Working Paper: CEPR ID: DP9353
Authors: Klaus Desmet; Jordan Rappaport
Abstract: Gibrat's law, the orthogonality of growth with initial levels, has long been considered a stylized fact of local population growth. But throughout U.S. history, local population growth has significantly deviated from it. Across small locations, growth was strongly negatively correlated with initial population throughout the nineteenth and early twentieth centuries. This strong convergence gave way to moderate divergence beginning in the mid-twentieth century. Across intermediate and large locations, growth became moderately positively correlated with initial population starting in the late nineteenth century. This divergence eventually dissipated but never completely. A simple-one sector model combining the entry of new locations, a friction from population growth, and a decrease in the congestion arising from the supply of land closely matches these and a number of other evolving empirical relationships.
Keywords: Gibrat's Law; Growth across Space; Long-term Development; Settlement of US; Spatial Distribution of Population
JEL Codes: N91; N92; O18; R11; R12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| initial population size (J11) | population growth (J11) |
| entry of new counties (R19) | population growth (J11) |
| age of locations (Y91) | population growth (J11) |
| population growth (J11) | initial population size (J11) |
| shift away from land-intensive production (Q15) | population growth (J11) |
| increase in returns to agglomeration (R11) | population growth (J11) |