Bjorn’s Corner: Production costs.

By Bjorn Fehrm

By Bjorn Fehrm

11 September 2015, © Leeham Co: In connection with our articles, there a numerous reader discussions around the development and production costs of new aircraft families. It’s not easy to understand how these costs arise, how they are booked in the OEM’s accounting and how they can be compared. Time for a primer.

I will not duplicate a course in company accounting, but it can be worth the read to understand how costs are created, accounted for and what we as externals can observe via aircraft industry economic reports .

I will focus on Airbus and Boeing. These are good examples of the different ways of collecting and showing costs in the global aircraft industry.

Development costs

Development costs are handled in two ways in the aircraft industry. The first and simplest, is where development costs are booked as costs when they occur (costs are expensed in the quarter). This means development costs that have been accrued in the quarter are shown as development costs in the financial report for the quarter and finally in the annual report. The alternative way is to defer development costs to inventory as e.g. “tooling cost”. The capitalized cost will later be amortized over, fof example, the first 400 aircraft. This is allowed under International IFRS accounting rules but not under US GAAP.

Development costs are predominantly costs for engineers working on the design of the aircraft, i.e., man-hour costs. It is not too difficult to estimate these costs. An engineer working for a year in Airbus or Boeing costs roughly $0.1m, so 10,000 engineers working on the design of an A350 or 787, cost the company ~$1bn per year. There are other costs like material, equipment and tooling, but man-hour costs are dominant in today’s design of aircraft. One gets a good understanding of a project’s “burn-rate” by looking at how many people are engaged in the project and assuming $0.1m per person and year.

It is now clear why project delays are so feared. A delay means the allocated man-hours, say five years into the project, have been consumed but the work result is not where it should be. Some key certification points have not been earned or design problems remain to be solved that should have been solved. Reaching the needed work result means more man-hours and therefore cost.

So to summarize; for the development of an aircraft “time is money” to the tune of roughly $0.1m per person and year. The costs are taken and shown in the time-period when they occur under US GAAP. Under IFRS it’s up to the company to decide if the costs are expensed (taken in the quarter) or capitalized to inventory, where they are amortized over a production run.


Production and preparation for production (like creation of production tooling, etc) is handled in two ways, with a mixed alternative possible.

The first is to expense costs when they occur. This is called Unit Accounting. The actual accrued costs are then shown in the quarterly and yearly financial reports as Cost of Goods Sold (COGS). This is the way that Airbus handled production costs until the A380 and A350 projects. For these two projects, select early customer orders are handled like Boeing’s program accounting (see below), but only for the aircraft inside the customer order. For all other orders, the actual production cost is expensed in the quarter. With the exception for these initial customer deliveries, the quarterly and yearly results are in close fidelity with how money has flown in the company during the time period. From an external viewpoint, it means that the quarterly and yearly bottom line includes all the revenues and costs that have been accrued during the time period. There is little or no overhang to the future.

From an observer’s point of view, it is difficult to know how much the different aircraft programs have cost the company. Airbus shows its revenues and costs on a division level but not on a program level. It is therefore not possible to know how much the first A350s cost Airbus to produce; it is hidden together with the margins that Airbus makes on other programs like the A320 and A330.

Boeing handles tooling and production costs differently. It packages every major aircraft program into its own accounting system called “Program Accounting.” Based on experience and knowledge of the production “learning curve” for similar aircraft programs, Boeing estimates the average production cost for an “accounting block” (for the 787 right now, the first 1,300 aircraft). When Boeing sells an aircraft like the 787, it takes the aircraft sales revenue and deduct the Program Accounting’s average production cost for the first 1,300 units as COGS (Cost Of Goods Sold). The result is the revenue, cost and profit which is show in the quarterly and yearly reports.

The method shows a higher profit per aircraft for the initial production run than if Unit Accounting was used. For late production units the pendulum swings the other way. Unit accounting shows a higher profit per aircraft, as program accounting is now adding the amortizing of the deferred costs to the current production costs to form COGS.

