Abstract
The suburbanization of poverty in the United States coincided with surging housing costs and growing rent burdens. Although much of the existing literature on eviction focuses on housing insecurity and displacement in cities, there is good reason to suspect that it has become an equally common phenomenon in the suburbs. This study evaluates changes in the frequency of urban and suburban evictions over time across seventy-four large metropolitan areas. Multilevel models show that the number of suburban evictions has steadily risen over time, even as urban evictions have been stable. The cases of Cleveland, Seattle, and Tampa demonstrate that the increase in suburban evictions is concentrated in pockets of the suburbs. Additional analyses reveal large geographic and racial inequalities in suburban displacement.
The geography of poverty in the United States has shifted dramatically over the last several decades. Whereas high-poverty neighborhoods were previously an urban phenomenon characterized by pockets of concentrated, often racially segregated disadvantage, poverty has since grown more diffuse. Most poor American households now live in the suburban periphery, particularly older inner-ring suburbs (Allard 2017). In this article, we analyze how the geography of eviction has and has not changed in response to these sociodemographic shifts. We ask three central questions. First, has eviction become more common in the suburbs as poverty rates have increased? Second, has the geographic concentration of eviction within suburbs shifted over time? That is, has there been a generalized shift in suburban eviction or is it concentrated in certain communities? Third, do Black and Latino suburban tenants face eviction more often than their White peers? Answering these questions allow us to both determine the trajectory of suburban evictions over time and, with an eye to the central theme of this volume, explore variation and inequality within and between U.S. suburbs.
To address these questions, we draw on the records of 5,611,800 eviction cases filed across seventy-four metropolitan areas between 2000 and 2016. We find that eviction counts have remained stable in urban spaces but increased significantly in suburbs. A sizable gap in evictions in typical urban and suburban neighborhoods in 2000 had narrowed by 2016. This has not played out uniformly across suburbs, however. Instead variation is considerable in the concentration of suburban eviction over time. We leverage the cases of Cleveland, Seattle, and Tampa to demonstrate how these patterns have played out differently across metropolitan contexts. We show that eviction cases have been increasingly concentrated in a subset of suburban neighborhoods, widening gaps between poor and affluent suburbs; we document large Black-White and Latino-White disparities in the risk of eviction in the suburbs. These disparities hold in many urban spaces as well but are as large or greater in the suburbs of many metropolitan areas in our sample.
Our findings bear implications for the study of poverty and housing stability. We offer evidence that those aiming to understand and address patterns of residential displacement in the United States must look beyond the urban core. Recent work has stressed the durable concentration of eviction in specific neighborhoods and even buildings (Rutan and Desmond 2021). This study offers a contrasting pattern: one in which sociodemographic shifts lead to changes in eviction patterns. The growing suburbanization of eviction calls for concerted, metropolitan-level policy response that aims to stabilize households at risk of eviction and mitigate growing inequalities between suburbs.
LITERATURE REVIEW
The suburbanization of American poverty marks a fundamental sociodemographic shift in metropolitan population structure. Until the 1980s, low-income households were heavily concentrated in a small subset of urban neighborhoods (Jargowsky 1997; Massey and Denton 1993). The suburbs have always been more diverse than often presumed (Lewis-McCoy et al. 2023, this issue; Kruse and Sugrue 2006), but the suburban poor made up a relatively small share of overall poverty. Yet beginning in the 1990s, the suburban poor population began to grow substantially, both in absolute terms and as a share of all low-income metropolitan households (Kneebone and Garr 2010; Jargowsky 2003). Most poor individuals now live in American suburbs (Allard 2017). Although the growth in suburban poverty was originally more diffuse, the suburban poor are increasingly concentrated in certain neighborhoods, creating stark inequalities between pockets of poverty and affluence (Lichter, Parisi, and Taquino 2012; Kneebone, Nadeau, and Berube 2011).
Despite clear evidence that suburban poverty has expanded, agreement is limited on why this may have occurred. Some scholars suggest that the decentralization of low-wage work has drawn poor households out from cities; others suggest that the phenomenon is driven primarily by households becoming poor in place as macroeconomic conditions deteriorate (Raphael and Stoll 2010; Cooke 2010). A third branch of thought points to the role that housing costs have played in pushing low-income households from high-cost urban markets and toward naturally occurring affordable housing in suburban communities (Rosenthal 2008; O’Flaherty 1996). The suburbs—particularly inner-ring suburbs with older housing stock—offer a potential reprieve from rising rents in cities. The draw of affordable housing is a particularly strong predictor of suburbanizing poverty in northeastern and midwestern metropolitan areas and among Black and Latino households (Howell and Timberlake 2014; Madden 2003). Yet tenants in metropolitan areas throughout the United States are experiencing record high levels of housing cost burden, including in places where incomes have fallen (Myers and Park 2019; Colburn and Allen 2018).
A growing number of suburban households are experiencing forms of material hardship that can pose unique challenges in these spaces—and that may put them at heightened risk of eviction. Evictions are a form of residential displacement in which a landlord removes a tenant (renter) from their home, often through a legal court process. The millions of formal, court-ordered evictions that occur in the United States each year overwhelmingly stem from nonpayment of rent, meaning that they are a direct consequence of poverty and a dearth of affordable housing (DeLuca and Rosen 2022; Gromis et al. 2022). Eviction is a traumatic and destabilizing experience that can precipitate prolonged spells of homelessness and a series of downward moves into more hazardous housing and neighborhoods (Collinson and Reed 2018; Desmond and Shollenberger 2015). Losing one’s home can cause absences from work, potentially leading to job loss and exacerbating instability and disadvantage (Desmond and Gershenson 2016). Eviction is associated with adverse health outcomes for everyone in the household, parents and children alike (Himmelstein and Desmond 2021; Hatch and Yun 2021). Displacement stresses not only individual households but also the surrounding community, amplifying threats to public health such as crime and infectious disease (Kirk 2021; Benfer et al. 2021). If eviction is indeed increasingly common in poor suburban neighborhoods, then it would likely intensify growing socioeconomic inequalities among suburban neighborhoods.
