Toward Understanding Homelessness:
The 2007 National Symposium on Homelessness Research

Housing Models

Gretchen Locke, MA, Abt Associates Inc., Cambridge, MA
Jill Khadduri, PhD, Abt Associates Inc., Bethesda, MD
Ann O’Hara, Technical Assistance Collaborative, Boston, MA

[ Main Page of Report | Contents of Report ]

This report is available on the Internet at:

How to Obtain a Printed Copy

Disclaimer: This paper was developed for the National Symposium on Homelessness Research held on March 1-2, 2007. The Symposium was conducted by Abt Associates Inc. and Policy Research Associates Inc. under contract for the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services; and the Office of Policy, Development, and Research, U.S. Department of Housing and Urban Development. The paper presents the views and opinions of the respective author(s) and does not necessarily represent the views, positions, and policies of the federal government.




This paper provides an overview of current housing and service models for programs serving people who are homeless and synthesizes the research on the efficacy of each model, what we know about which models work for whom, and the implications for preventing and ending homelessness. The authors begin with background on housing, poverty, and homelessness, including a discussion of changes in the policy and program context within which programs for homeless people operate that have affected housing models since the late 1990s. They then review the recent literature — both descriptions of program models and research on outcomes — focusing first on housing models for families and then on housing models for single individuals with disabilities. Finally, the authors suggest implications for preventing or ending homelessness and directions for future research.

[ Go to Contents ]


Housing is related to homelessness both as a cause and as a solution. Some families and individuals become homeless explicitly because of a housing crisis related to their extreme poverty and a lack of available housing at rents they can afford. For example, a family cannot pay the rent, is evicted, and can find no alternative housing or an individual is released from an institutional setting and cannot find affordable housing. Others are precariously housed with friends or relatives and lose their shelter in a crisis that involves issues such as mental or physical health or domestic violence. Still others have complex service needs and may spend years on the streets or cycling in and out of the shelter system for reasons that have little to do with their housing options.

Housing and Poverty

A well-documented shortage of affordable housing for the poorest American households — those with incomes below 30 percent of area median income, or roughly the poverty level — contributes to the flow of people into homelessness. The U.S. Department of Housing and Urban Development’s (HUD) “worst case needs” reports show that there are millions of families and individuals homelessness who have low incomes, have no public subsidy to help them with their housing costs, and are paying more than half their incomes for rent (HUD 2005).[1] Quigley and colleagues have shown that increases in over the past two decades are largely the result of increasing income inequality and a related increase in demand for low-cost housing (Quigley& Raphael, 2000; Mansur et al., 2002). Recent U.S. Census data cited by the Center on Budget and Policy Priorities indicate that more than 8 million households with incomes below 80 percent of the local median pay more than 50 percent of their incomes for rent. The number of such rent-burdened households has increased by 33 percent since 2000.[2] For some families and individuals, a severe rent burden is a temporary situation related to a short-term loss of income, but for many others it represents an untenable situation that can end in homelessness.

A recent Welfare-to-Work study evaluated the Housing Choice Voucher Program, the largest mainstream rental housing subsidy program, randomly assigning some welfare families[3] to receive housing assistance and others to a control group that did not receive assistance. Among those not using housing assistance, 12.5 percent reported that they had been literally homeless during the previous 12 months — that is, living on the streets or in a shelter — and 45 percent reported that they had at some point during the year been living temporarily with relatives or friends (Mills et al., 2006).

Regardless of the path taken to homelessness, the ultimate goal for every homeless individual and family is safe, affordable, and permanent housing. A system of housing and services for people experiencing homelessness has evolved to place people who become homeless into permanent supportive housing, to provide temporary emergency shelter to people who are homeless, and to provide time-limited housing to help people make the transition from homelessness to permanent housing. Services may include case management, mental health services, substance abuse treatment, or employment support to help them find and retain housing.

The topic of this paper is the various housing and service models that comprise programs for homeless people. At the same time, mainstream housing assistance programs have at least as important a role to play as the homeless service system in helping people to end their homelessness. Quigley and colleagues conclude that modest efforts to improve the availability and affordability of rental housing could substantially reduce homelessness in many communities (Quigley& Raphael, 2000; Mansur et al., 2002). Not surprisingly, in the Welfare to Work study cited above, using a housing choice voucher dramatically reduced both literal homelessness and the pattern of housing instability sometimes known as “couch surfing” (Mills et al., 2006).[4]

Almost 5 million units of federally subsidized rental housing reduce rent to 30 percent of a household’s income. These units can be used to help people exit homelessness, and sometimes they are, with documented success (Shinn et al., 2001). There is fierce competition for the limited subsidy slots from low-income people who are not homeless, however, as housing assistance is not an entitlement but instead is rationed through waiting lists (Khadduri & Kaul, 2005). The assisted housing stock has come under pressure in several ways over the past five years. First, the supply has been reduced because owners of roughly 300,000 privately owned, subsidized units have chosen to leave the subsidy program at the end of their contracts, have been subjected to foreclosure, or have come under enforcement action by HUD. In addition, there have been losses in the public housing stock because of deterioration or redevelopment. Finally, reductions in the number of households assisted by the voucher program has increased demand for public housing and project-based units from eligible households who might otherwise have received a tenant-based voucher.

Some formerly homeless families and individuals can rent mainstream housing without a subsidy because they are able to work and find jobs that lift them out of poverty, or because they live in communities where rental housing is relatively inexpensive, or both. However, many formerly homeless families and individuals do not gain the ability to pay for unsubsidized housing as part of their exit from homelessness. The shortfall of mainstream subsidized rental housing limits the ability of the homeless services system to achieve the goal of ending homelessness.

Housing Targeted to People Who Are Homeless

Within the homeless services system, three broad types of housing are targeted specifically to homeless families and individuals: emergency shelter, transitional housing, and permanent supportive housing. These three types — and specific models within each type — differ in their physical configuration, the expected tenure of the clients housed, and the degree of choice clients have in selecting where they will live. Each of these types also has its own set of funding streams. While a discussion of the financial models for developing and operating housing for homeless people is beyond the scope of this paper, it is important to note that the features of the different housing types may flow directly from the type of funding available.

Emergency shelters provide overnight shelter, often in a congregate setting. Some may be open during the day, as well. Services vary from minimal information and referral assistance to more intensive case management. Clients have little choice in the terms or conditions of a shelter stay, and the physical facilities may be less than ideal, especially for children. Transitional housing offers longer term but time-limited housing (typically 6 to 24 months), often in single household units or in smaller congregate settings with more intensive services. Clients may have some choice in where they live, depending on the scale of the program. Permanent supportive housing may be offered in these same physical configurations. It is targeted to persons with disabilities and offers intensive services on or off site, either by the same provider that operates the housing or through partnerships with community-based service providers. The level of choice about where to live depends on the program.

The approach to services varies among housing models. Services may be voluntary or required, on site or off site, intensive or limited — irrespective of the physical configuration, tenure conditions, or choice of location offered by the housing with which they are associated. The services offered may include housing search assistance, case management, support for finding and keeping a job, transportation assistance, mental health services, and substance abuse treatment. 

Housing and service models for programs serving people who are homeless have become more diverse since the 1998 National Symposium on Homelessness Research. Three papers prepared for the 1998 Symposium addressed aspects of housing models. One paper described approaches to emergency shelter, while another reviewed transitional housing strategies (Feins & Fosburg, 1999; Barrow & Zimmer, 1999). A third paper addressed the broader issue of reconnecting homeless individuals and families to the community, including approaches to fostering residential stability as well as employability and social connections (Rog & Holupka, 1999).

At the time, emergency shelter was viewed as an important first step in moving homeless people — especially families — to stable housing. At the same time, there was increasing recognition that not just shelter, but also services, were needed to help with that transition. The paper on transitional housing described the ambivalence of the policy and practitioner communities toward transitional housing. Its proponents argued that it was the best way to ensure homeless families and individuals received the services they needed to secure and maintain permanent housing. Detractors said it could be stigmatizing and ineffective if there was no next-step housing available at the end of the transitional program. The paper on reconnecting people who have been homeless with the community examined what was known about outcomes with respect to the different housing types and emphasized the importance of stable housing as a prerequisite to reconnecting to employment and social relationships. The paper also reviewed the substantial barriers to effective interventions.

As of the late 1990s, research on housing models and services was limited and inconclusive. Since then, housing and service strategies have evolved, and research and practice have delineated more sharply both the housing and the services components of housing models for homeless people. The key questions that have emerged are:

  1. How quickly and how successfully do homeless families and individuals move to permanent housing?
  2. Are supportive services voluntary or required, and does this make a difference in retention in the program and, ultimately, in housing success?
  3. How independent is the permanent housing; that is, is it a private apartment or group setting? Are others who live there also program clients? Is there on-site or off-site support? What role do these features play in retention and success?

The rest of this section describes the changes in the policy and program context within which programs for homeless people operate and how these changes have affected housing models since the late 1990s.

Changes in Context Since 1998

Changes in the design and resources of mainstream programs that serve low-income people have had a substantial influence on the evolution of housing models for homeless people during the past decade. At the same time, priorities and program emphases for funding streams targeted specifically to preventing or ending homelessness, especially HUD’s McKinney-Vento discretionary grants, have evolved along a number of dimensions. Finally, practices for serving homeless people have responded both to evidence and to changing philosophies and judicial decisions about how society treats its most vulnerable citizens.

