Inflation Forecasts Are Poor Compass for Fed

ECONOMISTS tell a joke on themselves about Albert Einstein arriving in heaven. It says something about their forecasting accuracy.

The famous scientist is first greeted by a person who claims he has an IQ of 190. ''Excellent,'' says Professor Einstein, ''we can talk about my notions of cosmology.'' A second visitor admits an IQ of 150. ''Fine,'' says Einstein, ''we can discuss the political problems of the world.'' A third greets the genius and confesses to an IQ of 75. ''Fantastic,'' Einstein replies. ''How much will the United States economy grow this year?''

Actually, economists today must be bright enough to master advanced mathematics and complex theories. But even with sophisticated techniques, forecasting the economy is not easy.

''Inflation is extremely difficult to forecast,'' Stephen Cecchetti, a Boston College economist, says, summarizing new research at a National Bureau of Economic Research conference in Cambridge, Mass., last weekend.

Mr. Cecchetti's conclusion is relevant to the decision a year ago by Federal Reserve policymakers to raise short-term interest rates. At that time, inflation was modest. But the Fed assumed that it would get worse and acted to slow the economy. Since then, inflation has remained in fact fairly steady. The rise in producer prices was 1.7 percent in 1994, up from 0.2 percent in 1933. Consumer prices rose 2.7 percent in both years. The deflator for the whole economy dropped slightly.

If Cecchetti is right, the Fed might claim some credit for this low inflation. He says a hike in interest rates can start to restrain prices in six months.

Statistics released this week show both wholesale and consumer prices up 0.3 percent in February. Spurred by a rising tab for food and energy, producer prices rose at a 3.9 percent annual rate in the first two months. A similar surge occurred in the first months of 1994.

Cecchetti observes that since late 1979, when inflation peaked at 13.3 percent, the goal of monetary policy ''has progressively shifted toward the reduction of the level and variation of inflation.'' The Fed downplays the likely impact of its interest rate hikes on output and unemployment, he holds.

But, to run a policy aimed at controlling prices, it is crucial that Fed policymakers be able to assess the future path of prices, Cecchetti notes. He examines 1970 to 1994 forecasts of Data Resources Inc., a prominent economic consulting firm based Lexington, Mass., and the consensus forecasts since 1982 of the 50 economists surveyed by Blue Chip Economic Indicators (Sedona, Ariz.). Looking at the last decade or so, the forecast errors have run between 1.2 percent and 2 percent.

Cecchetti finds that these inflation forecasts are slightly better than ''naive'' forecasts that assume current inflation rates will continue into the future. So, he says, Fed policymakers should be able to stabilize prices modestly by taking counterinflationary action based on sophisticated inflation forecasts.

Milton Friedman, a Nobel Prize-winning economist, has argued that economists' forecasts are so inaccurate the Fed would stabilize the economy better by pumping out a steady supply of new money and avoiding taking any countercyclical measures.

Donald Kohn, head of the Federal Reserve Board's monetary research staff, says Cecchetti's findings of high inflation forecasting errors ''overstate our inability to forecast inflation.'' The errors for the Fed's gross domestic product deflator forecasts run in the 0.25 percent to 0.5 percent range, he says. That's ''not nearly as bad'' as Cecchetti's findings.

Fed economists and policymakers, he says, do use econometric models that massage the statistics to make forecasts. But they add many other factors on a ''judgmental'' basis to those ''black box'' forecasts in making policy decisions.

But Cecchetti also finds that any rules for forecasting inflation -- say looking at the price of gold, or the price of commodities, or unemployment levels, or industrial capacity utilization -- are ''unstable.'' They work for some years, but not for other years.

Mr. Kohn notes that policymakers really have no choice -- they have to make decisions with the data they have.

''Life is tough,'' says Mark Watson, a Northwestern University economist, commenting on the Cecchetti paper.