Sage Journals: Discover world-class research

Abstract

Although existing scholarship notes that homelessness thrives in concentrated poverty, models estimating the association between the intensity of residential poverty segregation and local homelessness rates across communities remain absent from the literature. To fill this gap, the author considers this relationship for 272 homelessness Continuums of Care covering urban and suburban spaces spanning 43 states and the District of Columbia. Models suggest that poverty segregation is positively associated with the expected homelessness rate of a Continuum of Care, a relationship that remains significant when controlling for a range of established drivers of the condition. The author discusses this finding within a framework qualifying residential poverty segregation as both a cause and a consequence of the local prevalence of economic disadvantage that predicts homelessness via its relationship with disadvantage and unique spatial effects.

Keywords

homelessness segregation poverty inequality housing

In the American city and suburb, homelessness is governed partly by a spatial logic, a twofold system of geographic exclusion and seclusion that confines homelessness to neighborhoods of concentrated economic disadvantage (Herring 2014; Wacquant 2010). In communities across the country, persons experiencing homelessness are banished from areas typically reserved for nonpoor actors by means such as selectively enforced ordinances that criminalize activities associated with homelessness (e.g., panhandling) and the strategic placement of hostile architectural deterrents like antisleep benches and visible security cameras (Amster 2003; Beckett and Herbert 2009; Rosenberger 2020). Conversely, neighborhoods of concentrated poverty—spaces from which homeless episodes disproportionately originate (Culhane, Lee, and Wachter 1996; Tsai, Mares, and Rosenheck 2011)—offer the unhoused refuge from some forms of policing and surveillance while also constituting the locations generally reserved for the homeless service infrastructure: emergency shelters, soup kitchens, and the like (Brinegar 2003; Dear and Gleeson 1991; Harris 2017).¹ This infrastructure, however well intentioned, typically satisfies only short-term needs without addressing the long-term housing and resource stability of those seeking support (Desmond 2016; Poethig 2014; Vitale 2017). This arrangement suggests that poverty segregation should be positively associated with local homelessness rates across communities, as concentrated economic disadvantage provides the ideal environment for the production, management, and concealment of homeless episodes, generally without encouraging the development or maintenance of clearly defined organizational solutions promoting exit from the condition. The purpose of the present piece is to test whether this relationship is observable across homelessness Continuums of Care (CoCs), the local or regional administrative bodies responsible for the administration of homeless services in the United States.

I model associations between the intensity of residential poverty segregation and homelessness rates for 272 CoCs spanning predominately urban or suburban areas across 43 states and the District of Columbia.² Residential poverty segregation is defined as the extent to which households earning less than $15,000 annually are located in neighborhoods distinct from those inhabited by higher income households and is estimated via the bias-adjusted binary variance ratio index, R, using data from the 2014 to 2018 American Community Survey (ACS) Five-Year Estimates (Reardon et al. 2018).³ Following previous scholarship modeling rates of homelessness at the CoC level (e.g., Byrne et al. 2013; Fargo et al. 2013), I use a linear mixed modeling strategy that nests CoCs within states to estimate homelessness rates measured during the January 2019 Point-in-Time (PIT) Count, the biennial census of persons experiencing homelessness in the United States.

Models suggest that poverty segregation is positively associated with a CoC’s overall and sheltered homelessness rates, relationships that remain significant when controlling for a range of established drivers of the condition. I discuss findings within a framework qualifying residential poverty segregation as both a cause and consequence of the local prevalence of economic disadvantage. Poverty segregation is theorized to influence local homelessness rates via its interconnection with disadvantage and unique spatial effects that can negatively affect the self-efficacy of those experiencing homelessness and normalize the condition as an accepted social category for the housed.

Framework

The U.S. Department of Housing and Urban Development (HUD) estimated that 567,715 people were homeless on a single night in January 2019.⁴ These cross-sectional, midwinter counts have steadily declined from about 650,000 people in 2007, constituting a nontrivial shift in the desired direction. However, gains have been slower than anticipated: CoC-level 10-year plans to end homelessness, originally implemented in communities across the country with the support of the George W. Bush administration, were expected to be completed by the late 2010s (i.e., toward the conclusion of the longest period of economic expansion in U.S. history) (Harris 2017; NLIHC 2014, 2015; Roberts 2012). In this context, the observed 13 percent decrease in homelessness between 2007 and 2019 is something less than a total success.⁵

Despite public opinion polling reflecting widespread support for increased public aid to combat homelessness and develop a more robust affordable housing stock—and more than two decades of research demonstrating the cost-effectiveness of housing-first initiatives that subsidize housing expenses while providing optional support services—homelessness remains and is concentrated in densely populated areas of the country: roughly 80 percent of the American homeless population resides in CoCs covering primarily urban or suburban spaces (Gillespie et al. 2020; NLIHC 2019a; HUD 2020; Vitale 2017). In practice, nonpoor, housed residents tend to prioritize keeping homelessness out of their neighborhoods over implementing meaningful solutions, ostensibly because of practical fears such as increased crime rates and the potentially negative impacts of visible homelessness on property values, but also because of the classist assertion that those experiencing homelessness do not belong in their spaces (Dear and Gleeson 1991; Gilderbloom, Squires, and Wuerstle 2013; Lee and Farrell 2004; Oakley 2002). This creates a strategic opportunity for local political actors who, concerned that pursuing actual solutions to homelessness will violate the American doctrine of rugged individualism and ultimately be politically unpopular, commonly embrace strategies to obscure homelessness by concentrating it in high-poverty neighborhoods where nonpoor residents seldom venture (Gilderbloom et al. 2013; Harris 2017; Marcuse 1988).

Prior research (e.g., Dear and Wolch 1987; Lee and Price-Spratlen 2004) has established that the spatial distribution of homelessness within communities parallels the geographic distribution of poverty but has yet to explore whether the concentration of poverty is associated with the size of a CoC’s homeless population. To explain why we should expect a positive association between residential poverty segregation and homelessness rates across CoCs, I first discuss the forces that concentrate housed poverty and homelessness in the same neighborhoods. I then explain how poverty and its segregation affect the generation and management of homeless episodes. I frame this second part of the discussion in terms of the dual patterns of structural determination linking the arrangement of bodies and objects in physical space to structures of the economic order, that is, (1) economic effects on the spatial arrangement of bodies and objects and (2) spatial impacts on the economic order (Bourdieu [1991] 2018; Fogle 2011). In the case of the former, I discuss how economic stratification generates and spatializes both poverty and homelessness. For the latter, I consider how poverty segregation, itself driven by the local prevalence of economic disadvantage, could affect homelessness rates via spatial effects independent from the association between economic disadvantage and homelessness. I use this framework as a basis for hypothesizing a positive relationship between poverty segregation and homelessness that will be partly, but not completely, attributable to community-level measures of economic disadvantage.

