The Effects of the ACA Medicaid Expansion on Nationwide Home Evictions and Eviction-Court Initiations: United States, 2000–2016

Abstract

Objectives. To evaluate the effect of the Affordable Care Act (ACA) Medicaid expansions on national rates of home eviction and eviction initiation in the United States.

Methods. Using nationally representative administrative data from The Eviction Lab at Princeton University, we estimated the effects of the ACA Medicaid expansions on county-level evictions and filings from 2000 to 2016 with a difference-in-difference regression design.

Results. We found that Medicaid expansions were associated with an annual reduction in the rate of evictions by 1.15 per 1000 renter-occupied households (P < .001), a reduction of 1.59 eviction filings per 1000 renter-occupied households (P < .001), and a reduction in the average number of evictions by 46 (P < .05). We found additional evidence that increasing rates of African American residents in a county was associated with a greater rate of evictions filed, and increased rates of poverty and rent burdens relative to income were associated with more evictions both filed and completed.

Conclusions. Evictions decreased after Medicaid expansion, demonstrating further evidence of the substantive financial protections afforded by this coverage. The reduction in the eviction filing rate suggests that Medicaid expansion could be reducing evictions by preventing the court proceeding entirely.

One in 8 poor renting families in the United States faced rents they could not afford as of 2013, and similar numbers reported that they may soon be evicted.¹ The majority of poor renting families in the United States now allocate more than half of their income to housing, far exceeding the 30% rent burden deemed affordable by the National Housing Act (Pub L No. 73-479).¹ In this housing market, eviction has become commonplace among low-income populations. However, little is known about the immediate causes of eviction. Past research has shown that impoverished neighborhoods and those with a greater rent burden face an especially acute risk.^2,3 Evictions are particularly frequent in Black neighborhoods, in part because of the concentration of Black families in segregated, disadvantaged communities.⁴

Evictions can have devastating consequences for the individuals and families involved. Evictions increase the likelihood of homelessness, contribute to the perpetuation and worsening of poverty, and are linked to deteriorating health outcomes. Eviction hinders attempts to secure new housing as landlords often deny potential tenants with a history of eviction.^5,6 These processes can drive families into subpar housing conditions and into high-crime and underresourced neighborhoods, creating further instability for children and adults alike.² As a whole, evictions are associated with an increased risk of a mental health problem, including depression, anxiety, and even suicide, and raise the risk of worse physical health outcomes for both parents and their children.^7–11

Public health practitioners and medical care providers are renewing consideration and investment in housing as an important social determinant of health. The social determinants of health, those that lie outside of the individual’s choices or biological makeup, are understood to determine 75% of a person’s health outcomes, in part by shaping one’s ability to engage in healthy behaviors or comply with medical advice.¹² As a social determinant of health, housing can affect health through multiple pathways. Inadequate housing can worsen health through conditions such as overcrowding, peeling paint, pests in the home, and mold.^13–16 Unstable housing also hinders health. Individuals with unstable housing are less likely to have a usual source of care and are more likely to postpone needed medical care, use the emergency department, or have a hospitalization. Studies show that housing-assistance programs improve health and psychological well-being, increase medication adherence, and improve health care access.^7,17

We investigated whether the expansion of Medicaid coverage under the Patient Protection and Affordable Care Act (ACA; Pub L No. 111-148) reduced the risk of home eviction, thereby stemming the tide of devastating health and social consequences that would follow. The ACA enabled states to extend eligibility for public health care coverage to all adults with incomes up to 138% of the federal poverty level in complying states. Millions of low-income adults gained Medicaid coverage under the expansions, which provide health care access and financial protections from unaffordable health care expenses. In addition to improving health outcomes, the state expansions have been demonstrated to improve financial well-being.¹⁸ Medicaid expansion has been shown to reduce poverty, payday loan borrowing, and medical debt.^19–23

By covering medical expenditures and improving financial well-being, state Medicaid expansions under the ACA could be associated with a decrease in the likelihood of eviction. Tenants’ improved ability to make rent payments may affect the course of the proceedings at 2 intervals. First, the eviction might never be initiated in court. If tenants are better able to remain current on their rental payments after gaining health care coverage, they may be less likely to face the threat of eviction. Second, the eviction may be resolved during the court process. A formal eviction proceeding begins when a landlord files an eviction in court, requesting a hearing to be scheduled. The process usually entails a number of notices to the tenant, which provide the opportunity to stop the eviction process by paying back any owed rent before the hearing. Once filed with the courts, the eviction proceeding can be halted either through a landlord–tenant agreement or through a court judgment as to the sufficiency of payment. Tenants with newly obtained health care coverage may have a greater financial capacity to reach a favorable resolution.

