Ever since the Defense Department procurement scandals of the 1980s, the $600 hammer has been held up as an icon of Pentagon incompetence. Immortalized in the "Hammer Awards" that Vice President Al Gore's program to reinvent government gives out to waste-cutters, this absurdly overpriced piece of hardware has come to symbolize all that's wrong with the government's financial management.

One problem: "There never was a $600 hammer," said Steven Kelman, public policy professor at Harvard University's John F. Kennedy School of Government and a former administrator of the Office of Federal Procurement Policy. It was, he said, "an accounting artifact."

The military bought the hammer, Kelman explained, bundled into one bulk purchase of many different spare parts. But when the contractors allocated their engineering expenses among the individual spare parts on the list-a bookkeeping exercise that had no effect on the price the Pentagon paid overall-they simply treated every item the same. So the hammer, originally $15, picked up the same amount of research and development overhead-$420-as each of the highly technical components, recalled retired procurement official LeRoy Haugh. (Later news stories inflated the $435 figure to $600.)

"The hammer got as much overhead as an engine," Kelman continued, despite the fact that the hammer cost much less than $420 to develop, and the engine cost much more-"but nobody ever said, 'What a great deal the government got on the engine!' "

Thus retold, the legend of the $600 hammer becomes a different kind of cautionary tale. It is no longer about simple, obvious waste. The new moral is that numbers, taken as self-explanatory truths by the public and the press, can in fact be the woefully distorted products of a broken accounting system.

The root of the problem is as old as the Republic: Federal accounting has always been primarily concerned with making sure money was spent as Congress directed-not with making sure it was spent wisely. Historically, explained the Pentagon's deputy chief financial officer, Nelson Toye, DoD's bookkeeping systems were designed to "be able to satisfy the Congress that we were good stewards of the funds entrusted to us: We didn't overspend, we did spend it on what you asked us to, we didn't spend money to buy things you told us we couldn't buy." In the past, Toye said, "there has not been a requirement for DoD or any federal agency to routinely collect the costs of its assets and report those costs."

But a necessary change is under way, said Richard Eckhardt, deputy director of financial management for the Air Force Materiel Command, which does most of that service's shopping. "We've been very good at putting budgets together and writing budget justifications," he noted, "but in an era of declining budgets, we have to understand what our costs are." That means government must borrow business techniques to track the true costs of its activities.

The Air Force Materiel Command, for example, has broken its activities into eight "business areas"-from base upkeep to information systems to flight testing-and assigned to each a general as "chief operating officer." These generals, said Eckhardt, are "in different stages of developing cost-accounting systems"; of devising numerical measures for output (always difficult for the government, which doesn't sell anything); and of experimenting with "activity-based costing," a popular private-sector technique that pulls business processes apart to find the cost of every step.

Bookkeeping based on congressional appropriations makes such cost-finding immensely difficult. Functions that in practice are inextricably intertwined are often paid for by totally separate line items in the budget. New weapons are bought with one "color of money," existing weapons are maintained with another, and the personnel who operate them are paid with a third. In fact, to save administrative costs, military salaries and pensions are all paid from one central office. As a result, said Eckhardt, among commanders "there's a tendency to view military labor as free, because you're not making any expenditures from your installation [budget] to pay those people."

The National Reconnaissance Office, which runs satellites, has an even more confusing payroll: Some of its personnel belong to the CIA, some to various military services, some to the office itself-and each of these contingents is paid with a separate appropriation. The office used to use three incompatible accounting systems, too, but after 1995-96, when investigators found $4 billion languishing unspent in various accounts, it has introduced a unified system, using standard Momentum software from American Management Systems Inc. Vincent Dennis, deputy director for resource oversight and management, crowed that for the first time in the agency's history, "in March of '99, we will have an auditable financial statement." Even so, he admitted, with salaries coming out of three different congressional appropriations, "I'm not at the position where I can allocate those personnel costs."

Personnel is not the only cost arbitrarily broken up by the DoD's line-item budget. Many warships, planes, and other weapons systems depend on critical components-such as radars and anti-missile jammers-that were developed under separate programs financed by separate line items. Whether those subsystems are counted as part of the larger system's cost depends on what the meaning of cost is.

