We investigate whether there are systematic differences between forecasters in terms of their levels of disagreement and the accuracy of their forecasts, and whether these di¤erences are related to whether or not a forecaster efficiently uses their available information. We ?find that forecasters are not interchangeable. At any point in time, the level of disagreement between forecasters is more likely to be due to a given set of forecasters, as opposed to any randomly-selected set of forecasters. In terms of forecast accuracy, we also ?nd persistence, in that forecasters who are more (less) accurate in one period tend to be more (less) accurate in a subsequent period. Finally, we reject efficiency for around half of all forecasters at short horizons (depending on the variable in question), and ?find that efficient forecasters tend to be more accurate and less contrarian. Our results do not support the notion that contrarian forecasts stand apart by virtue of having superior information - knowing something that others do not.