It is common in the economics profession to think of forecasting as an art, or as a field of endeavor that is at best pseudoscientific. This view is unfortunate, though not without justification. A great deal of what forecasters have done has not been even pseudoscientific. The impostures and errors of the past--sometimes the consequence of detours by reputable economists into unfamiliar territory--tended to keep forecasting on the fringe of what was supposed to be the science of economics. Even when efforts met with success, those who succeeded often refused to disclose their methods, so that there was no way to determine if anything of scientific value had been achieved.
The fault, however, has not been all on the side of the forecasters. Partly it lay in the traditional notion of what constituted scientific method in economics. Much of what has been built onto the structure of this intellectual discipline through the generations has represented an unscientific retreat from reality. In some of its aspects economics became preoccupied with "principles of behavior" no better founded than the forecasts that were rejected. There was one important difference: The forecasts could be checked, but the principles could not. No discipline has a right to call itself scientific except as its theory meets the test of explaining what happens, and what will happen, in the world of things and people.
It is the custom among writers of books on economics--a custom not followed here--to conclude with a section on SYSTEM policy and control of business fluctuations. Sometimes these discussions are filled with brilliant, though perhaps overly abstract and impractical, ideas. Most of them ignore human relations and the mechanisms by which things actually get done. It may be perceived that control is possible only if the results of action can be foretold; but there may at the same time be a failure to perceive that persuasive foretelling is possible only in terms of the realities which tend to be ignored.
This is not to say that forecasting must be a final goal of action for . . .