A Cycle of Misery: The Business of Building Commercial Aircraft
"There are no historic precedents or current parallels for the magnitude of financial exposure risked by an American airframe company" – George Ball, managing director at Lehman Brothers, 1982
"You can't win, you can't break even, and you can't quit" – Jean Pierson, former CEO of Airbus
Boeing is once again in the news following an incident on Alaska Airlines Flight 1282, where an emergency exit door plug fell off mid-flight due to missing or improperly installed bolts. The aircraft, a 737-MAX9, is from the same 737 MAX family that suffered two fatal crashes in late 2018 and 2019.
Predictably, the incident triggered a new round of discussion about the decline of Boeing as an aircraft manufacturer. Nearly every source points to the same instigating event: the 1997 merger with McDonnell-Douglas, which changed Boeing from an engineering-driven company focused on building the best airplanes possible to one which focused overwhelmingly on financials and stock price.
As far as I can tell, this characterization is accurate. And yet it misses a big part of the picture: the brutal structure of the commercial aircraft industry that drives companies like Boeing to create things like the 737 MAX (an update to a 50-year-old airplane) instead of creating a new model from scratch. By unpacking how the commercial aircraft industry works, we can better understand Boeing’s behavior.
The risk of developing a new aircraft
Making commercial aircraft (large jetliners purchased by commercial airlines) is in some ways like any other manufacturing industry. A company develops a product and tries to sell it for enough to cover the costs of developing and producing it. If it’s successful and turns a profit, it goes on to develop new products; if not, it goes out of business. What sets the aircraft industry apart is the scale at which these things take place.
Developing a commercial jetliner is incredibly expensive. The budgets for new aircraft development programs are in the billions of dollars, and the inevitable cost overruns can drive development costs to $20-30 billion or more.
This is not simply a case of modern, bloated companies that have forgotten how to do things efficiently (though there is some of that): developing a jet airliner has always been expensive. Boeing spent anywhere from $1.2 to $2 billion to develop the 747 in the late 1960s (~$10-20 billion in 2023 dollars), and other manufacturers at the time like Lockheed and McDonnell-Douglas noted that their own new aircraft development costs were similar.
The cost of developing a new commercial aircraft can be a significant fraction of, if not greater than, the entire value of the company. The $186 million Boeing spent developing its first jet airliner in 1952, the 707, was $36 million more than the company was worth. When Boeing began development of the 747 in 1965, the company was valued at $375 million, less than a third of what it spent on the 747’s development. Most other programs aren’t quite so lopsided, but still represent enormous risk. The Boeing 777 cost an estimated $12-14 billion to develop at a time when Boeing was worth in the neighborhood of $30 billion. And when Airbus launched the A380 program, it budgeted $10.7 billion, half the value of the company (and much less than was ultimately spent). Aircraft manufacturers are frequently betting the company when they decide to develop a new jet model.
Spending billions of dollars on new product development isn’t unique to the aircraft industry: a new car model will cost billions of dollars to develop, as will a new drug. But both cars and drugs can spread their development costs over millions of product sales. The market for commercial aircraft, on the other hand, is much smaller; only around 1,000 large jets are sold every year. Aircraft manufacturers need to be able to make back the billions spent on product development, factories, and tooling on just a few hundred sales.
This creates a few difficulties for aircraft manufacturers. For one, it makes learning curves very important. As discussed previously, learning curves are the phenomenon that production costs (or some related measure such as labor hours) tend to fall by a constant percentage for every cumulative doubling of production volume: going from 10 to 20 units produced yields the same percentage cost decrease as going from 10,000 to 20,000.
High-volume products spend most of their time on a relatively “flat” portion of the learning curve, where doublings are farther and farther apart. If you’ve produced 1,000,000 of something, whether you make another 500 or 5,000 will make almost no difference in learning curve terms. But if you’ve only made 50 of something, making another 500 makes a huge difference in the level of cost reduction that can be achieved. Thus, if you only plan to sell a few hundred of something, a relatively small number of sales will have a large impact on how efficiently you’re producing and how profitable you are.
Commercial aircraft manufacturers rely on getting enough orders to push them far enough down the learning curve where they’re making enough money per plane to earn back the costs of developing it. Early planes might be produced so inefficiently that they’re sold for less than it costs to make them. This is typically in the neighborhood of 500 aircraft (and might be much more if a program goes way over budget); sell fewer, and the program will lose money.
The relatively small number of aircraft sales also creates intense pressure to accurately predict trends in air travel. Manufacturers need to skate where the puck will be: to develop the kind of aircraft that airlines will want for many years in the future.
