From Requirement to Runway, How the Pentagon Buys a Weapon

Ask a ten-year-old to design a fighter jet and you will have a drawing by lunch. Ask the Pentagon, and you will have a drawing by the time that ten-year-old has a mortgage. Somewhere in between, a few thousand people will have attended a few hundred reviews, and a congressman will have discovered that the factory making the wing is in his district.

Conservatives say we want a strong military and a government that does not waste our money. Those two goals meet in one place: the way the Pentagon buys weapons. It is the biggest, slowest and most argued-over shopping trip on earth, and nearly every complaint you have ever heard about it, late, over budget, gold-plated, has a cause you can point to on the map. So here is the walkthrough. It goes in order, the way a program goes, with what happens at each stop and why that stop can go wrong. Where the details have changed over the years, I say so, because the rulebook gets rewritten every few years and you should check the current Defense Department instructions before you quote a step as gospel.

Step one: a requirement is born

Every weapon begins as a requirement: a statement of what the military needs that it does not have. A fleet is aging out. A rival has built something that outranges ours. A commander on the ground cannot see over a hill.

Requirements are written by the military services and reviewed on the joint side by the Joint Requirements Oversight Council, a body that traces to the Goldwater-Nichols era of 1986. The idea was sound: before one service builds something, ask whether the others need it too, and whether it fits with what the whole force is doing.

Here is where the first trouble starts. Requirements are a wish list, and wish lists grow. Every office with a stake adds a feature. The plane must also fly slow, land short, hover, carry more, and cost less. Engineers call this requirements creep, and it is the quiet parent of most of the cost overruns that make headlines later. A requirement written to be reasonable can end up as a contract for a machine nobody can build for the money.

Step two: the money cycle

A requirement is a wish. It becomes real only when it gets money, and the money comes through a process called planning, programming, budgeting and execution, shortened to PPBE. It began as the planning-programming-budgeting system that Defense Secretary Robert McNamara brought in during the early 1960s and was later renamed to add execution.

In plain words:

1. Planning. The department decides what it wants the force to look like and what threats to prepare for.
2. Programming. It decides which programs fit inside a multiyear spending plan.
3. Budgeting. It turns the program into a specific budget request, which the President sends to Congress.
4. Execution. After Congress appropriates money, the department actually spends it, and a program manager finds out whether the money arrived when he needed it.

The trap is the calendar. This cycle runs about two years ahead of the money. A program that is planned today is spent on two years from now, and by then the threat has shifted, the technology has changed and the Congress has changed. That is slow by design, because a lot of people want a say before billions are committed. It is also slow by accident, because the fiscal year starts October 1 and Congress often fails to pass a defense appropriations bill by then, which forces the department to run on stopgap resolutions at last year's rates and bars it from starting new programs. How a budget becomes law, from the President's request to the checks explains that machinery from the other end.

Step three: Milestone A, the decision to try

With a requirement and a source of funds, the program meets its first big gate. The Defense Department's system uses three major decision points called milestones. A senior official called the milestone decision authority decides whether the program may pass through.

Milestone A is the decision to begin serious work on maturing the technology and reducing risk. Before it, the program is studies and concepts. After it, the program is spending real money on prototypes, demonstrations and cost estimates. The right question at Milestone A is blunt: are we trying to invent something, or build something we already know how to build? Programs that confuse the two are the ones that blow up.

Step four: Milestone B, the big commitment

Milestone B is the decision to begin engineering and manufacturing development. This is the point of no easy return. A contractor is selected and a design is frozen enough to start building test articles. The department sets its baseline: the cost, schedule and performance it promises Congress.

In theory, Milestone B should happen when the technology is mature and the estimate is realistic. In practice, there is pressure to approve early. A program that is delayed to mature the technology can lose its funding. A program that is approved with optimistic assumptions gets the money, and the bill comes later. Veterans of this process call that choice the optimism trap, and it has been repeated for decades.

The contract type matters more than anybody admits

At this stage the government picks a kind of contract, and the choice shapes everything.

• Cost-plus. The government reimburses the contractor's allowed costs and adds a fee. It makes sense when the work is uncertain and no one can price it. The risk is that the contractor has little reason to hold costs down, because costs are reimbursed.
• Fixed-price. The contractor agrees to a price. If the work costs more, the contractor eats the difference. It makes sense when the design is mature and the cost can be estimated. The risk is the opposite: when the work is really uncertain, a fixed price can bankrupt the builder or lead to a standoff.
• Incentive contracts. A middle road, which shares cost savings and overruns between the two sides by formula.

A famous example of fixed price gone wrong is the Navy's A-12 Avenger II, a stealth attack aircraft program that Defense Secretary Dick Cheney canceled in January 1991 after the contractors fell far behind. The government and the companies then argued in court for decades about who owed what. The lesson is not that fixed-price is bad. The lesson is that you do not fix the price of something you do not yet know how to build.

Step five: development and testing

After Milestone B comes development. Engineers build test aircraft, test ships, test missiles. The test results should tell the department whether the thing works.

Here is a reform worth knowing. In 1983, Congress created the office of the Director of Operational Test and Evaluation, an official who reports to the Secretary of Defense and to Congress and who is independent of the program office. The reason was simple. A program manager's career depends on his program surviving. An independent tester's career depends on telling the truth. Operational testing asks the question that matters to a soldier: not "does it meet the specification" but "does it work when ordinary troops use it in the field against a thinking opponent?"

Many programs overlap development, testing and production in order to save time, a practice called concurrency. It works when the design is stable. When it is not, you get aircraft that roll off the line before tests are done and then need expensive retrofits. Taxpayers pay twice, once to build and once to fix.

