Home to the Urban InstituteHelping America's Homeless



About the


Order This


    Chapter 1

    Homelessness has now been on the American policy agenda for close to two decades. In 1989, when the Urban Institute published America’s Homeless (Burt and Cohen 1989), policymakers and the public may have expected, or hoped, that we could end the crisis of homelessness relatively quickly. The decade of the 1990s has not fulfilled that expectation. Programs and services to help homeless people expanded dramatically in the 1990s, just as they did in the 1980s. At the same time, visible homelessness in many American communities does not seem to have diminished. How are we to think about the persistence of homelessness at the end of a decade of unprecedented prosperity, and at the dawn of a new millennium?

    Answers to this question are complex, because homelessness itself is complex. In this chapter, we explore some of these complexities. They form the critical context within which we can begin to interpret the information presented in following chapters. We will look briefly at some of the historical fluctuations in types and levels of homelessness within the United States. These include moments when homelessness has assumed the stature of a "social problem" and moments when it has not. We also undertake a fairly extended discussion of the meanings or definitions of homelessness as a condition experienced by some people.

    Given that homelessness stems, at base, from an inability to afford housing, we next consider the structural conditions of the economy, housing markets, labor markets, and related factors that influence people’s ability to afford housing. We also consider how people with different individual characteristics may be affected. Finally, we look at the ways in which the United States has chosen to address homelessness from the federal level. We save for the concluding chapter a discussion of what societies may be willing to do to reduce the probability that their members will become homeless. Such a discussion will take us into the realm of cultural norms and expectations about individual and societal responsibility for the well-being of individuals and households, and consider their practical extensions in the form of stronger or weaker, publicly funded, "safety-net" or "social-assistance" programs and supports. This discussion is best approached after the reader has absorbed the information about homeless people and homeless assistance programs presented in the intervening chapters.

    A Brief Historical Overview

    Many essays have been written in the last decade and a half examining the historical meaning and experience of homelessness in the United States (Hoch and Slayton 1989; Hopper 1991; Hopper and Baumohl 1994, 1996; Rossi 1989). European countries have also recently focused on the existence and meaning of homelessness in their midst (Avramov 1999). All of these writings conclude that one of the essential characteristics of homelessness as a phenomenon is its transience, instability, and flux. In this, homelessness as a whole resembles the trajectories of its individual members, who enter and leave homelessness,, sometimes repeatedly, sometimes only once, depending on their fortunes. We may distill the essence of these discussions to suggest that three elements, separately or in combination, characterize homelessness for some writers, with regard to some populations, at some times. These elements are the transience or instability of place, the instability or absence of connections to family, and the instability or absence of housing.

    Without Place

    In talking about homelessness, most writers have found it necessary and important to note that most societies have included larger or smaller groups of people who have no fixed place to live, and almost always are poor. Some have been cohesive groups for whom wandering is a way of life, such as nomadic tribes, circus and carnival workers, and many of today’s migrant farm workers. They have families, and usually some form of housing, but no fixed place. Others have been individual wanderers, such as peddlers or tinkers, who move from community to community in their own van or wagon. They have housing, but no family or place.

    Other collections of wanderers have been as changeable in their composition as they are in their location. In the United States, for instance, the last decades of the 19th19th century and first decades of the 20th saw an ever-changing population of single men following construction and industrial development opportunities across the country, and then returning to low-rent sections of cities when work was scarce. Most often, individuals in these groups were not "literally" homeless in today’s sense of the word, because they usually could afford a room in a cheap hotel or boarding house. But they were unattached to a particular place, and without family, and thus without a home in the larger sense of the word.

    Depending on economic conditions and cycles, these groups of transients swelled or diminished. In really bad times, local out-of-work laborers joined the itinerant core (houseless, but not placeless). During the Great Depression of the 1930s, when unemployment reached the astonishing level of one-quarter of the workforce, an official government survey in the United States estimated that at least 1.2 million people (or 1 percent of the total population) were homeless at a single point in time (mid-January 1933) (Crouse 1986). Further, the researchers felt this number was conservative. In addition to these patterns, intra- and international migration flows caused by war, famine, economic displacement (such as the failure of rural economies), and the economic draw of work in developed countries also contribute to houselessness, uprootedness, and large numbers of at least initially transient people flooding into receiving communities. Even homeless service systems can create this effect. For example, Milofsky and colleagues (1993) describe the effects on a small town of the influx of large numbers of homeless people relocating from nearby metropolises to obtain substance abuse recovery services.