Guessing wrong can be disastrous. Airbus lost enormous amounts of money when it misjudged the market for its enormous A380 (it only sold 251 planes, far short of what was needed to break even, and the last A380 rolled off the line in 2021). Airbus projected that the continued growth in air travel would create demand for an even larger aircraft that could move passengers cheaply between large hub airports. But in fact, international travel fragmented, and airlines increasingly fly direct flights between destinations using smaller, easier-to-fill aircraft like the Boeing 787. In the late 1960s, Lockheed and Mcdonnell-Douglas drove each other to ruin by each developing a new aircraft (the L-1011 and the DC-10, respectively) for what turned out to be a very small market: less than 700 aircraft were sold between them. Lockheed ended up abandoning the commercial aircraft market after losing $2.5 billion on the program (~$9 billion in 2023 dollars), and McDonnell-Douglas never recovered, ultimately selling itself to Boeing in 1997 as its market share declined and it was unable to fund the development of new aircraft.
But guessing right comes with its own problems. If a program gets delayed, that can cause lost orders, lack of confidence from the airlines, and ultimately drive customers into the arms of a competitor. Boeing originally planned to introduce its 787 in 2008, but it was delayed until 2011, driving customers to purchase the competing Airbus A330 (Airbus bragged that it sold more A330s after Boeing launched the 787 than before). Had Boeing delivered on time, its advantage over Airbus would have been enormous.
And too many orders for a new plane can be almost as bad as too few. In the late 1960s, Douglas choked on unexpectedly high demand for its DC-9: it couldn’t meet its delivery schedules, was forced to pay compensation to the airlines affected, and was almost forced into bankruptcy due to cashflow problems, resulting in a merger with McDonnell Aircraft. Boeing had a similar struggle when trying to rapidly ramp up production of its revised 737 (called the 737 Next Generation, or NG) in the mid 1990s: it was ultimately forced to temporarily stop production due to the chaos, resulting in late deliveries (and associated penalties) and a net loss for the year 1997, the company’s first since 1959. Boeing is estimated to have lost a billion dollars on the first 400 737NGs, despite them being derivative of an aircraft Boeing had been building since the 1960s.
Global events can quickly shift trends in air travel, completely changing the types of aircraft airlines want to buy. Airbus, for instance, had little initial success selling its first model, the A300. By 1978, four years after its introduction it had sold only 38 of them. But increased fuel prices from the second oil crisis, along with increasing competition from airlines, created a demand for a fuel efficient twin-engined widebody airliner, and only the A300 fit the bill. By 1979, sales had climbed to over 300. Similarly, deregulation of the American airline industry forced airlines to be much more competitive on price and focus on things like fuel efficiency and operating cost. This changed the calculus for the types of aircraft they were interested in buying, and boosted demand for planes like Boeing’s 737.
Often success comes down to luck as much as anything else. Airbus got lucky when spiking oil prices meant its A300 was suddenly in high demand and had no competition. Boeing got lucky with its 737 in the 1960s, which went into service more than two years after the similar DC-9, and only succeeded in part because of Douglas’s production delays. And Boeing got lucky again with the 747, which, not unlike the A380, was an enormous aircraft that few airlines really needed. It only succeeded because Juan Trippe, the founder Pan Am bought them on something of a whim (Trippe liked to have the latest airplanes and “saw no need for market analysis”). Other airlines then followed suit, not wanting to grant Pan Am the marketing benefit of having the largest airliners (though the 747 became more and more useful as international travel increased).
Aircraft manufacturers are faced with the unenviable task of trying to navigate this landscape while putting billions of dollars at risk. But of course, developing new products isn’t an option: like any other industry, competitors are trying to secure their own advantage by pitching their products to a limited number of customers. Aircraft technology is constantly changing, and airlines (in a brutal competition of their own) all want the latest technologies – better aerodynamics, lighter materials, larger and more efficient engines, and so on – that will reduce their operating costs and improve their passengers' experience. Losing even one order can be an enormous setback for an aircraft manufacturer, both because of the small number of customers overall and because sales tend to have momentum: a sale to one airline probably means more future sales to that airline (since there will be efficiencies from fleet commonality in things like shared maintenance and training), or sales to partner airlines, or sales to competitors who want to bet on the winning horse. Airlines are all too aware of the fact that too few sales can put an aircraft manufacturer on dangerous ground, making it risky for an airline to hitch itself to a loser.
