For these reasons, all adjustments for inflation should be considered as more of a guideline than a cut and dry assertion of absolute worth (for an in-depth exploration of this subject, I recommend the essay "Defining Measures of Worth: Most are better than the CPI").
Another subtle but important point: spending on Apollo spanned 14 years between 1960 and 1973, and annual inflation rates varied substantially throughout this period. Absent year-to-year data, you have to squash down all Apollo spending into a single year and adjust for inflation from there, functionally ignoring the fact that inflation values in 1970 were different than those in 1961. This new dataset greatly improves the inflation-adjustment calculation by enabling refined year-by-year adjustments for inflation.
I used two methods to adjust for inflation, each of which answers a slightly different question.
The first is NASA's New Start Index (NNSI), which is designed specifically for aerospace projects. I believe this is a far better method for inflation-adjustment for Apollo than using the Consumer Price Index (CPI), which is designed for household and consumer goods, not moonshots. You can think of the NNSI as answering the question: "What would NASA be spending on Apollo in 2019?"
The second method is to adjust the costs so that they occupy the same relative share of the nation's economy, or Gross Domestic Product (GDP), over time. In other words, if Apollo accounted for 2% of GDP in 1965, what is the equivalent of 2% of GDP in 2019? This approach answers the question: "If the U.S. were to allocate resources to a space project to the same extent as it did for the lunar effort, how much would NASA have to spend today?" Think of this as a statement of economic priority.
Running all the numbers on the annual cost data spanning 1960 - 1973, you get this:
|original $||Adjusted 2019 $||Relative GDP $|
|Spacecraft||8.1 billion||81.3 billion||194.8 billion|
|Launch Vehicles||9.4 billion||99.0 billion||243.4 billion|
|Development & Operations||3.1 billion||28.7 billion||66.9 billion|
|Direct costs||20.6 billion||209.0 billion||505.2 billion|
|Construction of Facilities, Salaries, & Overhead||5.2 billion||54.8 billion||136.2 billion|
|Total Apollo||25.8 billion||263.8 billion||641.4 billion|
|Robotic Lunar Program||907.0 million||10.3 billion||26.1 billion|
|Project Gemini||1.3 billion||14.1 billion||34.8 billion|
|Total Lunar Effort||28.0 billion||288.1 billion||702.3 billion|
Costs of the Apollo lunar effort, adjusted for inflation to 2019 dollars using the NNSI and relative GDP share. Detailed numbers are available in the source data. Source data.
To summarize: the entire lunar effort (with robotic missions and Gemini included) would cost $288 billion in today's dollars. If the U.S. prioritized a space project financially the same way it did with Apollo the 1960s, the nation would have to spend $702 billion to account for an equivalent share of GDP.
What costs should we consider as part of Apollo?
Since no accounting effort is wholly objective, I attempted to note every subjective decision I made regarding what to count, when, and where, via per-cell comments in the Excel spreadsheet version of the source data.
I took a generous interpretation of Apollo-related costs: if it fed into the lunar effort, or was likely to, I included it as part of Apollo. This occasionally means that there are small discrepancies between my annual project sums and those reported by NASA (mine tend to be slightly larger). I also included spending on the Saturn project pre-dating the actual start of Project Apollo, as those were used primarily for the lunar effort. My totals for launch vehicle development ($9.4 billion in then-year dollars) are therefore slightly higher than those reported by NASA ($9.1 billion).
I also attempted to normalize costs across accounts in the summary tables of the source data. NASA's accountants frequently changed their internal accounting practices for Apollo, making it otherwise hard to compare spending year-to-year. For example, in the very early 1960s, the costs of the Saturn rocket engine development were grouped with the costs of Saturn rocket engine procurement. In later years, procurement costs for the engines were included with the direct costs of the rocket, and development was considered separately. I account for this in my summary tables, and further acknowledgments of these sorts of minutiae are included in the Excel spreadsheet.
A common cost discrepancy ($25 billion vs. $20 billion) is explained by the following. Throughout the 1960s and 1970s, NASA's budget was divided into three primary accounts: Research & Development (R&D), Construction of Facilities (CoF), and Administrative Operations (AO)—later renamed Research & Program Management (R&PM). Nearly all of the detailed cost data we have for Apollo comes from the R&D accounts, which paid for obvious things like the development and production of Apollo spacecraft, the Saturn V launch vehicle, mission operations, and project integration. Combined, these amount to $20.6 billion and are the source of the $20 billion figure commonly cited online.
Indirect costs should not be discounted, however. These include the costs to build Apollo's enabling infrastructure, including the space centers currently known as Johnson, Kennedy, and Stennis. Without these facilities and their attendant staff, Apollo obviously would not have happened. These data are harder to suss out from congressional budget justifications, and I instead rely on two internal reports prepared for Congress from NASA's financial office which breaks down facilities and overhead costs from FY 1961 - FY 1973. Facilities and overhead costs for Apollo were included in (aptly-named) "Construction of Facilities" and "Administrative Operations" and were considered separately during the congressional appropriations process.