The difference between the real production cost and the Program Accounting’s average cost, is called deferred costs. These are booked on deferred costs accounts and are shown in the balance sheet of the OEM’s reporting under Inventory. By following the movements of costs for this account (actually two accounts, one for production and one for tooling, right now at about $30bn combined) and Boeing’s comments on deferred cost development, it is possible to understand the real production cost curve of the 787.

Real production costs

The real production costs of an aircraft like the 787 or A350 are very high for the first units. The financial data which is presented in connection with Program Accounting makes it possible to understand the real production costs of the 787 with good precision. I will give examples of figures that can be deduced from such data. The key take-away is not the figures but that fact that Program Accounting gives the external person a chance to understand the true production costs of an aircraft whereas unit accounting doesn’t.

Now to the example and how one can understand the production cost of the 787. To reach a production cost estimate, one has to assume a learning curve (Wikipedia link for those interested). The learning curve describes how different costs decline when production volumes increase.

The average production cost for the first 50 787 units can be estimated from Boeing’s accounting as $450m a copy (the first units cost over $1bn). The production cost of unit 50 can be estimated to $200m. With a learning curve of 85%, which is the normal in the industry (and which seems to be followed by the 787 after the initial troubled aircraft), the cost for aircraft 100 shall be 15% lower, or about $170m per aircraft. Double that again to unit 200 and we are at $145m. Right now, unit 330 goes out the door, which should have a production cost of around $125m. These values can be a bit off but not by much; the accounting around Boeing’s deferred production costs makes the figures pretty predictable.

As can be seen, Program Accounting gives the external observer a good idea of the real production costs in the aircraft industry. The market’s appraisers publicize the present market values for newly produced aircraft. This gives a good idea of the net prices for the aircraft. Right now a new 787-8 is valued at $120m, so Boeing would make a loss of about $15m per copy and go to cost/revenue break even next year. This is in line with Boeing’s financial guidance.


Airbus and Boeing represent the two different ways of how the costs of production aircraft can be presented. Of the two, the more involved method, Program Accounting, contains a number of assumptions over future costs and the learning curve effect. At first glance this makes the understanding of the true costs of the aircraft program less transparent.

As the practice leads to the company presenting the deferred costs each quarter, combined with statements of how fast the cost break even point will be reached, Program Accounting enables an analysis of the production costs that Unit Accounting doesn’t.

24 Comments on “Bjorn’s Corner: Production costs.

  1. Paradigms and (fault)mechanisms from Software engineering have long made it into the realm of hardware manufacture.
    Thus I’d like to bring the mythical man month forward as a good and educative read:
    ( take the link to the preferred file format from there )

    Specifically note the notes on the cost of delayed exposure and the delayed fixing of defects. Incurred costs and time consumed grow exponentially with programm progress
    and bringing more manpower to a late project ( which makes it later still )

    • Not just the Mythical Man Month essay, everything Fred Brooks wrote should be mandatory reading for anybody in a management role.

  2. “An engineer working for a year in Airbus or Boeing costs roughly $100,000”

    either they are paying insanely low salaries, or you are not talking about “fully burdened costs” but just raw average salary. real cost per employee includes things like health care, facilities, training, indirect labor (managers, secretaries, HR, Security etc) and a whole host of other costs. quick rule of thumb is actual cost of direct labor employee is 1.8-2x salary.

    • Hi bilbo,

      first I wanted to have a very round figure.

      Secondly, we talk about the large numbers of design engineers that comes in when you start the detailed design at the OEM or supplier such as parts design, coding software or QA work. It is also a function where these work, a lot of locations for the OEM and sub-suppliers are at low cost locations. Initial design and program management are typically done by more experienced engineers and at higher cost locations, but these constitute low numbers in comparison.