If an aircraft manufacturer is able to successfully navigate this landscape, the reward is a paltry profit: between 1970 and 2010, Boeing, the most successful commercial aircraft builder, averaged just over 5% annual profit. Unsurprisingly, fierce and costly competition and miserly profits have gradually driven competitors from the space, leaving just Boeing and Airbus (and, if you’re feeling charitable, Bombardier and Embraer). Companies like de Havilland, Dassault, Lockheed, Douglas, Convair, Glen Martin, have all been driven out or forced to merge. Looking at the history of jet aircraft manufacturing in 1982, John Newhouse estimates that, of the 22 commercial jet airliners that had then been developed, only two, the Boeing 707 and the Boeing 727, were believed to have made any money (though he notes that the 747 might eventually make that list as well).1
The result of these difficulties is that aircraft manufacturers think very carefully about developing a new aircraft: the risks are large, the rewards small and uncertain. It's often a much safer bet to simply develop a modification of an existing model – keeping the same basic airframe and adding more efficient engines or a tweaked wing shape, or stretching it to add more passenger capacity. Revising an existing model can cost just 10-20% of designing a new plane from scratch, and can provide nearly as many benefits. The typical new aircraft might be 20-30% more fuel efficient than existing designs, but Boeing was able to squeeze a 15-16% improvement with the revised 737 MAX. And updating an existing model is also cheaper for airlines, which don’t have to retrain their pilots to fly the new aircraft.
To see what this sort of calculus looks like in practice, let's look at the history of the Boeing 737, which has been revised and updated repeatedly since it first flew in 1967.
Evolution of the Boeing 737
Boeing first developed the 737 in the mid-1960s as a short range, small capacity aircraft to flesh out its product line and prevent Douglas from having the entire low-end of the market to itself with its DC-9. Initially, it was not particularly successful, nor was it expected to be: Douglas had a 2-year head start with the DC-9, and Boeing’s earlier 727 was already serving much of that market, albeit with 3 engines instead of the more efficient 2 engines of the 737 (all else being equal, the fewer engines an aircraft has, the less expensive it will be to operate). In fact, the program was very nearly canceled shortly after it was launched due to poor early sales. To minimize development cost and time, the 737 was designed to share as many parts as possible with the earlier 707 and 727.
Early 737 performance was lower than expected, so Boeing developed an “advanced” version with improved aerodynamics in 1970. But even with these improvements, Douglas’ struggles building the DC-9, and an order for a military version of the 737 (The T-43A trainer), sales were still sluggish. The aircraft was being built at just a rate of two per month, and until 1973 the program was on the verge of being canceled (during this period Boeing nearly went bankrupt due to cost overruns on the 747 program, and had to lay off 75 percent of its workforce). The 737 was only saved because it was finally being sold for less than its production costs, but it was not expected to pay back its development costs.
But sales began to pick up in 1973, and production had reached five aircraft per month in 1974. By 1978, it had against all odds become the best selling jetliner in the world, a title which it retained from 1980 to 1985. Airline deregulation in the US had caused a shift in airline strategy: instead of directly connecting two cities with low-volume flights, they began to connect through hub airports, using smaller, cheaper to operate aircraft. The 737 fit the bill perfectly.
But Boeing’s competitors weren’t standing still. Douglas launched an updated version of its DC-9, the Super 80, with an improved version of its Pratt and Whitney engine that made it quieter and more fuel efficient than the 737. To counter the threat, and to deal with ever more stringent noise regulations, Boeing responded with the “new generation” Boeing 737-300, which began development in 1981. This version of the 737 added passenger capacity, improved aerodynamics, and had a new, more efficient high-bypass turbofan from CFMI (A joint venture between GE and the french company SNECMA).
Fitting such a large engine under the 737’s wing was a challenge. The 737 had been originally designed with low ground clearance to accommodate “second tier” airports with somewhat limited stair systems. Extending the landing gear to raise the plane would have required shifting the location of the wheel wells, which would have changed the structure of the plane enough to essentially make it a new aircraft. Instead, the engine was squeezed into the space available, giving it a telltale ellipsoid shape. This high bypass engine gave the 737-300 an 18% improvement in fuel efficiency versus older generation aircraft, and an 11% improvement over McDonnell-Douglas’ Super 80, while still keeping it as similar as possible to the previous 737.
But as the 737-300 took shape, a new challenger was emerging. Following the success of its A300, Airbus began development of the smaller A320, a direct competitor to the 737, in 1984. The A320 incorporated many advanced technologies, such as fly-by-wire, replacing heavy mechanical or hydraulic linkages between the aircraft controls and the plane’s control surfaces with lighter electronic ones. By 1987, the A320 had already racked up 400 orders, including a large order from Northwest Airlines, a longtime Boeing customer. It was clearly going to be a fierce competitor.
Some argue that Boeing could have (and should have) killed the A320 immediately by announcing a new “clean-sheet” aircraft. At the time, Boeing was working on a 737-sized aircraft called the 7J7, which used an advanced “unducted fan” (UDF) aircraft engine from GE. Theoretically, the 7J7 would have been 60% more fuel efficient than existing airliners, along with incorporating technologies like fly by wire. But the UDF engine had unresolved technical issues like high noise generation, and Boeing was concerned with how long it would take to get the 7J7 to market. Instead, Boeing developed another stretched out version of its 737 (the 737-400), canceled the 7J7 project, and began to develop an aircraft to fill the gap between its 767 and 747, the 777.
