Sometime this spring, the federal government lost nearly $200 billion in tax revenues between now and 1995. Whoosh, the money was gone -- just like that. The economists who estimate future taxes wiped it out. It got eliminated in what's charmingly called a "technical reestimate" tucked away in the latest budget figures released in mid-July. Almost no one noticed.

There is a juicy irony here. Even as the White House and Congress are locked in hard negotiations over the budget deficit -- and how much to raise taxes -- the green eyeshades have hacked a sizable hole in the tax base. For fiscal 1991 beginning in October, the loss is put at $27 billion. That's roughly equal to (or even a bit larger than) the tax increase now being informally discussed. For the years fiscal 1990 and 1995, the estimated revenue loss adds up to $195.5 billion.

Gulp.

This "technical reestimate" is the largest in recent memory and may be the largest ever. It is a major reason why deficit projections are now so much higher than a few months ago. To put it in perspective, its long-term implications may overshadow the cost of the savings and loan bailout. Sooner or later, the costs of the S&L rescue will stop. By contrast, the tax loss -- unless the new estimates prove wrong -- could continue indefinitely.

Budget forecasts are notoriously unreliable, but the typical mistakes involve faulty economic assumptions and unanticipated policy changes. Errors of this sort are always politically suspect. Democrats routinely condemn the White House for painting an overly optimistic economic outlook, and the White House regularly blasts Congress for overspending.

One reason the new tax shortfall has attracted so little attention is that neither party can make much political mileage from it. It's simply a goof. The people who estimate the budget for the White House (the Treasury Department and the Office of Management and Budget) and for Congress (the Congressional Budget Office) committed the same blunder.

How could such a giant glitch occur?

No one really knows. Oh, government tax experts have theories. Some of these partly explain the shortfall, but no one has a full explanation. What the tax experts know for certain is that the government isn't collecting as much in taxes as computer models had predicted. So the computers' predictions have been cut: put simply, they have been brought in line with reality.

In some ways, this story is a familiar one. Economic statistics are rarely as precise as they seem. The Central Intelligence Agency, for example, says the size of the Soviet economy is about half that of the United States. Now, some experts think a more accurate estimate may be a third. For years, the Labor Department calculated the consumer price index in a way that slightly overstated inflation. This added billions of dollars to programs like Social Security that are tied to the CPI. In 1983, the Labor Department altered its calculations.

A similar technical flaw is being cited as the main culprit of the tax shortfall. To estimate taxes, the government's experts need to know how much Americans earn in wages and salaries -- the biggest part of the tax base. That figure comes from the Commerce Department, which now says that its preliminary estimate for 1989 wages and salaries ($2.631 trillion) may have been about $50 billion too high. Lower wages mean lower income taxes.

Understanding what went wrong shows why getting good statistics quickly isn't easy. The preliminary estimate of wages and salaries is based on a monthly Labor Department survey of employment and average wages. But this survey covers only 300,000 of the nation's roughly 6 million business locations. Statistical techniques are then used to estimate total wages and salaries. More complete information on what all businesses pay is collected for the unemployment insurance program. However, this information becomes fully available only months after the end of the calendar year. The new information now indicates that the initial estimate was $50 billion too high.

This error translates into a much smaller tax loss -- perhaps $9 billion to $12 billion for fiscal 1991. That could account for half the total loss. The rest stems from other overestimates of personal and corporate income taxes and of Social Security taxes. The understanding of these changes is so skimpy that the Bush administration and the Congressional Budget Office disagree on their future implications. Although the administration projects the total revenue loss at $195.5 billion by 1995, the CBO puts the cumulative loss at a mere $109 billion. In essence, CBO thinks that some of this year's losses are one-time affairs.

What we have here is a huge wild card in the budget ordeal. It's conceivable that a future "technical reestimate" could reverse the tax loss. But unless that happens, the political impact is unavoidable. Everything from Medicare to tax rates could be affected. The politics of fashioning a budget deal become tougher. And the prospects that any deal will ultimately end the deficits become more distant.