Most literature on the prevalence and consequences of eviction focuses on urban spaces (Desmond and Shollenberger 2015; Lundberg and Donnelly 2019). A handful of recent studies, however, have directly examined the prevalence of eviction in suburban communities. One national analysis found substantial heterogeneity across metropolitan areas in urban-suburban differences in eviction rates, showing that suburban eviction rates were greater than urban rates in roughly one in every five metropolitan areas (Hepburn, Rutan, and Desmond 2022). Analyses in Lexington, Kentucky, and metropolitan Atlanta, Georgia, reveal suburban neighborhoods and buildings with large volumes of eviction filings, in some cases outnumbering filings from equivalent places in the urban core (Immergluck et al. 2020; Shelton 2018). Although these cross-sectional analyses reveal that displacement is common in some suburban spaces—on par with or even exceeding levels of eviction in urban communities—they reveal little about the ongoing process of the suburbanization of poverty and its relationship over time with eviction patterns. Have suburbs, often imagined as stable enclaves that fuel upward mobility, become more unstable and insecure over time, or have they always been precarious for tenants?
Reason to suspect that the suburbanization of poverty has resulted in more evictions in suburban communities is ample. The sprawl and street design of many suburbs requires households to own a car. This not only creates additional expenses that might stress household budgets but also makes a household more vulnerable to destabilizing events like job loss if their car breaks down (Roberto 2008). Low-income suburbanites are more likely to struggle with food insecurity than their urban peers, a problem exacerbated by the logistical challenges of reaching a more limited set of food pantries and nonprofits (Allard et al. 2017; Shannon et al. 2018). Poor suburban residents face much greater challenges accessing social service organizations, particularly those intended to alleviate hardship or promote upward mobility (Allard and Pelletier 2023; Murphy and Wallace 2010; Allard 2009). Indeed, most suburban towns lack a single social service nonprofit of any kind (Allard and Roth 2010). The suburban poor also face barriers in accessing health care and finding providers that offer affordable and effective care (Schnake-Mahl and Sommers 2017; Francis et al. 2009).
The challenges of suburban poverty may be compounded as poor households are increasingly sorted into poor places. Suburban municipal governments may be ill equipped to provide social services or alleviate rising levels of material hardship (Mattiuzzi and Weir 2020; Allard 2017; Simms 2023). Conversely, some towns, struggling to fund municipal services because of an eroding tax base, turn to punitive fines and fees to raise revenue. This strategy exacerbates the material challenges facing low-income households, especially those that are Black or Latino (Beck 2023; Pacewicz and Robinson 2021).
Sociodemographic change alone does not necessarily lead to a shift in eviction trends, however. Research on eviction in urban spaces reveals eviction patterns to be relatively stable and unlikely to shift substantially over time. In many cities, a limited set of landlords within a community file a disproportionate share of evictions, anchoring displacement in place (Sims and Iverson 2021; Teresa and Howell 2021). The stability of eviction hot spots across time indicates that eviction is a durable feature of neighborhood inequality (Rutan and Desmond 2021). Further, if the expansion of suburban poverty has been driven by households seeking—and finding—affordable housing (Howell and Timberlake 2014; Madden 2003), then the suburbanization of poverty may not be accompanied by the suburbanization of eviction. This uncertainty over how the level of suburban evictions may have changed leads us to our first research question.
Question 1. Has the frequency of evictions increased in suburban communities over time?
Suburban neighborhoods are heterogenous, both between and within metropolitan areas. Whereas some experience increasingly concentrated poverty, others have retained their affluence and exclusivity. Evictions and their consequences are likely distributed unevenly across suburbs, creating and reinforcing inequalities between communities. With this perspective in mind, we investigate geographic and racial dimensions of eviction that may have influenced the diverging destinies of suburbs over time.
In urban settings, evictions are often concentrated in space because a limited set of landlords account for large portions of all evicted households (Rutan and Desmond 2021; Teresa and Howell 2021). Although some work demonstrates that high volume eviction filers do operate in some suburban neighborhoods (Immergluck et al. 2020), it is not clear whether evictions would become more diffuse or concentrated over time. Suburban housing stock is different than urban housing stock in ways that make concentration appear less likely: suburban tenants are more likely to live in a single-family rental (SFR) than in a large apartment building in which many renters might face the threat of eviction. The relatively diffuse nature of suburban rental housing would seem to create a bulwark against the concentration of eviction.
Still, poor suburban renters are increasingly clustered in a subset of available neighborhoods (Lichter, Parisi, and Taquino 2012; Kneebone, Nadeau, and Berube 2011). Even as most suburbs have become more diverse, many communities remain racially and ethnically segregated (Fowler, Lee, and Matthews 2016; Orfield and Luce 2013). Suburbs in some parts of the country—particularly the Sunbelt and Southeast—have been a target for corporate landlords looking to expand their investment in SFR properties (Fields, Kohli, and Schafran 2016). These corporate landlords turn to eviction more quickly than small operator landlords (Gomory 2022; Raymond et al. 2021), and their increased involvement in suburban markets may drive growing instability. Regarding the geographic inequalities between suburban neighborhoods, we ask,
Question 2. Has the geographic concentration of eviction within suburbs shifted over time?
The threat of eviction disproportionally falls on Black and Latino Americans. Nationally, roughly one in every four Black renters lived in a county where the eviction rate for Black tenants was at least double that of their White peers (Hepburn, Louis, and Desmond 2020). At least part of this disparity may be explained by economic factors: relative to their White counterparts, Black renters on average have lower and more unstable incomes and more limited access to savings or other resources, such as family financial networks, that would allow them to weather financial hardships and avoid eviction (NLIHC 2022).