Step six: Milestone C and low-rate production

Milestone C is the decision to start production and deployment. The first step is low-rate initial production, a small batch meant to give the military enough units for realistic operational testing and to prove that the factory can build the thing consistently. It is the pilot batch.

If the low-rate units pass the operational tests, the program moves to full-rate production, where the factory ramps up to the quantities the military planned. Then the weapon is fielded, and the story of sustainment begins. Maintaining the weapon over decades costs more in total than buying it. A weapon's purchase price is the down payment.

Step seven: the guardrails Congress added

By now you may be asking: with so many steps, what stops costs from running away? The answer is a stack of laws, mostly passed after something went wrong.

• The Nunn-McCurdy Act of 1982. Named for Senator Sam Nunn and Representative David McCurdy, it requires the Pentagon to report to Congress when a major program's unit cost grows past set percentages over its baseline. Under the version in use for years, a "significant" breach is 15 percent over the current baseline or 30 percent over the original, and a "critical" breach is 25 percent over the current or 50 percent over the original. A critical breach forces a review. Since reforms in 2009, the program is presumed to be ended unless the Secretary certifies, in writing, that it is essential to national security, that no cheaper alternative exists and that the new cost estimates are reasonable.
• Independent cost estimates and testing. Created so that the people estimating cost and the people running the program are not the same people.
• Bid protests. A losing bidder can protest a contract award to the Government Accountability Office, which must rule within 100 days. It can slow a program by months, and it can also catch real mistakes.

Step eight: why the 1986 reforms mattered

In 1985, President Reagan appointed a commission, chaired by David Packard, a co-founder of Hewlett-Packard who had served as Deputy Secretary of Defense. Its formal name was the President's Blue Ribbon Commission on Defense Management. It reported in 1986, the same year Congress passed the Goldwater-Nichols Act.

The two efforts attacked different problems. Goldwater-Nichols was mainly about command: it strengthened the Chairman of the Joint Chiefs of Staff and pushed the services to fight as one team. The Packard Commission was about buying. It argued for a single senior acquisition official in the Pentagon, for streamlining the bureaucracy, for more use of commercial products where they would do the job, and for more stable, long-term planning of programs. Much of what is in the acquisition system today traces to the debate those two efforts set off.

The backdrop was the spare-parts scandals of the mid-1980s, when reports of the hammer and the toilet seat that cost hundreds of dollars made the Pentagon a punchline. The anger was bipartisan. A conservative who wants a strong military should be the loudest critic of waste, because waste is what the opponents of a strong military use to argue for a weak one.

A hypothetical walk through the whole path

To see how it fits, take an invented example. It is a hypothetical, with invented numbers.

Picture a service that needs a new transport helicopter. In year one, a requirement is written. By year three the requirement has grown from "carry twelve troops" to "carry twelve troops, land in a small clearing, fly twice as far, and operate in the arctic." In year five it passes Milestone A. In year eight, after a competition and a protest that adds three months, it passes Milestone B with a cost estimate based on optimistic assumptions, say 100 units at a promised $20 million each, or $2 billion. In year twelve, testing shows a rotor problem. Fixing it adds $2 million per helicopter. The new cost is $22 million each across 100 units, which is $2.2 billion, a 10 percent increase, under the Nunn-McCurdy threshold. Then the service cuts the buy from 100 to 80 because of budget pressure. Suppose $600 million of the $2.2 billion is fixed cost for design, tooling and testing, which does not shrink when the order does. That leaves $1.6 billion for 100 helicopters, or $16 million each. Eighty helicopters cost 80 x $16 million = $1.28 billion, plus the $0.6 billion fixed, for $1.88 billion. Divide by 80 and the unit cost is $23.5 million. Against the original $20 million promise, that is 17.5 percent higher, past the 15 percent line, and the lawmakers get a letter.

The point of the hypothetical is that no villain was needed. Requirements creep, optimism, a test finding and a smaller order did the damage, and each step, taken alone, was defensible.

What conservatives can say about the fix

There are three honest views on the right.

1. Buy commercial when you can. If a commercial product does the job, take it. That is the Packard Commission's argument, and it is also common sense.
2. Make failure cheap and early. Prototype quickly, fly it, break it, learn. Programs that find problems in year three are better than programs that find them in year twelve.
3. Let someone be accountable. The diffusion of authority across dozens of offices is the biggest enemy of cost control. If everyone can say no and no one can say yes, you get the process we have.

Those are arguments, not guarantees. The system is also trying to protect against real dangers, such as fraud and favoritism. A reform that cuts oversight can bring back the scandals that created it.

You can see how the readiness side of this connects in Seven Tests of Whether the Military Is Really Ready for a Fight, where late deliveries and aging equipment drag on the force. For the money, A Worksheet for Reading Any Defense Budget Number Without Being Fooled shows how to read any program cost. Congress's role in all of this is spelled out in Congress Should Write the Laws, and Other Radical Ideas From Article I, and the reasoning behind insisting on cost accountability is part of A Balanced Budget Is a Moral Argument Before It Is an Accounting One.

The bottom line

• The delay is built in. A requirement, a two-year budget cycle, three milestones and a long test phase mean a weapon is old news before it reaches the troops.
• Cost overruns have a pattern. Requirements creep, optimistic estimates at Milestone B, concurrency and cut orders. Ask which one you are looking at before you blame a villain.
• Contract type is a decision, not a detail. Fixed price for mature designs, cost-plus for real invention, and no fixed price on a thing no one has built.
• Support independent testing. An honest tester who reports to Congress is cheaper than a failed weapon in the field.
• Ask your representative about the calendar. A defense budget that is late every year is a tax on readiness. Tell Congress to pass it on time.