But as the A320 continued to encroach on the market and more longtime Boeing customers defected (such as United in 1992), it was clear that a replacement for the 737 was required. Many once again favored the development of a clean-sheet airplane (which Airbus believes would have been catastrophic for the A320, but Boeing was gun-shy about new aircraft after the 777. Though the program was on time and delivered an exceptional aircraft, costs spiraled out of control, up to $14 billion by some estimates ($28 billion in 2023 dollars) against a projected budget of $5 billion. Instead, Boeing launched the “Next Generation” (NG) 737, another update to the 737 airframe. The 737NG featured, among other things, a new wing design, a more efficient engine which reduced fuel costs by 9% and maintenance costs by 15%, and added “winglets” for improved aerodynamics. The 737NG also reduced part count by 33% compared to previous versions, while still retaining enough commonality to require minimal pilot retraining and staying within the FAA’s “derivative” rules. First delivered in December 1997, the 737NG became immensely popular, with the 737-800 version selling more than 5000 aircraft over the next 20 years (though as we’ve noted, ramping up production came with immense difficulties). However, this wasn’t enough to bury the A320, which also continued to sell well. Some Airbus people believe that a clean-sheet aircraft would have been catastrophic for the A320 in the late ‘90s.
By the early 2000s, the 737 and A320 had become the most important products in Boeing and Airbus’ offerings, and combined made up 70% of the market for commercial airliners. Once again Boeing began to consider a replacement for the 737 and initiated a project, Yellowstone, to explore clean sheet replacements for the 737 and other Boeing aircraft. But the findings weren’t particularly heartening: without a new, advanced engine (which wouldn’t be ready until 2013 or 2014), improvements in fuel efficiency would be at best 4%. And the technologies it would incorporate from the under-development 787, such as advanced composites, would be difficult to scale up to the high-volume production required for a 737 replacement.
Boeing had once again become gun-shy about new aircraft due to its experience on the 787, which had gone massively over budget and behind schedule. The new, financial-focused Boeing had been reluctant enough to approve the development of the 787, and was even more reluctant now.
But by 2010, with new engines like the Pratt and Whitney GTF and the CFM LEAP on the horizon, Boeing was leaning heavily towards a clean sheet 737 replacement. Boeing’s hand ended up being forced by Airbus. In 2011 Airbus began work on a re-engined A320 with significantly improved performance, called the A320neo (for “new engine option”), and used it to partly lure away a major Boeing customer, American Airlines (which split a major order between Boeing and Airbus). Airbus believed that Boeing would feel forced to respond with its own re-engine rather than lose more customers while it developed a clean sheet replacement. Customers, for their part, had lost confidence that Boeing could deliver a new aircraft on schedule after the 787 debacle, and also preferred that Boeing launch a re-engine with a better shot of being on time. A re-engine would have nearly all the benefits of a clean-sheet aircraft (~15-16% fuel savings versus 20% for a typical clean sheet), cost perhaps 10-20% to develop, and avoid the costs of airlines having to re-train pilots, as well as things like having to figure out how to produce composite parts in high volumes.
The rest, of course, is history. Instead of a new aircraft, Boeing developed yet another revision of the 737, the 737 MAX. Fitting even larger engines on the aircraft while keeping it similar enough to fall under the FAA’s derivative rules required shifting them very far forward and angling them up slightly, which slightly changed the performance characteristics of the airplane. To keep its performance similar to previous 737s, Boeing created software, MCAS, to try and emulate the behavior of earlier aircraft. The MCAS software, and its interactions with various sensors, ultimately caused two fatal crashes of 737 MAX flights.
I sometimes think about how the boundary of technological possibility is defined not just by mastery of the universe, but by the limits of the economy and the organizations that operate within it. If products are sufficiently complex, and demand is for such small quantities that there's a limited business case for them, we won’t get them, even if they’re physically possible to build.
Nuclear submarines seem close to this boundary: enormously complex weapons that only a tiny handful of organizations on the planet are capable of constructing. Jet airliners seem to be rapidly heading to this outer boundary, if they're not there already. Cost and level of technology required, along with the tremendous risk of developing them and the tiny number of sales on which costs can be recouped, have already whittled the number of providers down to essentially two (though perhaps China's COMAC might eventually add a third player), and there's no evidence that it's getting any easier.
Thanks to Ashwin Varma for reading a draft of this. All errors are my own.
If you’re interested in reading more about Boeing and the commercial aircraft industry, a reading list post with short reviews of the major sources I used for this piece is available for paid subscribers.
Newhouse notes that 15 years after its introduction, the 737 (which went on to be Boeing’s most successful aircraft by far) was “unlikely to make any money,” showing once again the difficulty of predicting trends in aircraft demand (though this is also probably because he assumed it would be replaced by a newer model).