Trends and Patterns of Homelessness

The number of people experiencing homelessness in the U.S. has declined modestly over the past ten years, driven in large part by reductions in the number of homeless veterans and individuals with chronic patterns of homelessness. Modest declines were also seen in the percentages of all people experiencing homelessness as individuals and of people experiencing homelessness as part of a family with children, although the decline for families with children was limited to those found in unsheltered locations. The number of people using shelters over a year as part of a family has increased somewhat.

There is considerable variation at the state and local levels. The largest states—California, New York, Florida, and Texas—account for about half of the national estimate of people experiencing homelessness, but have very different rates of homelessness (the share of the state’s population found homeless by Point-in-Time counts). States also differ dramatically in the share of the population that is found in sheltered locations rather than in places not suitable for human habitation.

At the local level, homelessness is a largely urban phenomenon, with more than 70 percent of people using shelter in urban, rather than rural or suburban areas. The Continuums of Care that cover the 50 largest cities in the U.S. account for about half of the total population of people experiencing homelessness on a single night in January. However, as with states, rates of homelessness vary substantially across cities, even within a single state.

Systematic collection of data beginning in the early 2000s has greatly improved our ability to understand homelessness. For example, during the Great Recession there was an expectation that homelessness would increase substantially. It did, but the increases were almost entirely for families with children rather than for people experiencing homelessness as individuals. We can see substantial reductions in chronic homelessness in recent years, which appear to reflect the emphasis placed on permanent supportive housing since the early 2000s.

What are the implications for policymakers and practitioners?

What we know about trends in homelessness and the data that can be used to understand them has implications both for national policy and for policymakers and practitioners at the community level. For example, we can conclude that additional resource and their effective use make a difference. The number of people who experience homelessness is not so large as to be insurmountable.

  • Adding substantial resources for ending homelessness for families with children and for individuals without chronic patterns of homelessness could reduce the numbers for these populations, as it has for chronic individuals and for veterans.
  • National and local policymakers should continue to focus resources on interventions that have been shown to be effective in addressing homelessness. Research has shown correlations between homelessness and housing vacancy rates, rent levels, and other housing market variables. Investment in mainstream rent assistance programs should be prioritized.
  • Trends indicate that investment in permanent housing solutions to homelessness may decrease homelessness. National and local policy-makers should continue to invest and encourage the adoption of these models.

With continuous improvements in the quality of PIT counts and the expansion of the percentage of providers covered by the HMIS in many communities, information is available to help communities better understand the local nature of homelessness. The information allows them to focus their resources on particular population groups and to design their homeless services systems and modify them as needed.

Communities should:

  • Make full use of the variety of local data sources, including Homeless Management Information System (HMIS) and Point-in-Time (PIT) data and also the program performance data required by HUD and data matches to health care and other systems.
  • Fully implement the HMIS, expanding it to cover all providers of shelter if it does not already, and work with service providers to address and resolve data quality issues—for example, by making sure program entry and exit dates are accurate, since this is important for understanding how the local homeless services system works.
  • Work on continuous improvements to the PIT counts, based on HUD guidance about best practices and by enlisting the help of social scientists and statisticians from universities and appropriate organizations as appropriate.
  • Consider special analyses of client-level HMIS data (available only at the community level) for system planning and performance measurement. HMIS data provide valuable information on where people were before they became homeless and can support analysis of how people move through programs in the homeless services system and patterns associated with their returns to homelessness. Virtually all communities now have fairly well-developed HMIS, but many do not maximize the value of the data collected.

Read the full research brief to see the underlying research evidence.

Last updated February 2018