      Re what is included in your man-hour, it varies from country to country but seldom you include health care and indirect labor in the countries and companies I have worked for. Management and support functions are accounted for as part of the development workforce in my thinking and are not counted as burden on an design man-hour.

      But in the end it is not if the man-year is 100k or 150k, the message is that the predominate cost in all development is man-hour cost, at the OEM or sub-supplier. Even with deliveries of hardware to prototypes etc they have a very high man-hour content, actual material costs becomes more visible as the learning curve has been driving down man-hour content in the product.

      • you make a good point regarding Healthcare costs, potentially a huge cost advantage for European countries. in good old ‘Murica, each company gets to balance the cost and quality of their employee’s health care coverage vs. Executive profit sharing bonuses.

        regarding sourcing engineering services in “low cost” regions, at least in the Software industry, most studies have shown that for tasks more complex than tech support call center type activities, the added supervision, travel, communication and time costs of outsourcing results in negligible or non-existent cost savings, however from a technical perspective it introduces a diversity of thought and access to sheer numbers of engineers that brings a different sort of value to the team.

  3. Hi Bjorn,

    Great post, very clear explanations! During the Airbus investor days this year, they of course announced that they are moving the early A350 deliveries (2014-2018) to contract accounting. Could you provide an update on this and how it affects profitability vs. cash flow for Airbus over that time period?


    • Hi,

      they do average cost for the first A350 per customer contract, i.e. for Qatar Airways they don’t charge the first units with the costs I indicated but the average cost of the 43 they have ordered. This has an effect for the first years but it depends on order size and delivery span for the order. For Vietnam Airlines which has ordered 14 and get them pretty early, less of an average effect. It is the same form of accounting as Boeing does but on an order level rather then the whole program (15-50 units in the average block instead of 1300). The reason is the same as for Boeing, the first off the line cost around $500m to produce and you want to spread that over a few quarters/years. Boeing spreads it over 10-15 years.

      According to UBS it means Airbus stuffs $300m during 2015 in the balance sheet inventory line instead of to the bottom line (seems low IMO).

  4. Well done Bjorn and no I don’t dread reading your submittal’s as they are as clear as a complex subject can be made.

    If they broke the program costs out of the heard could you not do the same thing with production costs method?

    Do we have a comparable program in Boeing and where they stood and how they did to compare it to the 787?

    Just from memory it seems the 787 is by far the worst program they ever did with break even further down the road.

    Or another way, typical costs for a 767 at that stage?

    • Hi TW,

      the closest comparable program to 787 where you could get the deferred costs would be the 777. One would have to get hold of Boeing financial filings for the relevant years where one can find the deferred costs relevant to the 777. For the 777 one does also know the learning curve improvement %, it was 84% i.e. 16% cost improvement when the production numbers doubled, the accounting block was 400, then 500 and finally 600. I don’t know of anyone who did the job but financial analyst should have.

  5. Bjorn, Please explain my fundamental misunderstanding. I ask, as a US CPA . Your article is conceptually solid, for a non-accountant, but your practical differentiation between develoopment and production appears substantially wrong, by many years and $Billions.

    For the 787, you say that Boeing would have started accruing production costs upon “preparation for production (like creation of production tooling)”. That would have been after the firm configuration, which Boeing announced on 9/23/05 (Wikipedia), and probably after the subsequent design freeze (date unknown).

    But my accounting training tells me that Boeing would have started accruing 787 program costs many years before, when the 787 became a real product, rather than just a research project. That would probably be much earlier than the 2003 announcement date (Wikipedia).

    For the current hot program, the MOM, Boeing may have already started accruing and defering project costs. If not now, soon. Boeing is likely close to finishing research, and defining MOM’s basics. After all, the popular wisdom is that Boeing will announce and offer the MOM within a couple of years.

    Remember management’s incentive to start accruing as early as possible, through the balance sheet, which helps today’s earnings and today’s stock price and todays’ management bonuses.

