Are economic forecasts significantly better than naive predictions? An appropriate test