Forced Displacement From Rental Housing: Prevalence and Neighborhood Consequences

Abstract

Introduction

Urban sociologists long have emphasized the importance of neighborhood disadvantage (Sampson 2012; Wilson 1987). Concentrated neighborhood disadvantage can have acute negative effects on children’s health, development, and cognitive performance (Sampson et al. 2008; Sharkey 2010), and living in distressed neighborhoods with high poverty and violent crime rates can harm adults’ physical and mental health as well as hinder their economic well-being (Sampson et al. 2002; Sharkey and Faber 2014). Since the earliest days of American sociology, urbanists have tried to understand how families end up in distressed neighborhoods, how they break out of them, and how residential opportunities—always in lockstep with social and economic ones—are extended to some and denied to others. City demographers long have explored the link between locational attainment and residential mobility (Lee et al. 1994; Simons 1968). However, researchers have focused almost entirely on voluntary mobility, overlooking how involuntary displacement—disproportionately experienced by low-income households—may be consequential to neighborhood selection.

Drawing on new survey data of Milwaukee renters, collected using novel techniques designed to capture respondents’ reasons for moving, this study corrects this oversight. In so doing, it advances our understanding of residential mobility, neighborhood inequality, and low-income housing in two ways. First, it provides a rigorous estimate of the prevalence of forced removal from housing among urban renters. By capturing formal evictions (processed through the court system), informal evictions (not processed through the court system), landlord foreclosures, and housing condemnation, our estimate is much more comprehensive than those based on eviction court records (Desmond 2012) or a single survey question about eviction (Mayer and Jencks 1989). We find that more than one in eight renters in Milwaukee experienced an eviction or other kind of forced move in the previous two years.

Second, this study suggests that renters who experience a forced move relocate to more-disadvantaged neighborhoods than those who move under less-demanding circumstances. Experiencing a forced move is associated with more than one-third of a standard deviation increase in both neighborhood poverty and crime rates, relative to voluntary moves. Across all models, the most robust and consistent predictors of neighborhood downgrades between moves are race (whether a renter is black) and move type (whether the move was forced). This study is among the first to examine the consequences of forced removal from housing. By providing evidence implying that involuntary displacement is a critical yet overlooked mechanism of neighborhood inequality, this study helps to clarify why some city dwellers live in much worse neighborhoods than others.

Intentionality Bias in Residential Mobility Research

For decades, the question of why families move has been central to the study of migration, urban demography, and city life. When social scientists began addressing this question around mid-century, they often collected data on forced moves. In Why Families Move, the foundational text of residential mobility research, Rossi (1955/1980:33) observed, “There are moves that are ‘induced’ or precipitated by eviction, by dwelling unit destruction through fire, other hazards, or demolition; or by conversion to nonhousing uses.” Focusing on four Philadelphia census tracts, Rossi classified fully 39 % of the moves in his study as forced.

While researchers gathered data on forced moves, they focused primarily on documenting and explaining voluntary mobility. Despite observing an extraordinarily high rate of involuntary displacement, Rossi (1955/1980) chose to emphasize how changes in family composition (e.g., marriage, having children) often led to intentional moves aimed at meeting new housing needs, referring to this process as “the major function of mobility” (p. 9). Other studies followed suit, investigating how voluntary mobility could be explained by life cycle changes in employment, family size, or housing conditions (e.g., Kendig 1984; Speare 1970), conceptually parsing voluntary mobility into categories and subcategories of moves (Clark and Onaka 1983), and exploring the degree to which households were able to realize their mobility expectations (Newman and Duncan 1979).

Researchers then expanded their vision, explaining voluntary mobility not only through individual-level factors, such as housing dissatisfaction or family expansion, but also through neighborhood-level conditions. As ecological push and pull factors were incorporated into models of voluntary moves, work on mobility preferences combined seamlessly with work on racial intolerance and residential segregation. Research on racial preferences, based primarily on vignette studies, repeatedly found whites to have the strongest preference to live alongside same-race neighbors, and blacks to have the weakest (Charles 2003; Pattillo 2005).

