This paper addresses two issues. To what extent can models
estimated on modern data be used to account for growth patterns in
the past? Can information on historical patterns help to improve
long-term forecasting of economic growth? We consider a
reduced-form statistical model based on the demographic dividend
literature. Assuming that there is a common DGP guiding growth
through the demographic transition, we use an estimate from
post-war global data to backcast the Swedish historical GDP growth.
The results indicate that the assumption of a common DGP can be
warranted, at least back to 1870. Given the stability of the
relationship between population and growth, we use the model to
forecast income for the next 50 years. We compare our approach to a
previous attempt to simulate the long-term Swedish growth path with
an endogenous growth model. Encompassing tests show that each of
the models contains independent information on the Swedish growth
path, suggesting that there is a benefit from combining them for
long-term forecasting.