Quantifying the Restrictiveness of Local Housing Authority Policies Toward People With Criminal Justice Histories: United States, 2009–2018

Abstract

Objectives. To quantify variation in the restrictiveness of local public housing authority policies related to the admission and eviction of people with criminal justice histories.

Methods. We conducted content analysis of housing authority policy documents for US cities with a population of 100 000 or more (n = 152). Factor analysis identified policy provisions to create a restrictiveness score (range = 0–8). We explored associations between restrictiveness scores and city-level measures of racial/ethnic diversity, racial/ethnic neighborhood segregation, ideology, and public housing scarcity.

Results. Eight policy provisions, 6 relating to consideration of mitigating circumstances, explained 71.0% of the variance in housing authority policy provisions related to criminal justice histories. We observed small but significant positive associations between restrictiveness scores and racial/ethnic diversity (r = 0.22) and neighborhood segregation (r = 0.18). There was no correlation between restrictiveness scores of housing authorities within the same state (intraclass correlation = 0.0002).

Conclusions. Housing authority policies vary substantially regarding the circumstances under which people with criminal justice histories can obtain and retain public housing. Exposure to constellations of policy provisions that might institutionalize health inequities and increase health risk among people with criminal justice histories can be quantified through a systematic process.

Incarceration rates in the United States are higher than in any other country in the world and have a disproportionate impact on Black Americans.¹ In 2016, Blacks were 5.9 times more likely than were Whites to be incarcerated.² Considerable research documents associations between incarceration and poor health,^3–5 and, not surprisingly given the disproportionately high rates of incarceration among Blacks, a growing literature attributes at least some of the magnitude of racial health disparities to mass incarceration.^6–9 While the impacts of incarceration on health are well-established, less clear are the mechanisms through which these impacts occur. Some of these impacts likely result from the ways that having a criminal justice history shapes access to critical social determinants of health.¹⁰

Individuals face numerous barriers to reintegration after incarceration, and housing poses a particular challenge.^11–14 Upon leaving prison, they simultaneously face financial challenges—45% of re-entrants do not earn any income in the first year after incarceration¹⁵—and a severe affordable housing crisis. Across the United States, fair market rents have increased faster than wages such that there is currently no state where full-time minimum-wage work is sufficient to rent an unsubsidized fair-market 2-bedroom unit.¹⁶ Government rental assistance programs are one of the few sources of affordable housing available to low-income renters,^16,17 particularly for Blacks—who are the head of household (i.e., leaseholder) of approximately 44% of the public housing units in the United States.¹⁸ Recent research suggests positive effects of rental assistance on psychological well-being, overall self-rated health, and access to health care.^19–21 However, public policies impose restrictions on the ability of people with criminal justice histories to access rental housing assistance and achieve housing security.^22–26

Given the well-established effect of housing on mental, physical, and behavioral health outcomes²⁷ and emerging evidence regarding the health benefits of rental assistance programs, specifically,^19,20 policies that affect access to rental assistance are likely to have significant health implications for people with criminal justice histories. Furthermore, given that Blacks are disproportionately likely to have been incarcerated,² housing policies that affect access to rental assistance based on criminal involvement are likely to exacerbate health disparities.⁹ The measurement of these policies and their implications for health equity are the focus of this study.

LOCAL HOUSING POLICIES AS STRUCTURAL MEDIATORS

Public housing units—subsidized residences managed by local public housing authorities with federal funding from the Department of Housing and Urban Development (HUD)—are home to approximately 1.2 million low-income households in the United States. While federal statute (24 CFR §982.553) specifies some circumstances in which people with criminal justice histories are restricted from public housing (e.g., 3-year ban following eviction from public housing for drug-related criminal activity), local housing authorities have tremendous discretion when setting policy to determine who gets admitted to public housing and the circumstances under which they can be evicted.^{22,24–26,28} These local housing authority policies are codified as provisions in housing authorities’ Admissions and Continued Occupancy Policies (ACOPs), which are typically hundreds of pages in length. Previous analyses of ACOPs have shown that most local housing authorities impose restrictions on people with criminal justice histories that are more restrictive than the federal minimum.^22–25 For example, this was demonstrated by Curtis et al. in a 2013 analysis of the ACOPs of 40 state and local housing authorities²² and by a 2015 Shriver Center report that analyzed more than 300 public housing policy documents, including ACOPs.²⁴

