Abstract: Did the economic status of US colonial settlers have an impact on their descendants many generations later? Using surnames, US censuses, and novel data on colonial migration, I follow the descendants of 45,000 English migrants, most of them bound migrants who arrived with little or no wealth and worked several years without pay. In 1850, descendants of bound migrants were 16% less likely than other US-born white men to own real estate and more than twice as likely to be illiterate. While the gap in real estate ownership does not fully close, they earn higher wages and are better educated in 1940. The gap in real estate ownership implies a parent-child correlation of 0.3 to 0.6 under different functional form assumptions, between 1750 and 1940. These findings imply that even in the United States, the status of ancestors predicted the real estate of their descendants for nearly two centuries; however, persistence in one dimension did not entail an overall disadvantage.
Abstract: Long-run persistence of economic status across multiple generations has been shown to be much higher than what parent-child associations imply. I argue that this inconsistency stems from using the wrong functional form to connect short- and long-run associations. I compile a dataset of all published multigenerational estimates to show that economic status across more distant generations approximately follows a power law. This result is expected if economic status is transmitted through many channels, such as wealth, location, occupation, and networks, that differ in their rates of transmission. Using linked US Censuses from 1850 to 1940, I then show that status explained by location and urban residence is transmitted from parents to children at much higher rates than the rest of occupational status. Finally, I propose a novel test in the mobility literature that measures the dispersion between transmission rates of different latent channels, which is immune to measurement error and temporal shifts in mobility.
Abstract: Intergenerational mobility is usually summarized by a single parent-child slope. That statistic is useful, but it implies that dynastic correlations should disappear within a few generations and that inequality-mobility patterns such as the Great Gatsby curve should be difficult to reproduce. We show that both problems are resolved by a minimal extension of the canonical AR(1) model: adding a permanent lineage component. Using linked full-count U.S. censuses from 1860 to 1940, we assemble 10.6 million father-son pairs in 2.8 million dynasties and compare the resulting AR(1)+lineage-effects model with standard autoregressive alternatives. Estimated only on 1860-1900 data, AR models predict that long-run persistence should vanish and underpredict the rise in occupational-score inequality observed in 1920 and 1940. The lineage-effects model, by contrast, matches both the correlation with the 1860 patriarch and the path of the Gini out of sample. We then derive closed-form steady-state expressions for inequality and persistence as functions of within-lineage reversion, lineage dispersion, and transitory shocks. The decomposition shows that dispersion in permanent lineage means, not the parent-child elasticity, is the main driver of long-run inequality and persistence. This yields a structural Great Gatsby curve and reframes mobility policy: reducing inherited lineage advantage has larger long-run implications than marginally increasing parent-child mobility.
Colonial Migration to the US (with Taylor Jaworski)
Early Life Events and Legal Decision-Making of American Judges (with Pablo Zarate)
Historical Development of Inequality of Opportunity in the United States (with Francisco Ferreira)