Hogarth and Makridakis, in acomprehensive review of the applied management and finance it erature, show that the record of forecasters using both judgement and sophisticated mathemati cal methods is not good. What they do suggest, however, is that certain forecasting techniques perform better under certain circumstances.
In short-term forecasting there is: considerable inertia in most economic and natural phenomena. Thus the present states of any variables are predictive of the short-term future(i.e three months or less), Rather simple mechanistic methods, such as those used in time series forecasts, can often make accurate short- term forecasts and even out-perform more theoretically elegant and elaborate approachesusd in econometric forecasting,"
Long-term forecasting methods, although difficult to judge because of the time la between the forecast and the event, do seem to be more amenable to an ca approach. In a comparative study of long-term market forecasting methods, Armstrong annan conclude that econometric methods offer more accurate long-range forecasts than do expert opinion or time series analysis, and that the causal methods improves as the time horizon increases.