Spatial Constraints on Poverty and Homelessness

I define residential poverty segregation as the extent to which households earning below some low-income threshold (i.e., those in poverty) are clustered in neighborhoods distinct from those occupied by more advantaged residents. I illustrate this concept in Figure 1. In a simple case with two groups, those in poverty and those not in poverty (Figure 1A), a low-segregation community observes a near random assortment of households such that knowing the poverty status of one unit is not meaningfully predictive of the status of another in its neighborhood (Figure 1B). In a highly segregated community (Figure 1C), the economic order is clearly inscribed on the larger community with distinct advantaged and disadvantaged neighborhoods, such that the poverty status of most units in a given neighborhood is the same. The concentration of low incomes is associated with the clustering of other disadvantages such as lower access to quality public resources, higher exposure to crime and violence, and, perhaps the ultimate expression of social divide, gaps in life expectancy relative to nearby nonpoor neighborhoods exceeding corresponding disparities between the United States and countries in the developing world (Ansell 2017; Phelan, Link, and Tehranifar 2010; Reardon and Bischoff 2011; Sharkey and Elwert 2011; Wolch and Dear 1993).⁶ Nonpoor actors seeking housing tend to avoid these neighborhoods in search of spaces that have higher quality public resources, are perceived to be safer, and are associated with higher social status (i.e., “good neighborhoods”).

Figure 1.

Conceptual diagram of communities characterized by low poverty segregation and high poverty segregation.

Over the past half century, income segregation in American cities and suburbs has been increasing along with economic inequality (Reardon and Bischoff 2011; Reardon et al. 2018). The social forces influencing local levels of income segregation are numerous, nonlinear, and complex, but the spatial philosophy of Pierre Bourdieu provides a useful shorthand for how to interpret their expected aggregate effect: to the extent that the physical layouts of cities and suburbs represent a spatial objectification of a stratified social order, sites in physical space represent physical materializations of capital within that stratified order.⁷ In competitive housing markets, high levels of income inequality imply an environment in which the most advantaged actors can easily outbid lower-income shelter seekers to claim housing in their most desired neighborhoods. The least desired neighborhoods—marginal spaces with scant economic value—are ceded to the poorest local (housed) residents who cannot afford units in other neighborhoods and who are often severely rent burdened and on the cusp of entering homelessness themselves (Desmond 2016, 2018; Snow and Anderson 1993; Wolch and Dear 1993).

Homelessness is concentrated in high-poverty neighborhoods through a series of exclusionary and seclusionary forces (Herring 2014; Wacquant 2010). Communities across the country maintain sets of ordinances that criminalize the correlates of homelessness (e.g., bans on camping, begging, loitering, sitting or lying in designated spaces, sleeping in cars, or sharing free food) (Amster 2003; NLCHP 2019; Vitale 2017). Such ordinances are most strictly enforced in prime spaces (i.e., those used by nonpoor actors) and are often complemented by hostile design features such as antisleep benches and visible security cameras that similarly encourage those experiencing homelessness to exist elsewhere (Amster 2003; Beckett and Herbert 2009; Rosenberger 2020; Snow and Anderson 1993). Exclusionary strategies force persons experiencing homelessness from all but the poorest neighborhoods, which are spaces often corresponding to (perceived or real) threats to their physical well-being. As a result, the majority of unhoused persons are unaware of outdoor locations where it is both safe and legal to sleep (Beckett and Herbert 2009; Vitale 2017; Western Regional Advocacy Project 2014). The lack of suitable outdoor sleeping quarters invites the criminalization of visible homelessness given that several CoCs do not maintain a stock of shelter beds sufficient to reliably house their entire homeless population, which leaves some people with few other options but to embrace dangerous sleeping arrangements or have their existence outlawed (NAEH 2017; NLCHP 2019).

The infrastructure supporting unhoused populations is similarly relegated to economically depressed neighborhoods via exclusionary processes. When homelessness reemerged on the public agenda in the 1980s, major U.S. city governments and private organizations responded by increasing local inventories of shelter beds (Brinegar 2003), but before the beginning of the 1990s, there emerged “a dangerous trend that place[d] the aesthetic concerns of select groups of business and property owners above the life-or-death needs of the homeless” and designated the placement of local homeless service infrastructures as among the most contentious topics on local public agendas (Safety Network 1988:2, cited in Wolch, Dear, and Akita 1988). Economically advantaged actors (e.g., business owners, longtime homeowners) have generally focused on keeping homelessness and homeless services out of the spaces they claim, which is consequential given that, in most cases, community and professional coalitions have considerably more leverage over local politics than persons experiencing homelessness and those representing the service industry (Gilderbloom et al. 2013; Hall and Yoder 2018; Herbert 2009). As a consequence, homeless services are typically excluded from nonpoor spaces.

The exclusionary processes that concentrate the service infrastructure in marginal spaces also facilitate a set of seclusionary features that qualify these same spaces as the most accommodating residential options for those experiencing homelessness. Given the legal and infrastructural arrangements described above, residing in marginal spaces minimizes police harassment and other forms of unwanted attention while increasing proximity to social services relative to those experiencing homelessness in low-poverty spaces. For these reasons, the latter case is relatively uncommon: residential and public spaces used mostly or exclusively by nonpoor actors tend to be sanitized of homelessness (Alexander-Eitzman, Pollio, and North 2013; Amster 2003; Daiski et al. 2012; Lee and Price-Spratlen 2004; Wright and Vermund 1999).

Poverty Segregation and the Generation and Management of Homeless Episodes

The relationship between the arrangement of objects and bodies in physical space and the economic order is characterized by two simultaneously occurring modes of structural determination, (1) economic effects on the spatial order and (2) spatial effects on the economic order, both of which are mediated by the internalized norms, dispositions, and resulting actions of all relevant actors (i.e., by the collective achievements of all stakeholders in a defined space, each possessing a sufficiently predictable habitus and likelihood of realizing her will).⁸ I diagram this relationship with respect to homelessness in Figure 2 and discuss its implications over the next two subsections. Using this framework, I arrive at the hypothesis that poverty segregation will be positively associated with CoC-level homelessness rates and that this association will be partly attributable to the local prevalence of economic disadvantage across communities, which produces both segregated poverty and segregated homelessness.

Figure 2.

General structure of the dialectical relationship between the economic and physical orders as it applies to homelessness.

Economic Impacts on the Generation, Legitimation, and Spatialization of Homelessness

At its root, homelessness emerges from local economic systems that permit, or are otherwise incapable of preventing, poverty deep enough to limit some persons from accessing housing (Liebow 1993; Wolch and Dear 1993). We should therefore expect a strong, positive relationship between the increasing prevalence of economic disadvantage and homelessness rates across CoCs.⁹ Widespread economic disadvantage, assuming it is long-standing, also indirectly legitimates the existence of homelessness by facilitating its enduring presence, which normalizes homelessness as an accepted social category via social psychological processes (Figure 2, path A → B → A). Research in neuroscience (e.g., Fiske 2009; Harris and Fiske 2006) suggests that members of extreme out-groups, such as those experiencing homelessness, are preconceived by others as subhuman and that reactions to being presented images of homelessness are consistent with neurological patterns indicating disgust. Such judgements can be prereflexive, implying that internalized socialization processes legitimating the inferiority of those experiencing homelessness affect both what we think with (i.e., our cognitive structures) and how we think of homelessness. They are also a function of an observer’s class position: actors situated higher in the economic order are more likely to blame those experiencing homelessness (and housed poverty) for their own conditions (Babjaková, Džuka, and Gresty 2019; Moore 2015). Although positive affect toward those experiencing homelessness tends to be higher for actors who interact with or see people lacking housing more frequently (Lee, Farrell, and Link 2004), exposure to homelessness is itself partly conditioned on one’s location in the economic order (i.e., economic class is positively associated with segregation from homelessness). Path A → B → A in Figure 2 should therefore operate such that high rates of economic disadvantage generate both homelessness and the social psychological conditions necessary to justify its existence. This relationship is key, as the widespread generation of homelessness without legitimation could also lead to its swift eradication via community action or government intervention, especially given the strong public consensus that more needs to be done to address the issue (NLIHC 2019a). However, if homelessness is legitimated as a social category, and can therefore be ignored or deprioritized on local agendas, the community-level association between economic disadvantage and homelessness rates should be clearly identifiable.

