Abstract
To study long-run large-scale early childhood policies, this paper incorporates early childhood investments into a general equilibrium heterogeneous agent overlapping generations model calibrated using US data. A randomized controlled trial evaluation of a short-run small-scale early childhood program in the model predicts effects on children’s education and income that are similar to the empirical evidence. A long-run large-scale program, however, yields over twice as large welfare gains as those of a short-run small-scale program even after considering general equilibrium and taxation effects. Key to this difference is that investing in a child not only improves her skills but also creates better ancestors for future generations.