While one group of researchers was designing studies to capture how tolerance for racial residential integration varied across different segments of the population, another group was assessing the degree to which households were able to parlay economic capital for residential capital in the form of neighborhood quality. This work gave rise to the “residential attainment model,” which perceived residential mobility as an expression of social climbing (Logan and Alba 1993; South and Crowder 1997). Spatial location was understood to be the outcome of an “individual-level attainment process” involving “upgrading from central-city slums to working-class neighborhoods to suburbs” (Logan and Alba 1993:243, 244). For members of minority communities, such upgrades often involved moving closer to white communities, a kind of “spatial assimilation” understood to be essential to the general process of assimilation itself (Massey and Mullan 1984). After adjusting for socioeconomic status, researchers found that blacks were less likely to reside in safe and economically prosperous neighborhoods (Logan et al. 1996; South and Crowder 1997). This fact led scholars to focus on structural impediments to residential mobility. Garnering support for the “place stratification model,” analysts demonstrated that housing market characteristics and racial discrimination prevented many blacks from escaping segregated neighborhoods (Massey and Denton 1993; Pais et al. 2012).

Researchers have continued to examine the relationship between residential choice and racial intolerance when explaining patterns of racial segregation that characterize the modern American city. Studies have compared individual attributes of city dwellers (e.g., white) with the characteristics of their neighborhoods (e.g., percentage white) to examine how aggregate patterns of racial segregation may be explained by individual migration flows (e.g., Quillian 1999; Sampson and Sharkey 2008). Most recently, analysts have expanded their focus to consider how mobility patterns are influenced not only by neighborhood characteristics but also by features of the broader metropolitan area (Bruch 2014; Crowder et al. 2012).

Although researchers made significant advancements in our knowledge of residential preferences and voluntary mobility, eviction and other forms of involuntary displacement receded into the background of their studies. In their influential review of the residential mobility literature, Clark and Onaka (1983:49) devoted only a single paragraph to forced moves—and small wonder: many researchers had altogether stopped collecting data on them. Instead of asking households to articulate their reasons for moving, researchers began inferring such reasons from aggregate patterns of neighborhood turnover (e.g., Quillian 1999). This research was essential to our understanding of migration patterns and enduring racial segregation. Still, demographers have yet to complement this aerial view of the city with ground-level analyses of everyday mechanisms that also explain residential mobility and neighborhood sorting.

At least since Why Families Move (Rossi 1955/1980), empirical and theoretical work on residential mobility has reflected an intentionality bias. To Duncan and Newman (2007:175, 174), for example, “moves are, for the most part, rational, deliberate, and planned. . . . The sequence begins with a desire to move and proceeds to crystallized intentions or plans and finally to the move itself.” This perspective takes as its reference point the experience of many middle- and upper-class households and considers families disadvantaged if they are unable to leave disadvantaged neighborhoods or to translate economic gains into residential ones (e.g., South et al. 2005). However, low-income families are not only priced out of moving—they also can be priced out of staying.

This study expands the literature on residential mobility, inequality, and neighborhood attainment by focusing on a subset of moves that migration scholars long have neglected: forced moves. The degree to which a move is voluntary is not a given. Researchers should reserve judgment of moves being self-actuated and intentional until such a condition is empirically validated. To that end, this study begins by asking, What is the prevalence of involuntary mobility among renters? It then investigates whether involuntary mobility influences neighborhood selection.

How Prevalent Is Involuntary Mobility?

If rates of forced mobility were trivial, the intentionality bias of residential mobility research would be understandable. However, several researchers have documented fairly high rates of involuntary displacement in American cities. More than one in three moves in Rossi’s (1955/1980) study was involuntary. Abu-Lughod and Foley (1960) classified 30 % of intraurban moves as involuntary, with 20 % owing to demolition, fire, or eviction. Ross (1962) found that the previous intracity move of 13 % of Boston residents was forced.

No study to date has produced an updated and comprehensive estimate of the frequency of forced removal in any major city. Drawing on eviction records, Desmond (2012) found that roughly 16,000 adults and children in Milwaukee are evicted through the court system annually and that in the predominantly black-populated inner city, one renter-occupied household in 14 is evicted each year. However, estimates of the frequency of forced removal based on eviction court records undershoot the mark, given that a court-ordered eviction is but one type of forced move that renters may experience. Tenants also may be forced to relocate through informal evictions—such as when a landlord simply tells a family to leave, or changes the locks—which can be less expensive and more efficient than formal evictions (Desmond 2012:95; Hartman and Robinson 2003). A landlord going into foreclosure or the city condemning a unit as unfit for human habitation also can provoke a forced move (Been and Glashausser 2009). These forms of forced mobility are not recorded in eviction records. This study moves beyond this limitation by employing a novel survey technique to capture multiple forms of forced moves within a sample generalizable to the rental population of Milwaukee.