"There are all types of costs," said the Pentagon's Toye, "and people need to be specific when they ask." A Tomahawk cruise missile, the kind that occasionally lands on an Iraqi target, costs about $750,000--if bought new in 1998, now that years of manufacturing experience have driven down the price. Any missile actually fired today, however, was bought at a higher price earlier in the production run, and has been stored, serviced and shipped across the seas, making for a total cost, by some estimates, of nearly $2 million a missile.

Conversely, said defense analyst Loren B. Thompson of the Lexington Institute, a conservative Arlington, Va., think tank, the $2 billion-per-plane figure cited by opponents of the B-2 stealth bomber includes the program's high research and development expenses--which must be spread over only 21 planes--plus spare parts, maintenance and future inflation. Said Thompson: "What would it cost me to build one more bomber? . . . $700 million."

Interpretations, admitted Toye, compound the problem: "It is indeed possible to go into two different program offices, and use the same terminology, and come out with some components in, or some components out, that weren't treated that same way in a different office." In other words, different agencies may apply the same technical definition of cost to the same weapon and come up with different numbers.

Under whatever definition, a weapon's cost rarely reflects the expenses of the headquarters that supervised its development, since those administrative offices are funded under their own line items. And many administrative offices, in turn, depend on support services--such as legal counsel and computer support-that are themselves financed by separate appropriations and are therefore often ignored in computations of a given office's cost of doing business. Unlike the private sector, said the Kennedy School's Kelman, "the government, in its own internal cost accounting, . . . typically doesn't fully account for overhead; sometimes it doesn't account at all for overhead."

"How do you allocate the cost of carrying inventory, for example?" asked John W. Douglass, recently retired assistant Navy secretary for research, development and acquisition. "Generally speaking, the service only pays the price of [buying] the part in their cost models. They don't show the cost of carrying that inventory."

Such accounting arcana are bread-and-butter issues for Douglass now that he heads the Aerospace Industries Association of America Inc., whose members want more military service contracts-which they can win only by showing they can perform a given service at lower cost than the military could do it in-house. But when the public and private sectors compete, said Bert M. Concklin, president of the Professional Services Council, differing accounting standards mean that "the government's costs are elusive, at best."

The Air Force Materiel Command conducts many such competitions, said Eckhardt, and it uses Pentagon and Office of Management and Budget guidelines to "take all those sources of overhead [and] make sure all the costs are included." The OMB's competition guidelines, for instance, start by accepting most federal cost estimates and then add on an estimated overhead rate. But the Pentagon's true overhead "may be more or less than the government rate," fretted Lisa G. Jacobson, director of Defense audits in the accounting and information management division of the General Accounting Office. "DoD's business operation seems to be very inefficient, in general."

Jacobson felt so strongly about Defense's inefficiencies that she took the unusual step of testifying, not as a GAO representative but as a private citizen, before the Federal Accounting Standards Advisory Board. She was hoping--vainly--to have the board require the Defense Department to report the price it initially pays for any given piece of equipment, implementing a common private-sector standard called "historical cost."

"I don't pretend that we have precise historical costs," said Toye, who represents the Pentagon on the accounting advisory board. "After six years and three months, we are [free to discard] records [of particular purchases]. But in terms of meaningful cost information, reasonable cost information, I believe we have that."

"I don't know how Nelson Toye can give you that data [on cost information]," complained one Senate aide. In an investigation of military books, he said, "we couldn't find most of the records"--not just records of transactions more than six years and three months old, but "of things that were just paid, or of things that hadn't been paid yet." And if the basic records are in such disarray--if the Pentagon cannot even account for the true cost of a hammer--then, critics warn, any attempt to install sophisticated commercial accounting will be a castle built on sand.

"I would disagree," countered Toye, "with the statement that there isn't hard, actual, auditable data out there. I believe it's there [even if] the department may not be able to summarize that and report it in ways that individuals want us to." The problem for the Pentagon is that those "individuals" of Toye's are the citizens.