In competitive markets, economic stratification also indirectly shapes the spatial order (Figure 2, path A → B → C), as housing costs are among the largest investments most owners and renters will make, and prices are set partly on the basis of the status assigned to particular neighborhoods (Bourdieu [1991] 2018; Desmond 2018).¹⁰ When competition is unequal, individuals and families of sufficient means tend to self-segregate in neighborhoods characterized by social statuses they (consciously or unconsciously) deem appropriate (Reardon and Bischoff 2011; Trounstine 2018). The least desirable spaces are ceded to the poor and homeless. In this respect, poverty segregation is a social segregation: it is a projection of the system of economic relations onto geographic space. These two orders, the economic and spatial, are mutually reflective, but they are not one and the same. The spatialization of poverty and homelessness has the potential to affect homelessness rates in ways distinct from the effects stemming from the economic order described above.

Spatial Effects on the Visibility and Experience of Poverty and Homelessness

The ability to reside in a desired neighborhood constitutes one of the central power struggles inherent to housing markets. In general, those looking for housing aim to optimize their proximity to desired goods and services, the symbolic value of the properties they inhabit, the amount of land over which they have control, the mix of people they are likely to encounter, and the visibility of that which is aesthetically pleasing (Bourdieu [1991] 2018; Fogle 2011). They also aim to keep away that which is undesirable (e.g., crime and visible poverty). In scenarios in which a sizable group of people in poverty are forced to bid for shelter against others with higher means, the expected high degree of residential poverty segregation legitimates the stratified economic order for nonpoor residents through their observance of relatively consistent profiles of the people who reside in certain spaces (e.g., given the relationship between racial identity and household wealth, see, e.g., Urban Institute 2017) and the semipermanence of the built environment (Figure 2, path C → B → A).¹¹ In other words, the spatialization of economic disadvantage creates a façade for advantaged residents who misunderstand the economic despair of others concentrated in these spaces as a naturalistic and inevitable outcome.

Because the spatial profits accruing to housed residents are partly conditioned on the absence of visible expressions of poverty, nonmarginal spaces tend to be arranged such that the poor and homeless are continuously threatened by probabilistic surveillance. Security cameras and spatial exclusion ordinances, for example, do not guarantee arrest or other forms of trouble. They only suggest that trouble might happen, especially if individuals experiencing homelessness draw attention to themselves. Such arrangements create a risk-reward scenario for those experiencing homelessness who, aware of the high-resolution, polynucleated form that poverty segregation typically takes in American cities and suburbs,¹² strategically design subsistence strategies that involve selectively traversing prime spaces or navigating around them to reach service nodes or other resources (e.g., day labor), while minimizing risks (Snow and Anderson 1993; Snow and Mulcahy 2001). In this regard, the clustering of homelessness in spaces of concentrated poverty has a certain breathability: persons experiencing homelessness can temporarily enter nonpoor spaces without encountering unreasonable risks, as long as their existence is transitory and purposeful. Thus, it is not immediately clear that poverty segregation drives homelessness by limiting resource access.

It is plausible that the clustering of poverty and homelessness in distinct neighborhoods facilitates social psychological normalization effects encouraging those without housing to accept their social positions as inevitable outcomes, driving a positive association between poverty segregation and homelessness across CoCs. Prior research suggests that virtually everyone prefers housing over homelessness and that people experiencing extreme poverty typically use whatever means available to escape or avoid the latter (Desmond 2016; Edin and Shaefer 2015; Liebow 1993; Snow and Anderson 1993; Snow and Mulcahy 2001). Over time, however, such struggles can become unbearable even for the most resilient poor people (Ehrenreich 2010). The causal nexus is messy, but if widespread economic disadvantage facilitates the spatialization of housed poverty and homelessness in the same neighborhoods, it stands to reason that clustering unhoused persons close to one another and other symbols of despair (e.g., cracked sidewalks) normalizes homelessness for those experiencing homelessness by creating an immediately observable, accepted social category corresponding to the bottom of the social order. This does not imply that persons experiencing homelessness in highly segregated communities give up on upward mobility, but it does suggest that constant reminders of one’s own marginality can negatively threaten a person’s ability to overcome it. Establishing the precise spatial mechanism(s) linking poverty segregation to homelessness (e.g., via statistical analysis) is beyond the scope of the present study, but it is important to note that if poverty segregation is positively associated with CoC-level homelessness rates, the relationship is likely partly attributable to the way symbols from the built environment become inscribed in the minds of those experiencing the condition.

The Present Study

I ask two novel questions about the relationship between poverty segregation and homelessness rates across urban and suburban CoCs:

Is the degree of poverty segregation observed in a CoC associated with its expected homelessness rate?

If an association between poverty segregation and homelessness rates is identified, can it be explained by the local prevalence of economic disadvantage, as the latter is theorized to drive both poverty segregation and homelessness?

I expect that poverty segregation will be positively associated with local homelessness rates in unadjusted models and that this relationship will be partly attributable to the CoC-level prevalence of economic disadvantage. Furthermore, consistent with prior research (e.g., Byrne et al. 2013; Lee, Price-Spratlen, and Kanan 2003), I expect that fully adjusted models will have nontrivial explanatory power, demonstrating that the local prevalence of homelessness is tightly related to community-level processes. Assuming that unique spatial effects stemming from poverty segregation affect local homelessness rates, I expect the relationship between poverty segregation and homelessness to remain significant in regression models controlling for a full set of CoC-level drivers of homelessness previously established in the literature that do not measure a degree of spatialization.¹³

Data and Methods

Data

The PIT Count is the biennial census of persons experiencing homelessness in the United States. During odd-numbered years, communities are required to count and report the number of persons experiencing sheltered and unsheltered homelessness on a single night during the last 10 days of January. Local enumerations are reported at the CoC level. CoCs range in size from the subcounty level to spanning multiple counties and do not always correspond to typical census geographies.¹⁴ To accurately merge PIT Count data with community-level demographic information, I use a publicly available crosswalk file that identifies the specific census tracts within each CoC.¹⁵ I gather tract-level demographic measures from the 2014 to 2018 ACS Five-Year Estimates, and I aggregate these data to the CoC level.¹⁶ Local average temperatures for January 2019 are estimated by overlaying HUD’s publicly available CoC shapefiles onto a map of weather stations reporting data to the PRISM Climate Group (https://prism.oregonstate.edu) and taking the simple average of all stations located within the boundaries of each CoC. Given prior research (e.g., Byrne et al. 2013) indicating that the drivers of homelessness vary across rural and nonrural settings, and the focus of the present study on the latter, I restrict this analysis to CoCs primarily responsible for managing homeless services in urban or suburban spaces, as designated by HUD. I also omit a set of CoCs that exist outside of the contiguous United States, Alaska, and Hawaii (e.g., those in Puerto Rico). Using this strategy, I derive a sample of 272 CoCs spanning 43 states and the District of Columbia. A map of the CoCs used for this analysis can be found in Appendix A.