    • Hi Bill,
      here from a 10-K filing from Boeing:

      Program Accounting
      We use program accounting to account for sales and cost of sales related to all our commercial airplane programs by the Commercial Airplanes segment. Program accounting is a method of accounting applicable to products manufactured for delivery under production-type contracts where profitability is realized over multiple contracts and years. Under program accounting, inventoriable production costs (including overhead), program tooling costs and warranty costs are accumulated and charged as cost of sales by program instead of by individual units or contracts. A program consists of the estimated number of units (accounting quantity) of a product to be produced in a continuing, long-term production effort for delivery under existing and anticipated contracts. To establish the relationship of sales to cost of sales, program accounting requires estimates of (a) the number of units to be produced and sold in a program, (b) the period over which the units can reasonably be expected to be produced, and (c) the units’ expected sales prices, production costs, program tooling, and warranty costs for the total program.

      As you can see there is no development cost included which comes after research and initial studies.

  6. Bjorn, Thanks for your response, which doesn’t answer the question. You apparently read “inventoriable production costs” narrowly, ie, production is cutting metal/composites. I read it expansively, ie, all the steps leading to cutting metal/composites. All those steps may go back many years for a huge and complex product like a commercial aircraft.

    I’ll keep looking for an answer. But any textbook answer will be quite general, and subjective. Only the finance insiders at Boeing know how they really implement it. The stock analysts will also have some idea. They may be honest about the history of the 787. I wouldn’t so much trust what they say about a current program like the MOM, because they are trying to make money in today’s stock market.

  7. In a post to an earlier related story, I had calculated that, in order to break even, B needed to shave an average of $ 53 to 60 MM on AVERAGE on the remaining 900 or so aircraft in their cost model.
    That means they need to save around $90 million on the last few aircraft of the full run of 1300.
    Ergo, if current cost is $130 MM as postulated in the article, the cost of the last aircraft in the run would have to be $40 MM all in. That’s a $90 million saving or a little over 75%, while the learning curve will only yield around 30% with the doubling and redoubling of the production run… I think B has a problem???

    • @JP Gorge:
      Why complicate things that much. Couldn’t it be explained this way: Boeing has invested about 30 billions on the B787 first 350 airplanes, they will have to more or less make a 30 million profit on each of the next 1000 planes to recoup this money. Probably very unlikely that they would make 30 million profit on each one, but some of them (specially the B787-10) could probably make 50 million profit. So it all depends on how many more (and at what profit margin) Boeing can build and sell the next 1000 B787 aircraft. If potentially they sell more than that, then it should be easier to recover all the money invested in it.

      • unfortunately it is much worse than that. you need to use the Net Present Value model to truly understand how much Boeing needs to profit to pay off that $30B hole they have dug. Just like a credit card, you need to figure the interest on the $30B and treat per unit cash flow as either a charge or a payment against the base debt. this means that Boeing needs to make much more than $30B in per unit profits over the next 20 years to pay off that $30B.

        of course, this completely ignores the fact that all future Boeing programs will benefit from the technology developed, infrastructure built and lessons learned in the 787 program.

        one of the big problems with program accounting is that it is easy to hide things that are really company wide expenses (fundamental R&D, infrastructure improvements etc) in a single loser program, which then makes other programs look better. when evaluating these things you have to not only look at the program in question, but also at the company as a whole to get a full story.

        • “of course, this completely ignores the fact that all future Boeing programs will benefit from the technology developed, infrastructure built and lessons learned in the 787 program.”

          This assumes that 787 technology is used in future programs.

          It’s far from certain that NSA will use CFRP barrels and MoM (or whatever it’s called this week) is being touted as a traditional Al fuse with CFRP wings.

          Is there any real-world significant benefit from the 787’s bleedless, all-electric architecture?

          Some of the 787 technologies and learnings may be very expensive in the long-term (if they’re never used again).

