The question of forecasting accuracy is, of course, one that has been the subject of numerous investigations over the last two decades. The present study contributes to this line of research in two ways. First, we introduce a new concept, called “forecast eﬀiciency,” that measures the extent to which information is incorporated into forecasts. This concept is closely related to concepts of eﬀiciency used in the analysis of stock and other ﬁnancial markets. The paper proves two readily testable propositions about eﬀicient forecasts. Second, the empirical part of the study examines forecast eﬀiciency by looking at forecast revisions (“ﬁxed-horizon forecasts”), rather than a series of forecasts of diﬀerent events (“rolling-horizon forecasts”) as is the case for most studies of forecasting. This new approach to estimation in certain circumstances will provide a more powerful test of forecast eﬀiciency. A number of ﬁxed-horizon forecasts are collected and these are tested for forecast eﬀiciency.
Nordhaus, William D., "Forecasting Efficiency: Concepts and Applications" (1985). Cowles Foundation Discussion Papers. 1015.