Is Gentrification a Carcinogen? Neighborhood Change and Cancerous Vehicle Emissions in Los Angeles County

Abstract

Neighborhood disadvantage erodes residents’ mental and physical health. But whether rapid reductions in disadvantage spurred by gentrification attenuate or exacerbate these effects remains unknown due to mixed theoretical expectations and empirical results. To help clarify these dynamics, I propose a novel hypothesis that casts gentrification as a carcinogen. As neighborhoods receive inflows of affluent, White residents, influxes of private vehicles may come with them. In turn, stationary residents become exposed to higher vehicular emissions, and their risk of cancer—especially lung cancer—climbs. As an initial empirical test of these theoretical possibilities, I link Urban Displacement Project data identifying Los Angeles County neighborhoods that gentrified during the 2000s to tract-level data on vehicle ownership and cancer risk profiles—the latter from the Environmental Protection Agency’s National Air Toxics Assessment. Descriptive regressions that include a lagged dependent variable and municipal fixed effects suggest gentrifying tracts’ levels of cancer risk factors increased by ∼0.5 standard deviations more than those of disadvantaged neighborhoods that did not gentrify. Sobel tests of mediation indicate nearly half of this association may be explained by a pathway related to increasing vehicle density. The study thus motivates future research leveraging individual-level data and quasi-experimental methods to solidify whether gentrification is indeed a carcinogen.

Keywords:

• Gentrification
• racial residential segregation
• health disparities
• carcinogens
• transportation

Acknowledgments

The author thanks the editors and anonymous reviewers for helpful comments. He also gratefully acknowledges the Mansueto Institute for Urban Innovation at the University of Chicago, USC Sol Price Center for Social Innovation, Joint Center for Housing Studies, and the National Science Foundation for support.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 The EPA characterizes these measures as capturing neighborhoods’ overall cancer risk rather than the risk of specific forms of cancer, and I follow suit, using the phrase “levels of cancer risk factors” throughout the manuscript. However, given that air toxins and vehicle emissions are the key input employed to assess spatial variation in cancer risk factor levels, I believe that the measure primarily captures difference in levels of lung cancer risk factors, specifically, and that readers should interpret the measure as such.

2 More minor caveats include the fact that indoor toxins, which could affect cancer risk via breathing/ingestion, are not factored into these estimates. Moreover, the assumptions and methods employed have changed slightly from one assessment to the next, given gradually improving emission inventories, an expanding set of air toxins modeled, and an evolving understanding of how air toxins affect human health, so year-to-year comparisons in tract-level cancer risk estimates should be interpreted with caution.

3 It is important to note that the 2002 EPA-NATA tract cancer risk estimates rely on slightly different modeling methods than the 2011 estimates, which may complicate the traditional LDV approach. However, for my analytical purposes, I assume that the two sets of estimates sufficiently overlap, and robustness checks that employ 2005 EPA-NATA tract cancer risk estimates—which more closely resemble the methodology of the 2011 estimates—as the LDV generate almost identical results (available upon request).

4 The vast majority of CDPs capture the boundaries of incorporated jurisdictions (e.g., Malibu, Santa Monica, Los Angeles). Some tracts, however, are located within CDPs, which do not correspond to jurisdictional boundaries and instead capture geographically contiguous tracts within nonincorporated areas. Despite having no legal status (services are often provided by the county and/or state rather than a separate municipal government), the census assigns each area a locally recognized name (e.g., Marina Del Rey). Within Los Angeles, CDPs have sharply divergent political, social, economic, and environmental conditions, conditions that may induce correlation in error terms among tracts within the same place (for example, due to unobserved differences in meteorological and topographical features that affect emissions). Clustering errors by census places helps mitigate this concern.

5 Assuming the same sample is used across all three models, the procedure used to recover the indirect effect estimate (i.e., multiplying (a) and (b)) is mathematically equivalent to subtracting the magnitude of the focal coefficient (the direct effect) in the model where the mediator is included from the magnitude of the focal coefficient when it is excluded (the total effect).

6 This index encompasses four evidence-based lung cancer risk predictors at the neighborhood level: % of tract residents who are Black, % of tract residents 25 or over with a bachelor’s degree or higher, tract median family income, and % of CDP (or city council district if within Los Angeles city) residents who reported smoking regularly. The first three variables are drawn from census 2000 data, and the latter variable is drawn from the Los Angeles Community Health Survey of 2007. All four variables are standardized relative to the full tract analytic sample (N = 2176) to have a mean of 0 and SD of 1. I reverse-code the standardized variables capturing tract % bachelor’s degree and median family income because these factors exert negative rather than positive effects on cancer risk. I then estimate a simple average based on the four standardized variables with valid data to create the index.

Additional information

Notes on contributors

Jared N. Schachner

Jared N. Schachner is a Postdoctoral Research Associate at USC’s Sol Price School of Public Policy, where he is affiliated with the Sol Price Center for Social Innovation. An urban sociologist by training, his research examines whether and how neighborhoods and schools mediate the intergenerational transmission of skills, health, and status.