Recent HUD memos have encouraged local housing authorities to be cautious when setting policy related to how criminal justice histories are used, given the disparate impact of these policies on Blacks and, therefore, the potential for these policies to violate the Fair Housing Act.²⁵ A 2015 HUD memo promulgated a rule that arrest is not sufficient evidence to prove that someone engaged in criminal activity and, thus, should not be used to trigger application denial or eviction from public housing (Appendix A, available as a supplement to the online version of this article at http://www.ajph.org). However, there are no mechanisms for HUD to monitor or enforce the rule.²⁸

Although no studies have assessed associations between ACOP policy provisions and health outcomes, Figure 1 offers an empirically informed diagram of the pathways through which these local housing policies could contribute to health disparities. ACOP provisions related to criminal justice involvement are conceptualized as a structural mediator of the relationship between criminal justice policies that have a disparate impact on Blacks (e.g., stop and frisk, structural determinant) and housing insecurity (proximal outcome) and poor health outcomes (distal outcome). Housing insecurity can lead to poor health in many ways, such as by increasing exposure to physically, socially, and sexually unsafe living situations; fragmenting social capital and support; producing chronic stress; and limiting time and space needed to manage chronic diseases.^19–21,27 Empirically documenting the impacts of ACOPs on health—in addition to other “invisible punishments” embedded in policies that extend the reach of the criminal justice system and have a disparate impact on Blacks²⁹—requires precise measurement of the policy exposure. Standard methods of policy exposure measurement are unlikely to adequately achieve this.

FIGURE 1—

Diagram of One Pathway Through Which Local Public Housing Authority Policies Could Mediate the Association Between Criminal Justice Policies and Health Disparities

In policy impact studies, the independent variable is typically a single policy measured as a dichotomous variable.^30,31 Although this approach is appropriate for measuring exposure to macro policies that have strong and independent influences on health, it is less well-suited for measuring exposure to constellations of micro policy provisions that, in aggregate, might increase health risk. As Tremper et al. suggest, such policy provisions might be better “measured at the ordinal or interval level, thereby enabling dose–response analyses, possibly enhancing statistical power to detect the law’s effects.”^30(p253) However, few policy impact studies in general, and no studies of public housing authority policies in particular, have measured policy exposure as a continuous variable.

STUDY PURPOSE

The primary objective of this study was to quantify variation in the restrictiveness of local public housing authority policies related to the admission and eviction of people with criminal justice histories. More broadly, the study sought to develop a methodological approach that can be used to quantify exposure to clusters of policy provisions that are likely to increase the health risk of people with criminal justice histories and, in turn, produce health inequities and perpetuate health disparities.

METHODS

We conducted a quantitative content analysis of ACOPs. We used factor analysis to identify a small number of ACOP policy provisions that captured restrictiveness toward people with criminal justice histories and developed a scoring procedure to quantify restrictiveness.

Sample Frame and Identification of Policy Documents

We used US Census data to identify all cities with 2015 population of 100 000 or more persons. These 302 cities were under the jurisdiction of 265 housing authorities, 57 of which did not manage any public housing units. This resulted in a sample frame of 208 housing authorities. We conducted Internet searches to identify ACOPs on these housing authority Web sites. When ACOPs could not be identified, housing authorities were e-mailed 4 times and called 3 times. We identified ACOPs for 152 (73%) housing authorities, which were in 40 states (range of ACOPs per state = 1–17). ACOP year ranged from 2009 (2 ACOPs) to 2018 (16 ACOPs) and more than half (78 ACOPs) were from 2016 or 2017.