Changes in income support and housing assistance.  In some communities, the implementation of the welfare reform legislation enacted in 1996 had important effects both on patterns of homelessness among families with children and on the way in which providers think about serving families. Cash assistance is now temporary, and families reaching their TANF time limit or sanctioned for failing to comply with TANF rules are among those particularly vulnerable to housing instability (Mills et al., 2006). At the same time, providers helping families exit homelessness focus increasingly on stable employment because of the temporary nature of assistance for those who do not work.[5]  Some providers hope to see their clients leave homelessness with a wage high enough to pay for unsubsidized housing because of the increasing difficulty of gaining access to assisted housing.

The Housing Choice Voucher Program, the mainstream program best suited to providing permanent housing for homeless families, has become less available for that use over time due to budget cuts and shifting program priorities that reduce advantages that people leaving homelessness once had in competing for the limited number of subsidy slots. Access to HUD’s assisted housing programs has become more difficult recently in many communities because “waiting priorities” for homeless families and individuals are no longer in effect. These priorities took two forms: (1) a “preference” on waiting lists for households experiencing homelessness that was equivalent to preferences for households with extreme rent burdens or living in substandard housing,[6] and (2) special allocations of vouchers reserved for clients of the homeless services system. In addition to the discontinuation of priorities for homeless people, admission policies have been tightened across assisted housing programs for people with criminal records or poor housing histories (whether previously homeless or not), making it more difficult to enter public housing and the voucher and project-based Section 8 programs (Khadduri & Kaul, 2005).

Furthermore, according to HUD data cited by the Center on Budget and Policy Priorities, the number of households assisted by housing choice vouchers fell by about 100,000 between 2004 and 2006.[7] Numbers of units in public housing and Section 8 projects declined starting in the mid-1990s. They were replaced by a comparable number of housing vouchers in the same communities, but more recently the number of vouchers has dropped as well, leading to an overall decline in the number of “slots” available in programs that permit people to pay no more than 30 percent of their income for housing.

For individuals, particularly people with disabilities, the picture is somewhat different. Supplemental Security Income (SSI) remains an entitlement, and providers have focused increasing attention on helping homeless people qualify for this important income source. Permanent supportive housing affordable to people who receive SSI continues to be produced by HUD’s Section 811 program for people with disabilities; by the HUD McKinney-Vento grant programs; and by resources under the control of state and local governments, including funding from state mental health systems.

Some 4 million people receive SSI, yet housing that people with SSI can afford remains in short supply compared with the need. O’Hara and Cooper (2005) compared SSI income to the average cost nationwide of renting a one-bedroom apartment. In 2004, on average, a person receiving SSI needed to pay 109.6 percent of his or her monthly income to rent a modest one-bedroom unit. Like homeless families, individuals attempting to exit homelessness have been affected by the reduced availability of housing vouchers, public housing, and units in Section 8 projects that would help narrow the gap between incomes and housing costs.

The only federal housing program that has produced significant numbers of additional rental housing units since the 1998 Symposium, the Low Income Housing Tax Credit, has rents set at a fixed dollar amount rather than as a percentage of a household’s income, and those rents usually are not affordable for households with poverty incomes.[8]  In addition, choosing to allocate tax credit resources to programs targeted to homeless people often means states must trade off using limited resources for people who are homeless against preserving or expanding housing for people who are low-income but not homeless.

Shifting priorities in HUD’s homeless assistance programs.  HUD funding for permanent and transitional housing for people leaving homelessness comes largely through two McKinney-Vento programs — the Supportive Housing Program (SHP) and the Shelter Plus Care (S+C) program.[9]  As of the early 1990s, each program had its own Congressional funding authorization. The SHP funded transitional and permanent housing as well as services. S+C provided permanent housing for persons with disabilities. S+C program funds could be used only for rental assistance, while services had to be leveraged from other funding sources.

Beginning in the mid-1990s, HUD received lump sum McKinney-Vento appropriations instead of separate appropriation amounts for the SHP and S+C programs. Following that change, providers sought greater amounts of funding for eligible activities — specifically, transitional housing and supportive services programs — from the SHP program relative to the amount requested for permanent housing from both programs. For providers, SHP funding was one of a limited number of sources of services funding. Transitional housing was an attractive option because many providers did not have expertise in the development or management of permanent housing. As a result, the shift in the mix of transitional vs. permanent housing changed substantially. Whereas at one point more than roughly 60 percent of total funding was dedicated to permanent housing, by the late 1990s that percentage had declined to only 20 percent.

To renew emphasis on funding for permanent housing, Congress responded by mandating that at least 30 percent of McKinney-Vento funding (exclusive of S+C renewals) be used for this purpose. Concurrently, HUD began de-emphasizing the use of HUD McKinney-Vento funding for services by offering various incentives for applicants to use HUD funds for housing activities and mainstream sources for services. Recent HUD policies have also given continuums of care (CoCs) flexibility to “reprogram” existing McKinney-Vento funding during the renewal application process, which has prompted some CoCs to monitor more closely the effectiveness and outcomes of their housing and services programs. Given the scarcity of both mainstream and McKinney-Vento funding for permanent housing for homeless people, many CoCs are now working to redirect funding toward permanent housing.

There has also been an increasing emphasis on serving homeless people who are disabled. Since 2001, HUD’s McKinney-Vento funding priorities have focused on addressing the needs of people who are chronically homeless. Through a federal interagency consultation process, chronically homeless people were defined as single individuals with a disabling condition who have been continuously homeless (on the street or in a shelter) for at least one year or have had at least four episodes of homelessness during the past three years. Many people meeting these criteria have histories of mental illness and co-occurring substance use disorders. There are human and public benefits to having this population stably housed: safe, secure, affordable places to live for people who are chronically homeless and less strain on costly emergency services and institutional care systems.

Changing views on participation in services.  At the same time that priorities were changing for public programs that serve low-income people in general and people who become homeless, so too were the models developed by practitioners for combining housing and services. Although not mandated by HUD, the model common in the 1990s in many communities emphasized providing services linked to a continuum of housing settings in which people moved from emergency shelter to transitional housing (typically for 6 to 24 months) and then to permanent housing. Requirements that residents participate in services to acquire and maintain housing were permitted, although not mandated, under HUD’s Section 811, S+C, SHP, and Housing Opportunities for Persons with AIDS (HOPWA) programs.

During the 1980s and 1990s, the difficulty that people with mental illness had in accessing scarce mainstream affordable housing resources prompted a number of mental health systems (including those in California, Connecticut, Massachusetts, New York, Ohio, and Pennsylvania) and their service providers to fund their own housing programs. While these initiatives helped meet the need for housing, many of these programs came with “bundled” supports; residents were typically required to accept the services offered with the housing program, and the services often were co-located with the housing.

Some homeless people met the service participation requirements of this type of housing and moved successfully (not necessarily sequentially) through the continuum. Many, especially those with serious mental illness and/or substance abuse issues, were less successful. Some advocates said that housing and services should not be “bundled”; that is, participation in services should not be a condition of obtaining or maintaining housing, and housing should not be used to induce people to comply with services. This was a particular concern of advocates for people with mental illness, who saw this model as a continuation of coercive practices under which mental health systems “exercise enormous control over the lives and behavior of people with psychiatric disabilities” (Allen, 2003; see also Diamond, 1996; Carling, 1993).

Many providers of transitional housing would not agree that services should be voluntary. Most transitional programs mandate participation in services, considering it their mission to set goals that move the resident towards self-sufficiency and to use program services to reach the goals.

The Supreme Court’s Olmstead decision.  A landmark legal decision also figured into the evolution in housing and services models for people with disabilities. In 1999, the Supreme Court held in Olmstead v L.C. that segregating people with disabilities in state institutions may be discriminatory under the Americans with Disabilities Act and that states may be required to provide community-based services rather than institutional placements for persons with disabilities. Regulations promulgated by the Department of Justice to provide guidance on implementing the Court’s decision clarified that: “A public entity shall administer services, programs, and activities in the most integrated setting appropriate to the needs of qualified individuals with disabilities (28 CFR, Section 130(d)).”

The Olmstead decision has had implications for both housing and services for people with disabilities, including those who are — or may become — homeless. On the housing side, states and the federal government have been encouraged to identify alternative housing for people in institutions who wish to live in the community. For example, HUD’s ACCESS program provided voucher assistance to a set of communities to test the concept of using vouchers to help non-elderly persons with disabilities move directly from nursing homes to permanent rental housing. The Olmstead decision is also credited with the creation of policies that promote the most integrated models of permanent housing for people with disabilities. For example, state policies encouraging sponsors of housing developed with Low Income Housing Tax Credits to set aside a percentage of units for people with disabilities have created integrated housing settings in California, New Jersey, North Carolina, Louisiana, and other states.[10]

On the services side, the Olmstead decision encouraged states to identify funding sources for community-based services. One source is Medicaid’s Section 1915(c) Home and Community-Based Services Waiver program. This program gives HHS authority to waive Medicaid provisions in order to allow long-term care services to be delivered in permanent housing in the community instead of in institutional settings. Certain subpopulations of homeless people with disabilities may qualify for these services, depending on the state’s Medicaid policies. Similarly, Medicaid’s Medical Rehabilitation Option is used in some states to provide case management, health, mental health, and substance abuse treatment services.

