The “digital century series” approach for the next-generation fighter could leave the Air Force with more costly small fleets of aircraft that exacerbate growth in O&S costs and force difficult tradeoffs between capability and capacity. For example, the relationship between total ownership cost and the number of aircraft in a fleet from Figure 10 suggests that five separate fleets of 72 aircraft each (for a total of 360 aircraft) would cost roughly $6.8 billion annually to operate and sustain (in FY 2020 dollars).
In contrast, a fleet of 360 aircraft of one type would cost just $3.0 billion annually. Alternatively, the annual ownership cost of five fleets of 72 aircraft each is roughly the same as the annual ownership cost of one fleet of 1,800 aircraft. While small fleets may be desirable for rapid integration of new technologies into the force and maintaining competition in the industrial base, this approach would likely lead to higher operation and sustainment costs and a smaller force than the Air Force could otherwise afford.
That was Todd Harrison’s report, “The Air Force of the Future: A Comparison of Alternative Force Structures.” There is an associated event which was published as an episode on the CSIS podcast titled “The Air Force of the Future.” Self-recommending.
Note that Todd Harrison’s estimation of O&S cost per aircraft has an equation $160M * TAI^0.5, where TAI = total active inventory. So every time you double the inventory, the unit O&S cost falls about 29 percent.
It’s hard to know in which direction causation flows. After all, lower cost systems will be bought in greater numbers. The Air Force never planned to inventory the F-16 or UAVs in the numbers they did. On the other hand, fixed costs associated with maintaining any system still suggests some causality from inventory to unit cost.
I would also mention that numerous other considerations are absent here, such as the degree to which components are common. The F-15 and F-16, for example, use the same basic engine. And these common components could be rapidly integrated around new designs which are destined for experimentation rather than inventory. Just because we develop 5 fighters doesn’t mean we have to field 5 fighters. It creates competition and allows us to discover information that wouldn’t have been available for a systems analysis. Indeed, as Armen Alchian reminds us:
Resources are not wasted when perfectly sound aircraft are developed and then not procured. In fact, such an outcome is a necessary result of an adequate development program. Failure of such an outcome is absolute proof of inadequate development.
This relates to another issue. Sustainment costs are not preordained by some law of nature described by a regression. Sustainment costs are a function of design choices. A system that focuses on simplicity of design and reliability will certainly have a step-function decrease in cost compared to one that drives in complexity and pushes through test & evaluation using costly engineering fixes, holding constant the inventory. It just so happens that simple/reliable systems are bought in greater number, and complex systems have reduced buys. In other words, the elasticity of demand is not zero. Government quantities purchased is sensitive to price, even if it is driven by a “requirements” analysis.
Nice post, Eric. Treating the historical correlation as causal would be a huge mistake.
The other missing dimension is capability, both planned and actual. The Analysis of Alternatives for a new aircraft assumes that we know what it will cost and what it will be able to do, but in practice neither of those is accurate, and the errors are larger at the high end. When we buy lots of simple, reliable planes, we also get about what we expected and (usually) see less cost growth, allowing us to buy the intended number. When we buy bleeding edge capability, we invariably get less capability than was promised, at higher cost, forcing a reduction in fleet size.