Generic Conditions for Forecast Dominance
share
Recent studies have analyzed whether one forecast method dominates another under a class of consistent scoring functions. For the mean functional, we show that dominance holds under simple conditions: Both forecasts must be auto-calibrated, and one forecast must be greater than the other in convex order. Conditions for quantile forecasts are similar but more complex. Unlike existing results, the new conditions allow for the case that the forecasts' underlying information sets are not nested, a situation that is highly relevant in applications.
READ FULL TEXT