1. Testing a Typology of Family Homelessness Dennis Culhane University of Pennsylvania Stephen Metraux University of the Sciences in Philadelphia Jung Min Park University of Illinois at Urbana-Champaign Jesse Valente State of Massachusetts Maryann Schretzman City of New York

2. Background  Singles typology experience  But families are different: - much lower MH/SA rates - not different from poor housed families - relatively homogeneous  Potential confounders – policy/program factors - use of shelter system as queue for subsidies - transitional shelter as a reform movement

3. Cluster Distributions: Persons and Shelter Days Consumed (Single Adults in Philadelphia)  Transitionals:  1.19 stays  20.4 days  Episodics:  3.84 stays  72.8 days  Chronics:  1.53 stays  252.4 days

4. Disability Condition & Veteran Status By Cluster (Single Adults in Philadelphia)

5. Methodology  Four jurisdictions – Philadelphia, NYC, Columbus OH, and Massachusetts  HMIS data – new admissions followed for two or three year periods  30 day exit criterion applied  Cluster analysis, specifying three cluster solution  Database merges to identify service histories

6. Health and Social Service Databases Merged  In one city: Medicaid, Mental Health, Substance Abuse, Child Welfare  In one state: Medicaid, Mental Health, Substance Abuse, Child Welfare  In one city: Child Welfare

7. Results: Cluster Solution (Massachusetts, family shelter users)  Transitionals:  1.0 stays  105 days  Episodics:  2.0 stays  195 days  Long-Stayers:  1.0 stays  444 days

8. Intensive Service Histories of Families

9. Income Sources

10. The Average Cost of Shelter Stays by Type (Massachusetts)  Transitional $11,550  Episodic$21,450  Long-term$48,440 Does not include McKinney-Vento funding or non-DTA public service contracts.

11. Summary  Cluster patterns are robust across sites  Most families (75%) leave quickly and don’t return  A small number (5%) return repeatedly  20% of families have long stays, using 50% of resources  BUT – unlike singles – long stays do not indicate personal barriers to housing stability

12. Conclusions  Policies and programs driving long stays  Characteristics of “graduates” may reflect selection effects of policies and programs  Most costly service users are not differentially service-needy  Need for reform

13. New Conceptual Framework Considerations  Least needy families should get least intensive intervention, and most needy families should get most intensive intervention  Most families have few service histories, and exit quickly; policies should support this preference (“rapid relocation”)  Various service and housing assistance packages for remainder (including transitional rental assistance), based on level of barriers  Convert existing facilities to supportive housing for corrections, addictions treatment and family unification

14. VolumeVolume Cost per Case Model Cost by Volume Service System for Addressing Housing Emergencies Prevention Supportive Housing Shelter Admission Diversion, Relocation and Transitional Rental Assistance Mainstream systems Community- Based programs

15. Future Research  Develop assessment tools for classifying families  Develop housing assistance models to relocate families and test efficacy by type of family  Use HMIS for assessment and tracking outcomes, possibly routinely check for service histories of families  Study selection behaviors of facilities, role of policies