    Without Family

    Studies of skid row populations in the 1950s and 1960s (Bahr and Caplow 1974; Bogue 1963; Wiseman 1970) provided a different lens on homelessness. The situations they studied were different. These studies described a population, mostly of single men, who were housed, lived steadily in a particular part of a particular city, but lived by themselves. That is, they did not live with any family members although they clearly lived in hotel rooms with many other people on the same and adjacent floors. Very few men in these communities would have been classified as literally homeless by today’s formal government definition, yet they were considered homeless by the people who studied them. Even the U.S. Census Bureau, as late as the 1980 decennial census, identified people who lived by themselves and did not have a "usual home elsewhere" (i.e., with family) as "homeless." This way of thinking about homelessness reflects a cultural expectation that the "normal" way to live is in a family, and that something is wrong when people live by themselves. "Home" in this usage implies people, not physical shelter.

    Homeless people, themselves, in the 1980s and 1990s drew a distinction between being sheltered or housed, and having a home. One of the great difficulties in gathering information about patterns and histories of homelessness comes from the distinction between "housing" and "home." A common question on homeless surveys, designed to determine how long a person has been homeless, asks in effect, "When was the last time you had a home or other permanent place to stay? " or "When was the last time you had a place of your own, such as a house, apartment, room, or other housing, for 30 days or more in the same place?"1 It is not uncommon for people to respond that they last had such a place a very long time ago. Some people say "never." Yet further probing uncovers that people have spent significant periods of time living in conventional dwellings where their tenure was insecure and/or they did not feel "at home." This could include accommodations such as hotel rooms they had paid for themselves. Thus their experience of literal homelessness was much shorter, or much more intermittent, than would be implied by their answer to the survey question. But their answers accurately reflect how long they have been without significant attachments to people.

    Without Housing

    Writers struggling with the concept of homelessness also recognize that individuals and families may become homeless "in place," so to speak. Fire, flood, or natural disaster could render people homeless by destroying their dwelling. Or circumstances could change the ability of families and communities to accommodate certain of their members, causing them to eject the most burdensome. Or people’s own behaviors may make it difficult for family and community to support them in housing. In the United States, official government definitions of homelessness focus exclusively on this "houseless" aspect.

    If money were no problem, these temporary losses of a dwelling could easily be overcome, without risk of becoming an actual or potential social problem. Essential elements of homelessness as a social problem are a level of poverty so extreme that homeless people cannot remove their homeless condition themselves, and an unwillingness or inability of society either to do it for them or to establish conditions that would not make them so desperate in the first place. Finally, even so apparently clear a condition as being houseless is specific to particular times and places. Residents of shantytown dwellings under a bridge or along a roadside, for example, are considered homeless in New York City, where such structures are not allowed, but are well-housed in many cities in the developing world, where such structures are the norm for millions.

    Today’s homeless in the United States are houseless, and some are placeless (transient). Many are also without family. But some significant number bring at least part of their family with them into homelessness, and maintain some reasonable degree of connectedness with housed family members. Over time, these "connected" people comprise a larger proportion of people experiencing homelessness than the isolates, because "connected" people stay homeless for relatively shorter periods and are replaced by others like themselves. In this respect American homelessness of the 1980s and 1990s differs considerably from that of the 1950s and 1960s, but perhaps not so much from the homeless experiences of families in local communities during hard times in the 19th century or during the Great Depression of the 1930s.

    Further Definitional Considerations

    The foregoing discussion notes historical circumstances that have been considered part of the phenomenon of homelessness at different times. Definitions of a phenomenon such as homelessness usually embody one or more social purposes. They are not neutral, but, rather, are constructed to affect public concern and action. Thus they are "programmatic." They may be trying to make a phenomenon visible, to get it defined as a "problem," and thus make it worthy of intervention and amelioration. They may be trying to do the opposite. They may be trying to influence the scope of action expected or demanded. Or they may be trying to influence the value placed on one way of living compared to others.

    Definitions are necessary, though, from several perspectives. From the perspective of immediate action, definitions identify who is eligible to receive whatever assistance is available specifically for homeless people. From a research perspective, definitions identify who should be counted and described. And from a policy perspective, definitions identify who should be planned for and what policies will be most relevant to the type of assistance needed.

    However, definitions of a phenomenon such as homelessness are difficult, and frequently require balancing two horns of a dilemma. If the definitions are too inclusive, they become useless; the phenomenon becomes too diffuse, ultimately covering too many people. With homelessness, this tendency is manifested by definitions that threaten to include the entire population in poverty, or everyone who is poorly housed. But if the definitions are too specific, they focus too exclusively on the homelessness of the moment. They can lead to policies and practices that are ameliorative but not preventive, that fail to address the larger question of desperate poverty and the pool of people at high risk for periodic bouts of literal homelessness.