The need to document the prevalence of involuntary displacement from housing is especially pressing in light of the current affordable housing crisis. Missed rent payments motivate the vast majority of formal eviction proceedings (Desmond 2012:101). A combination of flat incomes, soaring rents, and a shortfall of federal housing assistance has resulted in a surge in severely rent-burdened households (Collinson 2011; Steffen 2011). Today, roughly one-half of poor renting households spend at least one-half of their income on housing costs, with one-quarter dedicating upwards of 70 % of their income to pay rent and utility costs (Desmond 2015).

Does Involuntary Mobility Influence Neighborhood Selection?

Research on residential mobility has focused almost exclusively on what might be called “major moves”—those characterized by large shifts in ecological context. Studies have examined moves between high- and low-poverty areas (South et al. 2005), moves between neighborhoods with significantly different racial compositions (Quillian 1999), or moves that take families from the city to the suburbs (Sampson 2012). In directing our attention to movers making large leaps across racial and economic divides, we have not fully turned our ear toward the soft shoe falls of millions taking smaller steps across the urban landscape, inching toward the edges of the ghetto or slipping deeper into it. Crime and gang activity, an area’s civic engagement and its spirit of neighborliness, the quality of the housing stock—all these things can vary drastically from one inner-city block to the next (e.g., St. Jean 2007). There is a rich and meaningful microeconomy of differences among poor, segregated neighborhoods. In neglecting to explore incremental moves involving subtle yet significant changes in neighborhood context, current theories of residential mobility cannot explain the considerable diversity of neighborhood quality among city dwellers who confront similar barriers to upward mobility. Why do some city dwellers live in much worse neighborhoods than others who share the same racial identity and socioeconomic status? Investigating why they move may help us gain some purchase on this question.

A forced move carried out under critical and even traumatic circumstances is guided not by the aspiration to “move up in the world” but simply by the need to move somewhere else (DeLuca et al. 2013).¹ Tenants who receive an eviction judgment often are ordered to vacate in a matter of days. Many, lacking legal counsel, are confused by the eviction process—which, from first eviction notice to removal by sheriff, takes roughly one month in Milwaukee, although most tenants vacate before their landlord summons the sheriff—and are caught off-guard when the eviction squad raps on their door and orders them out. Moreover, tenants evicted through the court system bear the blemish of eviction on their records. After noticing that prospective tenants recently have been evicted, landlords often turn them away (Desmond 2012; Kleysteuber 2006). As a result, recently evicted tenants often apply for dozens of apartments and move into the first unit for which they are approved (Desmond et al. 2015). Taken together, these considerations lead us to expect that compared with those who have relocated voluntarily, renters who have undergone a forced move will experience a drop in neighborhood quality.

Data and Methods

The Milwaukee Area Renters Study

This article draws on the Milwaukee Area Renters Study (MARS). An original survey of 1,086 tenants in Milwaukee’s private housing sector, MARS was designed to collect new data on housing, residential mobility, eviction, and urban poverty. The study’s focus on the private rental market reflects the experiences of most low-income families who receive no state or federal housing assistance (Schwartz 2010). The MARS sample excluded renters living in public housing but not those in the private market in possession of a housing voucher. To bolster response rate and data quality, surveys were administered in person in English and Spanish by professional interviewers at tenants’ place of residence. For each household, interviewers surveyed an adult leaseholder or, should a leaseholder be unavailable, an adult knowledgeable about household financial matters.² According to the most conservative calculation (AAPOR Rate 1), MARS had a response rate of 83.4 %.

Interviews were conducted from 2009 to 2011. Households were selected through multistage stratified sampling. Drawing on census data, Milwaukee block groups were sorted into three strata based on racial composition. Block groups were classified as white, black, or Hispanic if at least two-thirds of their residents were identified as such. Then, each of these strata was subdivided into high- and moderate-poverty census block groups based on the overall income distribution of each racial or ethnic group in the city. Blocks were randomly selected from each of these six strata. When a block was selected into the sample, interviewers visited every household in the selected block, saturating the targeted areas. To focus on renting households, interviewers screened out owner-occupied dwellings. MARS also included an oversample of 100 recently evicted tenants who were randomly selected from closed Milwaukee eviction cases that occurred 12 to 24 months prior to the final fielding of the survey. After data collection, custom design weights for the regular sample and oversample were calculated to reflect the inverse of selection probability, facilitated by a Lahiri (1951) procedure, based on the demographic characteristics of Milwaukee’s rental population and adjusted to the MARS sample size. The Lahiri procedure allows the sampler to select probability samples (with a probability proportional to size) and to compute the selection probabilities for the resulting sample. Selection probabilities are then used to calculate the design weights for the overall sample. We use these custom weights throughout our analyses to facilitate estimates generalizable to Milwaukee’s rental population.

