(FORTUNE Magazine) – The U.S. budget surplus is one of the great unanticipated economic phenomena of the 1990s. Just a few years ago, the idea that the federal government could balance the budget--much less run a surplus--was considered pure fantasy. The surplus, one of the most beneficial things that's happened for the economy, has helped keep interest rates low and the stock market rising, and has kept the U.S. Treasury's demands on the bond market modest--creating plenty of opportunity for business to raise capital.

You'd think Congress would leave well enough alone, but battle lines have already been drawn for what promises to be the last big political fight of the millennium: what to do with the $1 trillion surplus projected to accumulate during the next ten years. The Republican majority has passed its $792 billion tax-cut plan, which President Clinton has promised to veto. The Administration is likely to propose spending increases in Medicare and other programs. A small, level-headed contingent of fiscal purists (among them Federal Reserve Chairman Alan Greenspan) wants to use the money to pay down the national debt. Where it all winds up at the end of the year is anyone's guess. The only problem is, the $1 trillion pot of gold they're fighting over may never materialize.

There are three reasons the surplus may turn out to be a mirage, and the first two are pretty obvious: No. 1, budget forecasts are often just plain wrong. The main budget forecasting authority, the Congressional Budget Office, has been consistently off the mark for the past five years. By underestimating tax revenue, CBO failed to predict the favorable swing to a budget surplus. Based on its track record, CBO expects to make an average error of 13% on its five-year projections. That means that a $234 billion surplus predicted for 2004 could turn out to be anywhere between a $16 billion deficit and a $484 billion surplus.

Predicting how the federal budget will behave is one of the trickiest things in all forecasting. For starters, you have to get the economy right, since GDP is one of the key variables in the budget forecasting model. "If you get off by a couple of tenths of a percentage point on GDP," says Christopher Swann, an economist at the forecasting firm WEFA, "depending on your time horizon, you can be off by $100 billion or more on your budget forecast." Next, current law and existing spending caps have to be taken on faith. And finally, thousands of estimates must be made about every federal government program in existence. One wrong assumption could bring the whole $1 trillion surplus projection crashing down.

Which brings us to reason No. 2: Can you really assume that Washington has suppressed forever both the political urge to spend and the desire to buy votes with tax cuts? True, for almost a decade now, Congress has controlled its spending impulse, with impressive results. Total discretionary government spending (spending other than on interest and Social Security) currently stands at 13% of GDP, its lowest level in more than 25 years. But even stronger determination will be required to keep it there in the future, especially given the current political debate and growing pressure to spend more on health and education. At the top of the Administration's wish list, for instance, is greater spending on Medicare, children's programs, defense, and environmental protection.

There's simply no margin for error. The $1 trillion projected surplus relies on discretionary spending remaining at 13% of GDP for the next ten years. Look at what happens if it merely inches up to 13.5% of GDP (still far below the 15%-plus levels of the 1970s and 1980s) and remains there. According to Saul Hymans, director of economic forecasting at the University of Michigan, the result is an $834 billion loss, both from greater government spending and from higher interest payments on increased federal debt levels over the ten-year period. That leaves an accumulated surplus of only $161 billion by 2009.

Then there's the impact of the Republicans' beloved tax cuts, which include across-the-board reductions in income tax rates as well as the end of inheritance taxes. Since lower taxes would lead to increased federal borrowing--and higher borrowing costs--any reduction in taxes turns out to have an even bigger impact on the surplus than is immediately apparent. According to Hymans, the Republican plan to cut taxes by $792 billion leaves a mere $88 billion accumulated surplus by 2009. While it's unlikely that the Democrats can get all the spending they hope for or that the Republicans can win all the tax cuts they want, there's a good chance that both parties will get something as the horse-trading begins.

The third reason to bet against the surplus forecasts is this: To a degree not commonly realized, the budget surplus is a byproduct of the highly unusual bull market that's been under way for the past few years. Unless the market keeps climbing well above its historical average, the $1 trillion surplus will just go poof!--even if the budget projections are right and Congress miraculously controls itself.

Take a look at the numbers. Four years of average annual stock market gains of 25% have substantially boosted the government's revenue from personal income taxes. In particular, much greater capital gains revenue, the growth of employee stock options, and more money from income earners pushed into higher tax brackets have lifted the effective tax rate (all personal income taxes collected by the government as a proportion of the total amount of wages and salaries earned in the economy) to 20.5%. Compare this with the average effective tax rate of 17.2% for the decade to 1995 and you have discovered a key factor behind the surplus. Next, look at what the CBO assumes. Its projection of a $1 trillion accumulated surplus relies on an average tax rate of 20%--a mere half a percentage point lower than during boom conditions. In effect, the CBO is betting the market's boom continues unabated.

But if the stock market grows at 10% (still more than the historical average rate of 8%) for the next decade, the average effective tax rate would likely recede to 17%, according to Hymans. Even if the tax rate falls back only halfway, to 18.5%, Hymans calculates that the entire $1 trillion accumulated surplus would vanish.

In the end, it's not the forecasts that are to blame but what the policymakers do with them. For anyone willing to take a look, CBO's budget forecasts come with a warning label--projections are dependent on "current laws and policies and providing that the economy performs as CBO assumes." It has also published a report on how the surplus changes under many different economic scenarios. "As a forecaster, you'd like to say to a policymaker, 'This is the best we can do,'" says Laura D'Andrea Tyson, dean of Berkeley's Haas School of Business and former chair of the Council of Economic Advisers. "There's a lot of uncertainty, so don't do anything based on these forecasts that would remove your room for discretion. Keep a good reserve. As a policymaker, you want to use five- or ten-year forecasts for a sense of direction. You don't want to lock in a difficult-to-reverse policy based on long-term projections. That will get you into trouble."

With so little to really count on in the budget forecasts, spending the projected $1 trillion now, in any form--or purposely reducing it by cutting taxes--may be as reckless as selling your house, quitting your job, and becoming a day trader. At least as a day trader you're risking only your own financial security.