WITH THIS fine book, Herman B. Leonard of Harvard's Kennedy School of Government has complicated our already vexed fiscal lives.

Most of the budget problems that preoccupy us today -- ruinous trade imbalances, runaway deficits, skyrocketing national debt -- are oppressively visible.

By contrast, Leonard's subject is the unbudgeted spending implicit, and therefore "quiet," in things done or not done. But quiet spending, whether present or promised, is very real.

Perhaps the most familiar example is the "tax expenditure." In simplest terms, the tax expenditure is a sum the Treasury might collect but forgoes, either accidentally, by leaving a loophole in the tax code, or for some reason of policy. It might be the credit a company gets for investing in a new machine. Or it might be the ordinary deduction for mortgage interest payments.

There has been a tendency among conservatives to resist this public accounting device. David Stockman tells us he found Ronald Reagan irritated by the very notion. The President regards "tax expenditures" as merely a fancy term for gratuitous theoretical claims to slices of our paychecks. That isn't quite the case, though as Leonard admits, it is a conceptual problem to establish just what basic system of tax liabilities " tax expenditures should be deducted from.

The important point about tax expenditures, whether expansively or conservatively defined, is that they are the equivalent of an actual disbursement of public funds.

There are many other forms of quiet spending, some less visible and more ingenious; Leonard walks us expertly and wittily through them.

Unquestionably the most lavish and costly form of quiet spending is unfunded federal pension liabilities. They are, says Leonard, "similar in character and magnitude to the explicitly recognized national debt," though it is not politically expedient to treat them that way.

The figures tell an impressive tale. As of September 1982, the civil service retirement system had net unfunded liabilities of some $575 billion, and the military retirement system unfunded liabilities of about $55 billion. In combination, they "amount to an implicit debt of $5,000 for every adult in the U.S." (And this is to say nothing of state pension liabilities of about $170 billion.)

The point is this: If budgeted at their true currently funded cost, pension liabilities would vastly raise spending and our consciousness of it. The same is true of Social Security. In theory an insurance program, Social Security has evolved into an intergenerational contract whose continuing solvency rests on the implied obligation of each succeeding generation to pay what the last has promised. It works well enough, so long as the number of payers grows pari passu with the number of payees. But if the actual unfunded cost of present benefit levels were included as a line item in the budget, we would have a very different picture of the "current" cost of Social Security. ALMOST INFINITE in variety and ingenuity, indeed, are the forms of quiet spending. There are industrial revenue bonds, which extend private incentives and benefits at the expense of federal taxpayers (through full exemption of interest income from taxes). There are many, many federal credit programs, which both make direct loans and guarantee them. (A 1982 Government Accounting Office count found some 424 such programs. Direct loans had reached some $15.3 billion net in 1983; loan guarantees some $34 billion. To the extent that credit subsidy is involved (i.e., a net difference between "market rate" interest costs and the usually lower interest costs offered by these government programs), we have another form of quiet spending.

Perhaps the quietest form of quiet spending is what is happening to the nation's deteriorating "infrastructure" (roads, bridges, sewers, etc.) in an era of fiscal stringency. When the depreciation of public capital assets goes unnoted in budgets (as it would not on the balance sheet of a private company), or when there is no continuing reinvestment in maintenance, future liabilities and deteriorating services quietly accrue. This is a hidden cumulative cost in the tens of billions of dollars.

Finally, one might mention the lease arrangements now popular with government units bumping up against bonded-indebtedness ceilings. Again, the forgiveness of taxes is usually the ultimate price of such "creative financing." And often -- as in the Navy's ship-leasing program -- the actual ultimate cost to the public is probably greater than it would be under more orthodox financing.

The forgoing should not, however, be taken to imply that Leonard is merely another green-eyeshade scold, deploring excess spending or "waste." He shows that the cost of government has been growing, but that is not his point. His message, rather, is that "quiet spending" in all its clever forms tends to defeat accountability. Accountability is, as he puts it, "silent on the question of scale"; but it is the only way we have to force efficient choices, or assure value, in the outlay of public moneys. Quiet spending flourishes out of sight, where taxpayers as "principals" exercise diminished control (or none) over officials acting as their agents.

Are spending subterfuges or evasions a new story? They are not. But no one has gathered the loose ends quite so dramatically or pointedly as Leonard. He has no pat solutions. He is skeptical of "mechanical" contrivances that would seek to restore fiscal control by imposing arbitrary limits here and there. For instance, his keen analysis of the likely perverse effects of the proposed constitutional amendment to mandate a balanced federal budget is must reading, along with much else in Checks Unbalanced.

Unfortunately, Leonard's is a work of such exceptional intelligence and analytical subtlety that it is likelier to reach experts than a wide general audience. Which would, in a way, defeat its purpose. We the people need the book more than it needs us, for any useful move to control quiet spending would have to start with a more attentive public.

As Leonard says, information about government, to be effective, must pass "the grandparent test": It must make sense to the average grandparent. "If," he writes, "our best efforts . . . still cannot pass the grandparent test, then the time has come to work on the grandparent. The first step is admitting that we have met the grandparents and that they are us."

Edwin M. Yoder Jr. is a syndicated columnist with The Washington Post Writers Group.