    At present, United States government policy and access to particular kinds of government-supported assistance are driven by a clear but narrow definition of homelessness, based on a person’s sleeping arrangements. "Literal" homelessness is defined on a day-by-day basis, and involves sleeping either in a facility serving homeless people, in accommodations paid for by a voucher from a program serving homeless people, or in places not meant for human habitation. This official definition narrows homelessness to a fairly small proportion of the precariously housed or unhoused population. About the only way to shrink the group further would be to exclude anyone in a shelter or transitional housing program, or using an emergency voucher, during the night of such use. The official definition is meant to help providers determine whom to serve, and to help planners calculate levels of service to provide.

    While conveniently precise from some perspectives (it is relatively easy to ascertain where a person slept on the previous night, and therefore easy to determine whom to help), the official definition is overly precise and therefore misleading from other perspectives. For example, if people can afford to pay for a motel room three nights a week, but sleep in the park the other four, and this goes on week after week because they cannot afford a room, it seems relatively meaningless to say they are homeless only on the nights they sleep in the park. Truly helping them would mean helping them achieve stable housing. Similarly, if someone has no stable place to stay, but obtains housing from a relative for two or three days until space in a homeless shelter opens up, it seems relatively meaningless to say the person is housed (because sleeping in conventional housing) and therefore not eligible for the homeless service.2 The ultimate failure of the official definition is that it does not help us address the larger issues of how to end homelessness for the long term. It only helps decide who should receive services at any particular time.

    Structural Factors Affecting Levels of Homelessness

    The "homeless population" will be larger or smaller depending on one’s definition of homelessness, regardless of anything real happening in society to increase people’s risk of being without place, without family, or without housing. But if we want to address the larger issue of ending homelessness, rather than just ameliorating the conditions of some homeless people for brief periods, then we need to try to understand why we have homelessness now, and why we have it here. Only then can we ask who will become homeless out of all the people vulnerable to the condition. Further, the answers should point toward solutions.

    Early in the 1980s, two opposing explanations of homelessness quickly gained rhetorical prominence. As many have described before (e.g., Burt 1992; Koegel and Burnam 1992), one explanation emphasized structural factors, the other individual factors, as "the cause" of homelessness. The structural argument focused on the impact of changing housing markets for low-income families and singles, changing employment opportunities for people with a high school education or less, and changing institutional supports for persons with severe mental illness (epitomized by drastic reductions in the use of long-term mental hospitalization). The effects of persistent poverty and racial inequalities were also offered as structural explanations for homelessness in the 1980s. In general, liberal-to-radical activists and analysts were most likely to focus on structural factors.

    The individual argument focused on personal problems and inadequacies, since homeless populations routinely report many more such problems than the general public. Individual factors include adult and childhood victimization, mental illness, alcohol and/or drug abuse, low levels of education, poor or no work history, and too-early childbearing. In general, conservative activists and analysts were most likely to focus on personal factors. But often joining them were clinicians and practitioners, who recognized the seriousness of the problems and wanted to help people deal with them.

    Over the years, most people have come to recognize that both structural and individual factors play a role in producing homelessness. The structural factors set the stage, without which fewer people would be homeless. They help to answer the question, "Why more homelessness now?" (Koegel and Burnam 1992). Then the individual factors help to identify "Who?"—which particular people subject to the worst combinations of structural factors are most likely to lose their housing. However, as the structural conditions worsen, even people without personal vulnerabilities other than poverty may experience crises that precipitate a homeless episode.

    The full effect of both structural and individual factors must be understood in the context still another set. That is the degree to which a society provides a safety net, in the form of social insurance (e.g., Social Security), social assistance (e.g., housing, food, child care, and other subsidies), and services (e.g., mental health services), intended to cushion the more-severe consequences of structural and individual factors. The better the safety net, the less effect either structural or individual factors will have on producing homelessness.

    In Burt’s work (1992) identifying causes of the growth of homelessness in the 1980s, she began with a model that incorporated structural and individual factors, along with safety net programs (figure 1.1), showing how all three may influence the ability to afford housing and, hence, the probability of homelessness. Housing affordability was, and still is, assumed to be the immediate cause of homelessness. If the cost of housing is too high for the incomes of households that need housing, then homelessness is likely to result for at least some of those households. Housing affordability is represented in the middle of figure 1.1 by the (potentially unequal) relationship between household resources and housing costs.

