Growing Up as Rents Rise: How Housing Affordability Impacts Children

Residential mobility

Residential mobility is linked to negative outcomes for children, beginning in early childhood. In a study of the impact of mobility on children in preschool years (age 5 and under), Ziol-Guest and McKenna (2014) examined data from the Fragile Families and Child Wellbeing study. Using OLS regression, they found that early childhood mobility was associated with behavioral problems (attention, internalizing and externalizing behavior) among 5-year-old low-income children. Similar negative impacts of mobility on young children were found by Cutts et al. (2011), who examined how mobility in early childhood affected child health through interviews with low-income caregivers with children younger than 3 years who were seen in seven U.S. urban medical centers between 1998 and 2007. Using multivariate regression, they found that multiple moves (defined as more than two in the prior year) was associated with very poor or fair child health and lower weight for their age (Cutts et al., 2011).

Studies of the impact of residential mobility on school-aged youth (approximately ages 6 to 17) have also found generally negative outcomes. Coley et al. (2013) examined longitudinal data on low-income children and adolescents in the Three City Study. Using three-level hierarchical linear modeling, they found that residential instability showed significant negative associations with children’s emotional and behavioral functioning, as well as negative associations with reading and math skills as measured by Woodcock-Johnson tests. Similarly, Voight et al. (2012) examined the association between residential and child academic outcomes through a longitudinal latent growth curve modeling analysis of reading and math achievement for students in 11 middle schools. They found that students who moved in early grades (K–2) had lower scores on state standardized math and reading tests in Grades 3 to 8 than nonmobile peers. Importantly, they note that they were not able to isolate the separate components of residential as opposed to school mobility.

Residential mobility is associated with poor mental health of adolescents. Haynie et al. (2006a) analyzed data from the National Longitudinal Study of Adolescent Health (also referred to as the Add Health data set) using multivariate logistic regression and found that the odds of adolescent girls of attempting suicide were 60% higher for movers in the year after the move than nonmovers. They found that 40% of this relationship was explained by differences in school attachment, isolation from peers, and peers’ suicide attempts between movers and nonmovers. Using the same data set (Add Health), Haynie and South (2005) also found recent movers exhibited significantly higher rates of involvement in violence (defined as fights or threatening others with a weapon) than nonmovers and that much of this difference was explained by composition of peer networks, specifically friends’ involvement in “deviance” (defined as drinking or skipping school), with similar findings of a later study by Haynie et al. (2006b), who conclude that the mechanism through which mobility influences children is through disruption of positive peer networks (Haynie et al., 2006b).

Childhood residential mobility has also been associated with long-term negative outcomes for adult mental and physical health. Bures (2003) examined data from the National Survey of Midlife Development in the US (MIDUS) of adults between the ages of 25 to 74 using multivariate logistic regression, and found that adults with greater self-reported residential instability (at least three moves in childhood) had worse reported global (physical) health in midlife, but no worse mental health. The poverty measure, however, was weak, consisting of self-reports of whether someone perceived they were “somewhat” or “a lot worse” off than other families growing up. Similarly, Oishi and Schimmack (2010) examined a sample of over 7,000 Americans who were followed for 10 years through the MIDUS study. Using logistic regression and a mediated moderation analysis, they found that residential moves were associated with a reduction in quality of long-term relationships for introverts. A significant limitation of this study, however, was that there were no controls for socioeconomic status.

Dong et al. (2005) argue that the negative effects of childhood residential mobility on adult outcomes can be explained through the moderating risk factors that children in more mobile families are facing. They examined the relationship between childhood residential mobility and health outcomes in adolescence and adulthood through a retrospective survey of Kaiser Permanente Members on the number of Adverse Childhood Experiences (ACES) including abuse, neglect, etc. Using multivariate logistic regression, they found that while residential mobility was associated with increased health risk (smoking and suicide), these relationships were moderated significantly after incorporation of ACEs into model. They conclude that “the apparent relationship between childhood residential mobility and negative health and social outcomes is mostly if not solely, due to the effects of underlying ACEs that occur with high frequency in extremely mobile families” (p. 1107).

In sum, studies find generally negative outcomes associated with residential mobility in childhood. Collectively, however, few studies isolate the cause of mobility (job change, job loss, preference), and none isolate the role of housing affordability problems in triggering residential moves. As a whole, then, these studies are unable to speak to the particular effects of affordability-induced residential mobility.