Lessons from evidence-based behavioral health practices.  In response to the perceived need for new models, policymakers and practitioners looked for housing and service approaches that had been tested and found effective, particularly for homeless people with mental illness and substance use disorders. The focus on evidence-based practices was particularly prevalent in the medical, mental health, and substance abuse treatment fields, in which the federal government, foundations, and researchers promoted clinical interventions that research studies had shown to be effective.[11]  Within public mental health/behavioral health systems, the assertive community treatment (ACT) model gained credence as an effective way to engage homeless people with mental illness and substance abuse issues, a population that had been particularly challenging to serve in the emergency shelter and transitional housing programs of the 1990s.

In its pure form, the ACT approach uses multidisciplinary teams trained in mental health and substance abuse treatment, employability development, medical care, case management, and life skills training to reach out to homeless people on the street and in shelters to encourage them to enter more permanent housing. The service approach is client-focused and separates housing and other supportive services; that is, clients do not have to accept supportive services as a condition of entering or retaining housing. While not all communities have the resources to implement the ACT model in its pure form, aspects of the model, such as the emphasis on meeting clients “where they are,” offering but not mandating services, and providing services in-vivo (either in the client’s home or in the community), have been adopted in many communities even though it is not clear that piecemeal application of what is designed to be an integrated model would be as effective as full implementation.

Emergence of “housing first” models.  The emphasis on permanent housing and on chronic homelessness, together with the success of approaches such as ACT to providing services to people with chronic mental illness and persistent substance abuse, encouraged a new paradigm for meeting the needs of this vulnerable population. In recent years, more providers have come to view the continuum of care not as a sequential series of placements but rather as a menu of options, any of which might be appropriate for any particular client. Among those options, housing first approaches are being tested that emphasize rapid placement in permanent housing with no or minimal transitional placements or service requirements. Community-based support and treatment (some using the ACT team model or variations on it) help people maintain their housing.

The rapid housing placement aspect of the housing first approach is being used for both individuals with disabilities and families, although the service approaches differ somewhat. Programs serving single individuals with disabilities tend to focus heavily on housing placement and retention, with minimal service participation required either to enter housing or to retain it. Programs serving families also focus heavily on housing placement and retention up front, while typically also establishing service plans. Service plans are initially focused on the housing search process and short-term case management; once the family is in permanent housing, the plans focus on longer-term case management. Similar to housing first programs for single individuals with disabilities, service participation in programs serving families is typically voluntary. Engagement is a central component in working with both populations. Providers working with families must also take into account the needs and safety of children in determining how ”voluntary” service participation should be. The vulnerability of children to dangerous or abusive parental behavior makes the issue of voluntary services different for families who become homeless with their children than it is for people who become homeless without accompanying children. Further, when family reunification or preventing the loss of custody is a goal, the parent needs to show credible progress to the child welfare system.

Debate continues over the effectiveness of the housing and service approaches associated with housing first and which elements of the model are most important. More broadly, the evolution toward community-based housing and services approaches, driven by funding priorities and emerging evidence-based practices, has spurred increased interest in identifying which housing and services approaches work best for whom, but so far has not resulted in a commensurate level of rigorous research to provide answers to these questions. In the next section, we describe further the evolution of housing models and review recent research findings on the implementation of these approaches and what is known about their outcomes for clients.

[ Go to Contents ]

Synthesis of Research Literature:  Findings and Discussion

Research indicates that housing with services, especially for homeless single adults with serious mental illness, increases housing tenure, reduces hospital stays, and reduces homelessness (Rog, 2004). However, conducting rigorous research on how this comes about and which models are most effective is extremely challenging. Random assignment studies are rare, and even well-matched comparison studies are difficult to construct and implement. Further, measuring both the interventions and the outcomes across programs is very complicated. Developing reliable, replicable measures of the housing provided and the services received is problematic given the diversity of program approaches, housing market conditions, staff capacity, and other variables that are beyond researchers’ control.

These factors make it very difficult to answer the question foremost in the minds of policymakers and program administrators: what works best for whom? In this section, we attempt to shed light on this complex question by describing a broad range of program models for families and individuals, from those providing short-term or transitional housing and services interventions to those designed to provide permanent housing and long-term supports. We review the unfortunately quite limited research findings on the outcomes of those models.

Housing Models for Programs Serving Homeless Families

Advocates for homeless families are quick to point out that most Americans underestimate the extent to which homelessness affects families. About 600,000 families and 1.35 million children experience homelessness each year, and about half of the homeless population are part of a family. A homeless family typically comprises a woman in her late 20s who becomes homeless together with young children (Burt et al., 1999). In many ways, homeless families are similar to other low-income families that are not homeless. Their limited incomes make it difficult to find and keep housing that is safe and affordable, they face stagnant wages for workers with few skills, and they may be affected by welfare time limits or sanctions under Temporary Assistance for Needy Families (TANF).

HUD’s Supportive Housing and Shelter Plus Care programs serve substantial numbers of homeless families.[12]  However, given the greater emphasis in recent years on addressing the needs of homeless single individuals, the need to devote a major portion of McKinney-Vento grant funds to renewing funding for existing grants rather than placing additional units under subsidy, and the reduced availability of mainstream assisted housing, fewer new permanent housing resources are available for homeless families.

Programs serving homeless families range from short-term assistance to shorten or avert shelter stays for families experiencing a crisis to long-term permanent supportive housing for families with complex supportive service needs. In addition, non-residential service providers, such as housing resource centers, housing locator services, and housing counseling agencies, may play important roles in helping people who are homeless or at risk of losing their housing to locate and retain stable housing.

In the following sections, we describe a number of approaches to assisting homeless families and review the evidence, where available, on the efficacy of each. However, services provided by residential programs for families are so diverse in their nature and intensity that it is difficult to identify a model used in different communities that links housing to a particular set of services in a particular way. This points to the need for rigor in classifying the housing and support services provided according to exact type and range, frequency, and duration.

Short-term assistance. Modest levels of financial assistance to families who are precariously housed or newly homeless have been used to help families that are experiencing a short-term crisis. For example, Portland, Oregon’s Transitions to Housing Program provides short-term emergency rent assistance to 400 individuals and families annually. The clients served may be homeless or at risk of homelessness; all have family-size adjusted incomes of no more than 20 percent of area median income. The average total assistance per household is $1,285. According to program data, this relatively small amount of assistance allows 70 percent of households to stabilize and remain in permanent housing for at least six months after intake (City of Portland, Oregon, 2004).

In the Minneapolis/St. Paul area, Hennepin County administers a state-funded Family Homeless Prevention and Assistance Program (FHPAP) through a network of providers. The legislation creating FHPAP established a set of intended outcomes, including preventing first-time shelter stays, reducing the length of shelter stays, and eliminating shelter reentry.

Prevention services are targeted to families that are threatened with housing loss because of nonpayment of rent, but for whom a resolution to the crisis is within reach. Case workers assess the amount of rent owed and the family’s resources, credit history, rental history, and other circumstances to decide how much assistance the program will provide and what the family can contribute. Case workers also work with families for up to six months on budgeting, determining whether a move is necessary (e.g., to a smaller apartment), and other issues to ensure the family remains housed.

According to program data cited by Burt and Pearson (2005), FHPAP’s screening system and prevention activities have reduced the average duration of shelter stays by one-half and reduced the daily census of families in shelter by 63 percent. The program’s 2003 annual report (also cited in Burt & Pearson) shows that 95 percent of families in the prevention component did not use shelter within 12 months. The average cost to the county per family was $472.[13]

Illinois, the District of Columbia, and Massachusetts also have begun testing strategies that divert families from becoming homeless or use short-term assistance to help them exit homelessness. Illinois’s Homeless Prevention Program, administered by the Department of Human Services (DHS), provides short-term rent and utility assistance and supportive services to families that are homeless or at imminent risk of homelessness. Program funds can be used for up to three months back rent to prevent eviction, up to two months rent or security deposit, and services such as housing location/inspection, job search, counseling, and case management. According to the DHS Web site, some 10,000 families were served by the program in 2004.

In the District of Columbia, all families entering the homeless services system go through a central intake center that focuses on resolving the crisis that is about to make the family homeless. Those who cannot be stabilized in their current housing but can stay there for at least 30 days are referred to a grant program that provides intensive case management, housing search assistance, and short-term assistance such as deposits and first month rents. Only families considered unlikely to succeed in a rapid housing placement are placed in emergency shelter. Factors considered in this assessment include current substance abuse, uncompensated mental illness, and whether the head of household has ever been employed and has ever been a leaseholder.[14]

In Massachusetts, the rising costs of emergency shelters and the need for additional space in even more costly motels had increased the annual cost of sheltering a homeless family to an average of $47,000 by 2004. In response, Massachusetts implemented several pilot projects to explore alternative approaches to helping families find or retain housing. According to state data, three pilot programs kept 1,119 families housed for the same cost as 63 shelter rooms. The Rental Assistance for Families in Transition (RAFT) program provided flexible funds for first/last month rents, security deposits, or utility payments. Some 436 families were assisted over a two-month period at an average cost per household of $1,365. Similar assistance was provided to families eligible for the state TANF agency’s emergency assistance program, helping another 476 families avoid homelessness or shorten a shelter stay, at an average cost of $3,080 per family. Finally, under the Shelter to Housing Pilot, 207 families were assisted with a one-time subsidy of $6,000 to cover rent and some stabilization services such as job search and household budget assistance. Two years later, 80 percent of the families were still housed (One Family, 2006).