Codebook and Coding

We carried out codebook development and coding in accordance with recommendations for conducting content analysis of policy documents.^30–32 Informed by research and scholarship about public housing authority admission and eviction policies,^22–26 we created preliminary coding categories and definitions. Five members of the project team then independently read 5 identical ACOPs from a geographically diverse sample of housing authorities, the categories were revised through discussion, and a coding instrument was created in Qualtrics. This process was then repeated 2 more times and a complete draft of the codebook was created.

Three coders then viewed a 1-hour, Web-based presentation about the project, reviewed the codebook, and participated in a 3-hour in-person training and facilitated discussion. Following this, the 3 coders independently coded identical ACOPs until sufficient interrater agreement statistics were achieved (mean agreement = 81%). Finally, the 3 coders jointly coded 15 ACOPs to ensure consistency in coding decisions and then independently coded the remaining ACOPs.

Analysis

We downloaded the coded data set from Qualtrics and imported it into R (R Foundation, Vienna, Austria) for analysis. We transformed nominal data into dichotomous, ordinal, and continuous variables to characterize policy provisions. We produced univariate statistics to describe the prevalence of policy provisions. We first investigated the underlying factor structure of 16 policy provisions (Table 1) through exploratory factor analysis. We then verified the factor structure through confirmatory factor analysis (details in Appendix B, available as a supplement to the online version of this article at http://www.ajph.org).

TABLE 1—

Variation in Local Public Housing Authority Policy Provisions Related to the Admission and Eviction of People With Criminal Justice Histories, Codified in Admission and Continued Occupancy Policies (ACOPs): United States, 2009–2018