    Figure 1

    Factors affecting whether a household will have sufficient income include job opportunities (a structural factor); personal factors, both individual and household, which appear in the box at the lower left of the figure; and government policy with respect to cash and non-cash benefit programs that are part of the safety net. Factors shown affecting the availability and cost of housing include fiscal and monetary policy, tax policy, interest rates, and government policy specifically affecting low-cost housing, including housing subsidies. The box labeled "housing market" incorporates policy influences plus the effects of personal incomes on the cost of housing. If the personal incomes of many people are stimulated by a booming economy, the housing market will be high, making it more difficult for people with lower incomes to afford housing.

    In addition to addressing Koegel and Burnam’s (1992) question, "Why more homelessness now?" structural factors can also address the question, "Why more homelessness here (as opposed to there)?" These, in turn, may shed some light on why we still have high levels of homelessness after a decade of significant economic expansion and a rising standard of living for most Americans. Burt’s 1992 analysis of differences in levels of homelessness among U.S. cities with 100,000 or more people showed homelessness in cities with very depressed economies. But cities with extremely rapid population growth due to expanding economic opportunities also showed significant homelessness. Only in the latter cities was cost of living a factor in predicting homeless levels.

    What appears to have happened in the latter cities, and in many parts of the United States in recent years, is that economically successful people put more pressure on housing markets, which drives up prices and increases housing’s scarcity relative to households that want it. The effects of such pressures are felt particularly at the low end of the market. Many analyses (e.g., Dolbeare 1999) show dramatic reductions in the number of housing units low-income households can afford, if they spend no more than one-third of their income for housing. All other things being equal, the fewer such units there are in comparison to households with incomes at or below the level used in these analyses, the more likely some of the households are to become homeless. Under these circumstances, homelessness will happen even to people whose only personal vulnerability is poverty. Thus, paradoxically, the success of the many plays a role in creating conditions conducive to homelessness for the (relatively) few who have been left behind by boom times.

    In 1996 36.5 million people lived in poverty in the United States—13.7 percent of the population. Only 2 percent needed to be homeless at any given time to produce the level of homelessness we report in chapter 2. It is not hard to imagine that rising housing costs coupled with stable wage levels for low-skilled workers could, in turn, combine with the periodic crises that beset poor households to push this many people into homelessness at any given time. To illustrate, in 1996–1997, according to the National Survey of American Families, 28 percent of families with children whose incomes were below 200 percent of the federal poverty level had been unable to pay their mortgage, rent, or utility bills at least once during the previous 12 months. This was true also for 17 percent of childless adults (Wigton and D’Orio 1999). This level of difficulty in affording housing, even among people with incomes up to twice the federal poverty level, bespeaks a large pool of people, including many households with children, who have some vulnerability to becoming homeless.

    The Government Response to Homelessness in the United States

    American government at all levels was active in the homeless arena during the 1980s and 1990s. In the mid-1980s, state and local governments greatly increased their commitments to homeless services. HUD’s 1988 study, collecting data in the first quarter of 1988, before anything but a trickle of federal money was actually in use, found that the total dollars committed to shelter services in the United States had climbed to $1.6 billion, from $300 million in 1984. HUD also found that the share of this much larger figure coming from state and local governments had increased to 65 percent in 1988, from 37 percent in 1984 (DHUD 1989).

    In 1987, homelessness had been visible long enough, was extensive enough, and had been documented often enough through research, that Congress inaugurated a major federal effort to address it through the Stewart B. McKinney Homeless Assistance Act. Most McKinney Act programs required their recipients to match federal dollars with a certain percentage of other funds. For states and localities that had long been spending their own tax dollars to serve homeless people, the match was easy to meet and did not require a major change of policy. But for states and localities that had never committed any of their own money toward programs for the homeless, the McKinney matching requirement stimulated creation of new statewide homeless task forces, coordinating committees, and funding programs. Thus the McKinney Act generated significantly more money for homeless services from state and local governments than had previously been available.

    Much of the McKinney Act money went, in the early years, toward fixing, renovating, or building structures that could serve as shelters. As a consequence, the number of shelter beds in the United States increased by about half in the Act’s first three years. (Burt 1992). Further, the types of shelter being built changed. McKinney funds made possible—for the first time—transitional shelters, primarily for families and the mentally ill, and specialized permanent housing with supportive services, for homeless people whose disabilities and handicaps made it unlikely they could ever house themselves. The Act supported demonstration projects that established the feasibility of bringing mentally ill, alcoholic, or drug abusing homeless people in off the streets and creating living environments that could keep them in stable housing.

    As the years have passed, the uses of McKinney Act money have shifted. After building new shelters and expanding the national capacity to shelter homeless people, the need for more shelter beds has waned. However, the needs still are pressing to staff shelters and offer rehabilitation services to homeless people through shelters and other mechanisms. The Department of Housing and Urban Development (HUD), which generally uses its money to build structures rather than run them, has found its homeless dollars going more and more for services to run programs than for "bricks and mortar."

