While Some Things Change, Do Others Stay the Same? The Heterogeneity of Neighborhood Health Returns to Gentrification

Abstract

Gentrification is associated with decreases in neighborhood poverty and crime, increases in amenities and services, among other benefits—all identified as structural determinants of health. However, gentrification is also associated with population-level replacement of the existing community, or threats thereof. Combining census data from the ten largest MSAs in the U.S. with tract-level estimates from the CDC-PLACES Project from 2013–14 to 2017–18, we explore how the changing socioeconomic conditions in gentrifying neighborhoods correlate with changes in neighborhood health. We find significant differences between gentrifying and non-gentrifying neighborhoods in their associations with neighborhood health. The sociodemographic changes occurring in gentrifying neighborhoods generally correspond with simultaneous decreases in aggregate health risk behaviors and negative health outcomes. However, these changes are heterogeneous and complex. Whether and how neighborhood health changes alongside other components of neighborhood change depends on whether gentrification occurs in majority Black, Hispanic, or White neighborhoods. Our findings provide preliminary evidence that the changes accompanying gentrification extend to neighborhood health, but the direction of influence varies by neighborhood composition, type of sociodemographic change, specific health outcome, and spatial spillover. We discuss theoretical implications for future work addressing the mechanisms driving changes in neighborhood health, and potential approaches that differentiate policy responses.

Keywords:

• Gentrification
• neighborhood change
• neighborhood health
• racial stratification
• urban inequality

Acknowledgments

The authors thank the editors and anonymous reviewers for their helpful comments throughout. We are grateful to the Population Studies and Training Center at Brown University, which receives funding from the NIH (P2C HD041020), for general support.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 See https://www.cdc.gov/places/about/index.html for information about the origin and evolution of the project, its methodology, and the measures it provides at multiple geographic levels.

2 Empirical definitions of gentrification vary, and there are limits to the aspects of gentrification that quantitative data can capture (e.g., changes to neighborhood norms and culture; etc.). As a sensitivity check, we conducted analyses using an alternative composite index of gentrification that captured neighborhood socioeconomic ascent via factor analysis, following past work (Candipan and Bader, 2022; Owens, 2012), and our main patterns held.

3 In alternative analyses, we performed models using a three-category neighborhood type measure that included tracts that were ineligible to gentrify.

4 We calculated the total 2010 population of all metropolitan tracts within census-designated MSAs (2003 Office of Management and Budget (OMB) definitions). The 10 MSAs are the following: Atlanta–Sandy Springs–Marietta, GA; Chicago–Naperville–Joliet, IL; Dallas–Plano–Irving, TX; Houston–Baytown–Sugar Land, TX; Los Angeles–Long Beach–Glendale, CA; New York–Wayne–White Plains, NY–NJ; Philadelphia, PA; Phoenix–Mesa–Scottsdale, AZ; Riverside–San Bernardino–Ontario, CA; Washington–Arlington–Alexandria, DC–VA–MD–WV.

5 The sample data and a replication code will be hosted on the author’s website and available to the research community.

6 We do perform a spatial ordinary least squares regression, but global Moran’s I tests indicate that conditioning for neighborhood context and MSA fixed effects does not remove the significant spatial autocorrelation in our models.

7 An alternative approach to the fixed effects specification would be to perform separate spatial regression models for each metropolitan area. We do also perform models without MSA fixed effects. These results are available upon request.

8 We tested a definition of majority White that relied on the raw majority (>50%) and patterns held. In addition to providing more reliable estimates, our preferred definition of majority White (>70%) more closely aligns theoretically with past work that does not use a simple majority for White but instead relies on thresholds based on the national racial composition and distribution of neighborhood racial composition (Galster et al., 2003; Owens & Candipan, 2019). We also performed robustness tests using a 2010 definition for neighborhood majority race, and broader patterns of effect heterogeneity held. Neighborhoods carry racial legacies that are durable and shaped over time; hence, we used an earlier year to capture neighborhood majority racial composition using an earlier time point.

9 Models in Appendix Table A1 are presented for illustrative purposes and differ from Table 2 in that they do not contain MSA fixed effects.

Additional information

Notes on contributors

Jennifer Candipan

Jennifer Candipan is an assistant professor in the Department of Sociology and faculty affiliate at the Population Studies and Training Center and the Spatial Structures in the Social Sciences (S4) program at Brown University.

Alicia R. Riley

Alicia Riley is an assistant professor in the Department of Sociology and core faculty in Global and Community Health at the University of California, Santa Cruz.

Janeria A. Easley

Janeria Easley is an assistant professor in the Department of African American Studies at Emory University.