This is a surprising title. Previous research on this topic has been diercted to the question: Does anyone forecast better? The answer has been no. Little success has been achieved in stating, before the event, who are likely to produce the more accurate forecasts. Even research that has examined forecasts after the event has failed to find significant differences.
Stekler analysed forecasts made by 32 forecasters from the Eggert Blue Chip Survey. The forecasts covered nominal GNP, real GNP, and the GNP deflator for each year from 1977 to 1982. Two accuracy criteria were used, RMSE and accuracy rankings. The results were surprising: Forecasters who were more accurate in one time period tended to be more accurate in the next; and forecasters who were more accurate in forecasting one variable were more accurate in forecasting others. These findings do not agree with prior research. Stekler cites three prior studies, none of which showed differences. To this list of studies, I would add Pencavel (1971), Fromm and Klein (1973), McLaughlin (1973), and Hatjoullis and Wood (1979); these studies also yielded negative results. If Steklers study is correct, it makes another question relevant: Why is one forecast more accurate than another? To study this, one needs to identify the methods used by each forecaster in order to determine how they relate to accuracy. Such work has not been undertaken often. Most research has compared methods directly. Nevertheless, studying the forecasters to infer which methods are best has an advantage because it reveals how the methods perform in practice. Bretschneider and Gorr (1987) make a promising start in this direction in their study of government revenue forecasting. However, the small differences in accuracy among macroeconomic forecasters suggest that research in this area will be difficult.