ACOP Policy Provision	Prevalence, % (Restrictiveness Scale Value)	Conceptualization of Variable
Admission
Drug-related criminal activity lookback period		Longer lookback period is more restrictive because more time needs to pass before a person is re-eligible for public housing (i.e., the duration of exposure of the risk factor for housing insecurity is greater when the lookback period is longer).
3 y or not specified	46.7 (0)
4–5 y	43.4 (1)
≥ 6 y	9.9 (2)
Violent criminal activity lookback period		Longer lookback period is more restrictive because more time needs to pass before a person is re-eligible for public housing (i.e., the duration of exposure of the risk factor for housing insecurity is greater when the lookback period is longer).
3 y or not specified	46.1 (0)
4–5 y	41.1 (1)
≥ 6 y	11.8 (2)
Drug-related eviction from public housing lookback period		Longer lookback period is more restrictive because more time needs to pass before a person is re-eligible for public housing (i.e., the duration of exposure of the risk factor for housing insecurity is greater when the lookback period is longer).
3 y or not specified	65.8 (0)
4–5 y	25.7 (1)
≥ 6 y	8.6 (2)
Index date for criminal activity lookback perioda		Release from prison or end of supervision is most restrictive because it is the date farthest from the date of the crime and results in the most time that needs to pass before a person is re-eligible for public housing (i.e., the duration of exposure of the risk factor for housing insecurity is greater when the lookback period is longer).
Crime, arrest, or charge	11.2 (0)
Conviction or not specified	82.2 (1)
Release from prison or end of supervision	6.6 (2)
No. of events that explicitly have the ability to trigger denialb		Higher number of events is more restrictive because the probability of an applicant not being eligible for public housing (i.e., exposure to risk factor for housing insecurity and its health consequences) increases with the addition of each event.
Mean	5.06
Median	5
Range	0–9
Arrests or charges explicitly given less weight than conviction		Absence of policy provision is more restrictive because a person can be wrongly arrested or charged and be found not guilty of the crime that would be grounds for denial of admission (2015 HUD policy memo stated that arrest should not be used to trigger application denial for public housing).
Yes	51.3 (0)
No	48.7 (1)
Mitigating circumstances explicitly considered		Absence of policy provision is more restrictive because explicit mention of this type of mitigating circumstance provides the opportunity to admit individuals and families that may otherwise violate admission eligibility criteria related to criminal activity.
Yes	86.2 (0)
No	13.8 (1)
Circumstances related to nature of the violation explicitly considered as mitigating circumstance		Absence of policy provision is more restrictive because explicit mention of this type of mitigating circumstance provides the opportunity to admit individuals and families that may otherwise violate admission eligibility criteria related to criminal activity.
Yes	50.7 (0)
No	49.3 (1)
Impact on family explicitly considered as mitigating circumstance		Absence of policy provision is more restrictive because explicit mention of this type of mitigating circumstance provides the opportunity to admit individuals and families that may otherwise violate admission eligibility criteria related to criminal activity.
Yes	40.8 (0)
No	59.2 (1)
Eviction
Family is explicitly permitted to remove member for any criminal or drug use activity		Absence of policy provision is more restrictive because explicit mention of this allowance provides the opportunity to admit individuals and families that may otherwise violate conditions related to criminal activity.
Yes	50.0 (0)
No	50.0 (1)
Mitigating circumstances explicitly considered		Absence of policy provision is more restrictive because explicit mention of this allowance provides the opportunity to not evict individuals and families that may otherwise violate conditions related to criminal activity.
Yes	52.6 (0)
No	47.4 (1)
Circumstances related to nature of the violation explicitly considered as mitigating circumstance		Absence of policy provision is more restrictive because explicit mention of this allowance provides the opportunity to not evict individuals and families that may otherwise violate conditions related to criminal activity.
Yes	38.2 (0)
No	61.8 (1)
Impact on family explicitly considered as mitigating circumstance		Absence of policy provision is more restrictive because explicit mention of this allowance provides the opportunity to not evict individuals and families that may otherwise violate conditions related to criminal activity.
Yes	34.9 (0)
No	65.1 (1)
Proof of good tenancy explicitly considered as mitigating circumstance		Absence of policy provision is more restrictive because explicit mention of this allowance provides the opportunity to not evict individuals and families that may otherwise violate conditions related to criminal activity.
Yes	27.0 (0)
No	73.0 (1)
Arrest is explicitly grounds for eviction		Absence of policy provision is less restrictive because a person can be wrongly arrested and not be guilty of engaging in criminal activity (2015 HUD policy memo stated that arrest should not be used to trigger eviction from public housing).
Yes	36.8 (1)
No	63.2 (0)
Testimony from rehabilitation centers can explicitly be used as evidence for eviction		Absence of policy provision is less restrictive because the provision enables the eviction of people with substance use disorders who are receiving treatment and might be no longer engaged in drug-related criminal activity.
Yes	32.9 (1)
No	67.1 (0)

Note. HUD = US Department of Housing and Urban Development. The sample size was n = 152.

^aACOPs that did not explicitly indicate the index date were coded as “conviction” because this is analogous to date of eviction being used as the index date for drug-related eviction from public housing according to federal law.

^bScored as the sum of 9 possible events.

Restrictiveness scale development.

For each policy provision, we assigned a numeric value to each coding option in which the least restrictive option was coded as 0 and the most restrictive option was coded as 1. For example, if an ACOP stated that arrests or charges were given less weight in admissions decisions than convictions, the policy provision was scored as 0. If the ACOP did not state that arrests or charges were given less weight, the provision was scored as 1. We calculated intraclass correlation (ICC) coefficient to examine whether there was a significant association between restrictiveness scores for local housing authorities in the same state.

Associations between restrictive scores and city-level factors.