Milwaukee is a strategic setting in which to investigate the experiences of urban renters for at least three reasons. First, the characteristics of Milwaukee’s residents (Pager 2007) and rental market (U.S. Department of Housing and Urban Development 2009) are comparable with many U.S. cities. Second, renter protections in Milwaukee also are fairly typical, especially compared with wealthy cities that have an economically diverse rental population (Manheim 1989), and available data suggest that neither Milwaukee nor Wisconsin has an unusual eviction rate compared with other cities and states (Desmond 2016). Third, studying Milwaukee not only expands sociological data on and knowledge of different urban environments; it also may produce findings more applicable to cities distinct from America’s exceptional global hubs (Small 2007). That said, future research is needed to evaluate the extent to which these findings are generalizable to settings beyond Milwaukee.

The MARS instrument comprised more than 250 unique items. The centerpiece of the MARS questionnaire was a housing roster used to obtain a two-year residential history from each respondent. To collect this history, interviewers employed a memory prop—a two-year calendar—to help respondents recall important events and features of their residential experience. Respondents were asked to list all the places they “lived or stayed for at least a month,” including other people’s houses, shelters, and correctional facilities. They also were asked to provide the addresses or crossroads of each residence. This information was geocoded using ArcGIS and an associated road network database. Then each current and past residence was assigned to a census block group—our neighborhood metric—and linked to aggregate population estimates.

Recording Reasons for Moving

Asking why someone moved is no simple task. Tenants often provide an explanation for a move that maximizes their own volition, and asking about involuntary displacement comes with its own set of complications because tenants tend to have strict conceptions of eviction. Take Rose and Tim, a couple whom the first author met while conducting fieldwork in distressed Milwaukee neighborhoods (Desmond 2012, 2016). They were forced to leave their mobile home after Tim sustained a back injury at work. Rose and Tim did not go to court but undeniably were evicted. (Their names appear in the civil court records.) Nevertheless, they do not see it this way. “When you say ‘eviction,’” Rose explained, “I think of the sheriffs coming and throwing you out and changing your locks, and Eagle Movers tosses your stuff on the curb. That’s an eviction. We were not evicted.” If Rose and Tim were asked during a survey, “Have you ever been evicted?,” they would have answered no. Accordingly, surveys that have posed this question underestimate considerably the number of families who experience eviction.

The MARS survey asked all respondents, “Since turning 18, have you ever been evicted?” But in light of how many tenants understand what constitutes an “eviction”—and thus reflecting the value of ethnographic fieldwork to survey design—the MARS Reasons for Moving Module goes far beyond that question. The module is structured around a series of yes/no questions about possible reasons for moving, beginning with involuntary removals and ending with voluntary moves. These questions are asked of each move that respondents experienced during the previous two years:

1. (1) An eviction is when your landlord forces you to move when you don’t want to. Were you, or a person you were staying with, evicted?

2. (2) Did you, or a person you were staying with, [leave after receiving] an eviction notice?

3. (3) Did you move away from this place because your landlord told you, or a person you were staying with, to leave?

4. (4) Did you move away from this place because you, or a person you were staying with, missed a rent payment and thought that if you didn’t move, you would be evicted?

5. (5) Did you move away from this place because the city condemned the property and forced you to leave?

6. (6) Did you move away from this place because (a) the landlord raised the rent? (b) the neighborhood was dangerous? (c) the landlord wouldn’t fix anything, and your place was getting run down? (d) the landlord went into foreclosure?

Respondents who answered no to question 1 for a given move would be asked question 2, and so on. If a respondent answered no to all these questions, she finally was asked, “I see that none of these reasons fit your case. Why did you move away from this place?” This approach minimized recall bias and allowed us to collect accurate data on the motivations for moves. It also allowed us to capture a variety of forced moves.

We organized tenants’ reasons for moving into three categories: forced, responsive, and voluntary moves.³Forced moves are initiated by landlords or city officials (e.g., building inspectors) and involve situations in which tenants have no choice other than to relocate (or think as much). These include formal and informal evictions, foreclosures, and housing being condemned. Responsive moves are motivated by housing or neighborhood conditions. These include rent hikes, a deterioration in housing quality, escalating violence in the neighborhood, domestic violence, and relationship dissolution. Voluntary moves are intentional and unforced relocations, often carried out ostensibly to gain residential advantage. These include moves to be closer to kin or work as well as housing and neighborhood upgrades.⁴

When reporting our findings, we begin by examining the prevalence of involuntary mobility among Milwaukee renters, offering estimates for all respondents (N = 1,086) as well as a subset of renters who moved at least once during the previous two years or “recent movers” (N = 580). We then model the association between move type and neighborhood quality, focusing on recent movers’ most recent move (the move that brought them to their current neighborhood) and comparing how forced, responsive, and voluntary mobility are related to neighborhood quality.

