Abstract
I study how geography shaped city formation and aggregate development in the United States prior to the Civil War. To guide my analysis, I first present a conjecture that cities' farm hinterlands fostered both city development and aggregate growth: the hinterland hypothesis. The hinterland hypothesis has rich implications on how various elements of U.S. geography -railroads, changes in U.S. political borders, increasing U.S. population, and international trade - affected city formation and U.S. growth. To quantitatively evaluate the hinterland hypothesis and its implications, I assemble a novel historical dataset on population, trading routes and agricultural productivity at a high spatial resolution, and combine it with a dynamic quantitative model of economic geography. I find evidence for the hinterland hypothesis by showing that the model can quantitatively replicate the key patterns of U.S. urbanization and city formation. Finally, I conduct a series of counterfactuals in the model to quantify the effect of geography on cities and growth, guided by the implications of the hinterland hypothesis. Results indicate that railroads were responsible for 8.2% of urban population in 1860 and for 27% of real GDP growth between 1830 and 1860. The effect of international trade was similar in magnitude, while population growth slowed down urbanization and GDP growth. The effect of political border changes was small during the period.