  8. For Boeing 787, wasnt a lot of production outsourced ( along with some design), so its likely the total production cost is far far higher than just Boeings share ?
    What about the wing plant in Japan, the fuselage barrel plant in Italy ? They are not owned by Boeing so are ‘off their books’. Then there was the Vought barrel factory in South Carolina, Boeing bought them outright, so where does that large cost figure?

  9. They still have to buy those parts not made in Boeing plant, and those mfgs have to make a profit. Cost Boeing more it would seem

    Having seen the state private relationship thing the State always paid more for their side of the partnership than using own employees.

    Boeing probably saw a short term gain vs long term project costs that stay high and you have to try to manage through another company to bring down.

    And what are they doing to do, have someone else make the wings?

    And none of it addresses the overhead Boeing failed to understand (or did not want to admit) it would take to manage all those providers let alone scattered all over the world so you could not use one team to deal with multiple suppliers, they had to form teams to get to all of them and as I recall it included all the specialty types (finance, logistics and technical ) as it was all intertwined. Phew.

    Probably the biggest driver in the whole mess was complete failure to understand you can just snap parts together at an assembly plant.

    You not only have to have the specialty to under4stand if its been designed right, you also have to manage problems and some downright complete failures (Chance Vought)

    that in turn led to billions invested in an all new facility that is also attached to the books when they could have done it in Everett (or course mgt does not talk about that part)

  10. Hi Bjorn

    I am interested to hear that you believe program accounting gives greater transparency program by program. The concern I have is that this method introduces subjectivity into what should bean objective exercise. In simple terms the costs booked are an estimate over the project life with a profit margin tacked on top. The OEM has vast scope to manipulate any number in terms o f learning curve rate, production cost, margin ( a figure kept secret) and accounting block. This allows the OEM to almost decide when they want to book profits.

    Boeing was accused of manipulating 777 program numbers to avoid having to realise a loss back in the 90s and I think there is some suspicion that they may be sailing close to the wind with the 787.

    Accounting should be about keeping score and not about profit manipulation. Douglas found that out the hard way back in the 60s

    • Hi Bob,

      I said it is easier for an EXTERNAL person to get the handle on production costs. Re your other points re program accounting I agree. I have done enough programs (small and large) in my days to know how one can present what one want to present (within margin) and program accounting leaves a lot of freedom to lean on parameters. I have also read the writeups re the 777 problems and yes I agree, it will be tough for Boeing to recover $30bn over the remaining 970 787. My article was about presenting the principles and I wanted to keep it clean of my opinion on things.

      Re the 787 the most likely event is not the announcement of a forward loss but a further extension of the accounting block. There is a lot of milage in the 787 platform for many years to come so this is plausible.

      • Hi Bjorn, thanks for that

        The other thing that struck me was how impenetrable the Airbus accounts are by programme. I suppose this has been used to obscure the programmes with problems such as a400 and a380. At the same time the a350 appears so far to be a robust programme with very substantial opportunities to make serious money, something also masked.

        I am surprised that the main product lines are not reported segmentally. I suppose the argument goes that the development cost of the a380 (say) has already been written off and as such is not relevant to an investor decision. At the same time it should give a much clearer understanding of programme ROI and not just FCF.

        The Airbus approach to accounting at least makes things a tad easier to focus on the critical aspect of likely cash flow generation.

        • The Airbus approach to accounting at least makes things a tad easier to focus on the critical aspect of likely cash flow generation.

          What you note is the most basic objective of reported accounting.
          A _conservative_ illumination of a corporations financial whereabouts. A snapshot.

          Reporting on future perspective is a different layer.

          What we see practiced from the US side first ( commerce and politics ) interweave fact and fiction as closely as possible. Fake information as close to the source as possible. ( prominently practiced for handfeeding lawmakers in the runup to the iraq war
          with foundational information for decission making.
          Stockmarkets work on the same principle.

          • Lectures from Europe on economics and foreign policy… priceless!

Leave a Reply

Your email address will not be published. Required fields are marked *