Nearly three years ago, the Federal Reserve made major changes in the way it manages monetary policy. Today, Federal Reserve officials, financial market analysts and academic economists looking back at those three years are unable to agree why the Fed's changes produced such startling results.

The disagreements are important because they mean that plotting the future course of monetary policy is likely to remain a highly uncertain enterprise.

At a conference last week at the Jackson Lake Lodge sponsored by the Kansas City Federal Reserve Bank, the wide-ranging disputes over monetary policy emerged once more in a series of polite, but pointed, academic exchanges. Federal Reserve Chairman Paul A. Volcker was one of the participants, though he kept his own counsel throughout the meeting.

There was no disagreement whatsoever on one point: that the Fed's policy of reducing the rate of growth of the money supply to slow inflation has worked.

The unanswered question is why the improvement in inflation has come at so much greater cost in terms of lost output and high unemployment than monetarist economists and the Reagan administration expected.

Or put another way, why has the new approach to monetary policy -- in which the Fed seeks to control money growth by controlling the availability of bank reserves rather than by pegging certain interest rates -- produced such high and volatile interest rates and more volatility in money growth rates to boot?

These same questions, of course, have been asked repeatedly over the last three years as the Fed's policy has helped produce two recessions, an unemployment rate close to 10 percent and the current prospect of only a weak and limited economic recovery.

Much of the discussion at the conference focused on the critical role of expectations and their impact on interest rates. If the Federal Reserve and the Reagan administration are pursuing credible anti-inflationary policies, why are long-term interest rates -- which generally are supposed to reflect inflation expectations plus a "real" rate of return related to risk -- so stubbornly high?

Phillip Cagan, a Columbia University economist, noted, "it seems plausible that the 'credibility' of a policy would have a major influence on expectations. . . . But, granted its current popularity among economists and dramatic implications, what has credibility done for us as an explanatory device? Consider that we do not know how to measure it, certainly do not know how to produce it, and have only the foggiest notion of whether, or to what degree, it is absent or present . . ."

Asked Cagan, "does the current anti-inflationary monetary policy possess this credibility? Apparently not. Current bond yields belie it, by not implying declining inflation over the next 10 years.

"But we have an announced policy of disinflation, and the administration seems determined to persist -- at least until the next election . . . ," Cagan said.

"If we have not yet achieved credibility for our anti-inflation policy with back-to-back recessions and disaster in the union strongholds of autos and steel and satellite industries," he added, "I shudder to think what it would take."

Cagan went on to make a point that has serious implications for whether monetary policy can reduce inflation other than through depressing the economy. "I want to suggest a different view," he said, "that short-run changes in policy almost never have credibility until they are viewed as permanent, and that takes time."

If investors do not think government policies add up to a credible anti-inflation program, and many obviously do not these days, then they will continue to demand high inflation premiums when they invest their money. For about a year now, long-term rates have remained very high compared to current levels of inflation. These high rates have taken a major economic toll.

Frank Morris, president of the Federal Reserve Bank of Boston, offered one concrete example of the lack of credibility in current policy. Morris, who is chairman of the Federal Reserve system's pension fund advisory committee, said that the latest projections of the fund's professional investment managers call for real gross national product to grow only 1 percent to 2 percent a year between now and 1985, and despite such weak growth, call for inflation to continue at about 10 percent a year.

This whole issue, in the opinion of many of the conference participants, turns on whether expectations are formed on the basis of recent experience or the confidence that policy goals of the government will be achieved. If, as Cagan maintained, they are "backward looking," then the only way to change them is to change the nature of current experience. In other words, to lower inflation expectations one must first lower inflation, presumably by depressing the economy.

Carl E. Walsh, a Princeton University professor currently working at the Kansas City Fed, suggested another factor that is contributing to higher than expected interest rates -- the change in the Fed's basic monetary operating techniques.

By allowing greater interest rate volatility since October 1979, the Fed has made investments inherently more risky, he argued. This increase in risk in turn has made the demand for money less sensitive to the level of interest rates.

Walsh went on to say that he does not know just how important "quantitatively" this explanation may be. But because some basic relationships in financial markets have, in his opinion, changed as a result of the shift in the Fed's way of doing business, economists basing forecasts on the set of relationships that existed before 1979 may come up with the wrong numbers.

Not surprisingly, no one at the conference seemed satisfied with the actual economic situation facing the country -- and the Federal Reserve -- these days. Some of the participants, such as John Taylor, a Princeton professor who was on the staff of the Council of Economic Advisors in the Carter administration, still hold out hope that a shift in expectations can lead to lower inflation without significant increases in unemployment.

But others, including Yale's Nobel laureate James Tobin, are much less sanguine. "For the last 10, 15, or 35 years economists and policy makers have been frustrated by their inability to break the stubborn connection of output and price levels or of unemployment and inflation by any combination of the conventional monetary and fiscal tools of macrostabilization," Tobin said.

"Right now, most people mournfully agree, if we want more output and employment in 1983 than the standard forecast," he continued, "we will have to accept a higher price level, a higher year-to-year inflation statistic."

President Reagan during his campaign and for the first six months or so of his administration held out the hope that persistently slower money growth would change inflation expectations and lead to lower inflation with little, if any, loss of output or increase in unemployment.

It didn't happen that way, and as the experts gathered here demonstrated, they are nowhere close to agreeing exactly why.

If Tobin is right, then the Fed's dilemma is clear: seeking faster economic growth by allowing faster money growth would probably mean an end to the improvement in inflation.

New nostrums such as the one offered recently by some Senate Democrats -- which would put a limit on the level of real interest rates, if one were able to pick the right nominal interest rate and decide upon the proper rate of inflation to subtract from it to get a real interest rate -- would be little help.

Only if there is some magic eventually to be found in the realm of expectations can Tobin be proved wrong. So far, all the figures seem to be running his way, unfortunately.