How Much Will JSF Cost?

When the Pentagon's top buyer appeared before the Senate Armed Services Committee, most observers expected Ash Carter to tell lawmakers just how much each F-35 costs and how much the plane is likely to cost over time. That didn't happen. Winslow Wheeler, a bipartisan conagreassional defense budget expert now at the Center for Defense Information, penned a detailed analysis and commentary picking apart the Pentagon's numbers and their underlying assumptions. Winslow's commentary follows.

When the Pentagon’s top buyer appeared before the Senate Armed Services Committee, most observers expected Ash Carter to tell lawmakers just how much each F-35 costs and how much the plane is likely to cost over time. That didn’t happen. Winslow Wheeler, a bipartisan conagreassional defense budget expert now at the Center for Defense Information, penned a detailed analysis and commentary picking apart the Pentagon’s numbers and their underlying assumptions. One area sure to spark disagreement is his discussion of F-35 production. This plane is supposed to be the first  advanced fighter built on the closest thing to an assembly line since World War 2. The Pentagon and Lockheed Martin say it will not be hand built, as was the F-22. But Wheeler argues that stealth materials will make it virtually impossible to build F-35s relatively quickly and efficiently. Winslow’s commentary follows. — Colin Clark

Ashton Carter, undersecretary of Defense for acquisition, technology and logistics, and Christine Fox, director for Cost Assessment and Program Evaluation, presented new unit cost estimates for the F-35 during two recent hearings in the Senate and House Armed Services Committees.  Those estimates are extremely optimistic and very incomplete.

The F-35 unit cost estimate is incomplete because the $114 million to $135 million “Average Procurement Unit Cost” (APUC) Carter and Fox announced, in “then=year” dollars, to buy 2,443 aircraft does not include any research, development, test and evaluation money for the F-35.  The best available estimate of those additional development costs is about $60 billion (to add to the estimate of $278 to $329 billion to produce the F -35s).  Including those costs would add about $25 million to the cost of each aircraft, making the Carter-Fox total program unit cost somewhere between $139 million to $160 million.

It may be that Carter and Fox are unwilling to testify to a total program unit cost because they are unwilling to inflict further “sticker shock.”  Presumably, the official, more complete numbers will be made available later in April when the Defense Department releases its new Selected Acquisition Report (SAR), now about 18 months late.  What Carter and Fox thought they had to gain by delaying the more complete revelation does not merit speculation; their existing (incomplete) production unit cost estimates have little to do with reality.

The 2011 budget request for the F-35 plans to buy 43 aircraft for $8.654 billion in procurement funding.  That makes for a production unit cost for the 2011 buy of $201 million per plane.  In his March 24 testimony to the House Armed Services Committee, Carter stated that the unit cost “will decrease significantly” from this level as purchases increase and production processes “optimize.”  This is consistent with conventional wisdom that there exists a “learning curve” for aircraft production that progressively shrinks unit cost steadily as production proceeds.  Thus, Carter and Fox argue, F-35 unit production costs will come down from the currently unsettlingly high number of $201 million each down to the $114 to $135 million band.

The last 50 years of DOD aircraft cost history, especially of “stealth” aircraft, do not treat the Carter-Fox estimates — and the prevailing conventional wisdom –very politely, however.  The absence of any progressive “learning curve” in unit cost has been thoroughly demonstrated by the analysis of Franklin “Chuck” Spinney, using actual procurement data.  (Read this  work at  In the case of the F-35, we can test the likelihood and amount of “learning curve” reduction in the unit cost by comparing the F-35 at this point in its program history to its closest relative, the F-22.

Indeed, the F-22 program is an excellent precursor for the F-35.  Both aircraft are “fifth generation” aircraft that combine “stealth” with complex long-range radar systems (the F-35 adds an extra emphasis on air-to-ground functions). Both rely heavily on extensive computerization (the F-35 includes significantly more software).  Both programs employ concurrent development and production (the F-35 schedule incorporates even more production before the end of development).  Both are from the same prime contractor and to a large extent the same aviation bureaucracy in the Pentagon (the F-35 adds two bureaucratically required complications: STOVL and carrier operations).  There are no other contemporary US aircraft with a more closely related design, production, and bureaucratic heritage.  Due to its more complex nature, the schedule and cost of the F-35 can be expected to experience more delays and increases in the future than the F-22 did.  In other words, using the F-22 “learning curve” should underestimate future F-35 developments.

Based on annual reports from the office of the DOD Comptroller, “Program Acquisition Costs by Weapon System,” showing annual appropriations for F-22 production, we can track the annual amounts paid for F-22 production.  Some interesting points emerge:

First , unit procurement cost for the F-22 roughly leveled out, with little cost reduction (learning) thereafter, by the fifth production year of the program.  Across all F-22 procurement, the average unit cost based on actual appropriations calculates to $197 million per copy.  At year five, the unit procurement cost was essentially the same: $201 million per unit.  (Note also, at that point in the program, year five, a total of 54 F-22’s had been acquired.)

Second, toward the end of the program, the learning curve went backwards as unit procurement costs went back up.  When the learning curve did so, it was during the three year period of the F-22’s “Multiyear Procurement Buy,” which is thought by conventional wisdom to reduce costs, not increase them.