Dependent Variable

The outcome variable of interest is the number of persons experiencing homelessness per 10,000 residents living within the boundaries of a CoC. Because the forces that physically concentrate persons experiencing sheltered homelessness are distinct from those that concentrate people experiencing unsheltered homelessness,¹⁷ I disaggregate homelessness into its sheltered and unsheltered subtypes and model the determinants of homelessness for all three outcomes (i.e., unsheltered, sheltered, and overall homelessness rates). Because of the heavy right skews of the distributions of all three outcome variables, I use their natural logarithms in all regression models.

Poverty Segregation

My goal is to accurately measure the extent to which households within each CoC are unevenly distributed across neighborhoods by poverty status to assess if poverty segregation is associated with local homelessness rates. I use census tracts as proxies for neighborhoods, and I calculate poverty segregation across tracts for each CoC. For this analysis, households earning less than $15,000 in annual income are defined as being in poverty, and all other households are defined as not in poverty. This operationalization is justified by prior research (e.g., Culhane et al. 1996) suggesting that homelessness is disproportionately generated in neighborhoods where typical incomes fall considerably short of the official poverty line ($21,330 for a family of three in 2019). In the supplemental appendix, I reestimate all models presented here using 14 alternative definitions of poverty segregation. Supplemental models collectively demonstrate that homelessness is most closely associated with the segregation of the lowest income households from everyone else.

I calculate poverty segregation using the bias-adjusted binary variance ratio index, R, which explicitly accounts for the upward bias that arises when constructing segregation measures using sample data, an important issue given the lowering of ACS sampling rates over recent iterations (Reardon et al. 2018). Using this measurement strategy involves an important trade-off relative to using simpler and more common segregation measures such as the index of dissimilarity. R has several appealing features that other common segregation measures do not, but its interpretation is less straightforward.¹⁸ Following Bischoff and Reardon (2014), I interpret R as follows: segregation ranges from a theoretical minimum of zero (no segregation) to a maximum of one (complete segregation). When segregation is low (i.e., close to zero), the poverty status of any unit within a neighborhood does not meaningfully predict the status of another randomly chosen unit in the same neighborhood (Figure 1B). As R approaches one, the poverty status of any unit in a neighborhood becomes more strongly predictive of the status of all others in the neighborhood, which more frequently share the same status (Figure 1C). In the hypothetical case of complete segregation, poverty status can vary across neighborhoods belonging to the same CoC as long as within-neighborhood homogeneity is maintained. The magnitude of this measure does not have any intuitive meaning, but it is useful for relative comparisons across geographic units.

Control Variables

A summary of all variables used in this analysis can be found in Table 1 and corresponding descriptive statistics are presented in Table 2. Control variables were selected to mirror community-level predictors of homelessness previously established in the literature (see, e.g., Byrne et al. 2013; Lee et al. 2003). Because of high correlations between two sets of predictor variables—(1) CoC-level poverty, unemployment, and public assistance use rates and (2) CoC-level median rents and renter median incomes—I create two composite measures using principal-component analysis (PCA): (1) an index of economic disadvantage explaining roughly 84 percent of the total variation in the first set of variables and (2) an index of renter market accessibility capturing 95 percent of the variation in the second set (see Appendix B). Because of heavy right skews in the distributions of some predictor variables and the potential for outlier observations to disproportionately influence coefficient estimates, some variables are log transformed when entered into regression models. I note these transformations where appropriate.

Table 1.

Summary of Study Variables.

Variable	Source	Description
Dependent variables
Unsheltered homeless persons per 10,000 residents	HUD PIT Count/ACS	Unsheltered homeless persons per 10,000 residents
Sheltered homeless persons per 10,000 residents	HUD PIT Count/ACS	Sheltered homeless persons per 10,000 residents
Homeless persons per 10,000 residents	HUD PIT Count/ACS	Homeless persons (sheltered plus unsheltered) per 10,000 residents
Independent variables
Poverty segregation^a	ACS	Segregation by annual household income dichotomized at $15,000
Economic disadvantage^b	ACS	Index measure of economic characteristics: poverty rate, public assistance rate, and unemployment rate
Rental market accessibility^b	ACS	Index of CoC rental characteristics: median rent and renter median income
Renter occupancy rate	ACS	Percentage of units occupied by renters
Vacancy rate	ACS	Percentage vacant units
Crowded rate	ACS	Percentage of units with more than one occupant per room
Percentage moved in last year	ACS	Percentage of residents in current housing unit for less than one year
Percentage black	ACS	Percentage black in CoC population
Percentage Hispanic/Latinx	ACS	Percentage Hispanic/Latinx in CoC population
Median age	ACS	Median age of population
Single occupancy rate	ACS	Percentage of housing units with single occupant
Tracts	HUD CoC shapefiles	Number of census tracts in CoC
Average temperature	PRISM	Average temperature for January 2019

Note: ACS = U.S. Census Bureau American Community Survey 5-Year Estimates, 2014 to 2018; HUD CoC shapefiles = U.S. Department of Housing and Urban Development geographic information system data sets denoting physical boundaries of Continuums of Care; HUD PIT Count = U.S. Department of Housing and Urban Development Point-in-Time Count estimates of persons experiencing homelessness in January 2019; PRISM = Northwest Alliance for Computational Science and Engineering PRISM Climate Group at Oregon State University.

Segregation variable constructed via bias-adjusted variance ratio index, R.

Indicates variables constructed via principal-component analysis (see Appendix B).

Table 2.

Descriptive Statistics.

Variable	Mean	Median	SD	Minimum	Maximum
Dependent variables
Unsheltered homeless persons per 10,000 residents	5.91	2.00	10.18	.00	69.54
Sheltered homeless persons per 10,000 residents	12.47	9.28	13.22	1.48	110.02
Homeless persons per 10,000 residents	18.38	12.76	17.34	2.05	110.56
Independent variables
Poverty segregation	.05	.05	.03	.01	.21
Economic disadvantage	.00	.13	1.59	–4.84	5.40
Rental market accessibility	.00	.22	1.38	–4.48	2.73
Renter occupancy rate	37.00	36.16	10.16	14.78	68.53
Vacancy rate	10.32	8.85	5.60	3.41	45.96
Crowded rate	2.96	2.14	2.17	.74	11.65
Percentage moved in last year	15.11	15.05	3.67	6.58	29.62
Percentage black	14.07	10.09	13.15	.59	78.65
Percentage Hispanic/Latinx	15.07	10.02	14.03	1.00	84.15
Median age	38.74	38.56	4.03	25.60	58.08
Single occupancy rate	28.27	28.50	4.99	12.86	46.97
Tracts	167.41	105.50	220.27	19.00	2,131.00
Average temperature	36.80	33.75	12.34	12.71	72.52
Continuums of Care (n)	272

Modeling Strategy

I assume that CoCs nested within the same state are not statistically independent from one another and are likely to share more similar characteristics than they do with CoCs located in different states (e.g., because of potentially related, state-funded programs addressing issues such as mental health among low-income populations; see Byrne et al. 2013). This type of clustering violates the independence assumption of ordinary least squares regression and must be explicitly considered in statistical models. To address this issue, I follow Byrne et al. (2013) and build a series of linear mixed-effects models with CoCs as the first level of analysis and states as the second. This strategy allows me to account for statistical dependence among CoCs nested within the same state by specifying random slopes and intercepts. For all models, I regress the natural logarithm of the outcome on a series of predictor variables, all of which are continuous and mean centered.