We produced Spearman correlation coefficients to explore associations between each city’s restrictiveness score and city-level measures. We assessed measures of city racial/ethnic diversity and racial/ethnic neighborhood segregation because they are indicators of sociodemographic context and integration. We obtained these measures from City Health Dashboard,³³ which generated estimates with American Community Survey data. Diversity was operationalized as an index in which 0 represents a city’s population comprising only 1 racial/ethnic group (minimal diversity) and 100 represents a city’s population comprising an equal proportion of all racial/ethnic groups (maximum diversity). Racial/ethnic neighborhood segregation is as an index in which 0 represents every census tract in a city being perfectly representative of the city’s overall racial/ethnic composition (absolute integration) and 100 represents every census tract comprising only 1 racial/ethnic group (absolute segregation). We assessed city ideology because it is an indicator of the sociopolitical climate in which housing policies are developed and implemented. We obtained ideology data from the American Ideology Project.³⁴ We assessed waitlist time for public housing (in months) and the percentage of public housing units occupied in each city by using HUD data¹⁸ because they are indicators of public housing scarcity, and ACOP restrictions could serve as a strategy to ration a finite resource.

RESULTS

Table 1 shows the prevalence of 16 policy provisions that relate to restrictions toward people with criminal justice histories obtaining and retaining public housing. In the admission domain, 3 policy provisions related to “lookback periods”—the amount of time that needs to elapse after a person is involved in criminal justice activity to be re-eligible for public housing. Longer lookback periods are more restrictive because more time needs to pass before a person is eligible. While 46.7% of housing authorities imposed a lookback period of 3 years or less for drug-related criminal activity, which is similar to the federal lookback period for drug-related eviction from public housing, 43.4% imposed 4- or 5-year lookback periods, and 9.9% looked back 6 years or more.

Almost half (48.7%) of the housing authorities did not explicitly state that arrests or charges were given less weight than convictions when defining what was considered proof of criminal activity. The inclusion of an explicit statement is an indicator of an ACOP being less restrictive toward people with criminal justice histories because it demonstrates commitment to implementation of the 2015 HUD rule that bans the practice. There was no significant difference (P = .99) in the prevalence of this policy provision in ACOPs published before and after promulgation of the rule.

Housing authorities also varied regarding consideration of mitigating circumstances when making admissions decisions for applicants with criminal justice histories. About half (50.7%) of ACOPs stated that mitigating circumstances related to the nature of criminal activity (e.g., seriousness) would be considered. A slightly smaller proportion (40.8%) stated that factors related to the impact of the admission decision on the applicant’s family would be considered as a mitigating circumstance. Explicit mention of these mitigating circumstances contributes to policies being less restrictive because they codify that housing authority officials have the ability to admit individuals who would otherwise be denied admission.^24,25

In the eviction domain, just half (50.0%) of ACOPs stated that a family was permitted to remove a person who engaged in criminal activity from the lease to avoid eviction. The absence of this policy provision is an indicator of restrictiveness because an entire family can be evicted because 1 family member engaged, or was suspected to have engaged, in criminal activity. More than one third (36.8%) of ACOPs explicitly stated that arrest was grounds for eviction, which is a direct violation of the 2015 HUD rule, and there was no significant difference (P = .99) in the prevalence of this policy provision in ACOPs published before and after promulgation of the rule. Only 52.6% of ACOPs stated that mitigating circumstances would be considered when making eviction decisions, compared with 86.2% for admissions decisions.

Factor analysis resulted in a single factor, comprising 8 policy provisions (Table 2), which explained 71.0% of the total variance (α = 0.88). The mean restrictiveness score was 4.07 (SD = 2.78). Table 3 shows the distribution of city ACOP restrictiveness scores by quartile rank. The most frequent restrictiveness scores were zero (32 ACOPs), 6 (29 ACOPs), and 7 (26 ACOPs). The ICC for restrictiveness scores for housing authorities in the same state was not significant (ICC = 0.0002; P > .99).

