Toward Understanding Homelessness: The 2007 National Symposium on Homelessness Research. Incarceration and Homelessness. Reentry Programs and Housing Outcomes:  What Do We Know?


The past decade has seen the development of new and innovative reintegration programs for returning prisoners, and funding for researchers and practitioners to examine the effectiveness of programs designed to facilitate community reintegration (and, as such, residential stability). Facilitating community reintegration often addresses the barriers to housing that were discussed in the previous subsection, but usually does not directly focus on preventing homelessness among persons reentering the community. New initiatives in this area are typically geared toward establishing comprehensive services. However, while these programs often include some type of housing assistance or housing referral, they rarely provide actual housing to returning prisoners. There has also been a body of research and evaluation that has followed this programming, but there is a paucity of research that examines homelessness related to reentry initiatives.

The existing research on these outcomes generally falls into three categories. First and most basic are needs assessments based on surveys and tracking of persons upon release from incarceration. A second category of this research tracks the use of services in homeless, criminal justice, and related public systems through matching administrative data. Finally, there are studies focusing on long-term outcomes measurement for persons in supportive housing and other housing/residential placements. These include a number of outcomes such as tenancy, homelessness, and recidivism. Each of these categories will now be briefly reviewed.

Needs Assessments/Surveys/Tracking

Most research relating homelessness to incarceration is based on cross-sectional studies or needs assessments that survey homeless individuals, individuals in prison, or those recently released from prison. A number of studies in this vein are also designed to track individuals over a period time. These studies, which typically do not contain comparison groups, are correlational in their examination of the relationship between homelessness and incarceration. The Returning Home study, conducted by the Urban Institute, provides a good example of how a large, multisite descriptive study of returning prisoners can shed light on important policy issues related to the needs of the formerly incarcerated. Returning Home documents the pathways of prisoner reintegration, examines what factors contribute to a successful or unsuccessful reentry experience, and identifies how those factors can inform policy. The Returning Home study has been implemented in four states, including a pilot study in Maryland (Visher, La Vigne & Travis, 2004) and full studies in Illinois (La Vigne et al., 2003), Ohio (La Vigne & Thomson, 2003), and Texas (Watson et al., 2003). The goal in each state was to collect information on individuals life circumstances immediately prior to, during, and up to one year after their release. At each site, Returning Home grouped the challenges of reentry along five dimensions: individual, family, peer, community, and state (Urban Institute, 2005). Returning Home survey instruments include a number of questions related to housing and homelessness.

Use of Services and Corresponding Cost Offsets/Cost Effectiveness

This approach primarily uses administrative data, and links incarceration records, shelter records, and other program records through personal identifiers such as name and social security number. Such studies are retrospective, and use data contained in administrative databases collected by various service providers, often large public entities such as city and state corrections, welfare, and mental health systems, for purposes of recordkeeping and coordination of care. Research using this design can establish a sequential history of services use, typically prior and subsequent to a particular intervention such as housing. And while this method provides a practical means of gathering data on large numbers of subjects, data matching across systems can be cumbersome, time-intensive, and expensive (Culhane & Metraux, 1997).

Administrative data can provide more accurate and detailed information on services use than any other means of data collection, and cost figures can be attached to these records of services use to provide estimates of resources expended and financial impacts of particular interventions such as housing. Culhane, Metraux, and Hadleys (2002) study of the cost offsets associated with providing housing and support services for persons who had extensive histories of homelessness and who were diagnosed with severe mental illness is the best known of a small body of research that ties costs to outcomes for housing programs serving persons with mental illness. Among their findings is that, in the New York City program that they examined, a supportive housing placement is associated with an annualized reduction of $12,145 in shelter, health, mental health, and corrections services. Such a reduction in service-related costs would recoup 94 percent of the cost of the housing and services. This suggests that housing stability can lead to a reduction in service utilization among the homeless mentally ill population, mostly by saving public resources through collateral reductions in services consumed by the erstwhile homeless individuals in other services systems.

With groups that are heavy users of criminal justice and homeless services, or who receive an early release to community-based housing, the case for fiscal savings, to go along with other less tangible benefits, is a promising approach to evaluating the effectiveness of housing and other homelessness prevention programs. The costs involved with maintaining people in prisons and jails have a median per diem cost of $60 and $70 (Lewin Group, 2004), respectively. These costs are high enough that an economic argument could be made for community-based interventions such as specialized reentry housing if such housing could be shown to reduce or avoid subsequent incarceration episodes. Culhane, Metraux and Hadley (2004) included jail and prison costs in their cost study and found reduced rates of incarceration after housing placement (as compared to a matched control group). However the overall levels of incarceration were low enough so that these reductions in incarceration made only a modest contribution to the overall cost savings associated with the supportive housing placements.