Understanding is limited on whether Black-White and Latino-White disparities in the risk of eviction are consistent throughout metropolitan areas or differ between cities and suburbs. On one hand, poverty and segregation are not as deeply interconnected in suburbs as they are in cities, which may alleviate some of the disadvantages that Black tenants experience (Massey and Denton 1993). On the other, landlords’ discriminatory tendencies appear more pronounced in suburban areas, which may lead them to evict Black tenants more readily (Fischer and Massey 2004). There is also no reason to think that Black and Latino renters who have moved to the suburbs escaped the basic economic pressures that may lead them to fall behind on rent more often. This leads us to our third question about racial inequalities in suburban evictions:
Question 3. Do Black and Latino suburban tenants face eviction more frequently than their White peers?
DATA AND METHODS
We analyzed patterns in evictions over time in metropolitan areas across the United States. An eviction is a moment of acute hardship and instability as tenants are forced from their home. We evaluated the prevalence of evictions based on the number of court-recorded eviction judgments, the final step in the court’s eviction process when a judge has dispossessed a tenant of their rental housing.1 We focused on eviction judgments as opposed to eviction filings because some landlords initiate the eviction process not to remove their tenants but to collect rent or exert power (Leung, Hepburn, and Desmond 2021; Garboden and Rosen 2019). Because eviction judgments are the final step in the legal process, we consider them to be a direct measure of displaced renter households. We examined forced moves among renter households and do not examine foreclosure or other proceedings against homeowners. Hereafter, we refer to evictions and eviction judgments interchangeably.
We drew on a large administrative dataset of eviction proceedings in court records provided by the Eviction Lab (Desmond et al. 2018). This dataset is constructed from individual-level eviction records that LexisNexis Risk Solutions collected from state and local courts between 2000 and 2016. Because these records are formal court procedures, they do not include either negotiated lease terminations between landlords and tenants or any illegal or under-the-table efforts that landlords make to remove their tenants. Research is limited on the prevalence of such informal evictions, although the distribution of informal evictions across neighborhoods appears highly correlated with the distribution of formal evictions, suggesting that eviction judgments are a strong representation of the moves that tenants are forced to make (Hepburn, Louis, and Desmond 2022; Desmond, Gershenson, and Kiviat 2015). The records were cleaned, stripped of duplicates and commercial cases, geocoded, and validated against publicly available data sources published by county- and state-court systems based on both case volume and case outcomes.2 The records do not consistently specify the reason that a landlord initiated the eviction process, but the vast majority of evictions are for the nonpayment of rent (DeLuca and Rosen 2022).
The national dataset of eviction records is assembled from a patchwork of state and local courts. Eviction regulations and legal procedures differ from state to state, including in the steps landlords must take to initiate proceedings, the time that can elapse between each step of the eviction process, and the protections provided to tenants (LSC 2021; Hatch 2017). Subtle differences in court and eviction procedures do not appear to affect the likelihood that an eviction filing becomes an eviction judgment (Sudeall and Pasciuti 2021). Case dispositions and outcomes were standardized by the Eviction Lab to ensure consistency across these legal contexts and to address challenges associated with analyzing court records (Porton, Gromis, and Desmond 2020; Desmond et al. 2018).
Sample Construction
We are interested in the changing prevalence of evictions within metropolitan areas. Thus we limited the Eviction Lab’s data to counties within the two hundred largest metropolitan areas by population.3 To ensure that our estimates were effective representations of eviction dynamics within metropolitan areas, we restricted our analytic set to years within metropolitan areas in which we observed valid eviction data for at least 50 percent of both the urban and suburban populations. We also required at least ten urban and ten suburban tracts in each metropolitan area. Further, because we intend to evaluate changes over time, we dropped counties with only one year of data. These three criteria restricted our analytic set to 234 counties across seventy-four metropolitan areas (1,373 county-years). We include a list of included metropolitan areas in the supplementary materials online.4 For the median metropolitan area in our analytic set, we observed eviction data for seven years, for 100 percent of its urban and for 100 percent of its suburban population. In total, we observed 2,729,831 judgments from 5,611,800 cases.
We adopted Elizabeth Kneebone and Alan Berube’s (2013) census-based definition of urban and suburban spaces. We considered tracts to be urban if they were contained within either the first principal city in the OMB name of the metropolitan area or any subsequent named city with a population greater than one hundred thousand. All other tracts in the metropolitan area were marked as suburbs. This definition is well suited to distinguish urban and suburban places as it is based on functional political boundaries that pertain to how space is governed (Terbeck 2020). Future research should explore alternative definitions that allow for distinctions between types of suburbs (Lewis-McCoy et al. 2023, this issue). In the current analysis, we aim to address the dearth of research on housing instability in suburbs broadly. Our analytic set included 6,279 urban and 10,689 suburban tracts.
Analytic Strategy
We analyzed changes over time in eviction judgments in cities and suburbs. We are primarily interested in evaluating the association between changes in neighborhood poverty and changes in evictions between 2000 and 2016. Because we do not have a complete panel of metropolitan areas across years, we cannot construct a simple descriptive measure of changes in suburban eviction counts. Instead, we fit a three-level negative binomial regression model in which tract-year observations were nested within tracts, which in turn were nested within metropolitan areas. This multilevel model allowed us to examine neighborhood level trends in eviction judgments, the differences between urban and suburban trajectories, and the influence of poverty change on eviction patterns. Multilevel models produce semi-pooled estimates that balance among the data where they exist; the multilevel framework is ideal for our data structure where longitudinal data are limited for some of the metropolitan areas (Raudenbush and Bryk 2002; Gelman and Hill 2007).
In building a model, we sought a parsimonious representation of change over time. We allowed for varying intercepts and coefficients and used a negative binomial model because the dependent variable was a count of eviction judgments.5 We estimated an individual intercept and coefficient for time for each tract, thus essentially fitting a growth curve model (Raudenbush and Bryk 2002). To ensure balance and promote model convergence, we centered the years in our analytic set in 2008, which means that any coefficients for the intercept should be interpreted as 2008 values. Formally, the level one model was as follows:
The dependent variable (Ytij) was the count of eviction judgments in year t in tract i of metropolitan area j. At level one, we simply modeled the change in eviction judgments over the years under analysis. The intercept (π0ij) was the predicted count of evictions in 2008, and the slope coefficient (π1ij) was the estimated rate of change in evictions between 2000 and 2016. We chose a loglinear parameter for time, as opposed to a curvilinear specification, because analyses with national data reveal largely stable levels of eviction over this period (Gromis et al. 2022).
