Some comments on the evaluation of economic forecasts