FORECAST SMOOTHING AND THE OPTIMAL UNDER-UTILIZATION OF INFORMATION AT THE FEDERAL-RESERVE

It is often found that forecasts are biased and/or inefficient, violat ing the often invoked rationality assumption. This paper proposes that forecasters may not be irrational, but have been misrepresented by ac ademic economists. Specifically, it is usually implicitly assumed that forecasters wish to minimize some function of the forecast error. Thi s formulation ignores the professional milieu of the forecaster. In pa rticular, if a forecaster is concerned about his reputation, then it c an be argued that this leads to a desire to ''smooth'' forecast revisi ons. The Federal Reserve staff's forecasts for real GNP and for inflat ion are tested for evidence of the proposed ''reputation effect.''