None of these programs for using short-term assistance for prevention or rapid exit from homelessness has been studied using a rigorous evaluation methodology that controls for family characteristics and examines long-term outcomes. Therefore, we do not yet know how to distinguish families who can benefit from this approach from other types of families, nor can we assess the costs and benefits of short-term assistance compared with more expensive approaches to placing families in permanent housing. Nonetheless, preliminary evidence from program records suggests that short-term housing assistance can play an important role in reducing and ending family homelessness.

Transitional housing.  Many communities continue to consider transitional housing to be an effective strategy for helping families secure and retain permanent housing. Since 2000, the Sound Families program sponsored by the Bill and Melinda Gates Foundation has supported the development of 1,100 units of service-enriched transitional housing for families that have experienced homelessness in Pierce, King, and Snohomish Counties in Washington. Several housing authorities in the region have allocated project-based housing voucher assistance to the Sound Families projects to make them affordable to homeless families and financially viable for project sponsors.[15]

Sound Families provides supportive services during the transitional housing stay and assistance in moving to permanent housing, which can be public housing or private rental housing supported by a tenant-based voucher subsidy. In some program sites, a “transition in place” option allows families to continue living in the same complex (if not the same unit) where their transitional housing unit is located.

Preliminary results from an evaluation of 10 sites participating in the Sound Families program (Bodonyi & Erwin-Stewart, 2005; see also show that, of 139 families interviewed at intake, 80 percent remained in transitional housing until they graduated from the program. The average length of stay was 12.7 months. The researchers found increases in employment, from 27 percent employed full- or part-time at baseline to 41 percent at exit. These outcomes compare favorably with other social programs. Receipt of TANF benefits declined from 62 percent at baseline to 46 percent at exit. Children benefited as well. Some 80 percent of parents said their oldest child was doing “very well” or “excellently” in school six months after exit compared to 52 percent who said so at intake. The proportion of children attending more than two schools in the previous year declined from 53 percent at intake to 5 percent at exit.

Some 86 percent of families secured permanent housing at exit from their transitional program, and 89 percent continued to reside in permanent housing six months after exit. Of the 14 percent who were evicted or asked to leave their transitional housing unit, most had mental health or chemical dependency issues the program was not designed to address. A pilot program to provide permanent supportive housing that does address these issues is under development.

Many transitional housing programs have a primary focus not on getting the highest-needs families ready for placement into permanent housing, but on other objectives, such as employment, income growth, and better life chances for children. Some transitional housing providers admit only families deemed able to take advantage of services the transitional housing program offers. They may screen out families with severe mental illness, current chemical dependency, or no employment history. Many transitional housing providers describe their programs in this way, and a recent analysis of patterns of use of emergency shelter and transitional housing in Philadelphia and Massachusetts suggests that such screening is common. The study found that the families with the longest stays in residential facilities within the homeless services system (most long stays were in transitional housing) were less likely to have histories of inpatient behavioral health treatment, had lower rates of disability, and had higher rates of employment than families with shorter stays (Culhane, 2006).

A recent survey of HUD-funded transitional housing programs for families shows that, although most programs screen out families with active substance abuse and about a third of programs would not accept families with severe and persistent mental illness, the families served have high needs. About one-quarter of parents take psychotropic medications for mental or emotional problems, and at least as many have histories of drug abuse. Addiction relapse is the primary reason for families being asked to leave transitional housing programs (Burt, 2006).

This study also collected information about housing outcomes for transitional program participants. On average across the 53 programs surveyed, 70 percent of families exited to permanent housing. More than one-third (36 percent) went to unsubsidized mainstream housing, while 22 percent left for housing with a rent subsidy, and 13 percent went to permanent supportive housing. A subsequent phase of the study will attempt to relate these and other outcomes to the different characteristics of transitional housing programs, such as whether the program provides scattered-site housing, whether it permits transition in place, and the program’s staffing levels (Burt, 2006).

Permanent supportive housing.  Permanent supportive housing for families may take several forms. The “transition in place” model mentioned above may offer security of tenure by allowing a family to stay in what is initially treated as a transitional placement while continuing with case management support and other linked services. HUD’s Shelter Plus Care (S+C) program provides permanent supportive housing to families if the head of household is disabled. S+C may be used to provide tenant- or project-based rental assistance. Regardless of the form of the rental assistance, the organizations that receive Shelter Plus Care funding must provide services that, in the aggregate, are equal in value to the value of the S+C rental assistance. S+C rental assistance is often used to provide the housing component of programs created by mental health or substance abuse treatment providers for their clients. 

A study of San Francisco’s Family Permanent Supportive Housing Initiative (FPSH) offers lessons from permanent housing programs designed to serve families with the kinds of supportive services needs that the Sound Families transitional housing programs had difficulty addressing (Nolan et al., 2005). The seven programs studied offer access to affordable permanent housing and voluntary services to address mental health and addiction issues as well as a variety of health and social services for adults and their children. The housing situations varied across the seven programs and included scattered-site units; buildings dedicated to homeless families; and “mixed” buildings housing low-income people, only some of whom had been homeless. Residents reported high levels of satisfaction with their living environments as well as with the services they received. According to the researchers:

No single program model appears to be significantly better than any other at helping tenants achieve the primary goal of housing stability, as long as the model succeeds in creating an atmosphere of respect and trust among tenants and staff and is able to provide the resources that tenants need.

According to program staff, services for children are an important component of permanent supportive housing for homeless families. Many mothers in these families have been separated from their children at some point in their lives, and housing stability with their children is an important objective for residents.

A 2006 study prepared by the National Center on Family Homelessness combined results from a number of studies of permanent supportive housing programs for families to identify client and program characteristics and client outcomes and to assess whether certain combinations of program characteristics are associated with improved client outcomes (Bassuk, 2006). The study examined 13 programs, all located in the San Francisco area or in Minnesota. The researchers assembled data on program context, housing arrangements, program control (that is, strictness of program rules for participation), the range of services available for adults and children, and the intensity of adult services (derived from the number of households per case manager and services per family per month). Participant outcomes in terms of residential stability, family reunification, and self-sufficiency were ranked as “high,” “medium,” or “low” for each program.

Although the authors caution that the analyses are limited by inconsistencies in the data collected across studies, the high control programs seem to have better reunification and self-sufficiency outcomes, but their attrition rates are high. By contrast, low control programs may have higher residential stability but are not as successful at helping families reunify or move to greater economic self-sufficiency.

Housing First for Families

Other lessons on serving homeless families may be drawn from the program operated by Beyond Shelter in the Los Angeles area. Beyond Shelter’s “Housing First” Program for Homeless Families began in 1988 and has been widely cited as a model for serving families with extensive supportive services needs ( The housing first approach in this program, as in other programs that use that name, emphasizes rapid placement in permanent housing while minimizing or avoiding transitional stays. Beyond Shelter helps families move from emergency shelters to permanent affordable rental housing scattered throughout residential neighborhoods and provides 6 to12 months of follow-up case management and services. Most families receive voucher assistance through a local housing authority, and the program provides assistance with moving expenses.

According to program administrators, three-quarters of the families served would be considered multi-problem families with unstable living patterns. Families and their case managers develop Family Action Plans to guide services. Services are provided by agencies other than the housing authority and focus on helping families retain their housing. Beyond Shelter has some aspects of a transition-in-place model, because services continue for a defined period after the housing placement. However, the families have security of tenure in their housing placement, which contrasts with many transitional programs that can evict families who do not cooperate with their services plan.

A two-year evaluation of Beyond Shelter's “Housing First” Program was conducted by local researchers from the University of Southern California as part of a Pew Partnership initiative. Data on 185 families were collected from April 1, 2000 to October 1, 2001, based on the Substance Abuse and Mental Health Administration (SAMHSA) Program Logic Model for Homeless Families.[16]  Outcomes identified by the model include increased residential stability, improved mental health functioning, reduced drug and alcohol use, and increased trauma recovery. For children, outcomes include reduced emotional and behavioral problems and improved school attendance.

The study found that more than 90 percent of the mothers who graduated from the program at the end of six months in permanent housing had achieved the short, intermediate and long-term goals identified in the SAMHSA model, and more than 80 percent of the children's goals were achieved. More than 80 percent of adults were employed, and others were enrolled in job training programs. Only 2.3 percent of those who entered the program with reported substance abuse problems had relapsed, and less than 1 percent of domestic violence survivors had returned to a dangerous relationship. Some 80 percent of children were enrolled in school during the evaluation period and 77 percent attended regularly.