School mobility

When children move as a result of affordability challenges, they often change schools. Numerous studies have found that switching schools is associated with negative impacts on student achievement, particularly in the early grades, and an increased risk of dropping out of school (see Mehana & Reynolds, 2004; South et al., 2007; Welsh, 2017). Welsh’s (2017) review of the research on school mobility concludes that while there may be some potential gains from school mobility when a child switches to a better performing school (see also Hanushek et al., 2004), those gains may offset the negative effects of the move itself.

As with the residential mobility literature, the research on school mobility does not differentiate between the varied causes of school moves—that is, whether moves were caused by residential mobility or due to other factors (transfer, choice, discipline, etc.), much less moves that were triggered by housing affordability problems. Welsh (2017) notes that researchers have found that anywhere between 60% and 70% of school moves are related to residential moves (p. 482), yet within this estimate, it is unclear what proportion were caused by housing affordability problems.

To understand the potential impact of affordability-induced residential moves that also trigger school changes, I do not review the entire body of school mobility research in detail here. Instead, I focus specifically on two studies that examined the effects of school moves that were accompanied by (and thus, presumably, caused by) residential moves, and thus can speak to the potential effects of affordability-induced school changes. The first such study was by Cordes et al. (2019), who examined the effects of combined school and residential moves on children’s academic performance. Drawing on longitudinal data on three cohorts of third graders in New York City public schools, and using both instrumental variables and fixed effects approaches, they found that, regardless of distance of move (short or longer distance), children who experienced both school and residential moves had worse outcomes. The second study, by South et al. (2007), examined the Add Health data from 1994–1995 to determine the reasons for higher rates of dropout (nearly double) among mobile students, defined as students who change both residences and schools, using multivariate regression. Similar to the previously cited literature on residential moves, they find that peer networks (specifically low-performing peers) explain most higher dropout rates for mobile students. They find few differences between mobile and nonmobile students in parent relationships or adolescent psychological well-being.

There also appears to be a concentration, effect of school mobility: Attending a school with many highly mobile students is associated with reduced student achievement, particularly for lower income students and Black and Hispanic students (Hanushek et al., 2004; South et al., 2007). While I could find no studies that have tied affordability directly to higher rates of school mobility in schools in high-rent burden neighborhoods, this is one plausible (as yet theoretical) pathway tying affordability to negative achievement.

Homelessness

High housing costs are clearly a leading cause of homelessness (Nisar et al., 2019; JCHS, 2019), and children experiencing homelessness suffer from an array of negative educational outcomes, including worse attendance, low academic achievement, and social isolation (Dhaliwal et al., 2021; Miller, 2011). According to Miller’s (2011) systematic review of the research literature on homelessness and educational outcomes, however, there has been some debate within the scholarship as to whether the outcomes are due to experiencing homelessness, or due to the low-income status that accompanies homelessness. Some studies comparing homeless youth to very low income housed youth have found few differences in outcomes. For example, Buckner et al. (2001) examined data on homeless and low-income housed single mothers in one city in Massachusetts. Using multiple regression, they found that homelessness itself did not predict student achievement beyond other risk factors that are also associated with poverty. However, it should be noted that the comparison group (housed mothers) included those families that were doubled up, and thus would have been classified as homeless by the federal McKinney-Vento Act’s more expansive definition of “homeless and highly mobile” students, which includes those children who are living in unstable and temporary housing situations (i.e., doubling up with family or friends, or living in motels or trailer parks). ¹ By contrast, Pavlakis et al. (2017) did find that homelessness independently impacted children: They examined data on students in fourth through seventh grade using OLS regression with fixed effects and found that, when compared to students who were consistently low income, students experiencing a homeless event (for 1 year) had low student math achievement growth. However, no relationship was found for chronic homelessness (two or more consecutive years).

Fantuzzo et al.’s (2012) study suggests that the residential mobility that accompanies homelessness, rather than homelessness itself, is particularly harmful to students. They conducted an analysis of data on third graders in Philadelphia using linear regression, and found that homelessness did not independently affect student achievement beyond the problem of mobility (which did); however, they did find that homelessness led to problems in both social and task engagement in school. They found that the students who experienced the greatest achievement and engagement challenges were those who experienced both homelessness and mobility.

