Losing Your Home Is Bad for Your Health: Short- and Medium-Term Health Effects of Eviction on Young Adults

ABSTRACT

U.S. cities are increasingly adopting antieviction policies predicated on the belief that evictions have negative consequences for families and communities. Yet the nature and duration of many of these consequences are relatively unknown. We add to the literature on the consequences of evictions by assessing the enduring effects of eviction on the self-reported health of young adults. Using the National Longitudinal Study of Adolescent to Adult Health (Add Health), we find evictions have both short-term (12 months) and medium-term (7–8 years) negative impacts on multiple measures of health. Individuals who experience an eviction are more likely to report being in poor general health or experiencing mental health concerns, even many years after an eviction. As state and local governments develop policies to reduce evictions, it is worth noting that any resulting decrease in evictions may have a positive impact on population health, making health professionals effective potential policymaking partners.

KEYWORDS:

• eviction
• health
• housing policy
• rental housing

Acknowledgments

The authors would like to thank Emily Forsee for excellent research assistance and the Levin Korean Student Fund for funding assistance to obtain the Add Health data.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1. Throughout this article, we use families synonymously with households and individuals.

2. This duration is 7–8 years because wave III was in 2001–2002 and wave IV (when we measure health) was in 2008, making a range of 6–7 years. The question asked about evictions in the previous 12 months, increasing that range to 7–8 years.

3. The weighted sample size is the same for all analyses except the multinomial logistic regression of mental health, where it is 16,923,611. The sample sizes for the multinomial logistic regression vary because one unweighted observation did not respond to the question about depression and had to be dropped from the analysis.

4. The effects of the eviction variables in both waves and most of the control variables are nearly identical in our preferred model specifications and in the robustness check using multiple imputation.

5. Despite concerns that self-reported measures of health are too subjective, researchers often use self-reports as predictors of future health care and mortality rates (Idler & Benyamini, 1997).

6. We are not implying it is normatively good to have depression or anxiety/panic disorder. Rather, we use the word good here to be consistent with our other variables, and it should only be taken as a middle category between excellent and poor health.

7. Add Health does not distinguish between Hispanic White and non-Hispanic White.

8. We included people who said they did not know if they had insurance as not having insurance. In another analysis not shown here, we dropped all 128 unweighted observations with an ambiguous answer to the insurance question and got substantively similar results.

9. Personal income is not a perfect measure of unemployment because a person may not be employed the entire year and/or would receive unemployment benefits, which may be included in this variable. However, individuals who are unemployed for at least part of the previous year are likely to generally have lower incomes than individuals with jobs the entire year.

10. The observed relationship between housing eviction and health without weights is largely similar to that of the analysis with weights, with two exceptions. Eviction in wave III is only statistically significant at p < .10 for the unweighted binary regression with mental health as the dependent variable. Most variables related to personal income in wave III and wave IV and educational attainment were not significant in the unweighted analysis. Add Health oversampled Black adolescents with parents having education levels of more than a bachelor’s degree, which might explain the observed differences between children’s income and education and health when using the unweighted data (Chen & Chantala, 2014; Harris et al., 2009).

11. The weighted sample size was the same for all analyses except mental health for multinomial logistic regression, which was 16,923,611.

Additional information

Funding

This work was supported by the Levin Korean Student Fund.

Notes on contributors

Megan E. Hatch

Megan E. Hatch is an associate professor of urban policy and city management. She studies the causes and consequences of policies that disproportionately affect vulnerable populations. Her research focus on rental housing includes evictions, landlord–tenant policies, source-of-income discrimination, and criminal activity nuisance ordinances.

Jinhee Yun

Jinhee Yun is a PhD candidate in urban studies and public affairs. Her research focuses on poverty and inequality. She studies the impacts of unequal access to opportunities, focusing on neighborhoods and community development.