Hennepin County, MN, developed a shelter screening and admission system to limit access to shelters to the families that need the most help. Pregnant or parenting teens, families with more than two children or with infants, and families receiving SSI receive priority for shelter space. Within one to three days, shelter guests meet with the rapid exit coordinator for an in-depth screening that focuses on housing barriers. The family is then referred to a separate rapid exit program where a caseworker works with the family to develop a housing stabilization plan. Continued shelter stay is contingent on the family cooperating with the caseworker and the plan. The caseworker focuses on helping the family find housing and coordinates with other service providers to address other needs. Follow-up continues for six months after the family leaves shelter. Some 88 percent of families served in the rapid exit component did not return to shelter within 12 months; the average cost per family for this component was roughly $800.

Without a comparison group drawn from a similar emergency shelter population or a population placed into transitional housing with tenure dependent upon cooperation with services, it is difficult to place these outcomes in context. Nonetheless, both the Hennepin County and Beyond Shelter programs seem to demonstrate that rapid placement into permanent housing is feasible for high-needs families.

Housing Models for Programs Serving Single Individuals

Over time, homeless assistance programs have served single individuals who are homeless for various reasons — from people who are working but experiencing a short-term crisis to those who are experiencing long-term homelessness and have complex service needs. Given the recent emphasis on addressing chronic homelessness, permanent and transitional housing programs serving single individuals who are homeless usually focus on people with a disabling condition such as mental illness, physical or medical disability, substance use disorder, or HIV/AIDS. Permanent housing funded under the McKinney-Vento Shelter Plus Care program can only be used for people with disabilities.

Emergency shelters often do not have an explicit focus on people with disabilities in their admissions process, but people with disabilities are heavily represented among those who use shelters frequently or for long periods. Homeless individuals with disabling conditions are considered particularly difficult to serve, especially if they have been homeless for extended periods and the symptoms of their disabilities have gone untreated.

Programs designed to serve these populations may take several forms. Because these are some of the most vulnerable people, some communities have emergency shelter programs specifically designed for them. Safe havens, which can be permanent or transitional housing, are designed to serve chronically mentally ill people who are homeless and who have been reluctant to enter the shelter system. Safe havens offer housing and make services available but in a low demand environment.

Transitional housing programs may provide “next step” housing to clients with substance use disorders after they have completed detoxification to prepare them for mainstream permanent housing without intensive supports. A permanent supportive housing program is sometimes used as a further “next step” for homeless individuals after a transitional program, but often is offered directly to homeless people coming from the streets or from emergency shelters under one or another variant of a housing first approach.

Research has shown that persons with severe mental illness who are offered the opportunity to live in permanent supportive housing experience reductions in shelter use, hospitalizations, length of stay per hospitalization, and time incarcerated (Culhane et al., 2002; Martinez & Burt 2006; Mayberg 2003). There has been some disagreement, however, on which model works best for this population: a model that requires clients to move through two or more housing placements before achieving permanent housing, or a housing first model that places clients directly in permanent housing with community-based supports.[17]

Safe havens. Safe havens may be the first step off the streets for some of the most severely mentally ill homeless people. The Ward Family Foundation (2005) surveyed safe haven programs to collect information on program characteristics and effectiveness in transitioning safe haven residents to permanent housing. Seventy-nine of the 118 programs identified (about 85 percent of which were HUD-funded) responded to the survey. The findings on program characteristics are consistent with what we expect safe havens to provide. The programs serve people who are extremely vulnerable — mentally ill and homeless — and rarely refuse admission to anyone who meets those criteria. Participation in services or activities is rarely imposed. Most programs (72 percent) have no limit on length of stay; with the average length of stay among programs surveyed 262 days.

The program administrators surveyed said that, overall, just over half their residents exit to some kind of permanent housing, while about 14 percent return to homelessness. The most common reasons cited for residents not moving to permanent housing are that the resident’s condition is too unstable (64 percent), the community lacks housing with appropriate supports (63 percent), and the community does not have subsidies to make the housing affordable (59 percent).

The researchers identified the characteristics of programs that had a high rate of successful referrals to permanent housing based on results from 15 programs that achieved an average referral rate of 85.2 percent. This compares to an average referral rate of 41.6 percent for the remaining 64 programs. The programs with higher successful referral rates were smaller, more likely to offer private rooms, and more likely to operate at full capacity. These programs were more likely to require that clients come from the street and be severely mentally ill, but were also more likely to refuse admission to clients with felony or sexual violence convictions. The proportion of programs with a rich variety of services offered on site appears higher in the group with higher referral rates. The programs with higher referral rates had only a slightly higher average annual cost per bed ($43,089 compared to $41,534 for those with lower referral rates).

Transitional programs.  The concept of transitional housing grew out of halfway houses for people released from prisons or mental institutions. HUD’s transitional housing program began as separate from the permanent housing program. Both were later brought under the Supportive Housing Program component of the McKinney programs, although each with its own set of distinguishing rules (Burt, 2006). However, there was not a strong theoretical framework for applying this concept to homeless individuals. Only recently, with transitional housing challenged by shifting federal funding priorities and by the housing first model, have researchers begun to create a theory of transitional housing that goes beyond the simple McKinney-Vento programmatic rule that a transitional housing stay may not last more than two years.[18]  Much of the research on outcomes for individuals participating in transitional programs focuses on comparisons of supportive housing programs serving homeless individuals with mental illness (who often also have co-occurring substance use disorders and other disabilities) with traditional mental health treatment without a housing component. There have been few studies of transitional programs that compare them to other housing models.

Analysis of data on transitional housing has emphasized the rate of placements in permanent housing. This is one of HUD’s GPRA performance measures for the McKinney-Vento programs, with a current goal that 61 percent of those exiting HUD-supported transitional housing be placed in permanent housing.[19]  An early study of the Supportive Housing Program, when it was funded as a demonstration, provided qualitative evidence that the housing and supportive services offered clients in transitional housing contributed to successful placement into permanent housing for 56 percent of clients studied (Matulef, et al. 1995).

Evaluations of local Supportive Housing Demonstration programs in Boston, Chicago, and Michigan also yielded promising findings on housing stability, although little change in the level of functioning of the clients served was observed. For example, as described in Brown (2004), in 1995, 114 undomiciled patients of a state psychiatric hospital in Chicago were randomly assigned to a supportive housing program (n=48) or to a controlled treatment (n=47) that provided links to whatever community service was available and no ongoing case management. According to data from case managers, experimental group participants were more than twice as likely to be housed. At six-month follow-up, none of the experimental group had returned to homelessness and 68 percent of the experimental group remained in supportive housing.

Permanent supportive housing. In contrast to the paucity of research on transitional housing programs for individuals with disabilities, a number of studies of permanent supportive housing have looked at both housing outcomes and service approaches.

A recent evaluation of the Connecticut Supportive Housing Demonstration Program examined the supportive housing concept in mid-sized cities such as New Haven and Hartford as well as in smaller communities such as New Britain and Middletown (Arthur Andersen, LLC and University of Pennsylvania, 2002). The Connecticut demonstration served people who were homeless or at-risk, many of whom also had mental illness, histories of addiction, or HIV/AIDS. The purpose of the study was to assess whether stable housing reduces the need for expensive social services over time, enhances residents’ quality of life, and allows residents to attend to employment and vocational needs. Connecticut’s supportive housing approach provides permanent housing in which participants hold their own leases in projects developed by the state with multiple state, federal, and private funding sources. Some services are offered on site, but participation is voluntary.

Findings on client outcomes from the 4.5 year study included that tenants decreased their use of costly acute care health services while increasing their use of necessary routine and preventive health care, compared to their use of those services before they were placed in permanent supportive housing. Tenants were satisfied with most aspects of the program, functioned at high levels, and were able to move toward greater independence. Employment increased modestly. Of some concern, however, was that about 39 percent of the tenants exited housing during the study period, and 36 percent of leavers (14 percent of all tenants) left under negative circumstances. The researchers found that the negative departures were associated with substance abuse, some aspects of functioning (e.g., self care, daily chores, using transportation), not working toward goals in life, unemployment, and social isolation, but cautioned that the small sample sizes did not permit conclusive analysis of reasons for leaving.[20]

The state of California has devoted substantial resources to serving vulnerable groups who had been inadequately served, including people who are seriously mentally ill and homeless, insufficiently housed, or returning from jails or prisons. The programs are known as AB2034 programs after the section of legislation that funded them. Mayberg (2003) found the programs resulted in reductions in homelessness, emergency room use, hospitalizations, and incarcerations. In a study prepared for the Corporation for Supportive Housing, Burt and Anderson (2005) found that clients with stable housing were more likely to stay enrolled in the program — that is, to stay engaged in mental health services. Housing approaches vary across the 53 programs operating in 34 counties; AB2034 funds can be used for housing development, securing dedicated voucher assistance from PHAs, or providing ongoing rental subsidies through Shelter Plus Care or state funds. The program has achieved promising outcomes in helping clients, including those deemed “hard to serve,” obtain and retain housing, The researchers note that:

Programs with a high proportion of consumers who are homeless, recently incarcerated, or diagnosed with a co-occurring substance use disorder have similar outcomes to other programs… [T]he data [also] show that those who disenroll from the AB2043 programs are no more likely than current enrollees to have lived on the streets, been incarcerated, or have a diagnosed substance abuse disorder.

