The author of this brief examined forecast accuracy conditional on forecaster disagreement of nine variables within the SPF compared to a benchmark model using the Giacomini and White forecast accuracy test. The most general finding is that where there is a conditional difference in forecast accuracy between the SPF and the benchmark model, the SPF becomes more accurate than the benchmark model as forecaster disagreement rises in many of the examined cases. Additional evidence emphasizes the importance of using all four historical realizations of the data to test the forecasts against rather than relying on just a single realization for forecast accuracy tests.

View the Full Brief