Sometime in the 1970s and 1980s, a strange thing happened to the defense research and development process. An enterprise that had put a nuclear submarine to sea with 16 long-range nuclear-tipped ballistic missiles in four years, and taken a spyplane from drawings to operational missions in two years, seemingly became incapable of developing bigger than a primary trainer in less than 20 years.

Producing the Air Force's next bomber in 15 years (from its 2010 restart) is considered a challenge, even though in many respects it will be a smaller B-2 and the requirements have been ruthlessly pared down.

This glacial pace has many consequences, all negative with one exception. It raises costs. In-service systems must be rebuilt and upgraded. Fewer programs can be funded and executed, so the knowledge base grows weak and rusty. Talented people leave for more dynamic industries.

The one exception is that the companies that carry out this work make money. It's easy to wonder whether R&D that is directly funded by the customer has become a revenue stream in its own right, so contractors have little incentive to plan or offer quick programs, or to finish them on time.

As Willie Sutton did not say ( is this column's friend) when asked why he robbed banks: “That's where the money is.” Analyst Todd Harrison of the Center for Strategic and Budgetary Assessments believes R&D spending could equal or even surpass procurement in the coming downturn. The procurement-to-R&D ratio (see graph) was between 2:1 and 3:1 in the Cold War with a couple of spikes toward 3.5. In the 1990s and 2000s, it was between 1:1 and 1.5:1.

R&D contracts are sole-source. They are cost-plus, with fixed margins and incentives. They support expensive engineers and their more expensive bosses, rather than hourly-paid factory workers.

Another attractive feature of R&D, from a business viewpoint, is that it is a low-capital business. Cubicle farms are cheaper to buy, and easier to divest, than factories full of machine tools, tape-layers and autoclaves. That factor may be growing in importance with the advance of modeling, simulation and additive manufacturing: You can buy much of the capital equipment for an R&D program from Dell.

That makes a case for suggesting contractors have an incentive to drag out R&D, but it is not a conclusive one. R&D margins are fixed: A negotiated fixed-price production contract can and usually does have higher margins. Wall Street cares about margins like Willie Sutton cared about banks.

And if production margins are pay dirt, support can be an even richer lode. One model that is used in performance-based logistics is a fixed-term contract that will be renewed as long as the contractor offers an agreed, lower price. The contractor's strategy is to offer a target price that he knows he can beat. The customer sees a lower price—a win-win situation—but a savvy contractor can run the margins through the roof, with relatively little capital.

So the contractor does have some incentive to get to production—with some caveats. One is the near-parity relationship between R&D and procurement spending, which reflects both long R&D programs and truncated or canceled production plans. Contractors need to make money on R&D, and those with relatively small production programs (Northrop Grumman, for instance) are insisting on higher margins.

Also, once a system is in production, the contractor has no incentive to make low-margin bids on R&D for block upgrades or performance improvements. That's the dark downside of spiral development.

And it's not as if R&D is fixed-price, let alone self-funded, as it is in commercial aerospace. Underbidding R&D, or making decisions based on underestimated R&D, will not kill you, even though it may not be the optimal use of your resources. Indeed, it may be better to underbid R&D than to let a competitor do it, because defense is zero-sum and the money you receive does not go to your competitor, but supports your overheads instead of his.

Also, there is little incentive to terminate a program that is not going well, or that requirements are making less relevant. You see that in the commercial sector, where Airbus dropped its initial A350 in favor of a bigger aircraft aimed at a market segment above the 787, and Boeing has continuously revised its 777X to stay above the A350.

Customer-funded R&D may not be the only culprit behind quarter-century development cycles, but it is certainly a factor. And it's surprising that (as far as I know) nobody has proposed, let alone conducted, a Packard Commission-like investigation into the increase in development times. Could it be that nobody really wants to know what the answer is, let alone have it come out in public?