Some studies of homelessness have used the McKinney-Vento Act’s broader definition (“homeless and highly mobile” [HHM]), which as noted previously includes those children who are doubling up with family or friends, or living in motels or trailer parks. These studies have found consistently more negative academic outcomes for children. For example, Obradovic et al. (2009) analyzed longitudinal Minneapolis Public School data on students in second through fifth grades using linear mixed models, and found that HHM students demonstrated slower rates of academic growth in reading and math compared with their low income but residentially stable peers (Obradovic et al., 2009). Similarly, Herbers et al. (2014) studied the impact of HHM status on students’ rate of academic growth in reading and math, also using longitudinal data from the Minneapolis Public Schools. Using hierarchical linear regression, they found that HHM students had lower initial levels of oral reading scores in first grade, and lower rates of growth in reading and math than their lower income housed peers. Furthermore, for HHM students, early scores were more predictive of later achievement than students in other risk groups. Similar findings were reported by Cutuli et al. (2013), who, using linear mixed models on the Minneapolis data, found that HHM students scored below their lowest income housed peers on reading and math in third grade, with gaps persisting through eighth grade. They also found significant drops in performance the immediate year after homelessness status on both reading and math.

Low et al. (2017) looked specifically at the subgroup of HHM students in one school district who were living in “doubled up” situations through data from one school district. Using linear regression, they found that doubled up youth had lower GPAs, were more likely to be truant, and were less likely to graduate on time, but no more likely to be referred for disciplinary problems. However, while the study was conducted in a district that was greater than 80% low income, the study did not explicitly control for socioeconomic status in the analysis, so the conclusions should be taken with caution.

In sum, the literature indicates that homelessness is a risk factor that puts children at greater risk for low achievement. Furthermore, homelessness and mobility together are implicated in both lower rates of achievement and lower rates of academic growth. One key limitation in the literature is that studies often fail to distinguish reasons for homelessness, and homelessness can be caused by other reasons than housing affordability challenges, including job loss, eviction for nonfinancial reasons, as well as personal factors such as divorce, discharge from foster care, release from prison, mental health problems, or a substance abuse problem (Lee et al., 2010; Nooe & Patterson, 2010). The specific effects of affordability-induced homelessness on child outcomes thus cannot be separated out.

Eviction

Several researchers have sought to understand the effects of residential mobility that is caused specifically by eviction, which is typically (although not always) caused by inability to afford rent. As rental costs have risen, eviction rates have risen as well: According to data from the Eviction Lab (2018), eviction filings rose from just over 1 million in 2000 to 2.3 million in 2016. As Desmond and Kimbro (2015) write, as the affordable housing crisis continues to worsen, “involuntary displacement from housing is likely to increase” (p. 296). Indeed, in the five states the Eviction Lab has tracked during the pandemic, there have been over 250,000 evictions, despite moratoriums on eviction imposed by the Centers for Disease Control and Prevention (Eviction Lab, 2020).

Eviction has been shown to disproportionately affect families with children: Desmond and Gershenson (2016) found in their analysis of renters in Milwaukee, using discrete hazard models of eviction, that “a renter’s likelihood of eviction increased with his or her number of children, net of socioeconomic factors, race, and rental payment history” (p. 372). They theorize that children place families at increased risk of eviction for several reasons:

Because children can cause added stress on property, disturb neighbors, and attract unwanted state scrutiny by child welfare agents or law enforcement officers, landlords may be more likely to evict large families who fall behind in rent than smaller families or adult-only households.” (pp. 372–373).

Schafft (2006) finds, through qualitative interviews with low-income mobile parents, that eviction often triggers subsequent residential instability for families. As one example, Schafft (2006) writes:

A parent and child may face eviction, lose housing, and then be put up at a motel for several weeks by the Department of Social Services (DSS). They then move in temporarily with another family member until more permanent housing can be found. These stop-gap housing situations are not only stressful for the families involved, but also almost inevitably lead to more movement. (p. 225)

While eviction studies have largely focused on adults (Desmond & Shollenberger, 2015), they do offer some insights into potential impacts of eviction on children. For example, Desmond and Kimbro (2015) examined data from the Fragile Families study on low-income mothers with young children (ages birth to 5) using propensity score analysis (propensity score weighting and nearest-neighbor analysis). They found that mothers with an eviction within the last year were more likely to report a drop in income, as well as higher levels of depression and parenting stress. They also found mothers who experienced eviction were more likely to report that they and their children were in “poor” health (Desmond & Kimbro, 2015). They note that these results are particularly concerning for child outcomes, given that the evictions occurred when children were very young (Desmond & Kimbro, 2015).

Evictions also can result in a decline in neighborhood quality for families, which can also negatively impact children as well (as discussed in more detail later in the neighborhood opportunity subsection of this review). Desmond and Shollenberger (2015) examined data on evictions in Milwaukee and, using lagged dependent variable regression, found that individuals who were evicted ended up in more disadvantaged neighborhoods (in terms of neighborhood poverty and crime rates) compared with those who moved but are not evicted. This decline in neighborhood quality is likely because many landlords do not take residents with a history of eviction, and as a result “evicted tenants are pushed to the very bottom of the rental market and often are forced to move into run-down properties in dangerous neighborhoods” (Desmond, 2012, p. 118). Many families also lose many if not most of their possessions through eviction. While these studies suggest that mobility caused by eviction could be particularly harmful to children, overall, as Desmond and Shollenberger (2015) note, there is little research on the impact of eviction on children.