In this study, we analyzed the effect of Medicaid expansion on outcomes at both intervals of the eviction process and the frequency of eviction filings initiated in court as well as the frequency of completed evictions. Specifically, we evaluated the rate of evictions per 1000 renter-occupied households, the rate of eviction filings initiated in court per 1000 renter-occupied households, and counts of the total number of evictions and number of eviction filings. We hypothesized decreases in both completed evictions and in evictions initiations filed in court as a result of Medicaid expansion.

METHODS

In this study, we used publicly available data provided by The Eviction Lab at Princeton University.²⁴ It is the most comprehensive data set to date on evictions in the United States.²⁵ These data collate directly collected county court records as well as data sets of eviction records purchased from LexisNexis Risk Solutions and American Information Research Services Inc.²⁵ The data are presented as county-by-year observations, spanning 2000 to 2016. In total, we observed nearly 83 million eviction records throughout this time period. In our data, we observed 30 states (including the District of Columbia) that expanded Medicaid by January 2016 and 21 states that did not expand Medicaid.

We examined the impact of the ACA Medicaid expansion on evictions by using 4 specifications of the dependent variable. We evaluated the rate of eviction per 1000 renter-occupied households to assess the expansion–eviction relationship while directly accounting for population size of renter-occupied households. Similarly, we assessed the rate of eviction filings initiated per 1000 renter-occupied households. Eviction filings capture evictions initiated by the landlord in court that may not result in a completed eviction taking place. Finally, we examined the total number of evictions and the total number of eviction filings as count variables, which do not directly adjust for population size. With each specification, we evaluated the eviction outcome annually in each county.

We used a difference-in-difference research design that compared the average change in evictions before and after Medicaid expansion in expansion states to the average change in evictions in nonexpansion states over the same time period. Our primary identification strategy assumed that, absent Medicaid expansion, the average number and rate of evictions would have changed similarly between expansion and nonexpansion states. We tested this assumption in the Appendix (available as a supplement to the online version of this article at http://www.ajph.org), and found similar eviction trends over the time before expansion.

In our main models, we employed year fixed effects, county fixed effects, county population size, and county-specific linear time trends to control for potentially confounding county characteristics or co-occurring time trends that could have an impact on eviction outcomes. County fixed effects represent 3144 unique counties. Year fixed effects represent 17 unique years. County population size controls for the effects of larger or smaller populations on the likelihood of eviction outcomes. Although Medicaid policy is implemented at the state level, we used county-level outcomes and controls because eviction records are generated at county-level courts and because the economic and social trends that may have an impact on evictions, including employment rates and population density, occur at the more local level. The analytical specifications for the model are available in the Appendix (Section A1).

We additionally examined a set of extended models in which we analyzed the effects of demographic covariates shown in the existing literature to predict or increase the risk of incurring an eviction. We used the fixed-effects difference-in-difference models described previously and added county-level measures of the poverty rate, the share African American, and the rent burden relative to income (specified as the county median rent-to-income ratio). Each demographic measure was provided in the Eviction Lab data and was originally from the US Census.²⁵ As in the primary models, the extended models incorporated fixed-effect controls for independent socioeconomic trends while exploring how these demographic characteristics relate to our eviction outcomes. The analytical specifications for the model are available in the Appendix (Section A2). We assessed the parallel trends assumption underlying our difference-in-difference research design, and found no significant difference between expansion and nonexpansion states in trends of eviction outcomes across the time period before policy implementation (Appendix, Section A3).

RESULTS

We found that the ACA Medicaid expansion was associated with a reduction in the rate of home evictions per 1000 renter-occupied households in states across the United States, as well as a reduction in the rate of eviction filings per 1000 renter-occupied households. We observed a visible divergence in the rate of eviction after 2014, the year when most states implemented Medicaid expansion (Figure 1). We saw that in nonexpansion states there was an increase in the eviction filings in 2014 that did not occur in expansion states, indicating a relative decrease in the rate of eviction filings associated with Medicaid expansion (Figure 2). The trends in the eviction rate and eviction filing rate were comparable between expansion and nonexpansion states before ACA implementation, with eviction rates trending in the same directions during the 2002 to 2013 prepolicy period (Figures 1 and 2). In 2013, the rate of home evictions in both expansion and nonexpansion states was approximately 17 evictions per 1000 renter-occupied households. By 2016, the eviction rate fell to fewer than 14 in expansion states while rising slightly in nonexpansion states. The eviction filing rate increased in nonexpansion states from 36 in 2013 to 37 filings per 1000 renter-occupied households. By contrast, the eviction filing rate decreased over this same period in expansion states, from 27 to 26 evictions filed per 1000 renter-occupied households.