At level two we allowed both coefficients (π0ij and π1ij) to vary as a function of tract characteristics. Formally, the model for the intercept was as follows:
In equation (2), we modeled the number of evictions in the tract in 2008, the intercept-year in our data, as a function of suburban status (SUBURBij) and a vector (Xkij) of tract-level socioeconomic variables (ethnoracial majority, poverty rate, poverty rate squared, percent children, percent female headed households, percent foreign born, and percent high school graduates) and housing market characteristics (median rent, number of renter households, vacancy rate, percent of federally subsidized housing units, and median housing age) that have previously been associated with eviction. Aside from the subsidized housing measure, all covariate data were collected from the 2006–2010 American Community Survey (U.S. Census Bureau 2016). We excluded 1,051 tracts for missing ACS data. We calculated the proportion of housing units that were subsidized in a tract based on property-level data from HUD’s Picture of Subsidized Households (U.S. Department of Housing and Urban Development n.d.). We summed all subsidized units in a tract and divided by the total number of housing units in the tract.
Equation (2) distinguishes differences in eviction judgments across tracts at a single point in time. We also evaluated changes in eviction counts over time:
In equation (3), we modeled variation in a tract’s change over time in eviction count as a function of its suburban status (SUBURBij) and a vector of covariates (Zkij). The latter included the same covariates included in equation (2) but operationalized to reflect changes over time—between the 2000 Census and the 2012–2016 American Community Survey—to capture changing demographic and housing characteristics within the tract. Instead of an ethnoracial majority parameter, we measured change in percent of tract residents who identified as Black, percent who identified as Latino, and percent who identified as another race. We did not measure changes in the squared poverty rate or median housing age. We provide descriptive statistics for our sample and assess its representativeness in table A.1.
At the metropolitan area level (level three) we allowed for random variation—without predictors—for the terms for the intercept (β00j), the time-invariant suburb term (β01j), and the year term (β10j). As an example, the equation for the intercept can be written as follows:
All other terms were treated as fixed across metropolitan areas. Substituting across levels, the model can be rewritten as
In this formulation, the intercept (γ000) represents the predicted number of eviction judgments in the year 2008 in an urban tract with no racial majority and—because the predictors were mean-centered and standardized—average levels of the covariates in the vector Xkij. The γ010 and γ0k0 parameters adjust the baseline prediction up or down depending on the tract’s suburban status and the covariates included in vector Xkij. For years other than 2008, the term γ110 represents the log change in the number of evictions in an average urban tract. The parameters γ110 and γ1k0 adjusted the time trend up or down depending on suburban status and the values of the slope covariates. All of the coefficients can be interpreted as the percent difference in the count of eviction judgments when all other parameters are at their respective means.
Results from the regression model allowed us to determine general trends over time in eviction judgments in urban and suburban neighborhoods. However, they also elided variation between places. Appreciating the scale of this heterogeneity is, we argue, critical to both interpreting the significance of our regression findings and understanding the varying ways in which the geography of eviction shifted over time. To explore this variation more fully, we focused on three large metropolitan areas for which we had many years of eviction data. We chose Cleveland, Ohio, as an example of the thirty-three metropolitan areas where a majority of evictions occurred in the city early in the study period, but the suburban share of evictions increased over time. Seattle-Tacoma-Bellevue, Washington, (Seattle hereafter) was emblematic of the thirty-one metropolitan areas with a large suburban share of evictions throughout the period. Finally, we included Tampa-St. Petersburg, Florida, (Tampa hereafter) to represent the fifteen metropolitan areas where the urban share of evictions increased over time. We calculated tracts’ average eviction rate between 2000–2008 and again 2009–2016.6 Then, we mapped the average eviction rates from the two periods and calculated Local Moran’s I (Anselin 1995), a descriptive spatial statistic used to identify clusters of high and low values. In this application, high clusters represent areas of high instability and eviction. We describe how the geography of eviction shifted, if at all, from 2000 to 2016.
We further investigated the consequences of suburban evictions for tenants by evaluating two dimensions of inequality among suburban communities: geographic and racial.
We evaluated the extent to which the geography of suburban evictions has become more uneven over time by using a dissimilarity index to compare the distribution of eviction judgments to the distribution of renters across suburban neighborhoods. If all renters have a similar risk of eviction, then the index will be at a minimum. The index will be at its maximum (100), however, if evictions occur only within a few neighborhoods. We used our regression models to predict eviction counts in each tract-year. We used the 2000 and 2010 Decennial Censuses and the 2012–2016 American Community Survey estimates for renter households in 2000, 2010, and 2012–2016, respectively. We performed a linear interpolation to estimate the number of renter households in the intercensal years. We estimated the dissimilarity index (D) separately across all tracts (i) in urban and suburban places (p) in each of the metropolitan areas (m) and years (y) as
We evaluated racial inequalities in suburban evictions by estimating eviction rates by ethnoracial group in urban and suburban contexts. Eviction records do not record race or ethnicity but do include tenants’ first and last names and their addresses, among other details. We applied Bayes’ Rule to impute race-ethnicity based on tenants’ last names and the racial composition of the neighborhood where the eviction occurred (Hepburn, Louis, and Desmond 2020; Imai and Khanna 2016). The imputation process estimated the probability that a tenant is non-Hispanic Black, non-Hispanic Asian, non-Hispanic White, Latino, or of some other race-ethnicity. We estimated the total number of evicted tenants for each group by summing these probabilities. We used a linear interpolation of renter household heads by race—based on data from the 2000 and 2010 Censuses and the 2012–2016 American Community Survey—to calculate the denominator for these rates. Eviction records do not typically provide any information about household income, so we cannot adjust the estimated disparities for potential socioeconomic status differences among renters of different ethnoracial identities (for additional details of our procedure, see the online appendix).
