The contribution of this paper is twofold. First, it builds and
makes use of long-run data from Sweden on formal education that
have never been used to date. Second, it provides a quantitative
application of recent theoretical work on the link between
demographic changes and economic growth through their effect on
education. It concludes that changes in longevity may account for
as much as 20% of the observed rise in education over the period
from 1800-2000 via a horizon effect, but have little impact on
income growth over the period. On the contrary, changes in
population density and composition are central, mainly thanks to
their effect on productivity. Most income growth over this period
would not have materialized if demographic variables had stayed
constant since 1800.