Whatever the Supreme Court does with the Gramm-Rudman budget-balancing act, there are signs that the federal deficit is starting to shrink, primarily because Congress last year quit rubber-stamping Secretary of Defense Caspar Weinberger's military-budget requests. However, it's possible that the cheering about the shrinking deficit has started prematurely and could weaken whatever resolve remains on Capitol Hill to deal with the deficit problem.

President Reagan's budget message on Feb. 5 claimed that the deficit, without further action, would fall from about $206 billion in fiscal 1986 to $104 billion in fiscal 1991. You were supposed to conclude that the problem is therefore manageable without the bad medicine of a tax increase.

But a sense of euphoria about declining deficits didn't really take over until publication of a new budget analysis on Feb. 18 by the nonpartisan Congressional Budget Office. The CBO arithmetic is generally accorded more credibility than that of the partisan White House Office of Management and Budget.

The CBO, which only last August warned that the deficit would balloon from $200 billion in 1985 to nearly $300 billion by 1991, did an about-face. Although some of its calculations differed, the CBO came out at the same place as did the OMB, forecasting a prospective deficit of only $104 billion in 1991.

How could the outlook have improved so dramatically in just the seven months since August? Implementation of Gramm-Rudman, plus congressional determination in 1985 to slow down the mushrooming defense budget, "will reverse the recent trend of rising federal budget deficits," CBO explains.

Whether Gramm-Rudman will be effective or even operative after this fiscal year is still in doubt. Nonetheless, the first piece of the controversial sequestration process -- an $11.7 billion budget cut for the remainder of fiscal 1986 -- went into effect yesterday. This was unaffected by the U.S. Court of Appeals decision questioning the constitutionality of the law's automatic budget-cutting procedures although, at least theoretically, the Supreme Court could invalidate those cuts and order the money restored.

But the guessing is that, even if the Supreme Court should uphold the lower court decision against Gramm-Rudman, it would consider the first sequestration as a fait accompli. Thus, an essential piece of the CBO reasoning is that the initial $11.7 billion sequestration works out to be an effective $20 billion cut at annual rates out of future budgets. In turn, that means that the "baseline" deficit projected by CBO for fiscal 1987 -- the starting point before any other policy steps are taken -- comes down to $181 billion.

But even more important, CBO lowers its baseline deficit in each succeeding year. By 1990, the deficit would be only $120 billion compared with an estimate made last August of $285 billion for that year. Is that believable? Let me tick off some of the assumptions underlying this rosier CBO picture.

First, CBO offers a highly favorable economic guesstimate for 1986 and the following five calendar years. For 1986 and 1987, the economic assumptions seem reasonable, given the sharp plunge in the dollar and the collapse of oil prices, which are certain to help economic growth. But the CBO sees economic prosperity continuing unabated, a rock-steady 3 to 3.5 percent rate for the entire period through 1991. That means a business cycle with no recession over this whole period, while inflation holds at about 4.3 percent. Meanwhile, the scenario includes a continued decline in interest rates, sharply reducing the annual interest costs on the huge government debt.

That's a lot of favorable assumptions. CBO itself points out that merely an average recession in 1987-88, followed by an average recovery, would result in a $247 billion deficit in 1991 instead of its new baseline of $104 billion.

Then, CBO assumes no real growth in defense spending from now through 1991. That's a controversial call. There is no doubt that Congress last year rebelled against ever-rising defense expenditures. It passed a budget resolution that specified zero growth for defense in 1986, and 3 percent growth thereafter. Actual 1986 defense appropriations, moreover, fell short of the budget resolution target, and the 1986 sequestration will reduce the amount of new budget authority even more, below the 1985 level.

Because of the rigid deficit targets Gramm-Rudman sets for 1987 and beyond, the CBO concluded that the 3 percent real growth for defense specified in the 1986 budget resolution for 1987 and thereafter won't be achieved. Instead, the CBO instead cranked no real growth into its assumptions.

Some, including Robert Greenstein of the Center on Budget and Policy Priorities, think that, with or without Gramm-Rudman, the CBO assumption of zero growth in future defense budgets is a good one, unless Reagan is willing to come up with a tax increase. But opinion is divided on Capitol Hill. Many, like the Republican chairman of the Senate Budget Committee, Pete Domenici of New Mexico, will have to see it to believe it. Three percent real growth instead of no real growth for defense would add about $50 billion to CBO's $104 billion baseline for 1991.

The bottom line is that both Reagan and CBO may be on solid ground in setting the starting point for the fiscal 1987 deficit at $180 billion or so, an improvement of nearly $50 billion from the August 1985 projection of a $229 billion deficit. As economist John Makin said the other day, so long as we have to talk about deficits in the hundreds of billions of dollars, "It's good that the first number starts with a '1' instead of a '2.' "

But the rest of the CBO analysis -- for the years beginning in fiscal 1988 -- seems too euphoric. Can the president and Congress reach a compromise that actually will cut the deficit in later years without the pressures of Gramm-Rudman? It's a long way from 1987 to 1991.