Summary of the residential and school mobility pathway

The research literature finds support for the residential instability pathway as an indirect mechanism through which housing affordability struggles can influence child schooling and developmental outcomes. Studies have found affordability is associated with increased risk for residential and, consequently, school mobility, and that the disruptions associated with these types of moves (i.e., stress, transitions, etc.) negatively influence academic outcomes as well as behavioral and mental health outcomes. These effects are, some research suggests, particularly harmful for lower income students and students of color. There is some (weaker) evidence that child residential mobility also influences their mental and physical health as adults. There are, however, several significant gaps in the literature on both residential and school mobility when viewed through the lens of affordability. One gap in the literature is the lack of studies isolating the impact of residential or school mobility triggered specifically affordability problems; identifying the cause of moves is important to distinguish (i.e., residential moves for preference, for a new job, or school moves due to school choice) when seeking to understanding causal effects of moves. A second gap in the literature is on the topic of homelessness: While homelessness is the most serious consequence of affordability problems, research has yet to isolate the impact of homelessness that was specifically induced by affordability problems. A third gap in the literature centers on the impact of eviction on children specifically: While there is evidence that mobility caused by eviction can have lasting negative impacts on parents, specifically mothers, few if any studies have examined the impact of eviction on children, or how eviction might shape children’s schooling experiences.

Crowding

Affordability problems can lead multiple households or families to “double up” into a single residence to share rental costs (Lopoo & London, 2016). Crowding can affect child outcomes by creating a home environment “with constant stressors, overstimulation, and a lack of privacy” (Lopoo & London, 2016, p. 701). This can lead to lack of sleep, stress, and interpersonal conflict between parents, children, and other family members (Solari & Mare, 2012). Crowding can also strain the physical infrastructure of a dwelling through overuse. For children, crowding can also pose difficulty in finding a quiet space to study (Conley, 2001). Furthermore, crowding and doubling up can lead to the transmission of viruses, like COVID-19 (Conley, 2001; Solari & Mare, 2012).

Crowding can therefore lead to a number of problems that can impede a child’s academic development, and research has substantiated these negative effects. For example, Solari and Mare (2012) examined the impact of crowding on schooling outcomes through an analysis of PSID data using OLS regression with fixed effects. They find that a 1 SD increase in housing crowding is associated with a 2 percentile point reduction in math scores. Similar negative impacts of crowding were found by Lopoo and London (2016), who looked at the long-term effects of exposure to crowding before age 19 also through an analysis of PSID data using OLS regression with fixed effects. They found that crowding was associated with a 3 percentage point decline in the probability of high school graduation as well as a 0.19 year reduction in years of educational attainment by age 25. Conley (2001), also using PSID data, and employing OLS regression, found that children who lived in crowded conditions between 1968 and 1972 had a reduction in the years of schooling completed at age 25.

Dangerous housing conditions

High housing costs can also force families to live in unsanitary or unsafe dwellings. According to American Housing Survey data analyzed by Lew (2016) for the Joint Center for Housing Studies at Harvard University:

The lowest-income households . . . accounted for the largest share of renters reporting overcrowded conditions and physical housing problems such as toilet breakdowns, exposed electrical wiring, heating equipment breakdowns lasting six hours or more and the presence of rats in the unit. (para. 5)

Poor housing conditions have been shown to negatively impact child outcomes. Coley et al.’s (2013) study, mentioned previously, using three-level hierarchical linear modeling, found that children who resided in lower quality housing (leaking roofs, rodents, peeling paint, exposed wiring) had more emotional behavioral problems than those who lived in higher quality housing (Coley et al., 2013). They conclude that key mechanism through which these effects occur is maternal psychological stress (Coley et al., 2013).

Similarly, Kull and Coley’s (2014) study, also described previously, used multiple mediator path analysis on data from the Three City Study and found that housing problems (parent self-reports of having structural problems such as peeling paint, broken windows, rodents, etc.) predicted lower reading skills for adolescents but not younger children. Importantly, they found families who had higher cost burdens tended to live in homes with fewer structural problems (Kull & Coley, 2014), suggesting that families may voluntarily shoulder a higher burden as a tradeoff for better quality housing. This finding is consistent with other studies that have found that families who spend less than 30% of income on housing (and thus are not considered “cost burdened”) may have worse quality housing when living in expensive housing markets (Holupka & Newman, 2011).