The sum of the direct and indirect costs accounts for the $25.8 billion number that closely matches the total cost reported to Congress in 1973. However, I argue that this undercounts the total U.S. investment in Apollo, as there were other efforts pursued in service of the lunar goal throughout the 1960s. In particular, Gemini refined operations and rendezvous in low-Earth orbit in service of Apollo. The robotic lunar programs in the 1960s also mapped the lunar surface and provided crucial ground-truth in advance of astronauts. Just because they weren't included in the Apollo budget account in NASA's ledgers doesn't mean the money wasn't spent. I elected to include them as "related programs" and incorporated them into the total cost when speaking of the "lunar effort."
You may disagree with my subjective accounting decisions. It is for that reason that I included the raw data for each fiscal year as presented in the congressional budget justifications. You can reconstruct or re-interpret these data to your heart's content.
Why you shouldn't use the "Apollo Program Budget Appropriations" data source
The NASA History Office website hosts a detailed cost breakdown for Apollo, by year, which first appeared in the 2000 publication, Apollo By the Numbers: A Statistical Reference, by Richard Orloff, which itself compiles data first published the 1970s from the book, The Apollo Spacecraft: A Chronology, volumes I through IV.
Don't use this as a source. It contains fundamental errors, significant omissions, and presents an inaccurate picture of Apollo funding. It undercounts direct costs by nearly $1 billion and omits all indirect costs.
First, the data presented as "Apollo Program Budget Appropriations" are not actually appropriations. Instead, it is a mix of proposed spending (FY 1962, FY 1973) and contracted spending, or obligations for all other years. Congressional appropriations are not included at all. As a reminder: NASA requests funds via annual budget justifications submitted Congress. Congress then appropriates funding based on those requests. NASA obligates these funds in the form of contracts, eventually paying them out as expenditures.
Second, the table lists significant spending on "orbital reentry tests," "biomedical tests," and "high-speed reentry tests" in FY 1962. As far as I can tell, these are phantom expenditures. After doing some digging, I found that these numbers come from the FY 1962 budget request (see the previous paragraph) and don't represent actual money spent. These tests were probably incorporated into Project Gemini, which began after the FY 1962 budget estimates were prepared in early 1961.
Third, there are significant omissions for programs before FY 1964. It should be a warning sign to any attentive reader that this source claims that $0 was spent on the Saturn I rocket before FY 1964—an impressive claim of fiscal efficiency considering the rocket first launched in 1961.
Fourth, the annual values for "NASA Total" only include the space agency's R&D account. As discussed previously, it excludes the Construction of Facilities and Administrative Overhead accounts and thus understates NASA's annual budgets.
Fifth, it contains a fundamental arithmetic error for NASA's total budget from 1960 - 1973. It should be $41 billion given the numbers provided, not $56 billion as claimed.
Sixth (and finally), it makes no attempt to normalize annual accounting changes. This is why there are columns with a single value for items like "Saturn I-C," "Spacecraft," and "Flight Modules (the CSM and LM). As these are presented without context, it is impossible to compare one year to another.
There are more reasons, but is it worth further belaboring a point? Don't use this source.
Why this matters
Apollo demanded precision and attention to detail to succeed as it did. As space advocates and supporters, the least we can do is to maintain similarly high expectations for historical fidelity, which is why I find sloppy and inaccurate Apollo cost accounting so frustrating. There is no value, either historically or politically, to understating the resources required to succeed in that endeavor.
How much was spent on Apollo, and when, is relevant as NASA has once again been directed to return humans to the Moon. To properly evaluate the seriousness of this directive, it makes sense to compare its spending proposals to the one data point we have for a successful human lunar mission. How much money did it take to do it the first time? How was it spent? And, perhaps most importantly, when did the money show up?
To date, the White House has proposed an additional $1.6 billion for Project Artemis, on top of the $5 billion spent annually on the Space Launch System, Orion, and related ground systems. Compared to Apollo, this is a relatively modest investment. Looking forward, we should expect significant increases in spending associated with an accelerated lunar effort or adjust our expectations accordingly.
This page summarizes the costs of Apollo and includes charts highlighting the annual costs of major Apollo systems such as the Command and Service Module, the Lunar Module, and the Saturn V.
Project Apollo Cost Data Set (Excel spreadsheet)
Comprehensive Project Apollo annual cost data, non-inflation adjusted dollar amounts, program-by-program cost breakdowns, construction costs, and relative GDP adjustments are available to download as an Excel spreadsheet or to view as a Google spreadsheet.
A public Google Drive folder which includes NASA budget submissions to Congress, including every one between 1960 and 1973 and additional Apollo budget documents.