Long-Term Outcome Evaluation

One key limitation of the studies of services use and cost offsets using administrative data is that they rely on data already collected through other sources  data that are not collected for research purposes. In that context research is limited to examining, post hoc, questions that can be answered by the available data. An alternative to this is to build systematic and rigorous data collection procedures into housing programs. This permits data to be collected on an ongoing basis as part of providing services. Another, more expensive alternative is to incorporate larger, long-term evaluation components when developing new programming or when program outcomes seem promising. Long-term outcome or impact evaluations would involve examining the effects of programs in reducing recidivism and residential instability.

These studies usually track outcomes more than one year after release from incarceration. Although rigorous outcome or impact evaluations for reentry housing programs are scarce, a few jurisdictions have been building outcome data into their programs for a number of years. Maryland and California, for example, have developed statewide housing programs that serve persons with mental illness being released from prisons and jails.

Marylands Shelter Plus Care program, operating in 21 counties, provides tenant and sponsor-based rental assistance to persons with serious mental illness coming from jails. Case management and supportive services are provided. Outcomes tracked by the State of Maryland demonstrate that recidivism to jails is less than 7 percent. Only 1 percent entered hospitals and only 1 percent were homeless during the evaluation period (SAMHSA, 2003).

In 1999, California passed Assembly Bill 34 (AB 34) that created pilot programs in three counties that provided a range of services, including housing support, for persons diagnosed with mental illness and who are either homeless or recently released from jail or prison. Based on positive early results of the pilot programs, in 2000 the state legislature expanded AB 34 services through Assembly Bill 2034 by providing $55 million for implementation of 40 programs in 31 additional counties across the state. Today, AB 2034 programs serve more than 4,500 individuals in California.

Evaluation findings based on AB34 and AB 2034 suggest that the provision of housing to persons who have mental illness and are justice-involved can enhance residential stability and increase successful community integration (Burt & Anderson, 2005; Mayberg, 2003). Research findings also indicate that programs that serve the most challenging clients  clients with longer histories of homelessness and incarceration  produce similar housing outcomes as those programs that serve less challenging clients (Burt & Anderson, 2005). Essentially, the evaluation demonstrates that people with serious mental illness and histories of arrest or incarceration can achieve housing stability with adequate support.

More specifically, in their analysis of housing outcomes for currently enrolled clients who had been enrolled in an AB 2034 program for 24 months, Burt and Anderson (2005) found that of those who were homeless at enrollment (43 percent of all consumers), 35 percent were in permanent housing at 3 months and 66 percent were in permanent housing at 24 months. Another 16 percent were in semi-dependent/structured living settings at 24 months. Overall, the findings also show that the most successful AB 2034 programs were utilizing multiple housing strategies, ranging from partnering with housing providers and landlords to securing housing units (Burt & Anderson, 2005).

As a result of both the recent implementation of most reentry programming and the limited direct focus on homelessness prevention as a specific goal for these programs, outcomes measures related to reducing homelessness is currently lagging. As more detailed databases emerge that focus on specific aspects of housing, services, and participant performance, research based on these long-term evaluation models can explore more nuanced questions that go beyond basic outcomes of recidivism and homelessness and offer a more complex picture of factors that are key to facilitating the reentry process.

Limitations of Research on and Evaluations of Reentry Programs

Numerous challenges encumber the extant research assessing the effectiveness of housing-related programs for the formerly incarcerated. Most basically, there are problems with logistics related to the research itself. Rigorous experimental designs  including the use of comparison groups (randomly assigned or otherwise)  are rare in this research literature. There are also temporal restrictions of the research, where few studies examine outcomes beyond one year. Next, there is substantial variability among the outcome variables examined across studies. The numerous combinations of variables unique to each study render any comparison of housing models difficult. For example, some studies focus on a reduction in recidivism to the exclusion of any reductions in homelessness and vice versa. In another example, studies comparing types of supervision or service structure rarely control for specific attributes of the housing setting, such as structure type and number of units, or criminal justice history of the consumer. There are also problems of fidelity in that a common housing approach may manifest itself in different ways under different programs and circumstances. Foremost among the challenges here is the lack of any housing model specific to providing housing to persons who have been incarcerated. Due to this and other issues related to specificity, there is often difficulty in generalizing research findings from one program to others.

View full report


"report.pdf" (pdf, 242.57Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®