This paper provides evidence of the quality of private sector forecasts of the budget balance between 1993 and 2009 for a sample of 29 countries, grouped into advanced and emerging countries. We find large differences across the two groups: forecasts for advanced economies are much more accurate than for emerging economies and much less subject to a bias towards optimism (i.e. they are less likely to forecast a bigger budget balance than the realization). Forecasts for both groups, however, exhibit a tendency toward forecast smoothing: forecasts are revised slowly so that revisions to forecasts can be systematically predicted based on past revisions. This tendency proves costly around turning points in the economy when the budget balance moves sharply but the corresponding forecasts only adjust very slowly to the reality of the situation.