Homelessness: Better HUD Oversight of Data Collection Could Improve Estimates of Homeless Population
Fast Facts
The Department of Housing and Urban Development (HUD) found homelessness in the U.S. grew 3 years in a row (2017-2019). Rising homelessness in metropolitan areas drove the increases.
We found HUD’s count likely underestimated the homeless population. Organizations across the U.S. provide data for this inherently difficult count. HUD could improve its instructions to them, which in turn could improve data quality.
In addition, our statistical analysis found median rent increases of $100 a month were associated with a 9% increase in homelessness in the areas we examined.
We recommended ways for HUD to improve measurement of homelessness.
Unsheltered Homelessness in San Francisco, California
Street corner with blankets and furniture
Highlights
What GAO Found
Data collected through the Point-in-Time (PIT) count—a count of people experiencing homelessness on a single night—have limitations for measuring homelessness. The PIT count is conducted each January by Continuums of Care (CoC)—local homelessness planning bodies that apply for grants from the Department of Housing and Urban Development (HUD) and coordinate homelessness services. The 2019 PIT count estimated that nearly 568,000 people (0.2 percent of the U.S. population) were homeless, a decline from the 2012 count of about 621,500 but a slight increase over the period's low of about 550,000 in 2016. While HUD has taken steps to improve data quality, the data likely underestimate the size of the homeless population because identifying people experiencing homelessness is inherently difficult. Some CoCs' total and unsheltered PIT counts have large year-over-year fluctuations, which raise questions about data accuracy. GAO found that HUD does not closely examine CoCs' methodologies for collecting data to ensure they meet HUD's standards. HUD's instructions to CoCs on probability sampling techniques to estimate homelessness were incomplete. Some CoC representatives also said that the assistance HUD provides on data collection does not always meet their needs. By strengthening its oversight and guidance in these areas, HUD could further improve the quality of homelessness data.
To understand factors associated with homelessness in recent years, GAO used PIT count data to conduct an econometric analysis, which found that rental prices were associated with homelessness. To mitigate data limitations, GAO used data from years with improved data quality and took other analytical steps to increase confidence in the results. CoC representatives GAO interviewed also identified rental prices and other factors such as job loss as contributing to homelessness.
Estimated Homelessness Rates and Household Median Rent in the 20 Largest Continuums of Care (CoC), 2018
Note: This map shows the 20 largest Point-in-Time counts by CoC in 2018. GAO estimated 2018 homelessness rates because the U.S. Census Bureau data used to calculate these rates were available up to 2018 at the time of analysis. GAO used 2017 median rents (in 2018 dollars) across all unit sizes and types.
Why GAO Did This Study
Policymakers have raised concerns about the extent to which recent increases in homelessness are associated with the availability of affordable housing. Moreover, counting the homeless population is a longstanding challenge. GAO was asked to review the current state of homelessness in the United States. This report examines (1) efforts to measure homelessness and HUD's oversight of these efforts and (2) factors associated with recent changes in homelessness.
GAO analyzed three HUD data sources on homelessness and developed an econometric model of the factors influencing changes in homelessness. GAO also conducted structured interviews with 12 researchers and representatives of 21 CoCs and four focus groups with a total of 34 CoC representatives responsible for collecting and maintaining homelessness data. CoCs were selected for interviews and focus groups to achieve diversity in size and geography. GAO also visited three major cities that experienced recent increases in homelessness.
Recommendations
GAO recommends that HUD (1) conduct quality checks on CoCs' data-collection methodologies, (2) improve its instructions for using probability sampling techniques to estimate homelessness, and (3) assess and enhance the assistance it provides to CoCs on data collection. HUD concurred with the recommendations.
Recommendations for Executive Action
| Agency Affected | Recommendation | Status |
|---|---|---|
| Department of Housing and Urban Development |
Priority Rec.
HUD's Office of Special Needs Assistance Programs should conduct quality assurance checks on the PIT count methodology data it requires CoCs to submit and take actions as appropriate to ensure that HUD's standards for conducting valid and reliable PIT counts are met. (Recommendation 1) |
In 2023, HUD provided updated guidance to CoCs on PIT count methodology and responded to CoC questions on enumeration issues, such as rural counting and counting tents and vehicles as part of its standard periodic update to HUD's PIT count notice. HUD also updated its PIT count methodology submission questions. In February 2024, HUD officials told us they review CoCs responses to the questions with their PIT count methodology submissions to assess the quality of CoCs' data. In September 2024, HUD provided us its internal procedures on addressing data quality issues identified in their PIT count methodology submissions. According to the documentation and HUD officials, they evaluate CoCs'... data submissions for year-to-year variation and whether any data submission validation flags are being triggered. They consult with CoCs that have submissions that raise any data quality questions, and in the event errors are identified, HUD works with the CoC to correct its PIT count submissions. By implementing quality assurance checks of CoCs' PIT count methodology data, HUD addressed the recommendation to more consistently ensure that HUD's standards for PIT count data collection are being met.
View More |
| Department of Housing and Urban Development | HUD's Office of Special Needs Assistance Programs should provide more detailed instructions on using probability sampling techniques to complete the PIT count, such as by updating its Point-in-Time Count Methodology Guide to instruct CoCs on reporting measures of error and bias in PIT count results. (Recommendation 2) | In April 2023, HUD published a webpage entitled "How to Use Sampling within a CoC to Conduct an Unsheltered PIT Count - HUD Exchange". This webpage provides tools implemented in 2020, 2021, and 2023, to support CoCs in developing a sampling approach, estimating a total count of the number of people within their geographies who are experiencing unsheltered homelessness on the night of the PIT count, and extrapolating the count. These resources provide guidance on various sampling approaches, selecting and weighting samples, and extrapolating results that meet HUD's standards. The tools explain the challenges and statistical limitations of the various sampling approaches, including the...
|
| Department of Housing and Urban Development |
Priority Rec.
HUD's Office of Special Needs Assistance Programs should assess and enhance the usefulness of its assistance to CoCs' data collection efforts. (Recommendation 3) |
In March 2023, HUD officials told us they were developing an outreach strategy to work with CoCs on PIT count methodologies. The intent of this outreach strategy is to talk to CoCs about their methodologies, answer questions, and determine if additional guidance or assistance is needed. As of February 2024, officials told us they still intend to assess their outreach efforts but their ability to fully implement this effort is currently limited. Meanwhile, HUD provided evidence it has developed a random sample of CoCs to provide one-on-one outreach and assistance on PIT count data and methodology submissions. HUD also provided evidence of consultations between the Office of Special Needs...
|