Results

Unadjusted Models: Does Poverty Segregation Predict Homelessness Rates?

In Figure 3, I summarize the expected, unadjusted association between a CoC’s degree of residential poverty segregation and local homelessness rates, estimated via models including state-level random intercepts and slopes.¹⁹ A 1 standard deviation increase in poverty segregation is associated with a 30.9 percent increase in a CoC’s expected total homelessness rate and a 32.2 percent increase in its expected sheltered homelessness rate. The estimated association between poverty segregation and unsheltered homelessness rates is null. Marginal R² values, measuring the explanatory contribution attributable to fixed effects only, are .12 and .14 for models predicting total and sheltered homelessness rates, respectively, and virtually zero for the model predicting unsheltered homelessness. Collectively, these unadjusted models suggest an affirmative answer to the first research question with an important caveat: across urban and suburban communities, homelessness, particularly sheltered homelessness, tends to be more prevalent in CoCs where poverty is more highly concentrated.

Figure 3.

Estimated relationships between residential poverty segregation and Continuum of Care–level homelessness rates (95% confidence intervals displayed).

Does the CoC-Level Prevalence of Economic Disadvantage Explain the Association between Poverty Segregation and Homelessness?

In Figure 4, I retest the association between poverty segregation and homelessness while including an indexed term for a CoC’s level of economic disadvantage summarizing its poverty rate, unemployment rate, and rate of public assistance use (see Appendix B). Adding this control reduces the magnitude of the association between poverty segregation and sheltered homelessness by 35 percent and the association with overall homelessness by 44 percent relative to unadjusted models, but poverty segregation remains a meaningful predictor of sheltered and overall homelessness at the p < .01 level. A reduced effect size is expected given the strong conceptual relationship between the CoC-level prevalence of poverty and its segregation. However, the significant, adjusted relationship between poverty segregation and homelessness that remains when accounting for economic disadvantage broadly indicates that poverty segregation caries its own social inertia that can drive CoC-level homelessness rates independent from the effect that widespread economic disadvantage has on homelessness.

Figure 4.

Estimated relationships between residential poverty segregation, local levels of economic disadvantage, and Continuum of Care homelessness rates (95% confidence intervals displayed).

Fully Adjusted Models: What Are the Community-Level Correlates of Homelessness?

In Figure 5, I summarize results from fully adjusted regression models estimating CoC-level homelessness rates. All else equal, a 1 standard deviation increase in poverty segregation is associated with a nontrivial 12 percent increase in a CoC’s expected rates of sheltered and overall homelessness. I find marginal evidence suggesting the adjusted relationship between poverty segregation and a CoC’s unsheltered homelessness rate is negative, but combined with previous models indicating a null relationship between poverty segregation and unsheltered homelessness, it is not immediately clear that this relationship is practically meaningful. Model controls generally operate as expected. A CoC’s expected unsheltered, sheltered, and overall homelessness rates are positively associated with the local prevalence of economic disadvantage, indicating that homelessness tends to be more prevalent in places with higher rates of poverty, unemployment, and public assistance use. Rental market accessibility is negatively associated with all three outcomes, though only marginally so for the model predicting unsheltered homelessness, suggesting that homelessness rates are systematically lower in CoCs with lower median rents and a more robust demand for lower income units. A CoC’s single occupancy rate, which has been theorized to predict homelessness because of the limited income streams typically associated with single-earner households and the increased competition these residents face for low-cost units (Byrne et al. 2013; Desmond 2018), is positively associated with homelessness rates for all three outcomes, but this effect is statistically significant only for models predicting sheltered and overall homelessness rates. Other things equal, CoCs containing more census tracts (which have systematically larger populations) tend to observe lower rates of sheltered and overall homelessness but higher rates of unsheltered homelessness. Although local average temperatures do not meaningfully predict overall homelessness rates, they do predict particular expressions: all else equal, warmer CoCs tend to observe more unsheltered homelessness, and sheltered homelessness is more common in places experiencing colder winters. The negative association between the percentage of a CoC’s population identifying as black or African American and all three outcomes deserves increased attention, as it is counterintuitive given the increased risk for entering homelessness faced by black Americans (HUD 2020), but this exploration is beyond the scope of the present project. As in previous research (Byrne et al. 2013; Lee et al. 2003), CoC-level homelessness rates are highly predictable using community-level characteristics. This set of 13 predictor variables, along with state-level random effects, explains roughly half of the variation in unsheltered homelessness rates across CoCs and more than half of the variation in sheltered and overall homelessness rates (see Appendix E).

Figure 5.

Continuum of Care–level drivers of homelessness (95% confidence intervals displayed).

Discussion and Conclusion

This analysis has two basic implications for homelessness research. First, I confirm more than three decades of scholarship arguing that homelessness is driven mainly by issues of economic insecurity and housing affordability (see, e.g., Byrne et al. 2013; Desmond 2016; Liebow 1993; Rossi 1989; Wolch and Dear 1993). Second, I provide new statistical evidence establishing a positive association between poverty segregation and homelessness rates across urban and suburban CoCs. Homelessness thrives in concentrated poverty partly because when poverty prospers, its expressions, including homelessness, tend to be concentrated in geographic space. Models presented here further suggest that poverty segregation also carries its own social inertia stimulating homelessness independent from the local prevalence of economic disadvantage and other previously established CoC-level drivers of the condition.

The precise spatial mechanism(s) linking CoC-level poverty segregation to homelessness rates remain poorly understood. I propose three potential links in no particular order. First, poverty segregation might limit resource access for those experiencing homelessness if such resources are located in prime spaces or in distant locations separated by advantaged neighborhoods. I do not find strong support for this mechanism, as previous scholarship (e.g., Desmond 2016; Edin and Shaefer 2015; Snow and Anderson 1993; Snow and Mulcahy 2001) suggests that housed and unhoused persons in deep poverty typically engage in whatever struggles they can to improve their conditions, including strategically navigating class-segregated environments in order to reach needed resources. Second, experiencing homelessness in concentrated poverty (compared with relative class integration) surrounds those in the condition with constant reminders of their own marginality, which itself shapes their collective habitus and might legitimate homelessness for those experiencing it. It stands to reason that, other things equal, being immersed in the reminders of one’s own social inferiority prevalent in the poorest neighborhoods (e.g., old, chipped paint and other signs of economic neglect) has negative impacts on a person’s self-efficacy, which, in turn, might limit the proactive strategies used by those experiencing homelessness. Last, the segregation of homelessness in areas of concentrated poverty might constitute a de facto local solution to homelessness that permits episodes to go unchecked (e.g., Harris 2017; Marcuse 1988). Poverty segregation allows homelessness to be largely ignored by influential and engaged local political actors who might otherwise vocally disapprove of visible homelessness existing in their preferred spaces or the implementation of social programs that actually solve the problem. It also normalizes homelessness as a legitimate social category for nonpoor residents, who come to view the condition as an inevitable outcome limited to other social groups tacitly understood as inferior (i.e., poverty segregation shapes the habitus of nonpoor residents toward understanding the homeless as a group beneath them in the social order). It is plausible that, all else equal, normalizing homelessness in the collective consciousness of housed actors via its spatialization might downgrade the importance of meaningful antihomelessness initiatives on community agendas and limit other contributions that nonpoor actors can make (e.g., volunteer service or charitable donations). Although I do not explicitly test these ideas here, further exploration of such mechanisms constitutes a potentially fruitful avenue for future research.

