Using panel data for OECD countries, this study investigates the extent to which changes in government spending on education, health and other areas influence long-term growth. The results suggest that, if total government spending is kept unchanged, increasing expenditure on health, education and transport raises long-term GDP growth. In contrast, government spending on housing is found to weaken long-term GDP growth. The error-correction specification used allows assessing adjustment speed which, consistent with intuition, is estimated to be slow. According to the econometric results, it takes more than five years for half of the effect of a change in the structure of government spending to be reflected in longterm growth.