Summary of the living environment pathway

Both crowding and dangerous housing conditions lead to worse outcomes for children: residential crowding is linked to lower math and reading achievement, and lower rates of high school graduation. Furthermore, dangerous housing conditions are linked to worse maternal mental health, as well as higher rates of child behavioral problems. However, the relationship between the level of housing cost burden and the likelihood of suffering from these problems is less than straightforward, as research has found that low-income families with low cost burdens suffer from the most substandard housing.

Neighborhood opportunity

In my search of the literature, I found just one empirical study that directly linked housing affordability (specifically housing cost burdens) to neighborhood opportunity and then to child outcomes: Kull and Coley’s (2014) longitudinal study, described previously, using data from the Three City Study. Through their multiple mediator path analysis, they found that higher housing cost burdens were associated with better neighborhood quality (as measured by the poverty rate and unemployment rate of the census tracts the families lived in). They found, overall, that neighborhood quality predicted better child behavioral outcomes as well as reading scores for children over time. They conclude, “One explanation for these findings is that the resources and opportunities afforded to children living in more advantaged neighborhoods have wide-reaching impacts and may promote positive socio-emotional functioning and cognitive skills” (Kull & Coley, 2014, p. 34).

While there are relatively few studies making the direct connection between affordability, neighborhood contexts, and child outcomes, there is evidence that affordability does shape the kinds of neighborhoods families can access, primarily through exclusionary zoning that limits lower cost housing in high-income, opportunity-rich neighborhoods (Massey & Rugh, 1997). There is, in turn, is a significant body of evidence that shows that the kind of neighborhood a child grows up in can affect child outcomes: Peer-reviewed research syntheses of the body of research on “neighborhood effects” have concluded that growing up in a racially segregated, high-poverty neighborhood negatively affects schooling and developmental outcomes (Leventhal & Brooks-Gunn, 2000; Minh et al., 2017; Sampson et al., 2002; Sastry, 2012). As Sampson et al. (2002) conclude in their research synthesis: “The range of child and adolescent outcomes associated with concentrated disadvantage is quite wide and includes infant mortality, low birthweight, teenage childbearing, dropping out of high school, child maltreatment, and adolescent delinquency” (p. 446). Research further shows that neighborhoods matter most for students who are lower income (Wodtke et al., 2016).

Arguably the most rigorous and prominent study of the impact of neighborhoods on children is the study of the federally sponsored Moving to Opportunity (MTO) demonstration program. The MTO program enrolled 4,600 low-income families living in severely distressed public housing across five cities between 1994 and 1998, and randomly assigned families to one of three groups: a control group (families who stayed in public housing); a treatment group (families who were offered a Section 8 voucher that had to be used for housing in a low-poverty neighborhood); and a “Section 8” group (families who were offered a housing voucher that could be used anywhere). The outcomes of children were tracked by an evaluation team over 15 years (Gennetian et al., 2012). The MTO evaluation was not able to estimate the effects of being provided with housing assistance (as will be discussed in the last section of this review) given that all participants received either a voucher or lived in public housing; rather, by studying outcomes of families who have been provided assistance but lived in different types of neighborhoods, the evaluation was able to document how neighborhoods matter for low-income families. Indeed, it is precisely because MTO was able to control for affordability—and the endogeneity concerns raised earlier—that it was able to draw conclusions about neighborhood effects on families. This study is significant in understanding the causal impact of neighborhoods, because, as Gennetian et al. (2012) write, the random assignment “broke the link between family preferences and neighborhood environments, and it thus provides us with the opportunity to overcome the standard self- selection concern and identify the causal effects of neighborhoods on child and youth outcomes” (p. 138).

The MTO evaluation team published several peer-reviewed studies on the impact of MTO on child outcomes, which are too numerous to review for this present synthesis. Instead, I describe the evaluation team’s 2012 peer-reviewed summation of their final analysis of the MTO longitudinal dataset on participating youth (aged 10–20 as of 2007; Gennetian et al., 2012). Analyzing their data using intent-to-treat and treatment-on-treated estimation with OLS regression, the evaluation team found (somewhat surprisingly) that participation in MTO—specifically moving from very poor to less poor neighborhoods—had no impact on either reading or math achievement (as measured by the Early Childhood Longitudinal Study-Kindergarten Cohort [ECLS-K] as well as supplemental questions from the National Educational Longitudinal Survey–1988 [NELS] assessment for high school students). They also found that participation in MTO also had few long-term effects on educational attainment or employment outcomes, or on child physical health, mental health, or risky behavior, although female children did experience some improvements in mental health and reductions in risky behavior.