Null associations between poverty segregation and unsheltered homelessness deserve a nuanced interpretation. Although I do not recover a meaningful relationship between poverty segregation and unsheltered homelessness rates, I suspect that this result might be partly a consequence of measurement error. If those experiencing unsheltered homelessness prefer to render themselves invisible in order to avoid surveillance or harassment, then higher poverty segregation might provide more concealment opportunities (e.g., in abandoned buildings), including from street volunteers conducting local PIT Counts. With this in mind, the relationship between poverty segregation and unsheltered homelessness across CoCs also deserves further attention.

Limitations to this analysis are considerable and include its cross-sectional design and issues associated with sample selection and variable operationalization.²⁰ Generalizing results of this research should be done carefully. Although I find a statistically meaningful relationship between poverty segregation and homelessness rates for those CoCs administering services in predominately urban and suburban areas, these units do not represent all American cities and suburbs.²¹ Although this analysis was carefully designed, results should be retested regularly, especially as higher quality data become available.

I conclude by directly returning to the core concerns of this study. I add to the existing literature by providing evidence of a positive association between poverty segregation and homelessness across CoCs that is partly, but not completely, attributable to the former’s relationship with the local prevalence of economic disadvantage. Furthermore, the relationship between poverty segregation and homelessness is robust to a full set of established CoC-level drivers of homelessness, which, along with state-level random effects, explain nearly 60 percent of the variation in homelessness rates across CoCs. Although poverty segregation is a spatial manifestation of the stratified economic order, findings support the idea that it carries its own social inertia that meaningfully drives homelessness.

Supplemental Material

sj-pdf-1-srd-10.1177_2378023121996871 – Supplemental material for On the Relationship between Poverty Segregation and Homelessness in the American City and Suburb

Supplemental material, sj-pdf-1-srd-10.1177_2378023121996871 for On the Relationship between Poverty Segregation and Homelessness in the American City and Suburb by Paul Muniz in Socius

Footnotes

Appendix A

Appendix B: Dimension Reduction

Because of high correlations among some potential CoC-level predictors of homelessness and correspondingly high variance inflation factors in multiple regression models, I perform PCAs for two sets of variables that are conceptually related. First, strong correlations exist among CoC-level poverty rates, rates of public assistance use, and unemployment rates (all measures logged). The first principal component of the PCA including these three variables explains 84.18 percent of their total variation, and all variables load in the positive direction. Given that the use of public assistance programs—defined here to include Supplemental Security Income, cash public assistance, and the Supplemental Nutrition Assistance Program—indicates the presence of economic hardships that are conceptually similar to the underlying constructs being measured by poverty and unemployment rates, I refer to this first principal component measure an index of economic disadvantage. Second, I conduct a PCA on two rental market characteristics: (1) CoC median rent and (2) the natural log of renter median income. I interpret these measures as interrelated products of rental market supply and demand (i.e., rent levels generally increase proportionally to what current renters can afford). Both of these factors load negatively on the first principal component, which explains 95.5 percent of their total variance. Given that these are correlated measures that determine rental market affordability, with lower median renter incomes and median rents implying higher scores, I refer to this measure as an index of rental market accessibility. Using this technique allows the preservation of information while explicitly addressing issues of multicollinearity that could otherwise threaten the stability of model estimates.

Appendix C. Unadjusted Mixed-Effects Regression Models Predicting Rates of Homelessness per 10,000 Persons in a Continuum of Care Net of Residential Segregation

	(1)	(2)	(3)
	Unsheltered Homeless Rate (Logged)	Sheltered Homeless Rate (Logged)	Total Homeless Rate (Logged)
Poverty segregation	2.92 (5.78)	10.00*** (1.65)	9.64*** (1.51)
Constant	.63*** (.21)	2.30*** (.08)	2.63*** (.09)
Observations	272	272	272
Marginal R²	.00	.14	.12
Conditional R²	.43	.42	.51

***

p < .01.

Appendix D. Mixed-Effects Regression Models Predicting Rates of Homelessness per 10,000 Persons in a Continuum of Care Net of Residential Poverty Segregation and Economic Disadvantage

	(1)	(2)	(3)
	Unsheltered Homeless Rate (Logged)	Sheltered Homeless Rate (Logged)	Total Homeless Rate (Logged)
Poverty segregation	–5.68 (6.19)	6.78*** (1.84)	5.72*** (1.71)
Economic disadvantage	.23*** (.07)	.09*** (.03)	.12*** (.03)
Constant	.65*** (.20)	2.32*** (.08)	2.64*** (.09)
Observations	272	272	272
Marginal R²	.03	.15	.15
Conditional R²	.44	.44	.52

***

p < .01.

Appendix E. Fully Adjusted Mixed-Effects Regression Models Predicting Continuum of Care–Level Homelessness Rates

	(1)	(2)	(3)
	Unsheltered Homeless Rate (Logged)	Sheltered Homeless Rate (Logged)	Total Homeless Rate (Logged)
Poverty segregation	–9.67* (5.72)	4.15** (1.75)	4.04** (1.69)
Economic disadvantage	.35** (.14)	.12** (.05)	.20*** (.05)
Rental market accessibility	–.27* (.15)	–.12** (.05)	–.19*** (.05)
Renter occupancy rate	–.01 (.02)	.01 (.01)	.00 (.01)
Vacancy rate	–.13 (.30)	–.13 (.11)	–.14 (.10)
Crowded rate	.39 (.35)	.24** (.12)	.24** (.12)
Percentage moved in last year	1.86*** (.55)	.08 (.20)	.20 (.20)
Percentage black	–.34** (.15)	–.14*** (.05)	–.22*** (.05)
Percentage Hispanic/Latinx	.09 (.19)	.03 (.07)	.05 (.07)
Median age	.01 (.04)	.01 (.02)	.01 (.02)
Single occupancy rate	.06 (.04)	.07*** (.01)	.07*** (.01)
Tracts	.28** (.12)	–.12*** (.04)	–.11** (.04)
Average temperature	.04*** (.01)	–.02*** (.01)	.00 (.01)
Constant	.63*** (.14)	2.27*** (.05)	2.62*** (.05)
Observations	272	272	272
Marginal R²	.36	.46	.46
Conditional R²	.50	.55	.57

p < .10. **p < .05. ***p < .01.

ORCID iD

Paul Muniz

Notes

Author Biography

Paul Muniz is a visiting assistant professor of sociology at Bucknell University. His current research focuses on the determinants of urban and suburban homelessness and the relationship between the social order and the arrangement of objects and bodies in physical space in the present-day United States. His work has been published in outlets such as Research in Higher Education.

References

Alexander-Eitzman

Ben

Pollio

David E.

North

Carol S.

2013. “The Neighborhood Context of Homelessness.” American Journal of Public Health 103(4):679–85.

Amster

Randall

. 2003. “Patterns of Exclusion: Sanitizing Space, Criminalizing Homelessness.” Social Justice 30(1):195–221.

Ansell

David A.