Modeling the Neighborhood Consequences of Involuntary Mobility

Because poverty and crime rates are among the most important indicators of neighborhood quality (Sampson 2012; Wilson 1987), we take these as our outcomes. Our measure of neighborhood poverty rate, based on the American Community Survey (ACS) (2006–2010) rolling averages, is the percentage of families below the poverty line. Our measure of neighborhood crime rate, based on Milwaukee Police Department data (2009–2011), is the sum of all counts of all National Incident-Based Reporting System (NIBRS) Group A and B offenses per 1,000 people.⁵ Because our crime data are local, models estimating changes in neighborhood crime across moves are restricted to moves within Milwaukee. However, models estimating changes in neighborhood poverty rate, based on national data, allow us to examine all recent moves within the United States.

Controls

We account for a number of factors previous research has associated with neighborhood quality. Because blacks are more likely than other racial and ethnic groups to reside in disadvantaged neighborhoods (South and Crowder 1997), we account for race. Because previous research has linked one’s ability to escape or avoid disadvantaged neighborhoods to their socioeconomic status, we control for highest level of education. Additionally, because studies have shown that housing assistance can influence renters’ location choice (Greenlee 2014), we also observe whether respondents received any form of assistance in their previous residence.⁶

Families with children are at especially high risk of housing discrimination and eviction (Desmond et al. 2013). Accordingly, we observe whether a respondent lived in a single-mother household at her previous housing unit as well as whether a respondent had a child sometime in the previous two years. In addition, we observe whether a respondent obtained a criminal record before her/his most recent move because the mark of a criminal record could limit tenants’ housing and neighborhood options (Thacher 2008). Since residential advantage is thought to increase over the life course (Rossi 1955/1980), we also control for age.

Last, we control for three additional life shocks. We observe whether a respondent had experienced in the last two years but before their most recent move (1) the dissolution of a “serious relationship,” (2) a sudden stoppage of public benefits (e.g., welfare sanction), or (3) being laid off or fired from a job. Renters who experienced a forced move were more likely than other movers to have experienced a stoppage in public benefits (4 % vs. 1 %) and job loss (19 % vs. 16 %); they were also less likely to have experienced relationship dissolution (13 % vs. 20 %). Summary statistics for all variables are displayed in the appendix (see Table 4).

The average variable used in this study had 2.4 % missing data. No variable had missing values that exceeded 9 % of responses. In our multivariate analyses, we use multiple imputation to preserve observations with some missing data (Allison 2002). We imputed missing data through regression equations that estimate missing values for one variable by treating all other variables in the data set as regressors. We used logit models to impute binary variables to avoid bias potentially created when rounding estimates generated by linear methods (Horton et al. 2003). We replicate our models on 15 imputed data sets and report the aggregated results (Rubin 1987).

Results

The Prevalence of Forced Removal From Rental Housing

In Table 1, we report several estimates of the prevalence of involuntary mobility among Milwaukee renters. The first estimate, “Evicted as an Adult,” reflects the percentage of renters who reported having ever “been evicted” since age 18. More than one in eight (13 %) of all renters (N = 1,086) reported having been evicted as an adult. Roughly 1 in 14 white renters (7 %) reported having experienced an eviction, as did 1 in 10 Hispanic renters (10%) and almost 1 in 5 black renters (19 %). Using t tests to detect statistical differences, we find that black renters were significantly more likely than white (p < .001) and Hispanic (p = .01) renters to report having been evicted at some point in adulthood.

Because evictions are not the only kind of forced move that families experience, and because many tenants who experienced an eviction may not report it as much, survey questions that simply ask respondents whether they have been evicted are poor estimates of the frequency of forced mobility. Accordingly, we draw on the results of the MARS Reason for Moving Module to provide a more comprehensive estimate of the prevalence of involuntary residential mobility among renters. These data capture all moves that occurred within the previous two years. More than one in eight Milwaukee renters (13 %) experienced at least one forced move—formal or informal eviction, landlord foreclosure, or building condemnation—during this time period. The rate of recent involuntary mobility was significantly higher for Hispanic renters than for white and black renters (p < .001). Nine percent of white renters, 12 % of black renters, and fully 23 % of Hispanic renters experienced a forced move in the previous two years.