    New Policy Orientations

    For their first few years, during the Reagan administration, McKinney Act programs were funded significantly below their authorized levels. The Bush administration supported some additional funding, but it took the installation of a new Democratic administration in 1993 for homeless policy in Washington to take a new turn.

    The first actions of the Clinton administration were to increase funding for McKinney Act programs. In 1995, funding hit more than $1 billion. A second major policy shift was initiated by the Department of Housing and Urban Development (HUD). HUD officials recognized that homeless services had grown and developed haphazardly, without providing assurance to homeless people in any community that needed services of a particular type would be available. The new administration promoted the related ideas of a "continuum of care" and "coordinated services." That is, the administration wanted communities to make care available at each level or stage on a continuum that might be needed by homeless people. This included intake and assessment, emergency shelter, transitional shelter, and specialized services such as alcohol and drug treatment. Further, the administration wanted homeless clients to be able to access the various services through a coordinated system—with all providers in a community knowing about available services and how to help their clients receive them.

    As a result, new applications for funding under the various homeless programs run by HUD had to show how the new services fit into a continuum of care and how they would be coordinated with other services. These requirements pushed local communities and service providers to talk to each other and learn more about available community resources. Interaction and increased cooperation between local governments and community service providers in their communities was one result—to the benefit of both local service networks and their homeless clients. Chapter 10 explores the nature of these networks.

    Another policy change in the early 1990s was increased attention to rural homelessness. In 1995, the Rural Economic and Community Development (RECD) administration of the Department of Agriculture sponsored four regional conferences on rural homelessness (Burt 1995). The conferences pursued several objectives. They were designed to identify the nature and needs of the rural homeless, and how they might differ from the urban homeless; to learn how available federal programs were being used, and barriers to their use; and to explore how RECD programs and initiatives, in particular, could be structured to fill gaps in the network of available services. In addition, the National Survey of Homeless Assistance Providers and Clients (NSHAPC) covered rural as well as urban areas, giving, for the first time, a systematic view of rural homelessness in relation to urban homelessness. This is described in chapter 7.

    The 1990s also saw significant advances in knowledge about what does and does not work in helping people while they are homeless, and then to leave homelessness. In 1999, the Departments of Health and Human Services and Housing and Urban Development sponsored a symposium to summarize and synthesize a decade of learning from program evaluations (Fosburg and Dennis, 1999). The chapters in the summary volume focus on a wide variety of program types, from street outreach, to case management, to permanent housing plus supportive services. They make clear there are some things we know how to do and do well, but other needs where we are only just beginning to think about what might work. The greatest successes have come in programs helping people with the longest histories of homelessness and the highest levels of disability. The answer, succinctly put, is "housing" (Shinn and Baumohl 1999). Subsidize their housing, and they become—and stay—housed. The greatest gaps lie in the areas of short-term homelessness and family homelessness, which overlap considerably but are not identical. These gaps may exist because only recently have we come to recognize that a very large number of people experience short-term homelessness in the course of a year or several years. Programs, and evaluation of them, are just beginning to grapple with the implications of these findings.

    New Evidence for an Old Phenomenon

    A final thrust of policy in the 1990s emerged because of new evidence that the number of people who experience homelessness during the course of a year or longer could be as much as six times the number homeless at any given time (Burt 1994; Culhane et al. 1994; Link et al. 1994, 1995). Further, these studies also documented that the proportion of the U.S. population experiencing homelessness in the past five years could be as high as 3 percent (7 to 8 million people—Link et al. 1995, 1994), and that even the proportion experiencing homelessness in the past year approaches 1 percent of all Americans and a higher proportion of poor Americans. We make our own estimates of annual homelessness in chapter 2, where we come quite close to those just described.

    All this evidence has pushed policymakers to begin distinguishing between policies and services appropriate for alleviating short homeless episodes and those appropriate for people whose homelessness is long-term, and to recognize that the same set of policies and services might not be needed, or appropriate, for all homeless people. Some states and localities, for instance, have initiated "diversion" programs that help people experiencing short-term homelessness. These programs screen applicants for shelter services and identify those who could be helped to maintain or regain their housing with a little cash, negotiations with a landlord, or other immediate assistance.