RESULTS
We modeled the number of eviction judgments in urban and suburban neighborhoods between 2000 and 2016 using two versions of the regression model as detailed. In table 1, we present the results of these models. The first model included no control variables for either the tract’s intercept or slope aside from the number of renter households. In this model, we predicted the number of eviction judgments in 2008 to be 39.3 percent lower in a suburban tract than in an urban one (1 – e–0.500 = 0.393). The coefficient for the year term in this model was near zero and not significant, indicating that the number of evictions in an average urban tract was expected to be relatively constant over time (e0.002 = 1.002). By contrast, eviction counts were predicted to increase significantly over time—by 1.3 percent per year (e0.002+0.011 = 1.013)—in the average suburban tract.
Multilevel Negative Binomial Regression Estimating Evictions in Urban and Suburban Contexts, 2000–2016
In model 2, we added an array of socioeconomic and housing market predictors as covariates. After controlling for these neighborhood characteristics, the gap in the number of eviction judgments between urban and suburban neighborhoods in 2008 was smaller. A suburban neighborhood was predicted to have 23.7 percent fewer evictions in 2008 relative to an otherwise similar urban neighborhood (1 – e–0.270 = 0.237). The coefficient for year in model 2 remained non-significant and near zero, again suggesting that the typical urban tract has a steady number of eviction judgments from year to year (e0.003 = 1.003). By contrast, the interaction term is again positive and significant (e0.003+0.011 = 1.014), indicating a 1.4 percent increase in eviction judgments in a suburban tract each year. In figure 1, we plotted the trend in eviction counts in the average urban and suburban neighborhood between 2000 and 2016.
Predicted Eviction Counts
Source: Authors’ tabulation.
Even in the early 2000s, when the gap in eviction judgments between urban and suburban neighborhoods was at its largest, evictions were not infrequent in the suburbs. We predicted, for the year 2000, 8.04 evictions in the typical suburban tract relative to 11.47 in the typical urban tract. Over time, counts in both contexts converged to 9.93 evictions in the typical suburban tract and 11.94 evictions in the typical urban tract by 2016. This convergence is largely due to the increasing frequency of eviction in suburban neighborhoods rather than changes in urban neighborhoods.
We used results from model 2 to predict the proportion of each metropolitan area’s total evictions that occurred in the suburbs in each year. We plot changes in the suburban share of evictions in figure 2. Between 2000 and 2016, the share increased in fifty-nine of the seventy-four metros in our sample. In 2000, 42.5 percent of evictions occurred in the suburbs in the median metropolitan area; this increased to 45.2 percent by 2016. Notably, even early in the period, in thirty-two metropolitan areas most evictions occurred in the suburbs. By 2016, this was true of thirty-four metropolitan areas. Housing insecurity and displacement have become common in both urban and suburban communities.
Shift in Suburban Share of Evictions, 2000–2016
Source: Authors’ tabulation.
Case Studies
Although the regression results allow us to appreciate general trends across the United States, they fail to capture heterogeneity in how these trajectories played out across metropolitan areas. We now turn from general trends to closer examination of changes to the geography of eviction in three metropolitan areas: Cleveland, Seattle, and Tampa. These case studies demonstrate large variation in the patterns of suburban evictions.
Cleveland is a prime example of a metropolitan area that has undergone decades of transformation and yet remains beset by persistent racial and geographic inequalities. A legacy industrial powerhouse whose population and employment opportunities were gutted by deindustrialization, Cleveland has, like many other Rust Belt cities, experienced an uneven economic resurgence driven by investments in what are referred to as the Eds and Meds sectors (Nuemann 2016). Although this growth has benefited affluent workers, many lower-income households were shut out of high-opportunity industries such as education, financial services, and health care. Between 2000 and 2016, Cleveland experienced a pronounced, 9.6 percentage point increase in the share of poor population living in the suburbs.7 Despite these transformations, Cleveland overall remains a highly segregated metropolitan area (Krysan and Crowder 2017, 266).
Across the metropolitan area, roughly 10,800 renter households were evicted each year. Beneath this stability, however, was a large shift in where evictions occurred: the share of evictions in the suburbs rose from 43.4 percent in 2000 to 55.3 percent in 2016. We map the changing geography of these evictions in figure 3, using cluster analysis to identify sets of high (or low) displacement tracts that are located near other high (or low) displacement tracts. These tracts are distinguished from the tracts that are on-trend and have eviction rates near the metropolitan average.8 On the left side of the figure, which illustrates the geography from 2000 to 2008, are three clear clusters of neighborhoods with high eviction rates (above the metropolitan average), all almost entirely within the city of Cleveland. There are several prominent clusters of low eviction rates (below the metropolitan average), including the inner-ring suburbs of Rocky River and Fairview Park to Cleveland’s west and Shaker Heights to its east.9 These communities have relatively high median incomes and median property values and much lower levels of poverty than many urban neighborhoods. We could describe this map simply: unstable city, stable suburbs.
Eviction Rate Clusters in Cleveland Metropolitan Area
Source: Authors’ tabulation.
From 2009 onward, however, the geography of eviction in Cleveland was much more diffuse as eviction rates in many suburban communities rose and rates in many urban neighborhoods fell. On the right side of figure 3, the three high-eviction clusters are still present, but they each now include suburban neighborhoods, straddling the city line. Inner-ring suburbs such as East Cleveland, Euclid, and Warrensville Heights each witnessed large increases in eviction races. These communities had much larger shares of Black residents than neighboring communities did and experienced sharp declines in median incomes in real terms, from a drop of $9,500 in East Cleveland to one of roughly $17,500 in Warrensville Heights. After 2009, the portions of the high eviction clusters nearest downtown—previously experiencing exceptionally high eviction rates—saw levels of eviction at or below the metropolitan average. Although some suburban communities emerged as instability hot spots, other neighborhoods (such as Rocky River, Fairview Park, and Shaker Heights) experienced few evictions. This divergence in the fortunes of suburban neighborhoods suggests that intra-suburban inequality—inequality between suburbs—is rising within the Cleveland metropolitan area.