TABLE 2—

Eight Policy Provisions That Characterize the Restrictiveness of Local Housing Authority Policies Toward People With Criminal Justice Histories, Codified in Admission and Continued Occupancy Policies (ACOPs): United States, 2009–2018

ACOP Policy Provision	Factor Loadinga
Arrests or charges explicitly given less weight than conviction (admissions)	0.68
Mitigating circumstances explicitly considered (admissions)	0.65
Family is explicitly permitted to remove member for any criminal or drug use activity (eviction)	0.78
Mitigating circumstances explicitly considered (eviction)	0.87
Circumstances related to nature of the violation explicitly considered as mitigating circumstance (admissions)	0.93
Impact on family explicitly considered as mitigating circumstance (admission)	0.92
Proof of good tenancy explicitly considered as mitigating circumstance (eviction)	0.97
Impact on family explicitly considered as mitigating circumstance (eviction)	0.99

Note. The sample size was n = 76.

^aResults of confirmatory factor analysis. Factor loadings based on a single factor solution.

TABLE 3—

Distribution of Scores for Restrictiveness of Local Housing Authority Policies Toward People With Criminal Justice Histories, Codified in Admission and Continued Occupancy Policies (ACOPs): United States, 2009–2018

Quartile	No. of Cities	ACOP Score Range
1	41	0–1
2	46	2–5
3	29	6
4	36	7–8

Note. 1st quartile = least restrictive; 4th quartile = most restrictive. ACOP restrictiveness scores and quartile ranks for all cities in sample (n = 152) provided in Appendix C (available as a supplement to the online version of this article at http://www.ajph.org).

There were small but significant positive correlations between restrictiveness scores and city racial/ethnic diversity (r = 0.22; P = .007) as well as racial/ethnic neighborhood segregation (r = 0.18; P = .02), wherein greater restrictiveness scores were associated with more diversity and more segregation. Restrictiveness scores were not significantly associated with city ideology (r = −0.07; P = .37) nor waitlist time for public housing (r = 0.007; P = .93) or the percentage of public housing units occupied (r = −0.0008; P = .99).

DISCUSSION

Consistent with previous research,^22–25 we found that most housing authority policies are more restrictive toward people with criminal justice histories—in terms of lookback periods of crimes that can trigger admission denial or eviction—than required by federal law. Given that stable housing is a critical social determinant of health,^10,27 and given the disproportionate representation of Blacks in the criminal justice system,^1,2 these policies could contribute to health inequities.^6–8

The current study extends previous research by demonstrating how ACOP policy provisions cluster together and by identifying a small number of provisions that explain a substantial proportion of the variance in housing authority policy related to criminal justice involvement. The restrictiveness scoring methodology developed in this study can be used in future research to test associations between housing authority policy restrictiveness and health outcomes among people with criminal justice histories. More broadly, this approach could serve as a template for quantifying exposure to constellations of policy provisions in other areas (e.g., employment policy, financial lending policy) that influence access to health-promoting resources among people with criminal justice histories.

Although small in magnitude, ACOP restrictiveness scores were positively and significantly associated with the covariates of city-level racial/ethnic diversity and racial/ethnic neighborhood segregation, and post hoc analysis revealed that these 2 covariates were slightly associated with each other (r = 0.16; P = .06). These findings raise questions about the dynamics through which the racial/ethnic climate of a city might influence policy design or vice versa. For example, it is plausible that a city policy environment conducive to the design of ACOPs that impose broad restrictions on people with criminal justice histories (who are disproportionately Black²) would also be conducive to policy decisions that foster racial/ethnic neighborhood segregation. It is also plausible that ACOP restrictions toward people with criminal justice histories influence patterns of racial/ethnic segregation within a city. While speculative, the study findings highlight areas for future research about how other city-level factors (e.g., crime rate, housing affordability) are associated with public housing policies.

