Abstract
We test how government revenue and expenditure depend on economic activity, elections, and ideology. We show how the use of fiscal forecasts makes it possible better to understand the determinants of fiscal variables and to separate fiscal policy rules from discretionary policies. The approach is illustrated using a unique, unpublished Swedish data set of fiscal forecasts and forecasts of economic activity. Revenue varies positively with nominal earnings, expenditure varies negatively with real GDP. We find partisan effects, but no political business cycle effects. Revenue and expenditure are lower with non-Social democratic governments. The partisan effect on revenue is stronger than on expenditure. Using another unique data set, we find that there are autonomous decisions behind the reaction of expenditure, but not of revenue, to activity.