Do housing and neighborhood characteristics impact an individual’s risk of homelessness? Evidence from New York City

Abstract

Most existing homelessness research either connects aggregate levels of homelessness to housing market and economic characteristics, or analyzes the personal traits of chronically homeless individuals and those receiving formal institutional support. Little is known about the characteristics of individuals in the general population who become homeless, especially their housing and neighborhood contexts. This article assesses the relationship between an individual’s odds of experiencing homelessness and their housing, personal, and neighborhood characteristics using data from The New York City Longitudinal Survey of Well-Being, a representative panel of New York City adults. These data are leveraged to specify a series of multilevel logistic panel regression models. Findings suggest an individual’s housing conditions, particularly whether they are doubled-up or in a rent-controlled unit, and traditional risk factors such as mental health issues and drug use, help predict future homelessness. Results suggest that well-known individual characteristics common among unhoused individuals are accompanied by housing and economic factors that drive a path to experiencing homelessness.

Keywords:

• Homelessness
• housing
• neighborhoods
• poverty

Acknowledgements

I thank Matthew Maury and Sophie Collyer for their assistance with data coding and interpretation and the Columbia Population Research Center for providing me access to the New York City Longitudinal Survey of Well-Being. I also thank Irwin Garfinkel for his feedback on an earlier draft of this article and three anonymous reviewers for their insightful comments. A previous version of this paper was presented at the 2019 annual conference of the Association of Collegiate Schools of Planning.

Disclosure statement

No potential conflict of interest was reported by the author.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 As noted by Frederick et al.,(2014) a widely accepted definition of housing stability does not exist among housing researchers. The authors suggest the possibility that the term should be understood as multidimensional, and should be assessed based on traditional metrics of housing access and related factors such as educational status, financial status, the use of harmful substances.

2 See Linton et al., 2017 for an exception.

3 The New York City Longitudinal Survey of Wellbeing, also known as Poverty Tracker, was developed by the Columbia Population Research Center with funding from Robin Hood and in collaboration with the Center on Poverty and Social Policy. Access to the data, as well as codebooks and additional information, can be found at https://cprc.columbia.edu/content/new-york-city-longitudinal-survey-wellbeing

4 Importantly, studies relying on samples drawn from institutional settings yield insights about the non-private household population, which is a subgroup traditionally challenging for general population surveys like the one used in the present study to capture.

5 Poverty Tracker data are geocoded at the zip code level based on the mailing address of each respondent. This level of granularity is not ideal for analysis of neighborhood context, especially in a densely populated urban area such as New York City. However, it is required to maintain respondent anonymity.

6 This rate of homelessness is challenging to situate among benchmarks, as it spans across four years, includes a portion of households who have multiple experiences of homelessness, and only tabulates respondents (as opposed to all those living in a household) as homeless. U.S. Department of Housing and Urban Development-led point-in-time counts during the study period typically report around 1% of New York’s population as homeless (see https://files.hudexchange.info/reports/published/CoC_PopSub_CoC_NY-600-2015_NY_2015.pdf).

7 Sensitivity analyses were conducted using lags of two time periods for variables that might be suspected of having more delayed impacts on future homelessness such as material hardships and levels of neighborhood gentrification. In each instance, the model strength was not improved over the final model selected, and the direction and significance of covariates did not change.

8 Other race omitted due to chart size limitations

Additional information

Notes on contributors

Tyler Haupert

Tyler Haupert is an Assistant Professor Faculty Fellow of Urban Studies at NYU Shanghai. His research focuses on the social, economic, technological, and regulatory mechanisms contributing to racial segregation and exclusion in advanced economies, with particular interests in mortgage lending, housing policy, neighborhood change, and homelessness.