Six of the 8 ACOP policy provisions retained after factor analysis related to explicit mention of mitigating circumstances under which people with criminal justice histories could potentially obtain and retain housing. Although housing authority officials can still take mitigating circumstances into consideration if the ACOP does not explicitly state their ability to do so, the inclusion of these provisions might be indicators of a broader housing authority culture that believes that criminal justice histories should not translate into persistent social disadvantage. While such a culture could improve access to public housing among people with criminal justice histories and the health benefits that stable housing confers, there are potential consequences of codifying discretion in this way.

An ACOP policy architecture structured around discretion likely results in inconsistent admission and eviction decisions as housing authority officials make decisions on a case-by-case basis. This unpredictability could have implications for people with criminal justice histories and their health. Qualitative research suggests that the discretion held by housing authority officials can prompt some people with criminal justice histories to engage in a labor-intensive process to make a case for why they are worthy of tenancy.²⁶ These processes can be stressful, have taxing effects on health and relationships, and also internalize the stigma of criminal justice system involvement.³⁵ For others with criminal justice histories, inconsistent housing authority admissions decisions can result in beliefs that policies are more restrictive than they actually are and deter housing applications among those who are eligible and in need.²⁶ The codification of discretion might also allow the individual biases of housing authority officials (e.g., racial prejudices) to have substantial influence on who obtains and retains housing.

That discretion is central to ACOP policies related to criminal justice system involvement is further implied by the finding that there was no correlation between restrictiveness scores of housing authorities in the same state. This suggests that local housing authorities—and potentially the locally appointed Boards of Commissioners to which they report—exercise autonomy when setting policy and do not simply adopt the same policies as their counterparts in neighboring jurisdictions. Relatedly, that housing authority restrictiveness scores were not associated with measures of public housing scarcity suggests that ACOP restrictions toward people with criminal justice histories are not primarily being used to ration the distribution of a scarce resource.

The publication of the HUD rule prohibiting the use of arrest records in admission or eviction decisions was not associated with the presence of policy provisions related to the use of arrest records. This raises questions about the extent to which housing authorities are aware of, and take seriously, the rule. Because housing authorities are not required to restate federal rules in their ACOPs, the absence of an explicit statement that arrests are not considered in admission or eviction decisions should not be interpreted as meaning that a housing authority definitively considers arrests (although more than one third of ACOPs did explicitly state that arrest was grounds for eviction, a direct violation of the rule). While limited inferences can be drawn for this reason, the inclusion of a statement that arrests are not considered can be conceptualized as an indicator of an ACOP being inclusive toward people with criminal justice histories because it signals that a housing authority is committed to ensuring that the HUD rule is implemented in practice. As recommended in a 2018 Government Accountability Office report, HUD should update its Public Housing Occupancy Guidebook (last updated in 2003) to provide clear guidance for housing authorities to update their ACOPs on this and other rule changes.²⁸

Limitations

The study had limitations. First, it focused on policies “on the books,” which are not necessarily the same as how policies are implemented in practice.³¹ Housing authority officials presumably vary in how they interpret ACOPs and exercise discretion when making admissions and eviction decisions.²⁶ Validation of the restrictiveness scale would enable it to serve as a better indicator of policy in practice. However, it would be very challenging to validate the scale because doing so would require detailed information about the frequency of, and reasons for, public housing admission denials and evictions across cities. These data are not available.¹⁴ Thus, the scale should be interpreted as more of an indicator of policy climate than measure of policy in practice.

Second, despite a rigorous search protocol, ACOPs were identified for 73% of the housing authorities in the sample frame. It is possible that housing authorities that did not post their ACOPs online or respond to requests are systematically different than housing authorities for which ACOPs were obtained. Third, analyses gave equal weight to all policy provisions and did not account for differences in the magnitude of impact that different policy provisions might have.

Conclusions

Housing authority policies vary substantially regarding the circumstances under which people with criminal justice histories can obtain and retain public housing. Exposure to constellations of micro policy provisions related to criminal justice histories can be quantified through a systematic process. Such quantification can make it possible to examine the extent to which, and mechanisms through which, policies affect health outcomes and health inequities.