2017. The Death Gap: How Inequality Kills. Chicago: University of Chicago Press.

Babjaková

Jaroslava

Džuka

Jozef

Gresty

Jonathan

. 2019. “Perceived Causes of Poverty and Subjective Aspirations of the Poor: A Literature Review.” Ceskoslovenska Psychologie 63(3):325–36.

Beckett

Katherine

Herbert

Steve

. 2009. Banished: The New Social Control in Urban America. New York: Oxford University Press.

Birch

Eugenie L.

Silver

Christopher

. 2009. “One Hundred Years of City Planning’s Enduring and Evolving Connections.” Journal of the American Planning Association 75(2):113–22.

Bischoff

Kendra

Reardon

Sean F.

2014. “Residential Segregation by Income, 1970-2009.” Retrieved February 14, 2021. https://s4.ad.brown.edu/Projects/Diversity/Data/Report/report10162013.pdf.

Bourdieu

Pierre

. 1977. Outline of a Theory of Practice. Cambridge, UK: Cambridge University Press.

Bourdieu

Pierre

. [1991] 2018. “Social Space and the Genesis of Appropriated Physical Space.” International Journal of Urban and Regional Research 42(1):106–14.

10.

Brinegar

Sarah J.

2003. “The Social Construction of Homeless Shelters in the Phoenix Area.” Urban Geography 24(1):61–74.

11.

Byrne

Thomas

Munley

Ellen A.

Fargo

Jamison D.

Montgomery

Ann E.

Culhane

Dennis P.

2013. “New Perspectives on Community-Level Determinants of Homelessness.” Journal of Urban Affairs 35(5):607–25.

12.

Culhane

Dennis P.

Lee

Chang-Moo

Wachter

Susan M.

1996. “Where the Homeless Come From: A Study of the Prior Address Distribution of Families Admitted to Public Shelters in New York City and Philadelphia.” Housing Policy Debate 7(2):327–65.

13.

Cunningham

Mary K.

Batko

Samantha

. 2018. “Rapid Re-housing’s Role in Responding to Homelessness: What the Evidence Says.” Washignton, DC: Urban Institute.

14.

Daiski

Isolde

Halifax

Nancy Viva Davis

Mitchell

Gail J.

Lyn

Andre

. 2012. “Homelessness in the Suburbs: Engulfment in the Grotto of Poverty.” Studies in Social Justice 6(1):103–23.

15.

Dear

Michael

Gleeson

Brendan

. 1991. “Community Attitudes toward the Homeless.” Urban Geography 12(2):155–76.

16.

Dear

Michael J.

Wolch

Jennifer R.

1987. Landscapes of Despair: From Deinstitutionalization to Homelessness. Princeton, NJ: Princeton University Press.

17.

Desmond

Matthew

. 2016. Evicted: Poverty and Profit in the American City. New York: Broadway Books.

18.

Desmond

Matthew

. 2017. “How Homeownership Became the Engine of American Inequality.” The New York Times Magazine, May 29.

19.

Desmond

Matthew

. 2018. “Heavy Is the House: Rent Burden among the American Urban Poor.” International Journal of Urban and Regional Research 42(1):160–70.

20.

Edin

Kathryn

Shaefer

H. Luke

. 2015. $2.00 a Day: Living on Almost Nothing in America. New York: Houghton Mifflin Harcourt.

21.

Ehrenreich

Barbara

. 2010. Nickel and Dimed: On (Not) Getting by in America. New York: Metropolitan Books.

22.

Fargo

Jamison D.

Munley

Ellen A.

Byrne

Thomas H.

Montgomery

Ann Elizabeth

Culhane

Dennis P.

2013. “Community-Level Characteristics Associated with Variation in Rates of Homelessness among Families and Single Adults.” American Journal of Public Health 103(S2):S340–47.

23.

Fiske

Susan T.

2009. “From Dehumanization and Objectification, to Rehumanization: Neuroimaging Studies on the Building Blocks of Empathy.” Annals of the New York Academy of Sciences 1167:31.

24.

Fogle

Nikolaus

. 2011. The Spatial Logic of Social Struggle: A Bourdieuian Topology. Lanham, MD: Lexington Books.

25.

Gieryn

Thomas F.

2002. “What Buildings Do.” Theory and Society 31(1):35–74.

26.

Gilderbloom

John I.

Squires

Gregory D.

Wuerstle

Margaret

. 2013. “Emergency Homeless Shelters in North America: An Inventory and Guide for Future Practice.” Housing and Society 40(1):1–37.

27.

Gillespie

Sarah

Ballard

Katrina

Batko

Samantha

Peiffer

Emily

. 2020. “Addressing Chronic Homelessness through Policing Isn’t Working. Housing First Strategies Are a Better Way.” Urban Wire: Housing and Finance. Retrieved February 14, 2021. https://www.urban.org/urban-wire/addressing-chronic-homelessness-through-policing-isnt-working-housing-first-strategies-are-better-way?utm_source=urban_ad&utm_medium=twitter&utm_campaign=alternatives_to_policing.

28.

Hall

Andrew B.

Yoder

Jesse

. 2018. “Does Homeownership Influence Political Behavior? Evidence from Administrative Data.” Working Paper, Department of Political Science, Stanford University.

29.

Harris

Lasana T.

Fiske

Susan T.

2006. “Dehumanizing the Lowest of the Low: Neuroimaging Responses to Extreme out-Groups.” Psychological Science 17(10):847–53.

30.

Harris

Timothy

. 2017. “Neutralizing Homelessness: Federal Policy and the Depoliticization of Poverty.” Urban Geography 38(3):341–47.

31.

Herbert

Steve

. 2009. Citizens, Cops, and Power: Recognizing the Limits of Community. Chicago: University of Chicago Press.

32.

Herring

Chris

. 2014. “The New Logics of Homeless Seclusion: Homeless Encampments in America’s West Coast Cities.” City & Community 13(4):285–309.

33.

Hoffman

Lisa

Coffey

Brian

. 2008. “Dignity and Indignation: How People Experiencing Homelessness View Services and Providers.” Social Science Journal 45(2):207–22.

34.

HUD (U.S. Department of Housing and Urban Development). 2020. “The 2019 Annual Homeless Assessment Report to Congress.” Washington, DC: U.S. Department of Housing and Urban Development.

35.

Jargowsky

Paul A.

2013. “Concentration of Poverty in the New Millennium.” New Brunswick, NJ: The Century Foundation and Rutgers Center for Urban Research and Education.

36.

Lee

Barrett A.

Farrell

Chad R.

2004. “Metropolitan Neighborhoods with Sheltered Homeless Populations: Evidence from the 1990 and 2000 Censuses.” Washington, DC: Brookings Institution.

37.

Lee

Barrett A.

Farrell

Chad R.

Link

Bruce G.

2004. “Revisiting the Contact Hypothesis: The Case of Public Exposure to Homelessness.” American Sociological Review 69(1):40–63.

38.

Lee

Barrett A.

Price-Spratlen

Townsand

. 2004. “The Geography of Homelessness in American Communities: Concentration or Dispersion?” City & Community 3(1):3–27.

39.

Lee

Barrett A.

Price-Spratlen

Townsand

Kanan

James W.

2003. “Determinants of Homelessness in Metropolitan Areas.” Journal of Urban Affairs 25(3):335–56.

40.

Liebow

Elliot

. 1993. Tell Them Who I Am. New York: Simon & Schuster.

41.