    Homeless-Specific versus Generic Services—

    The Weakness of the U.S. Safety Net

    The vast majority of programs for homeless people discussed above, which have seen such growth in the last decade, are homeless-specific. That is, they are designed for homeless people, delivered to homeless people, and operated by programs and agencies focused on serving homeless people. Throughout the late 1980s and early and mid-1990s, services for homeless people were the only programming for poor people into which the federal government was putting, as opposed to cutting, money. Therefore, homeless-specific programs became something of a "growth industry." Homeless service providers argued that since they knew and worked with homeless people, they could help homeless people better than generic service providers. As a consequence, we now have health care, job training, GED preparation, substance abuse treatment, and even apartment complexes (transitional housing facilities) specifically for homeless people. We also have residential facilities for formerly homeless people with disabilities.

    This structure has its advantages and disadvantages. There can be little question that generic services—those that serve the general population rather than focusing on a particular user group—have not been either interested in serving homeless people or easy for them to use. These generic services include various cash assistance programs (Aid to Families with Dependent Children, General Assistance, Social Security Income, Social Security Disability Insurance), in-kind benefit programs (Food Stamps, Medicaid and state medical assistance, federal, state, and local housing subsidies), and problem-focused services (mental health, alcohol, and drug treatment services). Most of their disinterest can be attributed to lack of resources to assist even their "usual" clientele. Many of these programs faced cuts in funding and/or restrictions on eligibility during the period we are considering, rendering them less able to provide a previous level of service. Applying for these programs can be prohibitively difficult for people who cannot readily acquire or maintain needed documentation, and agencies do not go out of their way to make the process any easier. Another part of the problem is that homeless people often are perceived as "hard to serve," and generic agencies do not take the trouble unless required to do so. Some generic programs actually changed rules in ways that reduced the eligibility of many homeless people. This included many state general assistance programs and Supplemental Security Income at the federal level. Thus homeless people themselves, and people vulnerable to homelessness, have actually lost important sources of income, especially in the late 1990s.

    On the other hand, homeless-specific services help people who are homeless now. The federal definition of homelessness and eligibility for assistance with federal homeless dollars make this specialization necessary. Only in rare instances have even the most far-reaching local "homeless plans" extended themselves to reduce homelessness, prevent homelessness, or change the conditions that push people into homelessness. Thus homeless programs usually remain Band-Aids, even though they make the homeless condition less onerous for a lot of people. When they are able to help people leave homelessness, it is often because they have succeeded in making up for failures of many systems in the past, including school systems, child welfare systems, juvenile and adult justice systems, mental health systems, and substance abuse treatment systems. Further, even though they absorb a good deal of money in the context of homelessness, their funding is small compared to the resources potentially available to generic services. And they have not stanched the flow of people into homelessness.

    The obvious policy challenge still is, after a decade and a half, "What will we need to do to end homelessness?" The answer is at one level obvious—make housing affordable. But how do we get there, if we do? We return to this question in the concluding chapter, where we explore the pushes and pulls on the American political scene that will shape any approaches to solving homelessness.

    Sources of Data for this Book

    To help generate useful approaches to thinking about homelessness in the United States for the future, this book describes homeless people and homeless service systems across America toward the end of the 1990s. It is based on data from the National Survey of Homeless Assistance Providers and Clients (NSHAPC), conducted in 1996.3 This survey builds upon and extends the methodology first employed in the Urban Institute’s 1987 study, reported in America’s Homeless. It offers the first opportunity in almost a decade to look at homelessness from a national perspective that is based on information collected directly from homeless assistance programs and the homeless—and other—people who use them.

    NSHAPC is unique in its scope and coverage.4 First, it covered the entire United States, offering a systematic view of suburban and rural homelessness as well as the more familiar homelessness found in central cities.5 Second, it offers the only national data for the 1990s on which to base estimates of the size of the homeless population and changes since 1987 It also provides the opportunity to create national estimates of homelessness at a single point in time and over the course of a year. Third, it offers comparisons among currently homeless, formerly homeless, and never-homeless people who use homeless assistance programs other than shelters, such as soup kitchens or drop-in centers. Fourth, NSHAPC’s rich information about client histories and characteristics lets us examine patterns of homelessness, childhood and adult antecedents of homeless episodes, and changes in the adult homeless population between 1987 and 1996.

    Finally, NSHAPC offers a new and comprehensive look at homeless assistance programs nationally and in the 76 metropolitan and rural areas that comprise the study’s sampling locations. This information lets us see the effects of public policy that encouraged organizing homeless service systems into continuums of care. Looking at this new evidence, and integrating it with other major advances in knowledge about homelessness gained in the 1990s, this book concludes with a discussion of implications for how we might want to think about and address homelessness in the future.

