(FORTUNE Magazine) – DO YOU remember David Stockman's famous warning that America was in danger of having ''$200 billion deficits as far as the eye can see''? That prediction was supposed to shock us when, as Ronald Reagan's budget director, he made it back in 1983. Today it is the official forecast of the Clinton Administration -- and that's assuming all goes according to plan when Senators and Congressmen square off in mid-July to concoct a compromise on the President's budget. As the charts on the next page show, past official forecasts have been lousy. Why does Washington keep screwing up? A big reason is the bizarre way it builds budgets. The process permits -- even encourages -- obfuscation and mendacity. It lacks any effective mechanism for checking the wildfire growth of entitlement spending. And it seems totally incapable of killing off programs that have outlived their usefulness. For all Bill Clinton's talk about no gimmicks, he quickly adapted to this smoke-and-mirrors world. His February budget proposal -- billed as a $500 billion deficit reduction package -- took credit for $45 billion in savings already ordered by previous legislation. Clinton's plan also included enough funny math and dubious assumptions that the real bottom line, according to the Congressional Budget Office (CBO) -- the designated scorekeeper in matters fiscal -- would have been deficit reduction of just $355 billion over five years. Remarkably, the House so far has rejected much of Clinton's proposed new spending, trimming many programs and derailing a few of his other favorites, such as research on magnetic levitation trains. The Senate has made similar cuts. As a result, both houses have produced budgets more likely to come closer to that $500 billion target. Even so, the goal of deficit reduction keeps moving away like Tantalus's grapes. That's partly because the pharisaic way Washington measures the five- year impact of tax and spending proposals encourages accounting gamesmanship and bad policy choices. Under normal budgetary conventions, ''cuts'' in spending, and thus in the deficit, are tallied as cuts from a ''baseline'' of what spending would be if no changes in policy were made. So even when Clinton's plan cuts $20 billion next year, spending still rises by $38 billion.

Ever wonder why the R&D tax credit is temporary and gets extended each year? The answer is that making the credit permanent would force budgeters to add the full five years of lost revenue -- $10.2 billion -- to the deficit. House members, to their credit, did so, but the Senate made the credit temporary again, meaning only one year -- $1.7 billion -- had to be counted. The result? The Senate was able to funnel the $8.5 billion ''saved'' to other spending, even though the popular credit will doubtless be extended. Then there are the numerous half-baked proposals for budget savings. Example: replacing the current quasi-private student loan program with direct federal lending. The Administration initially claimed that would save $4.3 billion over five years. But wait! The bean counters realized their calculations hadn't included administrative expenses. When they did, the savings dropped to $2.1 billion. Even that makes no allowance for the chance that the Department of Education -- which isn't exactly run like GE -- might approve some bad credits. Another glaring flaw in the budget process is its inability to contain entitlements -- programs where Congress simply defines who qualifies and then opens the money spigots. Examples range from welfare and unemployment to farm subsidies and Social Security. Since 1963 entitlements have grown from 30% of federal spending to more than half. Among the programs out of control: Social Security disability benefits, which the CBO estimates will increase by 44% over the next five years; Medicare spending, which has doubled over the past seven years; Medicaid payments, which have doubled in just three years. The final area where the process fails us is in getting rid of bad old spending programs. Take the Rural Electrification Administration. When the REA was set up in 1935, one-fifth of American workers were farmers and only one in nine farms had electricity. Today, virtually all rural areas are wired, and a scant 2.5% of the work force is in farming. But the REA, still going strong, lent out $1 billion last year and wrote loan guarantees for $821 million, with much of the money going to suburban and resort areas. Yet instead of plowing this old weed under, the Administration proposed to pluck a petal or two, for average savings of $109 million a year. In fact, everything else is likely to survive in this summer's budget -- the strategic wool and mohair program (designed, before the age of polyester, to ensure that American soldiers stay clothed), the honey subsidy (designed to keep the nation in bees -- and beekeepers), and the nation's 32-billion-cubic- foot helium reserve, which employs 200 workers in Amarillo, Texas. Says CBO director Robert Reischauer: ''It's true these things don't add up to a rounding error. But from the standpoint of the average American, it's a legitimate argument to say, 'If you're going to have a stupid program like that, you're not going to raise my taxes.' '' DESPITE the problems, there is some good news to report about Washington's budget process. The 1990 budget act established caps for discretionary spending, which Congress directly controls, unlike entitlements. To general surprise, most congressional committees have learned to live with them. ''I call it George Bush's final revenge,'' says Rudolph G. Penner, director of economic studies at KPMG Peat Marwick and former director of the CBO. Another welcome reform from the 1990 act is the pay-as-you-go concept. This commonsense idea -- practiced by every well-run household but considered a remarkable innovation in Washington -- requires legislators who would increase spending to find offsetting cuts. What's needed now is to go further. Gimmicks like ''temporary'' tax credits should be outlawed. The House took a stab at controlling entitlement spending in May, proposing caps like the ones on discretionary spending. If expenditures grew faster, the President would be required to propose countervailing cuts or additional taxes. This measure, which would have been better than nothing, was scuttled in the Senate. To root out old programs, Clinton should consider copying the Defense Base Closure and Realignment Commission, which is charged with the politically difficult task of eliminating obsolete military bases. The commission makes a recommendation to the President. Once he signs off, Congress can adopt or reject -- but cannot change -- the proposal. Some kind of presidential line- item veto would also help, though any such change will have to pass in the Senate over the dead body of West Virginia's powerful Robert Byrd. Optimists believe Washington is increasingly ready to slay the deficit monster once and for all. Says Leon Panetta, Clinton's budget director: ''For the first time, I sense that members on both sides of the aisle are finding it politically attractive to make tough votes on this issue.'' Let's hope so. According to the CBO's projections, even with the tax increases and spending cuts that the President proposed, the deficit drops to only $205 billion by 1997 and then resumes its sharp upward trajectory.

CHART: NOT AVAILABLE CREDIT: FORTUNE CHARTS CAPTION: REALITY ACTUAL BUDGET DEFICIT Truth is, the projections in every recent budget plan have been wildly off base. The five-year estimates shown below were prepared by the Congressional Budget Office, Capitol Hill's official scorekeeper. CBO is also the source of the 1993 deficit estimate.