Yujiang River Chen, Coen Teulings

CEPR Discussion Paper DP20201 05 May 2025 This Revision 05 September 2025

The high return to human capital on GDP per capita, reaching up to 50% across countries or
regions, is puzzling. We develop a spatial model in which regions are classified as either rural
areas or monocentric cities. In this framework, both human capital and the concentration of
employment in city centers drive knowledge spillovers. Regional land prices clear the market for
inter-regional labor mobility, leading to joint predictions for the public return to human capital and
for land prices. We test the model using data on wages and real-estate prices for 47 U.S. rural
areas and 34 CMSAs from 1979 to 2015. Our findings indicate that the public return on wages is
25% of the private return in rural areas, rising to 136% in cities. The return to human capital on
GDP per capita is 24% in the cross section. The rise in urban house prices is largely accounted for
by knowledge spillovers.