Nearly one-half (48 %) of all forced moves experienced by Milwaukee renters during the previous two years were informal evictions. Formal eviction was less common, constituting 24 % of forced moves, suggesting that assessments of the frequency of forced displacement based on (formal) eviction court records are considerable underestimates.

An additional 23 % of forced moves were due to landlord foreclosure. The MARS survey took place in the wake of the foreclosure crisis.⁷ Analyses of court records show that Milwaukee evictions did not increase after the foreclosure crisis: in fact, they actually declined slightly (Desmond 2012:126–127). There is good reason to believe, then, that the number of formal evictions we have recorded is not inflated owing to the foreclosure crisis. However, it is likely that some of the landlord foreclosures we captured would not have occurred in other times. As displayed in Table 1, if we exclude landlord foreclosures from the analysis, the percentage of renters who had experienced a forced move within two years of being surveyed falls from 13.2 % to 10.2 %. Excluding forced moves on account of landlord foreclosures causes the rates of involuntary mobility among white and black renters to fall from 9 % to 7 % and from 12 % to 10 %, respectively. However, its biggest impact is seen in the rate of involuntary mobility among Hispanic renters, which falls from 23 % to 14 % after landlord foreclosures are excluded (see Rugh 2015).⁸

Last, roughly 5 % of forced moves were caused by building condemnation. As far as we can tell, none of the forced moves observed in our sample were executed on the basis of eminent domain.

Our multivariate models focus on respondents who moved recently and examine the reason for their most recent move. Among respondents who had moved at least once within the previous two years (N = 580), 14 % were forced to move from their previous dwelling. This was true for roughly one in 12 white renters (8 %), more than one in seven black renters (15 %), and nearly three in 10 Hispanic renters (29 %). In a single open-ended question, we asked these respondents to explain what happened. One-half cited issues with the landlord, including personal problems, disputes, and retaliation for calling a building inspector. Among the other half, the majority (and 26 % of all forced movers) pointed to financial problems.

The American Housing Survey (AHS) collects data on the reasons why renters relocated with the question, “What are the reasons you moved from your last residence?”; the AHS reports this information with respect to the most recent move of renters who moved within the previous year. According to the 2009 AHS (Table 4-11), among renters nationwide who had moved in the past year, between 2.1 % and 5.5 % were forced from their previous unit because of private displacement (e.g., owner moved into unit, converted to condominium), government displacement (e.g., unit was found unfit for occupancy), or eviction. According to MARS, 10.8 % of the most recent moves of renters who had moved within the previous year were forced. Our estimate is larger—and, we believe, more accurate—because MARS captured informal evictions. When we exclude informal evictions, our estimate drops to 3 %. We believe that the AHS significantly underestimates the prevalence of involuntary removal among renters by relying on open-ended questions that do not adequately capture informal evictions that many renters do not consider to be “evictions.”

The Consequences of Forced Removal on Neighborhood Quality

We now investigate with multivariate analyses whether forced displacement may influence the types of neighborhoods into which families select. Table 2 displays the results of four LDV models focused on the most recent move of renters who relocated two years prior to being surveyed. In each model, renters’ current metric of neighborhood quality (poverty or crime rate) is estimated by conditioning on that metric in their previous neighborhood. This is indicated by the variable Lagged Dependent Variable. Models 1 and 2 estimate poverty rate. Models 3 and 4 estimate crime rate. Models 1 and 3 condition only on reasons for moving and previous neighborhood quality. Models 2 and 4 introduce controls.

Even after controlling for a suite of important factors, our models estimate that renters who experienced a forced move live in neighborhoods with significantly higher poverty and crime rates than those who moved voluntarily, conditioning on previous neighborhood poverty rate and crime rate, respectively.⁹ Experiencing a forced move is associated with more than one-third of a standard deviation increase in both neighborhood poverty and crime rates, relative to voluntary moves. Our full models estimate that, all else equal, renters who experienced a forced move wound up in neighborhoods with a poverty rate 5.4 percentage points higher and a crime rate nearly 1.8 percentage points higher than those of renters who moved by choice. Responsive moves were not found to bring about a significantly different change in neighborhood quality relative to voluntary moves.¹⁰

We also found evidence that black renters experienced significant increases in both neighborhood poverty and crime rates between moves relative to white renters, net of move type, socioeconomic indicators, and life shocks.¹¹ As in cities across the United States, blacks in Milwaukee live in more disadvantaged neighborhoods than whites and Hispanics. For example, among recent movers, the average black renter in Milwaukee lives in a neighborhood where 17 % of families live below the poverty line, compared with 9 % for the average white renter and 13 % for the average Hispanic renter. These differences are statistically significant at the p = .01 level.