    Sixteen types of homeless assistance programs were included (full definitions appear in the appendix):

    • Emergency shelters
    • Transitional housing programs
    • Permanent housing programs
    • Programs distributing vouchers to obtain emergency accommodation
    • Programs accepting vouchers in exchange for giving emergency accommodation
    • Food pantries
    • Soup kitchens
    • Mobile food programs
    • Physical health care programs
    • Mental health programs
    • Alcohol/drug programs
    • HIV/AIDS programs
    • Outreach programs
    • Drop-in centers
    • Migrant labor camps used to provide emergency shelter for homeless people
    • Other programs

    Defining Homeless Status

    Since homelessness is the core concept in the analyses in all subsequent chapters, it is important for the reader to understand how it was defined. Because NSHAPC sampled all clients—whether homeless or not—of the homeless assistance programs just described, it was necessary to classify the clients as currently, formerly, or never homeless, based on the answers they gave to survey questions. The following alerts the reader to the criteria used in differentiating homeless people from other program clients.

    The definitions used follow closely the official federal definition of literal homelessness, except that they are based on experiences during a week rather than on a single day (chapter 2 explains the reason for this approach).

    Currently Homeless: The following specific conditions were used to classify clients as currently homeless:

    • The client stayed in any of the following places on the day of the survey or during the seven-day period prior to being interviewed for NSHAPC:
    • 1. An emergency or transitional shelter, or

      2. A hotel or motel paid for by a shelter voucher, or

      3. An abandoned building, a place of business, a car or other vehicle, or anywhere outside.

    • Or:
    • 4. Reported that the last time they had "a place of [their] own for 30 days or more in the same place" was more than seven days ago, or

      5. Said their last period of homelessness ended within the last seven days, or

      6. Were identified for inclusion in the NSHAPC client survey at an emergency shelter or transitional housing program, or

      7. Reported getting food from "the shelter where you live" within the last seven days, or

      8. On the day of the interview, said they stayed in their own or someone else’s place but that they "could not sleep there for the next month without being asked to leave."

    Use of the first criterion (shelter use) classifies 34.9 percent of the sample as currently homeless. Criteria two (voucher use) and three (places not meant for habitation) add 1.7 percent and 9.8 percent, respectively, for a total of 46.4 percent. The five remaining criteria together add another 7.1 percent, for a final total of 53.5 percent of the sample classified as currently homeless. All but the final criterion meet the McKinney Act definition of homelessness; the last criterion adds only 0.3 percentage points to the final proportion classified as currently homeless, and was included because the survey itself treats clients in this situation as homeless.

    Formerly Homeless: Many clients who were not literally homeless reported having been homeless at some earlier time in their lives (22 percent of the sample). The circumstances used to classify clients as formerly homeless also meet the McKinney Act definition of homelessness. Clients were classified as formerly homeless if they:

    • Did not meet any of the conditions qualifying them as currently homeless but at some point in their lives had stayed in any of the following:
    • 1. An emergency or transitional shelter, or

      2. A welfare/voucher hotel, or

      3. An abandoned building, a place of business, a car/other vehicle, or anywhere outside, or

      4. A permanent housing program for the formerly homeless, or

    • Said they had previously had a period when they were without regular housing.

    Never Homeless: The remaining 24 percent of NSHAPC clients had never been homeless at the time of the survey, according to the criteria used here, and also said they had never been homeless. They are referred to throughout this book as never homeless.


    All data reported in this book have been weighted to be nationally representative.6 Without weighting, respondents to a survey such as NSHAPC represent only themselves. Weighting makes their responses representative of all people using homeless assistance programs in the United States in the fall of 1996.

    All data describing client characteristics have been weighted to be nationally representative of clients of homeless assistance programs during an average WEEK from October 18 through November 14, 1996; that is, the week before the client was interviewed. The client weights were constructed to assure that a person is not double-counted even if she or he used several homeless assistance programs during the course of the seven-day period. In addition, the client weight itself is used as the basis of estimates of population size found in chapter 2.

    Information describing programs and service locations is weighted to be nationally representative of homeless assistance programs and service locations on an average day in February 1996. In addition, the program and service location weights themselves have been used as the basis for estimating the total number of service locations, programs, and service contacts in the nation, as discussed in chapters 9 and 10.