There are several hypothesized reasons for the general lack of neighborhood effects found by the MTO evaluation. One is that over time, families tended to move back into poor neighborhoods. Indeed, the experimental group families were restricted to living in a low-poverty neighborhood for only 1 year (Edin et al., 2012): After 1 year the voucher became portable. As a result, many families ended up leaving their original units due to a variety of reasons (landlord interference, substandard housing, discomfort in neighborhoods). Indeed, as Ludwig (2012) writes, the experimental group initially (1 year after participation) lived in census tracts that were 14 to 17 percentage points lower in poverty rates than control; but this diminished to a 10 percentage points difference 5 years after baseline, and then diminished again to 5 percentage points 10 years after baseline.

A second reason for the lack of stronger neighborhood impacts is the mediating factor of schools: Even though children ended up in (slightly) lower poverty neighborhoods, children still attended highly segregated schools by race and poverty, with very low levels of academic performance (as measured by test scores), which likely contributed to few differences in academic outcomes between the groups. Thus, while neighborhoods may “matter” to an extent, it may be the combination of neighborhood and school quality that may matter most for child outcomes.

A recent longer-term follow up study by Chetty et al. (2016) followed MTO participants through adulthood. This study is distinct from prior work in that the researchers were able to follow children over a longer period, and they also distinguished children who were young at the start of the experiment (below age 13) from those who were older at the start of the experiment (age 13 or older). They found, using intent-to-treat and treatment-on-treated estimation, that moving to a low-poverty neighborhood before age 13 had large and significant positive impacts on both educational attainment and earnings. By contrast, they found no effects of moving at an older age (i.e., for children 13 or older at the time of random assignment), possibly due to the disruption caused by a move in adolescent years. They conclude that, because the greatest benefits occurred for children who moved when they were youngest, that “every extra year of exposure to a low-poverty environment during childhood is beneficial” (p. 858).

In sum, Chetty et al.’s (2016) research does suggest neighborhoods (particularly neighborhood poverty rates) have an impact on longer term educational outcomes, primarily for young children (under age 13). The final impacts evaluation by Gennetian et al. (2012) also illustrates that obtaining housing in more advantaged neighborhoods does not always lead to access to more advantaged schools, which can potentially moderate any positive impacts of neighborhoods in the shorter term. The impacts of school advantage on child outcomes is explored in more detail, below.

School opportunity

Because affordable rental housing is disproportionately located in lower income neighborhoods, affordability problems can limit low-income children to segregated, high-poverty schools that are unequal on a number of dimensions that affect student achievement (Ihlanfeldt & Mayock, 2019; Owens & Candipan, 2019; see also Rothwell, 2012). ² In this section, therefore, I first review the research on the impact of school segregation by race and class on students and then review research on the impact of several school quality indicators.

School segregation

As noted, one key way in which affordability may “matter” to child outcomes is by restricting families to housing in economically and racially isolated neighborhoods, which are, in turn, zoned to high poverty and racially isolated schools. While the literature on school segregation is too vast to systematically review in its entirety here, as with the neighborhood effects research, I draw on peer-reviewed syntheses of literature on school segregation in the United States published between 2000 and 2020. In my search of the literature, I identified three such syntheses (Mickelson & Bottia, 2010; Mickelson et al., 2013; Reardon & Owens, 2014).

These syntheses find, as a collective, that racially and economically integrated schools are beneficial to student achievement. Mickelson and Bottia (2010) examined the relationship between school racial and socioeconomic diversity and mathematics outcomes through an analysis of 59 studies published between 1990 and 2009 through narrative analysis (which as the authors note, “present a qualitative, holistic interpretation of the findings in the summarized relevant literature”; p. 1018), and vote counting (tallying the number of studies that report positive, negative, or null effects). They conclude that “mathematics outcomes are likely to be higher for students from all grade levels, racial, and SES backgrounds who attend racially and socioeconomically integrated schools” (p. 994). In a later analysis, Mickelson et al. (2013) conducted a meta-regression analysis of 25 studies published between 1990 and 2010 on the impact of racial composition on mathematics outcomes. They conclude that “minority concentration has a small statistically significant negative association with mathematics outcomes” (p. 142). They also find that the effect of racial composition increases as students move from elementary to high school.

Reardon and Owens (2014) find in their review of research that school integration has a positive impact on educational attainment for African American students, reduces racial disparities in achievement between racial and ethnic groups, and has long-term benefits in terms of income and employment for African Americans. They conclude that there is compelling evidence that socioeconomic composition of schools also impacts student achievement, as well.