The suburbanization of poverty in the Seattle metropolitan area started earlier than in Cleveland and continued during the study period. A central hub of the booming tech sector, rents and home prices in urban Seattle rose dramatically during the 1990s (Glick 2008). This was associated with the rapid dislocation of Black residents from the Central District, many of whom settled in the suburbs. In 2000, 60.8 percent of poor residents in the metropolitan area lived in the suburbs. This proportion had increased by 6.1 percentage points by 2016. As lower-income households were increasingly pushed to the suburbs, so was the risk of eviction (Thomas et al. 2019).
As figure 4 makes clear, this process was already well under way in the early 2000s. Eviction rates were low throughout the city of Seattle despite a pocket of high eviction risk in urban Tacoma. It was the suburbs between these cities—the corridor encompassing Kent and Federal Way—that saw the most evictions in the first half of the study period, as well as a cluster south of Tacoma. Between 2009 and 2016, as eviction rates fell across the metropolitan area (Thomas et al. 2019), the geography of eviction risk changed dramatically.10 The high-eviction area south of Tacoma largely disappeared, and the suburbs between Tacoma and Seattle experienced declining risk. Simultaneously, a new set of high-eviction areas developed north of Seattle in Snohomish County. This included a cluster of high-eviction tracts in Everett as well as a set of more rural high-eviction areas to the east. These high-eviction pockets had not existed in the earlier period. In Seattle, much of metropolitan eviction was already suburban, but the geography of displacement shifted markedly across suburbs over time.
Eviction Rate Clusters in Seattle-Tacoma-Bellevue Metropolitan Area
Source: Authors’ tabulation.
Tampa is a rapidly growing and racially diversifying metropolitan area. From 2000 to 2016, its population grew by 24 percent, from 2.4 to nearly 3 million, and the share of the metro population identifying as non-Hispanic White fell from 74.8 percent to 62.7 percent. Although most poor households (62.6 percent) already lived in the suburbs in 2000, by 2016 the suburban share of the poor population grew by 7.3 percentage points. Rapid population growth has swamped the supply of affordable housing: Tampa has among the largest affordable housing shortfalls in Florida (Shimberg Center 2019). The percentage of renters who are cost burdened increased from two in five households (38.9 percent) in 2000 to one in two households (48.9 percent) by 2016. In 2000, nearly two-thirds of evictions (64.5 percent) occurred in suburban communities. The suburban share decreased by 0.7 percentage points to 63.8 percent in 2016.
Of the three case studies, the geography of eviction in Tampa was the most stable over time (figure 5). The left panel of figure 5 shows the two primary clusters of high-eviction neighborhoods early in the period: a small one centered in St. Petersburg and a large one stretching across Tampa’s eastern border with several suburban communities, including University, Temple-Trace, and East Lake-Orient Park. Aside from a mix of high- and low-eviction rate neighborhoods to Tampa’s south, few other patterns are discernible. Most communities in the metropolitan area had eviction rates near the region’s average, which shifted from 3.5 to 3.2 percent from the early years to the late ones. By the end of the period (right panel), the eviction clusters in both St. Petersburg and east Tampa grew considerably. The large cluster of high instability in eastern Tampa includes more urban and suburban neighborhoods. Although poverty rates grew in the suburban portions of these clusters, the increases in tract-level poverty were largest in the urban neighborhoods. Another notable change in the later period is the formation of more clusters of low instability suburban communities, creating a sharper contrast with the expanding instability along Tampa’s inner-ring suburbs.
Eviction Rate Clusters in Tampa-St. Petersburg-Clearwater Metropolitan Area
Source: Authors’ tabulation.
Dimensions of Intra-Suburban Inequality
Although eviction has become an increasingly common suburban phenomenon, it has affected metropolitan areas in different ways. As we saw in the case studies, small clusters of neighborhoods emerged as places with high levels of instability. Because evictions remained infrequent in many other suburban neighborhoods, the concentration of evictions within suburban hot spots may have increased during the study periods. To test this observation, we use the dissimilarity index to compare the distribution of evictions across suburban tracts with the distribution of renters. If all renters face a similar risk of eviction, the index will be at a minimum. If evictions are concentrated within just a few neighborhoods, however, then the index will be its maximum value, 100.
The results again demonstrate that how suburban eviction changed over time followed no single trajectory. In thirty-six of the seventy-four metropolitan areas, the dissimilarity index increased, at a median of 3.4 percent. In these places, evictions became more concentrated in a subset of suburban neighborhoods relative to the distribution of renters. For instance, several high-eviction suburban clusters emerged in the Cleveland metropolitan area even as large swaths of the suburbs still saw limited evictions. Because the increase in evictions was concentrated in those inner-ring pockets, the index for Cleveland increased by 23 percent, from 26.1 to 32.2. In the remaining thirty-eight metropolitan areas, however, the index declined. In many of these areas, the rising suburban share of evictions coincided with an expanding geography of eviction. Renters were evicted more frequently from in a greater number of suburban neighborhoods. This finding underscores the variety of metropolitan experiences with the suburbanization of poverty.
Eviction was not only increasingly common in some suburban communities, but also disproportionately experienced by Black and Latino tenants. By and large, racial disparities in eviction risk were largest for Black suburban tenants, who were 61 percent more likely to be evicted than their White peers in the median metropolitan area. Black suburban renters were evicted at higher rates than White renters in fifty-eight of the seventy-four metropolitan areas in our analytic set. These disparities were most extreme in the West, where Black suburban renters were evicted at nearly quadruple the rate as their White peers in the median metropolitan area (3.9 times in Seattle). Eviction rates were much closer to parity in Southern metropolitan areas. In the median metropolitan area (Durham, North Carolina), the eviction rate for Black suburban tenants was 7.1 percent higher than the rate for White.