    The Plan of this Book

    The remaining chapters explore issues pertaining to people who are homeless, and findings pertaining to homeless assistance programs and service systems in the United States. Chapters 2 through 8, analyzing the NSHAPC client data, include some that are purely descriptive and some that use statistical techniques to address complex questions such as population size or patterns of homelessness. Chapters 9 and 10 focus on the information collected through NSHAPC on homeless assistance programs and services. Specifically:

  • Chapter 2 addresses the question, "How many homeless people are there?" presenting a variety of estimates that are useful for different policy purposes.
  • Chapter 3 divides the homeless sample into subgroups based on family status (with children, alone, and with other adults) and sex, and describes each subgroup in terms of demographic, economic, health, childhood, and other characteristics.
  • Chapter 4 divides the homeless sample into subgroups based on history of alcohol, drug, and mental health problems, and describes each subgroup in terms of demographic, economic, health, childhood, and other characteristics.
  • Chapter 5 examines the situation of children of homeless clients, whether or not they live with the client. It also looks at the characteristics of the youngest homeless adults (those younger than 20, and those aged 20 to 24), and examines the role of childhood abuse, neglect, and out-of-home placement in their homeless careers.
  • Chapter 6 examines homeless episodes, or spells, in several ways. It first looks at subgroups among the homeless based in their pattern of homelessness, in two ways. It uses a simple cross-classification by length of current spell and number of homeless spells, and also uses cluster analysis to identify patterns. Three patterns of homelessness are identified by both methods: episodic (many spells, many of them relatively short), chronic (very long current spell, with relatively few other spells), and crisis/first-time homelessness. It then examines the meaning of "a spell," looking at continuity or breaks in clients’ current homeless episode, and describing the places that clients report staying during a homeless episode that would not, in themselves, be classified as a literally homeless situation.
  • Chapter 7 divides the entire NSHAPC sample into currently, formerly, and never homeless clients, and further divides these subgroups into those found in central cities, suburban/urban fringe areas, and rural areas. It then describes the subgroups in terms of many of the same characteristics as presented in chapters 3 and 4.
  • Chapter 8 uses the entire NSHAPC sample, again divided into currently, formerly, and never homeless clients. It uses logistical regression analysis to estimate factors associated with homelessness by comparing those who have ever been homeless (currently and formerly homeless clients) to those who have never had a homeless episode. Then, within the ever-homeless group, it uses logistical regression analysis to identify factors that differentiate clients who are currently homeless from those who are not now homeless but have had at least one homeless episode in their past.
  • Chapter 9 provides estimates of the number and types of homeless assistance programs in the United States in 1996, and compares these estimates to those from the late 1980s to identify the amount and type of change in homeless services during the 1990s. It also describes homeless assistance programs in terms of their size, geographic location, and populations served.
  • Chapter 10 looks at the issue of systems. It explores the way in which homeless assistance programs are configured, whether standing alone or clustered together with other programs of different types. It also looks at the service offerings by sampling area, assessing the degree to which the policy goal of a continuum of care has been realized in communities of different sizes and types.
  • Chapter 11 summarizes major findings, and draws out their implications for programs and policies in the coming decade. In doing so, it also examines American norms and values for personal and collective responsibility for individuals’ well-being.

  • 1. The first wording was used in the 1987 Urban Institute study (Burt and Cohen 1989); the second wording was used on the National Survey of Homeless Assistance Providers and Clients (Burt et al. 1999).

    2. These examples may seem extreme, but they have happened not infrequently in the course of applying the federal definition of homelessness.

    3. Twelve federal agencies sponsored NSHAPC: the U.S. Departments of Housing and Urban Development, Health and Human Services, Veterans Affairs, Agriculture, Commerce, Education, Energy, Justice, Labor, and Transportation, and the Social Security Administration, and the Federal Emergency Management Administration. The Bureau of the Census carried out the sample design and all aspects of the data collection. Martha Burt of the Urban Institute, along with representatives of the sponsoring agencies, was involved from the study's earliest stages in study design and development of the client survey instrument. She also directed the Urban Institute's work to analyze the NSHAPC data and write up the results for the sponsoring agencies, along with other authors of this book.

    4. For a more complete description of the NSHAPC methodology, see the appendix.

    5. The study used 76 primary sampling areas, including the 28 largest metropolitan statistical areas (MSAs) in the United States; 24 small and medium-sized MSAs, selected at random to be representative of geographical region (northeast, south, midwest, west) and size; and 24 rural areas (groups of counties), selected at random from a sampling frame defined as the catchment areas of Community Action Agencies, and representative of geographical regions. In New England the actual areas sampled were parts of counties.

    6. A full explanation of weighting procedures may be found in Burt et al., Homelessness—Programs and the People They Serve: Technical Report, Appendix D.

    Helping America's Homeless, by Martha Burt, Laudan Y. Aron, and Edgar Lee, with Jesse Valente, is available in paperback from the Urban Institute Press (6" x 9", 363 pages, ISBN 0-87766-701-2, $29.50). To order call (202) 261-5687 or toll-free 800.537.5487.

    Home to the Urban InstituteComments and questions may be
    sent via email.