In the last eight years, the number of Americans experiencing homelessness in the United States has increased by 40%, even as the supply of subsidized housing has doubled. Research shows that housing programs effectively support housing stability. However, system-level outcomes depend on whether housing reaches people likely to exit shelter quickly or those likely to experience longer-term housing instability. Systems that prioritize people with shorter expected stays reduce homelessness less than when they prioritize people facing longer-term instability. Using administrative data from Allegheny County, this policy brief and working paper examine how often housing does not reach those with the greatest need, identify and characterize those who do not receive services, and outline how to recognize these individuals before their homelessness becomes long-term.
Key Results
Using administrative data from Allegheny County, we have developed a predictive model that identifies—at the time of shelter entry—individuals likely to experience an extended shelter stay.
- Most shelter residents exit shelter on their own within weeks after starting their shelter stay.
- Among the highest-risk individuals flagged by the model, over 40% go on to experience extended homelessness—over three times the baseline rate.
- Allocating housing to individuals or families at the highest risk of prolonged shelter stays would prevent approximately 2.4 times more shelter days per unit than allocating to the average shelter resident.
- Making housing allocations based on the predictive model rather than current federal prioritization criteria would prevent 120% more shelter days.
- Shifting from retrospective to prospective targeting could substantially reduce overall homelessness without expanding the subsidized housing supply.
Why This Matters and What’s Next
Homelessness remains difficult to solve. These data highlight a core challenge: most individuals resolve occurrences of homelessness quickly, while housing supports take time to deploy.
Without thoughtful allocation, resources do not reach people with the greatest need. Allegheny County continues to develop approaches that prioritize resources for individuals at highest-risk. This principle already guides ACDHS programs such as 500 in 500 and the Allegheny Housing Assessment, which have successfully prioritized individuals for housing in times of acute need.
Questions or Feedback?
We welcome your questions and suggestions. To share feedback, email us at DHSResearch@alleghenycounty.us. If you’d like to stay informed, consider signing up for our newsletter. To learn how to use ACDHS data in your research, please visit our Requesting Data page. Thank you for your time and interest. Your engagement helps shape and improve how we share data that matters.