An evaluation of the Closer to Home Initiative offers insights into the outcomes of six programs designed to engage and house people with disabilities, long histories of homelessness, and repeated use of emergency services (Barrow et al., 2004). The six programs are located in four cities: three in New York and one each in Chicago, San Francisco and Los Angeles. The purpose of the study was to describe the program models, document implementation over time, and assess outcomes for an initial cohort of participants.

The Closer to Home programs fell into two general program models. Three programs attempted to engage long-term residents of shelters or lodging houses to encourage them to move to permanent housing.[21]  The results of the assertive engagement efforts at these shelter/lodging house programs were modest:

The programs developed relationships with most residents, provided a range of direct services, and initiated housing referrals for a substantial proportion of the individuals at the sites. But engagement in complex services and housing remained low, and most residents still lived at the sites two years later. Moreover, the predictive analyses failed to confirm that building relationships with long-term residents would improve housing outcomes — a key premise of these programs — but did show better outcomes for residents who had entitlements and who became engaged around housing. . . . [T]hose who had been homeless longest were least likely to be housed, indicating a need to prevent long-term homelessness at earlier stages.

The other three programs provided housing to adults referred from various community service providers; the emphasis in the services provided by these programs was on housing retention. The housing settings included buildings housing only program clients and buildings with a mix of program clients and other residents. In all cases, the service approach was characterized as “low demand,” although one of the three programs screened prospective residents for those willing to participate in services and accommodate its building’s “clean and sober” environment.

In the three programs that provided housing directly, housing outcomes were more promising than the outcomes of programs that focused on engagement and referrals to permanent housing. After two years, more than half (55 percent) of the residents in the shelter/lodging house programs were still at their original location, and 18 percent had moved to other temporary settings. Only 25 percent had moved to longer-term settings, defined by the researchers to include permanent housing, transitional housing, adult homes, or nursing homes. By contrast, in the programs that provided housing, 77 percent of residents remained housed, and a large majority of tenants were engaged in clinical or social services. Mental health referrals significantly increased housing stability, according to the researchers, who further conclude: “Across diverse housing approaches for homeless individuals with long-term homelessness and other barriers, housing works.”

Despite these promising findings, there have been concerns about the fact that departure rates from permanent housing are as high as they are. In 2004, according to data reported by HUD-funded permanent supportive housing providers, roughly one-quarter of residents in HUD-funded permanent housing that year left after stays of two years or less. A recent HUD-sponsored study explored the reasons residents may leave permanent supportive housing programs (Morris Davis and Company, 2006). The study focused on programs serving people with serious mental illness. The researchers examined patterns among participants in 28 permanent housing programs in Philadelphia. Based on patterns observed among a cohort of people who entered permanent housing in 2001, the authors estimate that only half of those entering permanent housing would maintain residency for three years or more. More than 10 percent of the 2001 cohort left within six months.

The researchers found that about one-third of leavers were “positive leavers” who went to stable alternative housing. The remaining two-thirds were non-positive leavers who went to congregate settings, institutional settings, homelessness, or other unspecified destinations. Some 61 percent left voluntarily, while the remaining 39 percent left involuntarily. Positive leavers tended to leave in order to improve their living situations. Negative leavers had more severe levels of mental illness, greater incidence of substance abuse, and higher supportive services needs. The study authors encourage initial and on-going monitoring of permanent housing residents to identify and address issues for those most at risk of leaving under negative circumstances.

Housing first. The recent interest in the housing first approach as applied to homeless individuals with disabilities has led to studies of programs that move the most vulnerable homeless people rapidly to permanent housing with limited or no transitional placements. A number of studies have been published on the Pathways to Housing program in New York City (Tsemberis & Eisenberg, 2000; Padgett, Gulcur & Tsemberis, 2006; Siegel et al., 2006). In the Pathways program, participants are offered scattered-site permanent apartments with limited or no transitional stays. Neighborhood-based, multidisciplinary support teams work with clients to maintain their housing and, if the client chooses, address other supportive services needs. A study comparing the outcomes of homeless persons with serious mental illness placed in community residential treatment facilities (where service participation and sobriety are typically required to obtain and retain housing) with those in the Pathways to Housing program found that the Pathways to Housing supportive housing approach resulted in greater housing stability. After five years, 88 percent of Pathways to Housing participants remained housed, whereas only 47 percent of the residents in the residential treatment system remained housed (Tsemberis & Eisenberg, 2000).

In a HUD-sponsored study, researchers examined outcomes in the Pathways to Housing program along with two other programs that have adopted the housing first approach — Downtown Emergency Services Center (DESC) in Seattle and Reaching Out and Engaging to Achieve Consumer Health (REACH) in San Diego (Pearson et al., in press). The three programs share some features: they serve clients with severe mental illness (including many with co-occurring substance use disorders) and long histories of homelessness; they offer permanent housing with access to a wide variety of services, but service participation is voluntary; and efforts to provide services continue even if the client leaves program housing for as long as 90 days. The housing types vary, however. While Pathways to Housing leases scattered-site units in privately owned buildings, DESC offers housing in several buildings the organization owns or controls. REACH (a program funded by California’s AB2034 program discussed above) has access to (but neither owns nor controls) a variety of housing units funded by Shelter Plus Care subsidies, project-based Section 8, and state funds. Some units are clustered in a safe haven and several downtown SRO buildings, while others are scattered site-apartments in complexes throughout the county. While REACH does not require service participation, a number of the housing providers associated with the program do have occupancy rules regarding alcohol and drugs, curfews, noise, and other issues.

The researchers tracked 25 to 29 clients at each site for 12 months to examine housing tenure patterns, among other outcomes. Overall, the programs had similar outcomes, but the findings reveal that there are nuances to housing stability. While a large majority of clients (84 percent) were still housed at the end of 12 months, not all had stayed in program housing throughout the tracking period. Across all three programs, 43 percent of the clients stayed in housing for the full 12 months. Some 41 percent experienced at least one departure to another living environment, but returned to program housing. The remaining 16 percent left or died during the follow-up period. The researchers did not observe substantial changes in clients’ mental health or substance use status, but this was not expected given the relatively short follow-up period. As has been seen in other studies, clients who entered housing from the streets and had more severe psychiatric impairment or co-occurring substance use disorders were more likely to leave.

The San Francisco Department of Public Health’s Direct Access to Housing (DAH) program offers another housing first approach. DAH provides permanent housing with on-site supportive services for formerly homeless adults, most of whom have mental health, substance abuse, and chronic medical conditions. The program is targeted to “high users” of the city’s public health system and describes itself as a “low threshold” program that accepts single adults into permanent housing without requiring service participation or abstinence from substance use. The housing consists of 876 units that include nine SRO hotels, three newly developed buildings, and one licensed residential care facility (or “board and care”). The program also secures blocks of units in several buildings owned by nonprofit providers. To access this large stock of housing, DAH has identified buildings that are vacant or nearly vacant and then negotiated with the owners to renovate the buildings in exchange for entering a long-term lease with DAH. When a building is ready for occupancy, DAH contracts with service providers to provide on-site services.

The DAH program pays particular attention to health outcomes, given that the program targets high users of emergency services. According to program data, emergency department use was reduced by 58 percent after program entry. In the two years after program entry, participants had 57 percent fewer inpatient episodes compared to the two years prior to program entry. Numbers of days of hospitalization also declined for participants with histories of mental illness and psychiatric hospitalization (Trotz, 2005).

Research comparing service approaches in permanent supportive housing. Researchers have attempted to tease out the roles of different housing and services models for permanent supportive housing in affecting tenure outcomes. One small-scale, New York City–based study conducted interviews with 224 residents from 10 developments financed by the Enterprise Foundation; the researchers also used case management data. Most residents in the study had their own apartments with kitchen and bathroom, and paid subsidized rents. On-site and off-site services were offered, but not mandatory. The study found positive outcomes in housing stability, as well as increased incomes and strong client satisfaction with services (Bayer & Barker, 2002).

In a larger study of homeless persons with severe mental illness served in the New York, New York Initiative,[22]  Lipton and colleagues (2000) followed a total of nearly 3,000 persons placed in high-, moderate-, and low-intensity housing for a period of five years. Intensity levels were determined by the researchers and refer to the degree of structure in the program, including the level of scheduling, house rules, and requirements for program participation. The degree of clients’ independence, including control over decisions about the living environment, activities, income, medications, and privacy, was also factored into intensity.

Clients placed in housing with different levels of intensity had somewhat different characteristics. Clients placed in high-intensity settings (30 percent of the sample) tended to be younger, referred from hospitals, and to have a history or diagnosis of substance abuse. Clients placed in moderate-intensity housing (18 percent of the sample) were more likely to be female and were least likely to have substance abuse problems. Individuals in low-intensity settings (52 percent of the sample) were more likely to be referred from city shelters where they had lived for four or more months.