FIGURE 1—

Trends in Rate of Evictions per 1000 Renter-Occupied Households in Medicaid-Expansion and -Nonexpansion States: United States, 2002–2016

FIGURE 2—

Trends in Rate of Eviction Filings per 1000 Renter-Occupied Households in Medicaid-Expansion and -Nonexpansion States: United States, 2002–2016

The results in Table 1 show precise estimates of the effect of the ACA Medicaid expansion on eviction outcomes, with year and county fixed effects, county-specific linear time trends, and a covariate controlling for county population size. We saw that, nationwide, the Medicaid expansions were associated with an annual reduction of 1.16 evictions per 1000 renter-occupied households (P < .01) relative to a pre-ACA baseline of approximately 17 evictions per 1000 renter-occupied households in 2010 (Table 1). This reduction is concurrent with the visible divergence we see in the eviction rate between expansion and nonexpansion states from 2013 to 2016 (Figure 1). Moreover, we found that the ACA Medicaid expansion was associated with a reduction of 1.66 eviction filings initiated per 1000 renter-occupied households (P < .01). In addition, we evaluated the effect of expansion on the total counts of evictions and eviction filings annually by county, which does not adjust directly for population size. We found a significant reduction in the total number of evictions by 49 (P < .05), but we did not find a significant effect of Medicaid expansion on the average number of eviction filings (Table 1).

TABLE 1—

Effect of Medicaid Expansion on Number and Rate of Home Evictions and Eviction Filings Nationwide: United States, 2000–2016

		Average Difference After Medicaid Expansion (vs Before Medicaid Expansion)
	Overall No.	Rate (SE; P)	No. (SE; P)
Evictions	41 339	−1.16 (0.39; .003)	−49.14 (22.64; .030)
Eviction filings	44 168	−1.65 (0.53; .002)	−52.87 (33.01; .11)

Note. The sample consists of county-by-year observations. Rates are per 1000 renter-occupied households. Robust standard errors are clustered on county. Model includes year fixed effects, county fixed effects, county-specific linear time trends, and a control for county population size.

Table 2 presents the results of our model with additional covariates. We found similar effect magnitudes on evictions outcomes when we included demographic predictors associated in the literature with increased eviction risk, suggesting the expansion–eviction findings were not driven by these other phenomena (Table 2). Our findings with respect to these demographic characteristics were less consistent across the 4 models than our primary results regarding Medicaid expansion. The percentage of the county that is African American was associated with significantly higher rate of eviction filings initiated (P < .05) but was not significantly associated with an increase in the completed eviction count or rate (Table 2). Higher county poverty rate and rent burden were associated with increased counts of evictions and eviction filings, but not with increased rates of evictions or eviction filings per 1000 renter-occupied households.

TABLE 2—

Extended Model of the Effect of Medicaid Expansion on Evictions: Evaluation of Poverty, Rent Burden, and Neighborhood Racial Characteristics, United States, 2000–2016

	Average Difference After Medicaid Expansion (vs Before Medicaid Expansion)
	Evictions, Rate (SE; P)	Eviction Filings, Rate (SE; P)	Evictions, No. (SE; P)	Eviction Filings, No. (SE; P)
Overall	−1.14 (0.39; .004)	−1.55 (0.53; .003)	−45.38 (22.64; .032)	−45.13 (33.12; .12)
% African American	0.24 (0.15; .11)	0.41 (0.21; .049)	2.51 (4.07; .54)	11.53 (6.68; .09)
Rent burdena	0.01 (0.03; .78)	0.03 (0.05; .52)	1.39 (0.70; .047)	1.64 (0.82; .047)
Poverty rate	0.02 (0.04; .64)	0.02 (0.05; .72)	2.41 (0.99; .015)	6.03 (1.93; .002)
R2	0.869	0.937	0.952	0.979
N	41 329	44 158	41 329	44 158

Note. The sample consists of county-by-year observations. Rates are per 1000 renter-occupied households. Percent African American, rent burden, and poverty rate are presented at the county level. Robust standard errors are clustered on county. Model includes year fixed effects, county fixed effects, county-specific linear time trends, and a control for county population size.

^aRent burden reflects the county median rent-to-income ratio.