School quality

Research has consistently shown that high-poverty and racially isolated schools are unequal in quality on a number of important dimensions that can impact student achievement (Owens & Candipan, 2019) which I review here (although this is not an exhaustive review). One such dimension is teacher experience: Research has consistently shown that schools with majorities of low-income students and students of color have higher proportions of inexperienced teachers (Goldhaber et al., 2015; Goldhaber et al., 2019; Lankford et al., 2002).Teacher experience has been demonstrated to impact student achievement: Rockoff (2004), for example, analyzed data from two New Jersey school districts using fixed effects regression and found a 0.17 SD difference in reading scores between students of beginning as compared to veteran teachers (with 10+ years of teaching experience). Rivkin et al. (2005) analyzed three cohorts of fourth graders in Texas through seventh grade and, using fixed effects regression, found that the impact of teachers on students improves over teachers’ first several years of teaching, a finding also supported by Buddin and Zamarro (2009) in their analysis of data on students in the Los Angeles Unified School District. ³

A second dimension is rates of teacher turnover: High-poverty schools have been consistently shown to have higher rates of teacher turnover, both in terms of annual as well as longitudinal turnover rates (Holme et al., 2018; Loeb et al., 2005). One of the few studies to identify the causal impact of teacher turnover on student achievement was conducted by Ronfeldt et al. (2012). Analyzing data on New York City elementary school students, and using fixed effects regression, they find that high rates of teacher turnover in a grade negatively affects the math and ELA for all students in that grade, not only the students who lost their teachers. These negative impacts are particularly strong in schools with higher concentrations of low-achieving and Black students.

A third dimension is school resources: Because school funding is often tied to local property tax revenue, higher income districts with more expensive housing can often fund their schools at a higher level (C. Turner et al., 2016). While there have been ongoing debates about the role of financial resources in student outcomes (Owens & Candipan, 2019), a recent causal study by Jackson et al. (2016) showed that school funding did have long-term impacts on students. They analyzed PSID data on children born in 1955–1985 through 2011, which they augmented with education funding data, using two-stage least squares difference in differences regression with fixed effects. They found that school financial resources were related to positive long-term outcomes in both income and educational attainment for low-income children.

These inequalities often culminate in differences in overall school performance, which can also affect students. Research has shown that schools in low-income neighborhoods have lower test scores, on average, than schools in higher income neighborhoods (Owens & Candipan, 2019). There are relatively few studies of the causal impact of attending a low-scoring school on students, in part because of the difficulty accounting for student self-selection into schools. One study was able to overcome these limitations through the use of quasi-experimental methods: Allensworth et al. (2017) analyzed a longitudinal data set from Chicago, and found, using propensity score matching and controlling for prior achievement and demographic characteristics, that there were positive benefits for students attending high-performing (high-ACT scoring and high-graduation rate) nonselective schools. They specifically found that students attending high-performing schools had increased levels of achievement (as measured by ACT scores) and were more likely to enroll in college and persist for at least 2 years (Allensworth et al., 2017).

Summary of the neighborhood and school opportunity pathway

Research indicates that housing affordability problems can limit families’ housing options to high-poverty, racially segregated neighborhoods, and that such neighborhoods have long-term impacts on both earnings and educational attainment. These neighborhoods are often zoned to lower income and segregated schools, which suffer from significant disparities on a number of dimensions of school quality, which in turn can depress academic achievement and attainment.

Expenditures on children

A study by Newman and Holupka (2014) indicates that parents with housing affordability problems reduce enrichment expenditures on their children. They examined the impact of housing cost burdens on expenditures on children, including necessities (food, health care) and enrichment (child care, school fees, toys, musical instruments, etc.). Analyzing the Bureau of Labor Statistics Consumer Expenditure Survey data on low-income families using propensity score matching, they found that housing costs were unrelated to parental spending on basics. However, they did find that there was an “inverted U-shaped relationship between enrichment expenditures and housing cost burden, indicating that these expenditures are lowest when the fraction of income spent on housing is either very high or very low” (Newman & Holupka, 2014, p. 98).

While there are few studies on the impact of enrichment expenditures on child outcomes, a study by Yeung et al. (2002) did find a positive relationship: Analyzing PSID and its Child Development Supplement data using structural equation modeling (SEM), they specifically found that the presence of enrichment materials in the home (toys, games, books, instruments) was linked to improved academic outcomes (Yeung et al., 2002; see also Kornrich, 2016). Furthermore, as Duncan and Murnane (2016) argue, extra financial resources also can buy access to enrichment activities such as summer camps, music lessons, and travel, which collectively may contribute to gaps in vocabulary and background knowledge between high and low income children.

