Russel Sage Foundation

While theoretical analysis of intergenerational mobility has a long tradition in economics, empirical evidence has lagged behind because of a lack of adequate data. Few survey data sets provide information on both parent and child outcomes, and those that do generally have small samples. To address this gap, economists Raj Chetty and Nathaniel Hendren have constructed new public use statistics on intergenerational mobility in the United States using tax data from more than 40 million children and their parents. The data set was made public in January 2014.

Chetty and Hendren have already explored some aspects of the data. For example, they have shown that upward mobility across geographic areas is correlated with factors such as racial and economic segregation, school quality, social capital, family structure, and middle-class income inequality. However, many additional issues can be analyzed with this data. For example, they have not explored downward mobility, possible poverty traps, county-level mobility, time trends, and causal determinants of changes in mobility, among other topics. Given the richness of this new data set, RSF issued a request for proposals to encourage other researchers to use the data for studying intergenerational mobility. The following projects are being funded under this initiative:

School Finance Equalization and Intergenerational Mobility: Does Equal Spending Lead to Equal Opportunities?

Barbara Biasi (Stanford University)

Biasi seeks to understand the effect of the distribution of school spending on intergenerational income mobility. She will exploit the variation that school finance reforms introduced in the distribution of per-pupil spending within each state to understand the effects on future economic opportunities of children. The correlation between average per-capita income and per-pupil spending across school districts within each commuting zone (CZ) and in each year will be explored, as well as the effect of school finance equalization reforms on intergenerational mobility. The dependent variables are CZ-level measures of intergenerational mobility. Other data sources come from the National Center for Education Statistics of the U.S. Department of Education, the NCES Longitudinal School District Fiscal-Nonfiscal (FNF) file, and the Bureau of Economic Analysis’ Regional Economic Accounts.

The Determinants of Intergenerational Mobility: An Intra-Household Perspective

Marianne Bruins (Yale University)

Bruins will address the question: “What geographic disparities in children's and parents’ time-use and household consumption coincide with variation in intergenerational mobility across counties in the United States?” She will analyze detailed information on children’s time-use and household consumption from the American Time-Use Survey (ATUS), and Chetty et al.’s data on intergenerational mobility across commuting zones. The research seeks to improve our understanding of intergenerational mobility, with particular interest in the variation in the time-use patterns of children born to parents in the lowest income quartile, and how they compare to those of children in the top income quartile in the same county. Observations in the ATUS will be matched at the county-level when it is available and at the metropolitan statistical area data otherwise. Bruins will also supplement her analysis with data from the Panel Study of Income Dynamics child development supplement (PSID-CDS).

Shocks in the Geography of Opportunity: The Foreclosure Crisis and College Enrollment

Jacob Faber and Peter Rich (New York University)

Faber and Rich will assess the hypothesis that wealth losses experienced during the foreclosure crisis may have muted or stunted the counter-cyclical college entry response related to diminishing job opportunities, and varied for families at different points in the income distribution. Research questions include: What is the relationship between the timing of the foreclosure crisis and family college enrollment decisions? How did the relationship vary across the income distribution? Primary data will come from RealtyTrac, with other sources being the Bureau of Labor Statistics’ Local Area Unemployment Statistics, and the Integrated Postsecondary Education Data System (IPEDS). Faber and Rich will use a fixed effects regression, including sensitivity analyses. Reverse causality and lagged effects will be tested for by varying the temporal relationships between the outcome and explanatory variables.

Neighborhood Conditions and Upward Social Mobility

Lindsay Fox and Joe Townsend (Stanford University)

Fox and Townsend will investigate the association between income segregation and mobility by exploring how neighborhood conditions—particularly variation induced by segregation—may mediate the relationship between segregation and intergenerational mobility. They will test how differences in neighborhood income levels, educational attainment, unemployment rates, and household type are associated with intergenerational mobility. They will merge the Chetty et al. data across CZs with measures of neighborhood conditions. Using a set of parametric smoothing methods applied to tract-level Census data, they have constructed measures of average neighborhood income, educational attainment, unemployment, and family structure that are specific to families of a given income in each metropolitan area. This will allow them to test if the CZ-level association between income segregation and mobility is driven by differences between CZs in the neighborhood characteristics of families at a given income level.

The Effects of Local Job Destruction on Youth Mobility

Anna Gassman-Pines and Elizabeth Oltmans Ananat (Duke University)

Gassman-Pines and Ananat will measure the effect of area job losses when a cohort is age 17 on college-going at the population rather than individual level, and explore the consequences of area job losses for later earnings inequality. They hypothesize that community-wide job losses realized at a critical life stage for a given cohort substantially affects its aggregate mobility. By leveraging the timing of job losses relative to this life stage, they seek to better identify the effects of macroeconomic change on intergenerational mobility. Ananat and Gassman-Pines will combine data constructed on mass job losses by month in every county in North Carolina and by quarter in every state in the U.S. with local area statistics on mobility by cohort to identify how changing local job opportunities are associated with the relationship between family income and college attendance. They will also use data from the North Carolina Security Employment Commission, a database that lists, by county and month, all job layoffs and closings in the state. Data on job losses are from the Bureau of Labor Statistics’ Mass Layoff Statistics. Data on state college tuition policies are from the NCES Integrated Postsecondary Education Data System.

The Effect of Violent Crime on Economic Mobility: Moving Toward Casual Inference through Instrumental Variables

Patrick Sharkey (New York University)

Sharkey will use an instrumental variables approach to identify the causal effect of violent crime on levels and trends in economic mobility. He seeks to address two questions: 1) Looking across CZs, how is the level of violent crime associated with the level of upward and downward income mobility? 2) Looking within CZs, how are changes in violent crime rates associated with changes in upward and downward economic mobility? He has identified four instruments for the level or change in violent crime in a city, metropolitan area, county, or collection of counties: 1) the proportion of the population that is male and between the ages of 15 and 24; 2) the proportion of the population that is female and between the ages of 15 and 24; 3) the timing and size of police agencies’ receipt of federal grants through the Community Oriented Policing program; and 4) the number of firefighters per capita.