DISCUSSION

The visual representation of eviction rates per 1000 renter-occupied households illustrates a clear divergence by Medicaid expansion status at the time of the policy’s enactment, with expansion states seeing lower rates compared with nonexpansion states (Figure 1). Furthermore, we found significant reductions in the rates of evictions, eviction filings, and the number of evictions in our fixed-effect models (Table 1). Our estimated reduction in the rate of eviction filings implies that Medicaid expansion may reduce evictions by preventing the process from ever being initiated. While we did estimate a statistically significant effect of expansion on the rate of evictions filed per renting household, we did not find a significant effect on the total count of evictions filed annually. For both completed evictions and eviction filings initiated, count measures were less precisely estimated, exhibiting larger standard errors. This result suggests that standardizing the outcome by the number of renter-occupied households may be an important component of interpreting eviction outcomes. Moreover, our results do not appear driven by any observable community characteristics. Point estimates of the effect of expansion on the rate of evictions and eviction filings were affected very little by the addition of demographic covariates.

We found that increases in a community’s share of African American residents were associated with higher rates of eviction filings, even after we controlled for poverty rate and rent burden (Table 2). Moreover, the magnitude of the racial effect was substantial: the effect of increasing the percentage of African Americans in a county on all eviction outcomes exceeded the effects of increasing the poverty rate or rent burden in a county (Table 2). Previous research has indicated that inner-city Black neighborhoods are especially affected by evictions, with almost half of all evictions in Desmond’s seminal article occurring in Black neighborhoods.⁴ Findings of the causal relationship between race and evictions have been mixed. Predominantly Black neighborhoods have higher rates of eviction, but Black individuals may not be any more likely to experience an eviction conditional on their other individual characteristics.^2,4

Increases in the county poverty rate and rent burden were both associated with larger numbers of evictions and initiations filed but were not associated with higher rates of these outcomes (Table 2). These findings suggest that neighborhoods with increasing shares of low-income residents or with growing rent burdens incur larger numbers of evictions and eviction filings, but these large numbers are not disproportionately higher per renter-occupied household. This pattern could be attributable to the fact that rent burden is often higher in more urban counties where increased numbers of evictions and eviction filings may occur among low-income residents but not be widely shared by the total population of renters, which often includes more affluent renting households.

Although we cannot directly assess which features of Medicaid expansion account for the lower eviction rates, our theoretical model and existing evidence point to 3 likely pathways. First, improved financial well-being resulting from decreased financial exposure to health care costs and medical expenditures may increase timely rent payments, thereby reducing eviction proceedings at both the filing and eviction stages. Second, access to health-improving medical care may alleviate poor health as a trigger for housing-related economic hardships.²⁶ Finally, improved health resulting from health care access may increase employment and earnings outcomes, thereby increasing financial stability and reducing evictions. Future research should examine the pathways by which gaining health care coverage may reduce evictions. Further research should also examine reasons behind the disparities in African American counties facing higher rates of eviction initiations.

Limitations

This study has several limitations. Our study could not account for any reporting differences or additional demographic changes that may have occurred during the timeframe that could have had an impact on the reported outcomes. Nevertheless, we have no reason to expect that these changes would occur differentially between expansion and nonexpansion states in accordance with the expansion timeline. We do not have individual insurance status and therefore could not assess the precise effects of gaining Medicaid on individual eviction outcomes. In addition, The Eviction Lab database cannot capture informal evictions that occur outside of the formal court proceedings process and may not include all eviction records. Some data limitations occur from the localized process of eviction records keeping, leading to low count estimates in some counties. These data collection difficulties may lead to less-precise estimates.

Public Health Implications

Housing is a critical public health issue.²⁷ Housing instability increases the risk of worsened physical and mental health outcomes and creates barriers to securing care and adhering to treatment guidelines. Eviction sparks an acute housing crisis for families, causing instability in the immediate term and introducing long-term housing challenges. Evicted families are more likely to accept unsafe and inadequate housing conditions because of damaged credit and rental histories and a heightened need to secure immediate shelter, leading to both acute and long-term risk of worsened health outcomes.

This study adds to the growing body of literature on the financial security afforded by health insurance. Such coverage affords a level of financial security that reduces eviction risk, increases housing stability, and improves health outcomes. This study provides evidence of the far-reaching consequences of Medicaid expansion, demonstrating that expansion reduced evictions both by preventing the initial filing and by reducing completed evictions. This evidence further demonstrates the program’s ability to improve well-being through housing stability among high-need individuals and families.