Although Newman and Holupka (2014) did not find a connection between affordability challenges and parent expenditures on basic needs, other research has documented a connection. According to data from the 2018 American Consumer Expenditure Survey, low-income families with housing cost burdens spent 38% less on food than low income families that were not cost burdened (JCHS, 2020, p. 30). Research has, in turn, established clear links between food insecurity and child outcomes. A systematic review of the research by Shankar et al. (2017) found that, for school aged children (ages 6–13), food insecurity is linked to health problems, behavior problems, academic problems, and greater risk for anxiety and depression, as well as suspension from school (Shankar et al., 2017).

Parental mental health

When parents have difficulty affording their rent and/or struggle to meet other basic needs due to high rent costs (i.e., clothing, food, health care), they can experience significant stress, and this stress can negatively affect their children (Harkness & Newman, 2005). There is indeed evidence that affordability is potentially linked to higher levels of parental stress: The study by Coley et al. (2013), discussed previously, which analyzed data from the Three City Study using multiple mediator path analysis, found that one of the mechanisms through which mobility affected child outcomes (defined as emotional and behavioral functioning, and reading and math skills as measured by Woodcock-Johnson tests) was through higher rates of maternal distress related to the moves. Similarly, Desmond and Kimbro (2015), noted earlier, also noted that mothers who were evicted suffered from higher rates of depression, even years after their moves.

I did identify one study that directly examined the connection between affordability, parental stress, and child development outcomes: Warren and Font (2015) looked specifically at the influence of affordability struggles on the risk of maternal child neglect or abuse, analyzing data from the Fragile Families study on parents with children at ages 3 and 5, using nested linear probability and SEM. They found that housing affordability problems (defined as paying more than 50% of income toward housing costs) were associated with higher rates of maternal stress, but not a higher risk of child abuse and neglect. They did find that housing instability did increase risk for neglect and abuse; thus, to the extent to which housing affordability drives instability, they conclude that this can be an indirect trigger of neglect and abuse.

These findings on increased maternal stress due to affordability problems have implications for child academic and social development. Crum and Moreland (2017) found, in their analysis of data on young children collected as part of a parent/child intervention in Indianapolis (using SEM), that parental stress is linked to the development of problematic behaviors in children, as well as anxiety, emotional problems, and poor social coping. Similar negative effects of stress were found in Mistry et al.’s (2002) study, which analyzed cross-sectional data on low-income parents with children participating in the random-assignment New Hope experiment in Milwaukee that provided income and other supports to low-income adults who worked a minimum number of hours per week. Using latent variable SEM, they found that economic strain was associated with parent psychological distress and that distressed parents reported lower efficacy in parenting children. Distressed parents were also observed as displaying less warmth and affection in interactions with children, and teachers, in turn, reported such children as having more problem behaviors.

Another potential source of stress on parents linked to affordability problems is the strain caused by the accumulation of debt. Research suggests that housing affordability leads parents to try and cover living expenses through the accumulation of “unsecured” debt (credit cards, payday loans, etc.). This has been particularly true during the COVID-19 pandemic: According to data from one large online rent payment company, there was a 43% increase in the number of people paying their rent with credit cards in the first half of 2020 (Zego, 2020). This type of unsecured debt has been linked to worse child outcomes. In two different studies, Berger and Houle (2016, 2019) examined the association between debt and child emotional and behavioral outcomes via the National Longitudinal Study of Youth. Using different methodological approaches (OLS with fixed effects and hierarchical linear modeling, respectively) they found that higher levels of unsecured debt (credit card debt, money owed to businesses, banks or individuals, and/or medical debt), which is the type of debt most likely to accrue to low-income families struggling with paying rent, is associated with child behavior problems, particularly for low socioeconomic status and Black children. They hypothesize that the underlying reason for this relationship is the higher level of parental stress caused by unsecured debt: “Unsecured debt may induce stress, anxiety, or other adverse indicators of psychosocial functioning for parents, each of which is associated with poorer-quality parental behaviors, which are, in turn, negatively associated with child well-being” (Berger & Houle, 2016, p. 7).

Summary of the parental resources pathway

Overall, research indicates that financial strains on parents caused by housing affordability problems can affect child outcomes by reducing parents’ expenditures on basic needs and child enrichment, both of which are linked to academic achievement, behavior and mental health. Furthermore, research suggests that parents experiencing problems with affordability may have higher levels of stress and that such stress can negatively impact children. Affordability problems can also lead parents to take on higher levels of unsecured debt, which is also linked with child behavioral problems. However, more research is needed to better understand these connections.