Drawing on the results of Model 2, we could estimate the neighborhood poverty rate by race and move type by holding the lagged dependent variable at the group mean by race and ethnicity, holding all continuous control variables at the population mean, and setting all binary variables equal to 0. Doing so leads us to expect that, all else equal, a black renter who experienced a forced move will live in a neighborhood with a poverty rate of 20.2 %, while a black renter who moved voluntarily will live in a neighborhood with a poverty rate of 14.8 %. A white renter who experienced a forced move is expected to live in a neighborhood with a poverty rate of 13.4 %, while a white renter who moved voluntarily is expected to live in a neighborhood with a poverty rate of 8.0 %. In other words, we would expect the most-advantaged black movers in our sample (those who moved voluntarily) to relocate to neighborhoods with a higher amount of concentrated poverty than the neighborhoods to which the least-advantaged white renters (those who experienced forced displacement) relocated.¹²

Although demographers of residential mobility long have prioritized life cycle changes (Rossi 1955/1980), our analyses found the association between such changes—for example, having a child, relationship dissolution, or losing a job—and neighborhood quality to be insignificant after controlling for forced mobility. Troublingly, we did document a statistically significant association between neighborhood crime rate and single motherhood. Perhaps reflecting not only their economic vulnerability but also their vulnerability on the housing market in particular, owing to widespread family discrimination (U.S. Department of Housing and Urban Development 2010), renters who lived in a single-mother household in their previous residence wound up in higher-crime neighborhoods than otherwise similar renters.

Does Eviction Help to Reestablish Market Equilibrium?

If forced movers experience a drop in neighborhood quality, does involuntary removal help to reestablish market equilibrium? This question imagines a rental market in which some renters are living in units that are too expensive for them. When they are eventually forced out, they relocate to segments of the market “where they belong”: for example, in less-expensive units. However, the majority of forced movers in our sample (60 %) experienced either no change or an increase in rent between moves. In Table 3, we estimate tenants’ current monthly rent, conditioning on rent at their previous unit, reasons for moving, and all the control variables in our main analyses. The coefficients on forced moving in these models are negative, but this difference is not significant at the p < .05 level.

Why was the significant decline in neighborhood quality experienced by forced movers unaccompanied by a significant decline in housing cost? Because the distribution of rents in Milwaukee is considerably compressed, housing costs do not march in lockstep with neighborhood quality. According to weighted MARS estimates, the median rent for a two-bedroom unit in Milwaukee is $600; 10 % of those units rent at or below $480, and 10 % rent at or above $750.¹³ A mere $270 separates some of the least expensive units in the city from some of the most expensive. In Milwaukee’s poorest neighborhoods—block groups in which 40 % or more of families live below the poverty line—median rent for a two-bedroom apartment fetches $550, only $50 less than the citywide median. The median rent for a two-bedroom apartment at or above the 75th percentile in crime rate is the same as that for a two-bedroom apartment at or below the 25th percentile in crime rate ($600). Accordingly, it is possible for renters forced from their homes to relocate to neighborhoods with more poverty and crime but equivalent housing costs. Given the surge in extreme rent burden among low-income renters, it often is true that many are living in units they cannot afford, but it often is untrue that they are not already at the bottom of the market.

Rising housing cost burden and stark compression of rents are compatible trends. First, rising utility costs—which are unaffected by neighborhood variation—contribute to increases in housing cost burden. In Milwaukee and across the nation, most renters are responsible for utility costs, but since 2000, the cost of fuels and utilities has increased significantly, owing to global demand and the expiration of price caps (Carliner 2013). Second, in recent years the professionalization of the rental market, the spread of information technology, and the rise of rent-setting algorithms have made it easier for landlords to coordinate prices, which likely contributes to price compression (Alexander and Muhlebach 2009; Gilderbloom and Appelbaum 1987). Third, if landlords at the bottom of the market have not slashed rents to avoid the inefficiencies of missed payments and evictions, it is likely because it is less expensive to deal with the cost of those inefficiencies than to maintain properties, and it is possible to skimp on maintenance if tenants are perpetually behind, given that being in arrears prevents tenants from taking advantage of legal protections designed to keep their housing safe and decent (Desmond 2016).¹⁴ Future research is needed to more fully investigate the relationship between rising housing costs and price compression in urban rental markets. However, rents in poor neighborhoods historically have been similar to, or more than, rents for better housing in nicer regions of the city (e.g., Hunter 2013; Riis 1890/1997). In a way, then, we should not be surprised by the fact that rent is not significantly cheaper in distressed neighborhoods. In many cities, it has long been that way.