Using the same data from the same sources, we can conclude the following on the F-35:

First, since the second year, F-35 unit procurement costs have roughly leveled off at the same levels as for 2010 and 2011, $227 and $201 million per copy, respectively.

Second, by production year four (2010), 58 units have been produced, which compares to the 54 F-22 units that had been produced when that program achieved relative production cost stability.  If you argue the more complex F-35 requires more production experience to effect “learning,” the 101 units produced by the end of 2011 should easily suffice; thereby making the unit production cost $201 million per copy.

Carter, Fox, and other advocates of the F-35 will contend the F-22 cost experience is irrelevant.  First, the F-35 will see a much longer production run than the F-22, affording time and opportunity for learning and optimization of production – the old learning curve argument. Second, they will lean on how well they are solving the currently horrendous F-35 production problems: that is, all the out of station work; missing, late, and non-fitting components; redesign, etc., etc., etc. pointed out by the Government Accountability Office, the Defense Contract Management Agency, and the Independent Manufacturing Review Team: A litany of problems just like those of the F-22 production line.  Surely, they will say, once this legion of problems is addressed, we will see more efficient, cheaper production.

Not so fast.  Addressing many of the current assembly line problems assumes a stable design for the F-35.  We are a long way off from that; we may never get there.  (Indeed, we never got there with the F-22 and are still modifying produced units.)   Not only have recently uncovered design fixes not yet been incorporated into production, but there are certain to be many modifications imposed on the aircraft design as the F-35 contorts through is initial flight testing, now only 3 percent done.  In other words, a stable enough design to produce “optimally” is years off.

Moreover, once the current production “glitches” and fixes based on test flights are resolved, the production turbulence is not over.  As Spinney has pointed out for the F-18, F-16 and other aircraft, the changes never stop.  Engineering change proposals, upgrades in the form of new production blocks, product improvements, and new requirements from the user never end.  The F-16 is now in a “Block 50” modification, which about doubles the cost of the early block F-16s.  Modern tactical aircraft procurement programs never really allow a design to stabilize, a primary reason why Spinney found the “learning curve” to be illusory.

A downward learning curve for the F-35 is likely to be an even greater illusion; it will have no assembly line in the paradigm of World War II production.  The fabrication of stealth aircraft is inherently unsuited to assembly-line production – something I learned when I visited the Lockheed Forth Worth plant to observe the infinitesimally precise, hand-labor intensive riveting for the F-22 mid-fuselage section (essential to meet the stealth requirement).  Just riveting a single F-22 mid-fuselage costs 30,000 man-hours of hand labor.  An additional stealth cost burden is getting the stealth skin coatings right.  From the moribund Advanced Cruise Missile to the F-117 to the B-2 to the F-22, each and every stealth creation has had serious problems meeting its radar cross-section thresholds.

Overall, completed F-22s required almost $200 million in modifications in the 2010 Air Force budget alone, an expense that will continue and very probably grow.  It is this kind of ongoing turbulence that ensures the likely growing nature of future F-35 production unit costs; just for it to remain at $201 million per copy would be extraordinary.

What of the 2,443 F-35s for the US and the 730 for allies now scheduled?  The simple answer is that such a long production run will not occur.  The DOD budget has no room for the significant increase in annual production spending that the F-35 plan requires, even if the unit cost does not increase.  Inevitably, the annual production quantities will be squashed to a level the budget can accommodate; one expert predicted to me the annual buy will have to come down from 80 or more to 50 or less.

Indeed, for the last half century, higher than anticipated unit costs have led to production stretch outs which inevitably lead to further cost growth.  That is the inexorable “death spiral” that underlies progressively smaller production runs, higher costs, and shrinking, aging forces.  Nothing the Carter-Fox team has done is changing that.

The process has already begun.  Just this week, based in large part on the admitted $114 to $135 million unit cost, Denmark deferred its decision to replace F-16s for two years.  The United Kingdom, the Netherlands, Norway and Australia are all witnessing controversies that are likely to delay and/or truncate their F-35 purchases.  But most importantly, at home in the US, factions in the US Navy are openly horrified at the crushing costs of the F-35.  That, together with traditional Navy distaste for Air Force-dominated programs, may well lead the Navy to back out of the program, s it did in past decades with naval variants of the F-111 and the F-16.

So, the question is not whether F-35 production will shrink, but by how much.  Publicly admitted unit costs will go up; purchases by allies and ourselves will go down.  The costs will increase further, and so on.

F-35 unit cost is far more likely to stay around $200 million per copy, or to go up, than it is to reduce to the levels Carter and Fox now predict.

It also worth remembering that the $200+ million unit cost anticipated here is incomplete.  An accurate sticker price includes the total cost of development, testing, facilities and other factors amortized across the ultimate size of the fleet.  With the fleet size shrinking by some currently unknown, but very substantial, factor, the unit cost for the total program is sure to grow to even more horrifying levels.

Whatever that final unit cost may ultimately be, to predict it now will surely be met with gales of derisive laughter from the advocates of this ongoing disaster — gales that will last only until the actual bill arrives on their doorstep.

Something between $250 and $300 million?  Start laughing.