The disparities in eviction rates were generally smaller between Latino and White tenants, although heterogeneity among metropolitan areas was substantial. In the median metropolitan area, Latino renters were evicted 23.6 percent more frequently than their White peers. Latino renters were more than twice as likely as their White counterparts to experience eviction in fifteen metropolitan areas. Yet, in twelve metro areas—most located in the South—White suburban renters were evicted twice as frequently as their Latino peers. In contrast to Black and Latino suburbanites, Asian renters generally experienced eviction at similar rates as White renters.
Ethnoracial disparities in eviction were typically larger within cities than within suburbs. In the median metropolitan area for urban disparities, Black renters were evicted nearly twice as often (86.1 percent higher) and Latino tenants were evicted 28.7 percent more often than their White neighbors. These disparities were not always larger in urban spaces, however. In fifteen metropolitan areas, including Cleveland and San Antonio, Black-White disparities in eviction rates were more extreme in the suburbs than in urban areas. Suburban disparities between Latino and White tenants were larger in thirty-one of the seventy-four metropolitan areas including metropolitan areas such as Des Moines, Iowa, and Phoenix, Arizona.
DISCUSSION
The suburbanization of poverty has fundamentally transformed the geography of households experiencing hardship and housing insecurity in the United States. We find that evictions have become more common in the suburbs, even as they have remained largely stable in urban spaces. Low-income households may have moved to suburban communities seeking relief from surging urban rents, but in many places, they have not found stability. The annual number of evictions in a typical suburban neighborhood increased steadily over time, approaching the level of evictions in a typical urban neighborhood by the year 2016. This shift has occurred throughout the United States, from the inner-ring suburbs of Cleveland to the outlying communities of Seattle. The suburban share of eviction increased during the study period in fifty-nine of the seventy-four metropolitan areas in our sample. Eviction has never been a uniquely urban problem, but many suburban communities now experience eviction as frequently as in similar urban neighborhoods.
This shift was, however, far from uniform across metropolitan areas. To explore heterogeneity among metropolitan areas in these trajectories, we describe the changing geography of eviction in Cleveland, Seattle, and Tampa. Evictions in the Cleveland metropolitan area steadily expanded into inner ring suburbs. Eviction was already heavily suburban in Seattle at the start of the study period but became more frequent in outlying suburban communities that previously had relatively low levels of displacement. In Tampa, by contrast, the clusters of tracts with high instability remained anchored along its border with several suburbs, demonstrating a strong persistence in eviction hot spots. In each of these metropolitan areas, we also observed clusters of suburban communities where eviction was relatively rare. The suburbanization of poverty has not occurred uniformly across the United States and the rise in suburban evictions is no different. Just as is true within cities, understanding eviction risk and designing interventions that promote housing stability requires keen awareness of attention to these local particularities. One-size-fits-all solutions are unlikely to prove adequate.
The suburban shift in evictions may have either been spread evenly across suburban neighborhoods or concentrated in a few communities. We calculated dissimilarity indices to compare the distributions of evictions and renters in suburban areas by year. In about half of the metro areas in our sample, the dissimilarity index increased, suggesting that evictions were becoming more concentrated into a subset of suburban neighborhoods. Yet in the other half, the dissimilarity index decreased. In these places, instability was reaching a broader set of neighborhoods rather than becoming entrenched in a subset of suburban communities. Suburbs are not monolithic and even adjacent neighborhoods may have sharp differences in the risks of housing instability.
We find that Black suburbanites faced higher risks of eviction than their White peers in fifty-eight of the seventy-four metropolitan areas in the sample. Latino tenants also generally faced eviction more frequently than suburban White renters, but heterogeneity was greater between metropolitan areas: they were twice as likely as White renters to face eviction in fifteen metropolitan areas but half as likely in another twelve. Asian tenants faced generally similar levels of eviction as White tenants. As they grow and diversify, American suburbs are increasingly the site where ethnoracial disparities in poverty and housing are created and perpetuated. Housing access and stability has been a key driver of persistent ethnoracial inequalities in the United States since emancipation (Taylor 2019). Large ethnoracial disparities in eviction risk in the suburbs suggest yet another instance of these patterns, one that will only exacerbate inequalities given the severe and lasting negative consequences of eviction.
The growing concentration of eviction may place some suburban communities on a diverging trajectory from their peers. In each of the case studies, we observe suburban clusters of both high and low instability. Suburbs with large and increasing numbers of evictions must grapple with both the direct and indirect effects of displacement: heightened levels of need, unemployment, and homelessness; threats to public health and safety; schools with more absences and instability. The limited social service supports in these spaces may not be enough to meet increasing demand. Gradually, suburbs with large numbers of evictions may be forced to confront a cascading set of challenges that strain their political and financial capacity to respond. Meanwhile, other communities in the same metropolitan area may be able to maintain stability and see little change. Sharp inequalities, such as the ones we document in housing instability, may come to define American suburbs across several dimensions.
These challenges call for concerted policy responses across levels of government. Much of the tenant organizing capacity in the United States is concentrated in cities. If these organizations only pursue measures to address eviction and housing instability at the city level, such as through right to counsel or eviction diversion programs, then they are likely to leave suburban tenants, who have considerably less political power, behind. But no suburb—rich or poor—is an island: displacement and rising hardship have spillover effects.
Local governments that coordinate social services, develop equitable shares of affordable housing, and pass legislation to provide additional legal protections to tenants will be able to mount an effective regional response, stronger than any of them could manage on their own. Such an interjurisdictional approach requires a high level of cooperation. To avoid a patchwork response that would only amplify inequalities among tenants, county and state governments can introduce legal reforms to the eviction process, such as changes to notice requirements or case filing fees, provision of legal representation, or mandatory diversion, and provide robust funding for affordable housing, which benefits urban and suburban tenants alike. Nongovernment actors, such as foundations and social providers can support these public measures by redirecting or bolstering outreach in suburban communities where the safety net is thinnest (Allard 2009). By providing a robust and coordinated response, governments and nonprofits can create regions which are stable, resilient, and equitable.