Discussion

The findings of this study have several implications for how we think about residential mobility, neighborhood selection, and low-income housing. First, this study has generated a reliable estimate of the prevalence of forced mobility among urban renters. A significant proportion of Milwaukee renters—one in eight—were forced from their homes via eviction, landlord foreclosure, or building condemnation within the previous two years. That low-income families, the majority of them renters, have relatively high rates of residential mobility is well established (Ihrke and Faber 2012). We know much less about why this is the case. Our findings suggest that one reason why low-income renters move so much is simply because they have little choice in the matter. We can treat as intentional and motivated by a desire for residential improvement the many moves undertaken by renting households. Yet, the considerable frequency with which Milwaukee renters in general, and its black and Hispanic renters in particular, are forced to find new housing suggests that our models of residential mobility should be expanded to account for involuntary relocations.

Second, the results of this study provide evidence that experiencing forced dislocation may actuate a downward move: a relocation to a poorer and more dangerous neighborhood than we would expect had the move been voluntary. The trauma of being forced out of one’s home, the blemish of eviction that follows renters who were evicted through the court system, and the taxing rush to locate new housing likely combine to push tenants who relocate involuntarily into distressed neighborhoods. This study suggests, then, that forced removal may be a significant determinant of neighborhood selection. Decisions to enter and exit disadvantaged neighborhoods can be influenced by subtle institutional processes rendered invisible by standard demographic data that presume, rather than investigate, families’ reasons for moving.

Structural forces of the housing market help to determine not only where city dwellers end up, as with the place stratification model’s emphasis on housing discrimination, but also why they move in the first place, which in turn helps to determine into which neighborhoods they select. Although neighborhoods certainly function as sites of residential attainment (e.g., Crowder et al. 2012) or racial segregation (e.g., Krysan and Bader 2007), they also should be viewed as commodities and largely owned, in the case of the inner city, by those who do not live within their borders. Consequentially, we should treat market actors in general—and landlords in particular—as central players in our theories of neighborhood selection and mobility (Logan and Molotch 1987:33–34). These considerations do not suggest replacing conventional models that emphasize neighborhood preferences, residential attainment, or place stratification; rather, they imply incorporating fully the insights of these perspectives into an expanded model of neighborhood sorting.

The fact that forced mobility is so common in the lives of urban renters and consequential to neighborhood selection inspires the need for future research documenting the ramifications of eviction and estimating the prevalence of involuntary removal in other cities. We know very little about the effects of eviction and other forms of involuntary displacement on children and adults’ physical and mental health, material hardship, economic well-being, and social support (Desmond and Kimbro 2015). We know equally little about how forced mobility affects neighborhoods and schools. These questions remain unanswered even as the need to answer them grows more pressing with the rapid decline of affordable housing.

This study responds to the policy need to understand the prevalence and consequences of eviction and other forms of forced moves. Because the prevalence and the consequences of forced moves among renters were previously unknown, policymakers have been at a loss when attempting to assign importance to, say, anti-eviction policies vis-à-vis other priorities. This study underscores the need for policymakers to focus their attention on forced relocation, treating it as an important reason why some families move into disadvantaged neighborhoods. Because our study found forced moves to be both prevalent and consequential, one implication is that policymakers should devote more resources to keeping families in their homes or at least to softening the blow of eviction. That those who relocate voluntarily land in more-advantaged neighborhoods suggests that low-income families, to the extent that they are able, strive to escape distressed neighborhoods and that those who do move into distressed areas often do so under duress.

Acknowledgments

This research was supported by the John D. and Catherine T. MacArthur Foundation, through its “How Housing Matters” initiative. We thank Weihua An, Vicki Been, Rogers Brubaker, Michael Carliner, Jorge De la Roca, Kathryn Edin, Ingrid Gould Ellen, Marion Fourcade, Carl Gershenson, Jack Katz, Barbara Kiviat, Kristin Perkins, Adam Slez, Edward Walker, Bruce Western, and seminar participants at UCLA; Northwestern University; the University of Queensland; the Harvard School of Public Health; the NYU Colloquium on Law, Economics, and Policy; and the 2012 Annual Meeting of the Association for Public Policy Analysis and Management.

Appendix

Notes

References