Association for Public Policy Analysis & Management

In the past two decades, demand for college-educated workers has risen in many cities. While many welcome this trend, there is growing concern about the plight of less educated workers, who have not seen the same gains; indeed, households without any college-educated adults have seen incomes decline in almost all U.S. cities. We explore whether the increasing demand for college-educated workers has compounded the challenges of less educated households through increasing rents or, conversely, has generated new opportunities through positive labor market spillovers, improved city amenities, and increased neighborhood economic diversity. While previous work has analyzed skill sorting across cities, our interest lies in how less educated households remaining in these cities are faring.

This estimate the effects of increasing relative demand for college graduates on the well-being of less educated households along a number of dimensions, including housing costs, residual income, commuting costs, crowding, and exposure to higher human capital neighbors. To isolate the causal impact of shifts in labor demand for college graduates, we use an instrumental variables strategy motivated by Bartik (1991). We separately control for exogenous shocks to demand for workers with no more than high school educations. We distinguish between effects on incumbent households and effects due to changes in the composition of households in a given city due to migration using regression correction.

We also estimate effects on high-school only households relative to college graduate households in the same city, providing direct insight into effects on within-city inequality of well-being along a number of previously unexplored dimensions. We explore heterogeneity in effects by estimating quantile regressions to test if effects differ at the extremes of the distribution.

We answer these questions using two main sources of data. The first is Integrated Public Use Microdata Sample (IPUMS) versions of the 1990 Decennial Census, 2000 Decennial Census, and 2010-2014 5-year American Community Survey (ACS). The second combines restricted versions of the Census Decennial and ACS surveys with HUD administrative data on rental assistance receipt. We use unique individual identifiers developed by Census to match individuals in the 2000 Census and ACS 2010-2014 to individuals receiving housing choice vouchers or other types of rental subsidies from the Department of Housing and Urban Development. This novel data set will allow us to test whether housing subsidies help to insulate high-school only households from rising rents, allowing them to remain in potentially improving neighborhoods as rents rise.

Our results have important implications for understanding the distributional and welfare implications of increasing incomes of college graduates within cities. Our results are mixed. On the one hand, it appears that increasing demand for college graduates is actually increasing residual incomes among high-school only households. At the same time, it is increasing crowding among high-school only households and widening the differential between the human capital levels of the neighborhoods lived in by more and less educated households. In future drafts, we will explore how housing choice vouchers and other forms of subsidized housing shape these absolute and relative effects.