(FORTUNE Magazine) – When you turn on your television in the morning in search of the weather report, chances are you really want to know only one thing: Is it going to rain, or not? Small but important decisions ride on the answer, like whether you should allow extra time to get where you're going and whether you should wear your coat. Likewise, when most people seek out economists, the single most important question they want answered is, When will there be a recession? Here the decisions dependent on the answer tend to be big and important, such as whether it is a good time to expand your business, buy a house, or take a new job. Unfortunately, this one thing people really want from economists is the one thing economists can't give them.

Economic forecasters, in fact, are very bad at predicting recessions. At any given moment, they can't tell you with much certainty whether a recession will begin next quarter, let alone next year. Right now, for instance, there's abundant evidence that the economy is slowing down but no guarantee that it will achieve the so-called soft landing (a slowdown, with lower inflation, followed by renewed expansion) instead of slipping into recession. Hardly any forecasters warn of an imminent downturn, but the truth is that when the next one comes, they probably won't see it until it has already begun. To work the weather analogy one last time, it's as though weatherpersons could tell you whether it was going to rain only hours after the rain had started.

Think we're kidding? Look at the forecasting record of America's business economists in the last recession--the mild downturn of 1990. As that undistinguished year began (it was the second year of the Bush presidency--remember?), the Blue Chip consensus, an amalgam of forecasts by 50 or so of the nation's leading business economists, was predicting a slowdown in economic growth. So far, so good. The average forecast called for 1.7% real growth in 1990, and 2.2% in 1991. Then, as now, there was optimistic talk of a soft landing. As the year wore on, the forecasters stood pat. In July the Blue Chipsters were still predicting 1.9% growth for the year, rising to 2.2% in 1991.

That, of course, was the month the recession actually began. Not until November did a majority of the forecasters admit that a recession would even take place, and most were still convinced that GDP would be up in 1991 (it contracted 0.6%). By the time the recession had been officially recognized and the Monday-morning quarterbacking had gotten under way in earnest--by April 1991, say--the recovery and ensuing business expansion had begun. Says Maureen Allyn, chief economist at Scudder Stevens & Clark (she predicted a mild recession that fall): "When you're in the middle of it, the signals appear conflicting, and the data don't help. It's a big, messy world out there."

There are lots of reasons that economists can't predict recessions. The simplest has to do with the nature of forecasting itself. In spite of all the mathematical and statistical finery in which modern economists array themselves, what they are doing--more or less--is predicting that the near future will be pretty much like the recent past. That is usually what happens, and for that reason economic forecasts are usually fairly accurate when the economy is expanding, as it's been doing since early 1991. But it's also the reason they tend to miss turning points--those moments when something fundamental changes. Ultimately, as the late Karl Brunner, an eminent economist who taught at the University of Rochester, put it, "There is nothing in classical economic theory that will lead you to believe you can predict the future."

The best forecasters don't pretend to have all the answers. Each month DRI/ McGraw-Hill--the nation's largest forecasting outfit--ships its clients a fat book reviewing the economy's performance and explaining the firm's forecast. One chapter is always titled "Risks to the Forecast." It shows what might happen to the economy if they've missed something. The three sources of risk, says Roger E. Brinner, DRI's chief economist, are (1) an extreme policy shift in Washington, (2) an international incident, and (3) "manic-depressive mood swings in business or consumer behavior."

So we shouldn't be too hard on forecasters. Few would resurrect the 19th-century New York state law that imposed a $250 fine or a six-month jail term for persons "pretending to forecast the future" (there was an exception for ecclesiastical bodies). Just don't count on even the best of them to give you a reliable warning about the next recession.