Link

Bruce

Phelan

Bresnahan

Michaeline

Stueve

Ann

Moore

Robert

Susser

Ezra

. 1995. “Lifetime and Five-Year Prevalence of Homelessness in the United States: New Evidence on an Old Debate.” American Journal of Orthopsychiatry 65(3):347–54.

42.

Marcuse

Peter

. 1988. “Neutralizing Homelessness.” Socialist Review 18(1):69–96.

43.

Metraux

Stephen

Culhane

Dennis

Raphael

Stacy

White

Matthew

Pearson

Carol

Hirsch

Eric

Ferrell

Patricia

Rice

Steve

Cleghorn

J. Stephen

. 2001. “Assessing Homeless Population Size through the Use of Emergency and Transitional Shelter Services.” Public Health Reports 116:344–52.

44.

Moore

Peter

. 2015. “Wealthy Americans Take the Toughest Stance on Homelessness.” Retrieved February 14, 2021. https://today.yougov.com/topics/lifestyle/articles-reports/2015/11/23/wealthiest-americans-toughest-homeless.

45.

NAEH (National Alliance to End Homelessness). 2017. “Unsheltered Homelessness: Trends, Causes, and Strategies to Address.” Washington, DC: National Alliance to End Homelessness.

46.

NAEH (National Alliance to End Homelessness). 2020. “State of Homelessness: 2020 Edition.” Washington, DC: National Alliance to End Homelessness.

47.

NLCHP (National Law Center on Homelessness and Poverty). 2017. “Don’t Count on It: How the HUD Point-in-Time Count Underestimates the Homeless Crisis in America.” Washington, DC: National Law Center on Homelessness and Poverty.

48.

NLCHP (National Law Center on Homelessness and Poverty). 2019. “No Safe Place: The Criminalization of Homelessness in U.S. Cities.” Washington, DC: National Law Center on Homelessness and Poverty.

49.

NLIHC (National Low-Income Housing Coalition). 2014. “Homeless Assistance: Ten-Year Plans to End Homelessness.” Washington, DC: National Low-Income Housing Coalition.

50.

NLIHC (National Low-Income Housing Coalition). 2015. Ten-Year Plans to End Homelessness. Washington, D.C.

51.

NLIHC (National Low-Income Housing Coalition). 2019a. “Opportunity Starts at Home: The Need for Action on Housing Affordability.” Washington, DC: National Low-Income Housing Coalition.

52.

NLIHC (National Low-Income Housing Coalition). 2019b. “Out of Reach Report.” Washington, DC: National Low-Income Housing Coalition.

53.

Oakley

Deirdre

. 2002. “Housing Homeless People: Local Mobilization of Federal Resources to Fight NIMBYism.” Journal of Urban Affairs 24(1):97–116.

54.

O’Connell

James J.

2005. “Premature Mortality in Homeless Populations: A Review of the Literature.” Nashville, TN: National Health Care for the Homeless Council.

55.

Phelan

Jo C.

Link

Bruce G.

Tehranifar

Parisa

. 2010. “Social Conditions as Fundamental Causes of Health Inequalities: Theory, Evidence, and Policy Implications.” Journal of Health and Social Behavior 51(Suppl.):S28–S40.

56.

Poethig

Erika C.

2014. One in Four: America’s Housing Assistance Lottery. Washington, DC: Urban Institute.

57.

Reardon

Sean F.

Bischoff

Kendra

. 2011. “Income Inequality and Income Segregation.” American Journal of Sociology 116(4):1092–1153.

58.

Reardon

Sean F.

Bischoff

Kendra

Owens

Ann

Townsend

Joseph B.

2018. “Has Income Segregation Really Increased? Bias and Bias Correction in Sample-Based Segregation Estimates.” Demography 55(6):2129–60.

59.

Roberts

Joel John

. 2012. “Did America’s 10-Year Plan to End Homelessness Work?” Huffington Post, June 2. Retrieved February 14, 2021. https://www.huffpost.com/entry/did-americas-ten-year-pla_b_1394905.

60.

Roncarati

Jill S.

Baggett

Travis P.

O’Connell

James J.

Hwang

Stephen W.

Cook

E. Francis

Krieger

Nancy

Sorensen

Glorian

. 2018. “Mortality among Unsheltered Homeless Adults in Boston, Massachusetts, 2000–2009.” JAMA Internal Medicine 178(9):1242–48.

61.

Rosenberger

Robert

. 2020. “On Hostile Design: Theoretical and Empirical Prospects.” Urban Studies 57(4):883–93.

62.

Rossi

Peter H.

1989. Down and Out in America: The Origins of Homelessness. Chicago: University of Chicago Press.

63.

Routledge. 2016. “Habitus: Pierre Bourdieu.” Social Theory Re-Wired. Retrieved July 14, 2020. http://routledgesoc.com/category/profile-tags/habitus.

64.

Sharkey

Patrick

Elwert

Felix

. 2011. “The Legacy of Disadvantage: Multigenerational Neighborhood Effects on Cognitive Ability.” American Journal of Sociology 116(6):1934–81.

65.

Snow

David A.

Anderson

Leon

. 1993. Down on Their Luck: A Study of Homeless Street People. Berkeley: University of California Press.

66.

Snow

David A.

Mulcahy

Michael

. 2001. “Space, Politics, and the Survival Strategies of the Homeless.” American Behavioral Scientist 45(1):149–69.

67.

Stuart

Forrest

. 2013. “From ‘Rabble Management’ to ‘Recovery Management’: Policing Homelessness in Marginal Urban Space.” Urban Studies 51(9):1909–25.

68.

Trounstine

Jessica

. 2018. Segregation by Design: Local Politics and Inequality in American Cities. New York: Cambridge University Press.

69.

Tsai

Jack

Mares

Alvin S.

Rosenheck

Robert A.

2011. “A Geographic Analysis of Chronically Homeless Adults before and after Enrollment in a Multi-site Supported Housing Initiative: Community Characteristics and Migration.” American Journal of Community Psychology 48(3–4):341–51.

70.

Urban Institute. 2017. “Nine Charts about Wealth Inequality in America (Updated).” Retrieved February 18, 2021. https://apps.urban.org/features/wealth-inequality-charts/.

71.

Vitale

Alex S.

2017. The End of Policing. London: Verso.

72.

Wacquant

Loïc

. 2010. “Designing Urban Seclusion in the Twenty-First Century: The 2009 Roth-Symonds Lecture.” Perspecta 43:164–75.

73.

Western Regional Advocacy Project. 2014. “Without Housing: Decades of Federal Cutbacks, Massive Homelessness, and Policy Failures.” Retrieved February 14, 2021. https://wraphome.org/wp-content/uploads/2008/09/WRAPWithoutHousingfederalcutbacks2007report.pdf.

74.

Wolch

Jennifer

Dear

Michael

. 1993. Malign Neglect. San Francisco, CA: Jossey-Bass.

75.

Wolch

Jennifer R.

Dear

Michael

Akita

Andrea

. 1988. “Explaining Homelessness.” Journal of the American Planning Association 54(4):443–53.

76.

Wright

Talmadge

Vermund

Anita

. 1999. “Suburban Homelessness and Social Space: Strategies of Authority and Local Resistance in Orange County, California.” Pp. 121–44 in There’s No Place like Home: Anthropological Perspectives on Housing and Homelessness in the United States, edited by A. L.

Dehavenon

. Westport, CT: Bergin & Garvey.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.19 MB