Lipton examined tenure outcomes, classified as follows: consumers who were “continuously housed” either stayed in their initial placement or moved to another stable setting. Those classified as “discontinuously housed” became homeless, moved to an unstable setting, or were imprisoned. Consumers who died, were hospitalized for medical reasons, or could not be located were not classified. The study found that, for the sample overall, 75 percent, 64 percent, and 50 percent of consumers were continuously housed for one, two and five years, respectively. The risk of being discontinuously housed was highest in the first four months following housing placement, and this risk was greatest for those in high-intensity housing. The researchers also found that older age was associated with longer tenure, while a history of substance abuse was associated with shorter tenure. In addition, consumers referred from psychiatric hospitals were more likely to have poor housing outcomes regardless of the type of housing. The authors conclude:

Discussions about housing for this population have at times unnecessarily pitted the residential continuum model against the supported housing model. Although some individuals will initially benefit from normalized housing, others may require various degrees of structure, interpersonal intensity, and support. Varied types of housing are needed to meet the heterogeneous needs of a diverse consumer group.

[ Go to Contents ]

Implications for Preventing and Ending Homelessness

Implications for Families

The National Alliance to End Homelessness recently developed a set of Promising Practices to End Family Homelessness (2006). The strategies identified include some of the themes reflected in the research cited here:

Prevention.  The dilemma of prevention strategies is that it is hard to distinguish a person or family that will become homeless without an intervention from one who will not and, therefore, hard to target resources without the homeless services system taking on the whole burden of providing affordable housing for people with low-incomes. This is particularly the case for homeless families, who may be very difficult to distinguish from other low-income families with unstable housing and job histories and with some level of behavioral health problems. Burt and Pearson (2005) conclude that effective approaches include a single agency or system controlling the eligibility determination process, a community commitment to provide housing subsidies for a particular at-risk population (including funding that may come from a non-housing mainstream source such as the mental health or child welfare system), and having a system in place to provide feedback on success.

Short-term rental assistance mitigates the targeting dilemma for prevention strategies since people are not likely to create a housing crisis in order to get help with security deposits or one or two months rent. Increasing numbers of communities are likely to use this approach as part of their plans to end homelessness for this reason, because of its relatively low cost, and because of the limited availability of longer-term, mainstream rent subsidies.

Rapid placement into permanent housing.  Clearly it is best, particularly for children, if a family can limit the duration of shelter placements or bypass shelters altogether. Given the need for a safe and supportive environment for children, there does not appear to be an analog to safe havens that can be applied to families.

Rapid placement into permanent housing is as promising an approach for families as it is for individuals. However, it is less clear which of the features of the housing first model are relevant to families; for example, whether services should be completely voluntary or whether the family should be expected to enter into a services plan and to follow it after the housing placement.

Transitional housing for families as a housing and services model may well have a role to play in a community’s strategy to end homelessness. However, communities should be clear about the purpose and its precise role in their strategies. Is transitional housing to be targeted for those for whom rapid placement into permanent housing is not feasible — for example, because of active substance abuse or other issues on which progress must be made before public or private providers of mainstream housing will sign a lease? Or, is transitional housing a service-enriched living environment to be offered to those families most likely to use it to lift themselves out of poverty and to give their children better life chances — even though such families could go directly to mainstream permanent housing? Communities that make the latter choice should be aware that doing so can draw funds away from interventions more directly targeted to ending homelessness and should seek to fund this type of transitional housing through broader resources such as TANF or the child welfare system.

Mainstream housing opportunities and permanent supportive housing.  Mainstream permanent housing has a crucial role to play in preventing and ending family homelessness. Findings from the Sound Families program indicate that families with limited supportive services needs can be served effectively in public housing and voucher-assisted units. From the findings on Beyond Shelter’s programs and on the Family Permanent Supportive Housing Program, we can conclude that mainstream assisted housing can also be appropriate for multi-problem families when sufficient services support is provided. But is this mainstream housing or permanent supportive housing? For families, the line is blurred by the fact that most families with children need intensive services only for a limited time after placement into permanent housing.

This implies that more funding for mainstream assisted housing programs is needed. The alternative is to redirect existing resources (in particular, housing vouchers and the Low Income Housing Tax Credit) to provide access to affordable housing for people leaving homelessness. Which families need long-term intensive services has been little studied. HUD’s Shelter Plus Care program answers that question by making only families with a disabled head of household eligible for permanent supportive housing. However, lack of access to mainstream assisted housing may put pressure on communities to develop permanent supportive housing for families using the Supportive Housing Program or local and state resources.

Implications for Individuals with Disabilities

Prevention.  Targeting prevention programs may be less difficult for individuals than for families, because of evidence that interventions that include housing reduce the use of expensive medical services by people with certain types of disabilities. Such “high users” can be targeted, as California’s AB2034 program does, to avert their becoming homelessness. Other obvious targets, because they are at such high risk of becoming homeless, are people with disabilities leaving psychiatric hospitals and correctional institutions (Burt & Pearson, 2005).

Safe havens and housing first.  For programs serving individual persons with disabilities who become homeless, housing models with low-demand services have shown positive outcomes, especially for those who have been reluctant to enter or stay in transitional programs. Yet those with the most severe mental illness and substance abuse issues are still the most likely to leave, even from low-demand housing settings. Identifying risk factors in the program population is important as are services focused explicitly on retaining housing.

For communities with a sizable population of service-resistant individuals, safe havens can be an important part of a strategy to end street homelessness. The research evidence suggests that this approach can be costly, however.

Evidence also suggests that approaches that combine a low-demand approach with available intensive services help some succeed in permanent housing who otherwise would be at substantial risk of failing (Tsemberis & Eisenberg, 2000, Padgett, Gulcur, & Tsemberis, 2006; Siegel et al., 2006). Housing configuration seems to be less important than the service approach, although more research is needed to confirm this. Researchers have found positive housing retention outcomes in programs with a wide variety of housing configurations, from buildings dedicated to formerly homeless people with disabilities to mixed-occupancy buildings to scattered-site models. Services need to be available and adapted to the housing configuration. On-site support may work well in buildings where all the residents are program clients. In programs with scattered-site and/or mixed housing configurations, low client-to-staff ratios and frequent contact with clients are important in ensuring clients have sufficient support to maintain their housing. Balancing consumer choice and access to subsidies poses a policy dilemma in addressing the housing needs of homeless people who prefer scattered-site housing in their communities. There are not enough mainstream subsidies to meet the overall demand from people who are homeless and others of low-income, and people with mental illness or other disabilities may face greater barriers accessing the limited available subsidies.

[ Go to Contents ]

Recommendations for Future Research

The challenge facing researchers is that there are so many programs in the field, each influenced by its housing market, service delivery system, community funding, and institutional capacity. Authors of many of the multisite studies cited in the research findings above acknowledge that the researchers were not always comparing “apples to apples.” While programs in a multisite study may have similar overall approaches, the intervention can easily be different enough from site to site that the findings are difficult to compare. For example, differing credentials for case managers, varying landlord receptivity to housing homeless people, mixed housing types, different administrative procedures, or other factors can influence outcomes in ways that are difficult to observe or measure.

Thorough and accurate descriptions of both the service and housing interventions are crucial to expanding our knowledge. We need greater rigor in classifying exactly what the services are, how they are delivered, and how service approaches are linked to housing: how is the housing setting structured, what is the nature of the housing and services provided, and over what period of time? In addition to substantially improved methods for documenting and measuring the types and intensity of housing and services interventions, use of more rigorous experimental or quasi-experimental design studies would strengthen our knowledge of what works for whom. Given the challenges such studies entail, it is important to focus research efforts on the most critical questions. We offer several suggestions:

[ Go to Contents ]


Allen, M. (2003). Waking Rip van Winkle: Why developments in the last 20 Years should teach the mental health system not to use housing as a tool of coercion. Behavioral Sciences and the Law, 21, 503–521.

Arthur Anderson LLP, University of Pennsylvania Health System, Sherwood, K. E.; & TWP Consulting. (2002). 2002 Connecticut Supportive Housing Demonstration Program: Final evaluation report. Retrieved from

Barrow, S., Soto, G., & Cordova, P. (2004). Final report on the evaluation of the Closer to Home Initiative. New York: Corporation for Supportive Housing.

Barrow, S., & Zimmer, R. (1999).Transitional housing and services: A synthesis. In L. B. Fosberg & D. L. Dennis (Eds.), Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Bassuk, E., Huntington, N., Amey, C. H., & Lampereur, K. (2006). Family permanent supportive housing: Preliminary research on family characteristics, program models, and outcomes. Corporation for Supportive Housing.

Bayer, K., & Barker, M. T. (2002). Rebuilding lives: The impact of supportive housing. Enterprise Community Partners.

Bodonyi, J., & Erwin-Stewart, N. (2005, August). Evaluation of the Sound Families Initiative. Seattle: University of Washington School of Social Work, Northwest Institute for Children and Families.

Brown, J. G. (2004). Supported housing programs for the homeless mentally ill: A survival analysis. Unpublished master’s thesis, Drexel University, Philadelphia.

Burt, M. (2006, September). Characteristics of transitional housing for homeless families — Final report. Washington, DC: The Urban Institute.

Burt, M., & Anderson, J. (2005). AB2034 program experiences in housing homeless people with serious mental illness. Corporation for Supportive Housing.

Burt, M., Aron, L., Douglas, T., Valente, J., Lee, E., & Iwen, B. (1999). Homelessness: Programs and the people they serve: Summary report — Findings of the National Survey of Homeless Assistance Providers and Clients. Washington, DC: The Urban Institute,.