Regional policies that make it more difficult for landlords to pursue evictions or easier for tenants to find legal and material resources will likely reduce some of the substantial ethnoracial disparities observed in eviction risk. However, to simultaneously mitigate the risk of eviction for all tenants and alleviate the acute risks born by Black and Latino renters, local leaders must employ a targeted outreach strategy that brings anti-eviction measures directly to the people and places that experience eviction most frequently. Policymakers will need a comprehensive understanding of the particular issues facing tenants of color in their region to address ethnoracial disparities where they exist.
We suggest five areas for future research on suburban eviction risk and urban-suburban disparities. First, especially given heterogeneity between metropolitan areas, it will be important to both expand the scope of analysis and conduct more in-depth studies of single metropolitan areas. Although we were able to cover seventy-four metropolitan areas in our sample, we do not have enough data to include a number of large metropolitan areas with particularly significant suburban populations (for example, Atlanta, Detroit, San Francisco). Second, we encourage more research on the institutional role of landlords in creating inequality. Landlords hold the power to influence a community’s trajectory: they choose where to invest, what to charge for rent, whom to offer a lease to, and when to evict. What role do they play both in the suburbanization of poverty—in drawing new populations to the suburbs—and to the growing number of evictions in these spaces? Given the significance of SFR housing in the suburbs, it is important for future research to evaluate the ways in which new types of financial instruments and the expanding portfolios of corporate landlords—far more prevalent in some metropolitan areas than others—have affected changes in suburban poverty and eviction. Third, further exploration of the characteristics of suburban eviction hot spots could help us better understand which areas are at particularly high risk of increasing eviction. What sets these areas apart, and what could local leaders do to plan accordingly?
Fourth, we need more research that explores how the lived experience of eviction plays out differently for suburban and urban tenants. Suburban tenants have far less access to the social service nonprofits that could help avoid eviction in the first place through financial or legal assistance or mitigate eviction’s most harmful repercussions such as prolonged homelessness, job loss, and lasting health deficits (Allard and Pelletier 2023; Murphy and Wallace 2010). How does that affect their experience of the eviction process? How do suburban landlord-tenant courts process these cases differently? Fifth, we would encourage more research that examines how policies implemented in response to the COVID-19 pandemic affected urban-suburban disparities in eviction. For example, eviction moratoria implemented in response to the pandemic varied both between and within states (Benfer et al. 2022; Kneebone and Underriner 2022), cities being often more willing to implement additional protections than outlying areas. In addition, cities were able to directly access federal Emergency Rental Assistance funds whereas suburbs, due to their smaller populations, could not.11 Although direct access to funds yielded lower per capita funding, it did allow cities greater freedom in decision-making about how to distribute resources. Further research should explore how these variations in policy response and resource availability affected short- and long-term housing stability.
The last three decades have witnessed significant sociodemographic changes in suburban America as communities have become more diverse, poorer, and more unequal. We document here a further attendant change in housing stability: a growing number of suburban evictions. These evictions have been concentrated in a relatively small set of communities, places in which tenants may find few public or private supports as they face the risk of losing their home. These eviction cases have fallen more heavily on Black than on White renters and have resulted in an increasingly divided suburban landscape. These shifts have been more dramatic in some metropolitan areas than in others. Our findings highlight the need to confront poverty beyond urban spaces and to think of displacement as an increasingly suburban concern.
FOOTNOTES
↵1. The execution of an eviction writ is the last stage in the eviction process (Benfer 2021). However, reliable statistics on executed evictions are not widely available and, in many cases, tenants vacate the property after the judgment, which renders this stage moot (Desmond 2016).
↵2. County-year aggregate estimates were included if the total number of LexisNexis filings in a county fell between 87 and 114 percent of the courts’ publicly reported total. When public data were not available, we extrapolated the most recent total a maximum of two years and applied the same criterion. We exclude county-years where more than 60 percent of LexisNexis cases were either dismissed or missing outcomes.
↵3. This initial sample included 439 counties (2,786 county-years) that covered at least some portion of 142 metropolitan areas.
↵4. See the online appendix (https://www.rsfjournal.org/content/9/1/104/tab-supplemental).
↵5. Log likelihood tests confirmed a negative binomial distribution was more appropriate than a Poisson distribution.
↵6. We split the period into two discrete sections to evaluate net changes in the distribution of evictions over time. It is not a concern that the Great Recession falls during our cut point because analyses using national eviction data reveal only a very slight increase in evictions during the financial crisis (Gromis et al. 2022).
↵7. Authors’ calculation based on the 2000 Census and 2012–2016 American Community Survey.
↵8. The average eviction rate in the Cleveland metropolitan area was 4.1 percent in the early period and 3.9 percent in the later.
↵9. We include reference maps with place names in the supplementary materials (for Cleveland, see figure A.4).
↵10. The average tract’s eviction rate in the Seattle-Tacoma-Bellevue metropolitan area fell from 1.2 percent to 0.9 percent during the period (for a reference map with placenames, see online figure A.5.
↵11. In some cases, large counties including both urban and suburban areas directly accessed funds. In several cases, this created odd political arrangements wherein both central cities and counties received direct funding and established nonoverlapping jurisdictions. For example, residents of the city of Dallas applied to a city-level rental assistance program; those in the inner-ring suburbs applied to a county-level program, and those on the periphery (outside Dallas County) applied to the Texas statewide program.
- © 2023 Russell Sage Foundation. Rutan, Devin Q., Peter Hepburn, and Matthew Desmond. 2023. “The Suburbanization of Eviction: Increasing Displacement and Inequality Within American Suburbs.” RSF: The Russell Sage Foundation Journal of the Social Sciences 9(1): 104–25. DOI: 10.7758/RSF.2023.9.1.05. The JPB, Gates, and Ford Foundations, as well as the Chan Zuckerberg Initiative, funded this research. The Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number P2CHD047879 supported the research reported in this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Direct correspondence to: Devin Rutan, at drutan{at}princeton.edu, 225 Wallace Hall, Princeton, NJ 08544, United States.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
REFERENCES
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.