Rural-Urban Migration at High Urbanization Levels
with Matias Busso and Nicolás Herrera L. Regional Science and Urban Economics, 2021.
This study assesses the empirical relevance of the Harris-Todaro model at high levels of urbanization — a feature that characterizes an increasing number of developing countries, which were largely rural when the model was created 50 years ago. Using data from Brazil, we compare observed and model-based predictions of the equilibrium urban employment rate of 449 cities and the rural regions that are the historic sources of their migrant populations. We find little support in the data for the most basic version of the model. However, extensions that incorporate labor informality and housing markets have much better empirical traction. Harris-Todaro equilibrium relationships are relatively stronger among workers with primary but no high-school education, and are more frequently found under certain conditions: when cities are relatively larger; and when associated rural areas are closer to the magnet city, and are populated to a greater degree by young adults, who are most likely to migrate.
What is Different About Urbanization in Rich and Poor Countries? Cities in Brazil, China, India, and the United States
with Ed Glaeser, Nina Tobio and Yurean Ma. Journal of Urban Economics 98: 17-49, 2017.
Are the well-known facts about urbanization in the United States also true for the developing world? We compare American metropolitan areas with analogous geographic units in Brazil, China, and India. Both Gibrat’s Law and Zipf’s Law seem to hold as well in Brazil as in the U.S., but China and India look quite different. In Brazil and China, the implications of the spatial equilibrium hypothesis, the central organizing idea of urban economics, are not rejected. The India data, however, repeatedly rejects tests inspired by the spatial equilibrium assumption. One hypothesis is that spatial equilibrium only emerges with economic development, as markets replace social relationships and as human capital spreads more widely. In all four countries, there is strong evidence of agglomeration economies and human capital externalities. The correlation between density and earnings is stronger in both China and India than in the U.S., strongest in China. In India the gap between urban and rural wages is huge, but the correlation between city size and earnings is more modest. The cross-sectional relationship between area-level skills and both earnings and area-level growth are also stronger in the developing world than in the U.S. The forces that drive urban success seem similar in the rich and poor world, even if limited migration and difficult housing markets make it harder for a spatial equilibrium to develop.
Working paper version
Featured in the NBER digest (May 2016)
Press: Citylab, LiveMint (2016, 2017), PBS
Working Papers
Gender, electoral incentives, and crisis response: Evidence from Brazilian mayors
with Clemence Tricaud
While there is evidence of gender differences in policy preferences and electoral strategic behaviors, less is known about how these differences play out during crises. We use a close election RD design to compare the performance of female- and male-led Brazilian municipalities during the COVID-19 pandemic. We find that having a female mayor led to more deaths per capita early in the first wave of the pandemic — a period characterized by great uncertainty about the severity of the disease and the effectiveness of containment policies. In contrast, having a female mayor led to fewer deaths per capita early in the second wave — a period where this uncertainty was reduced, and when the 2020 mayoral election took place. Consistent with the evolution of deaths, we find that female mayors were less likely to implement commerce restrictions at the beginning of the period, while they became more likely to do so at the end. We also show that the second-wave effect coincides with a lower tendency of the population in male-led municipalities to stay at home around election day. Both the first and second wave effects are driven by municipalities whose mayors were not term limited, and thus allowed to run for re-election. These findings suggest that the gender differences we observe stem from female and male mayors reacting differently to electoral incentives. While electorally motivated female mayors were more likely to delay restrictive policies at the beginning, electorally motivated male mayors were more likely to open-up the municipality closer to the election.
Why does COVID-19 affect some cities more than others? Evidence from Brazil
This paper investigates what explains the variation in impacts of COVID-19 across Brazilian cities. I assemble data from over 2,500 cities on COVID-19 cases and deaths, population mobility, and local policy responses. I study how these outcomes correlate with pre-pandemic local characteristics, drawing comparisons with existing US estimates when possible. As in the United States, the connections between city characteristics and outcomes in Brazil can evolve over time, with some early correlations fading as the pandemic entered a second wave. Population density is associated with greater local impact of the disease in both countries. However, in contrast to the US, the pandemic in Brazil took a greater toll in cities with higher income levels — consistent with the fact that higher incomes correlate with greater mobility in Brazil. Socioeconomic vulnerabilities, such as the presence of slums and high residential crowding, correlate with higher death rates per capita. Cities with such vulnerabilities in Brazil suffered higher COVID-19 death rates despite their residents’ greater propensity to stay home. Policy responses do not appear to drive these connections.
The younger age profile of COVID-19 deaths in developing countries
with Annabelle Fowler and Nicolás Herrera L.
This study assesses the empirical relevance of the Harris-Todaro model at high levels of urbanization — a feature that characterizes an increasing number of developing countries, which were largely rural when the model was created 50 years ago. Using data from Brazil, we compare observed and model-based predictions of the equilibrium urban employment rate of 449 cities and the rural regions that are the historic sources of their migrant populations. We find little support in the data for the most basic version of the model. However, extensions that incorporate labor informality and housing markets have much better empirical traction. Harris-Todaro equilibrium relationships are relatively stronger among workers with primary but no high-school education, and are more frequently found under certain conditions: when cities are relatively larger; and when associated rural areas are closer to the magnet city, and are populated to a greater degree by young adults, who are most likely to migrate.
Rural Spillovers of Urban Growth
with Sam Asher, Idaliya Grigoryeva, and Paul Novosad
As cities in developing countries continue to grow rapidly, there is insufficient empirical evidence on how this affects growth in surrounding rural economies. We study the effects of shocks to labor demand in cities on village-level economic outcomes, using a new dataset with administrative data from multiple sources on the universe of urban and rural economies in India.
New draft coming soon
Gender-Segmented Labor Markets and the Effects of Local Demand Shocks
Gender segmentation in the labor market is widespread. However, most existing studies of the effects of labor demand shocks on local economies assume away gender. In this paper, I show that local labor demand shocks can lead to different outcomes depending on whether they favor male or female employment. I develop a spatial equilibrium model that features gender segmented labor markets and joint mobility frictions, which predicts that couples are more likely to migrate in response to male opportunities. As a result, positive shocks to local labor demand for men lead to population growth, increases in female labor supply, and housing demand growth. Meanwhile, equivalent shocks to labor demand for women lead to smaller inflows of migrant workers, and labor force participation is a relatively more important margin of adjustment in this case. I find strong empirical support for the model’s predictions in the context of Brazil during 1991-2010. Comparing the effects of gender-specific labor demand shocks, I show that male shocks produce a higher migratory response and make localities more populated and expensive. These results imply that place-making policies that create jobs for females are more likely to benefit residents while those that create male jobs are more likely to benefit immigrants and landlords.
Public Education Investment and Local Labor Markets. Evidence from a Large Federal Program
Do primary education investments improve local labor market outcomes? In principle, education could lead to higher local productivity, but potential benefits to local economies could be muted if workers migrate in search of better opportunities, or if shifts in the supply of skills outpace demand growth. I use a large program that redistributed public education finance across Brazilian municipalities (FUNDEF) as a source of exogenous variation to empirically study the effects of expansions in public education expenditure on attainment and labor market outcomes at the individual and the local economy levels. The program was successful at improving primary educational attainment levels for individuals and microregions (local labor markets). High-exposure individuals experienced higher wages and a higher likelihood of migrating to a different local economy. High-exposure microregions saw negative effects on the growth of average wages and other labor market outcomes with the exception of employment, suggesting that the increased supply of educated workers was not matched by demand growth.