Burt, M., & Pearson, C. (2005). Strategies for preventing homelessness. Washington, DC: U.S. Department of Housing and Urban Development.

Carling, P. J. (1993). Housing and supports for persons with mental illness: Emerging approaches to research and practice, Hospital and Community Psychiatry, 44, 439–442

City of Portland, Oregon. (2004, April 9). Sten commends mayor for commitment to end homelessness (Media Release).

Culhane, D., Metraux, S., & Hadley, T. (2002). Public service reductions associated with placement of homeless persons with severe mental illness in supportive housing. Housing Policy Debate, 13(1), 107–163.

Culhane, D. (2006, September 19). Using HMIS data for research and evaluation. Paper presented at the National HMIS Conference, Denver, Colorado.

Diamond, R. J. (1996). Coercion and tenacious treatment in the community: Applications to the real world. In D. L. Dennis & J. Monahan (Eds.), Coercion and aggressive community treatment. New York: Plenum.

Feins, J., & Fosburg, L. (1999). Emergency shelter and services: Opening a front door to the continuum of care. In L. B. Fosberg & D. L. Dennis (Eds.), Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Khadduri, J., & Kaul B. (2005). Permanent housing for homeless families: A review of opportunities and impediments. Paper prepared for U.S. Department of Health and Human Services, Homeless Families Typology Development.

Lipton, F. R., Siegel, C., Hannigan, A., Samuels, J., & Baker, S. (2000). Tenure in supportive housing for homeless persons with severe mental illness. Psychiatric Services, 51, 479–486.

Mansur, E., Quigley, J. M., Raphael, S., & Smolensky, E. (2002). Examining policies to reduce homelessness using a general equilibrium model of the housing market. Journal of Urban Economics, 52, 316–340.

Martinez, T. E., & Burt, M. (2006).The impact of permanent supportive housing on the use of acute care health services by homeless adults. Psychiatric Services, 57, 992–999.

Matulef, M. L., Crosse, S. B., & Dietz, S. K. (1995). National evaluation of the Supportive Housing Demonstration Program: Final report. Washington, DC: U.S. Department of Housing and Urban Development.

Mayberg, S. W. (2003). Effectiveness of integrated services for homeless adults with serious mental illness. A report to the Legislature. Sacramento, CA: California Department of Mental Health.

Mills, G., Gubits, D., Orr, L., Long, D., Feins, J., Kaul, B., et al. (2006). Effects of housing choice vouchers on Welfare families. Cambridge, MA: Abt Associates, Inc.

Morris Davis and Company, Inc., & University of Pennsylvania. (2006, March). Predicting staying in or leaving permanent supportive housing that serves homeless people with serious mental illness. Washington, DC: U.S. Department of Housing and Urban Development,

National Alliance to End Homelessness. (2006). Promising strategies to end family homelessness. Freddie Mac.

Nolan, C., ten Brooke, C., Magee, M., & Burt, M. (2005, August). The Family Permanent Supportive Housing Initiative: Family history and experiences in supportive housing. The Urban Institute and Harder + Company Community Research.

O’Hara, A., & Cooper, E. (2005). Priced out in 2004. Boston, MA: Technical Assistance Collaborative.

Olmstead v. L.C., 527 U.S. 581 (1999).

One Family. (2006, Fall). Housing first: An unprecedented opportunity. Retrieved from

Padgett, D., Gulcur L., & Tsemberis, S. (2006). Housing first services for people who are homeless with co-occurring serious mental illness and substance abuse. Research on Social Work Practice, 16(1), 74–83

Pearson, C. L., Locke, G., Montgomery, A. E., & Buron, L. (in press). The applicability of housing first models to homeless persons with serious mental illness. U.S. Department of Housing and Urban Development.

Quigley, J., & Raphael, S. (2000). The economics of homelessness: The evidence from North America (Working Papers). Berkely, CA: Berkeley Program on Housing and Urban Policy.

Rog, D. (2004). Building the case: Emerging research on solutions to end homelessness. Paper presented at the National Alliance to End Homelessness Annual Conference, Washington, DC.

Rog, D., & Holupka, S. (1999). Reconnecting homeless individuals and families to the community. In L. B. Fosberg & D. L. Dennis (Eds.) Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Shinn, M., Baumohl, J., & Hopper, K. (2001). The prevention of homelessness revisited. Analyses of Social Issues and Public Policy, 1, 95–127.

Siegel, C., Samuels, J., Tang, D., Berg, I., Jones, K., & Hopper, K. (2006). Tenant outcomes in supported housing and community residences in New York City. Psychiatric Services, 57, 982–991.

Spellman, B., Khadduri, J., Fennel, M., & Lord, C. (2006). A federal tax credit program for permanent supportive housing: Design and feasibility. Washington DC: National Alliance to End Homelessness.

Ward Family Foundation, Inc. (2005, July). Safe Haven Programs: Analysis of strategies and operating practices. Alexandria, Virginia: Author.

Tassos, J. (2006). Using the housing credit for supportive housing. Retrieved February 11, 2007, from

Trotz, M. Direct access to housing in San Francisco. (2005). Retrieved from

Tsemberis, S. & Eisenberg, R. F. (2000). Pathways to housing: Supported housing for street-dwelling homeless individuals with psychiatric disabilities. Psychiatric Services, 51, 487–493.

U.S. Department of Housing and Urban Development. (2005). Affordable housing needs: A report to Congress on the significant need for housing. Washington DC: Author.

[ Go to Contents ]


[1]     Worst-case needs households are unassisted renters with incomes below 50 percent of area median income and paying more than 50 percent of their income for rent and utilities. The estimates produced periodically by HUD (and similar estimates published by the Joint Center for Housing Studies of Harvard University) are based on the American Housing Survey, conducted for HUD by the U.S. Census.

[2]     See

[3]     Families selected for the program were current TANF recipients, recent TANF recipients, and those eligible for TANF.

[4]     See Quigley, Raphael, & Smolensky, 2001 for a similar conclusion based on econometric simulations.

[5]     The Earned Income Tax Credit, available only for workers, has become an increasingly important income support for families with children.

[6]     This preference was statutory and was repealed by the Quality Housing and Work Responsibility Act in 1998.

[7]     See

[8]     Some states have been able to put together the Low Income Housing Tax Credit and other funding sources to create permanent supportive housing for people with disabilities, but this is challenging because of the gap between the operating costs of housing and the rents that people with SSI income can afford. See Spellman et al. (2006).

[9]     Emergency shelters are supported primarily by the formula-based Emergency Shelter Grants program. A third McKinney-Vento funded program is the Moderate Rehabilitation for Single Room Occupancy (SRO-MR) program, but this program has been little used in recent years.

[10]    For further information on state LIHTC policies with respect to supportive housing, see Tassos (2006) and Spellman et al. (2006).

[11]    For example, see

[12]    As noted elsewhere in this paper, Shelter Plus Care serves families only if the head of household meets the S+C program’s disability criteria. Persons in families account for roughly 40 percent of persons served in the S+C program.

[13]    The study also notes that the county has an extensive data system that allows it to monitor provider performance to ensure that program contracts are awarded to providers that are achieving target outcomes.

[14]    Source: interviews conducted by one of the authors as part of a HUD-sponsored study of the costs of homeless services.

[15]    Public housing authorities administering the voucher program may “project-base” a portion of their vouchers. Voucher assistance, whether tenant based or project based, typically can be used only for permanent housing. The PHAs providing project-based vouchers to the Sound Families program have additional flexibility in this regard under the demonstration authority known as Moving to Work.


[17]    For a summary of this debate, see Brown (2004).

[18]    See Burt (2006). This study focuses on transitional housing for families, not individuals.

[19]    This indicator applies to all “participants,” including adult individuals, heads of families, and unaccompanied youth.

[20]    The study’s analysis of the financial stability of the supportive housing developments indicated that the costs of departures to the projects in terms of lost rental income were not great enough to cause concern.

[21]    The lodging house selected for study was not a typical shelter, but rather a former flophouse that had been leased by a nonprofit service provider offering lodging to single men in small cubicles. When the study was getting underway, the service provider had plans to redevelop the property as a drop-in center, and thus needed the lodging house residents to relocate.

[22]    The New York, New York Agreement was a response to the lack of housing for homeless persons with serious mental illness in New York City. Implemented in 1990, the agreement was to place 5,225 homeless persons with serious mental illness into housing and to develop 3,314 units of supportive housing. A third wave that will produce 9,000 units is just getting underway.

How to Obtain a Printed Copy

To obtain a printed copy of this report, send the title and your mailing information to:

Human Services Policy, Room 404E
Assistant Secretary for Planning and Evaluation
U.S. Department of Health and Human Services
200 Independence Ave, SW
Washington, DC 20201

Fax:  (202) 690-6562

You may also print the file below in PDF format: (Note that this PDF file is provided for the convenience of those who wish to print the report. The PDF file is not ADA compliant; use the HTML version above for the ADA-compliant version.)

Printer Friendly version in PDF format

Where to?

Top of Page | Contents
Main Page of Report | Contents of Report

Home Pages:
2007 National Symposium on Homelessness Research
Assistant Secretary for Planning and Evaluation (ASPE)
U.S. Department of Health and Human Services (HHS)

Last updated:  01/10/08