Evaluating Practice in a Pay for Success Context: Learning from a County-Level PFS Project Using a Mixed-Methods Design

Background and Purpose: In 2015, the first county-level social impact bond in the U.S. was launched in Cuyahoga County, Ohio. With a focus on homeless and housing unstable families with a child in foster care, the primary goal was to more rapidly address families’ housing needs, thereby allowing children’s stays in foster care to be reduced. Though any exit from out-of-home placement was valued, there was an implicit expectation that reunification was a desirable path to permanency for the majority of children. The initiative involved a homeless-serving agency to be assigned to identified families to work with them using a Critical Time Intervention approach and Trauma-Adapted Family Therapy. Families were offered case management services and housing supports.

Methods: The study used an RCT design in which newly referred and eligible families were randomly assigned to either the new service approach or existing services over a 2.5 year period. Programmatic records were linked to administrative data on homeless service use and child welfare records to monitor housing and child placement outcomes over time. The study enrolled approximately 270 families which were equally assigned to the two study groups.

Findings: The study effectively identified and recruited eligible families to participate and the treatment successfully engaged a high proportion of families. The treatment increased the proportion of children who exited out-of-home placement to reunification, as compared to the control group and the general child welfare service population. Though the treatment group experienced higher recidivism (returns to care) than the control group, they still had a greater proportion of children remaining exited to reunification. Notably, children of African-American parents in the treatment group had significantly higher reunification rates as compared to similar families receiving usual services.

Implications: Though the present approach increased reunification generally, it did so particularly for children of African American families. Given the well-documented racial disparities in child welfare in regard to reporting, foster care, and placement outcomes, this finding could hold promise for better addressing the circumstances of African American families. Though not designed specifically for families of color, the intervention appears to be particularly useful in increasing the odds that housing unstable African American parents would reunify with their children. Better understanding the mechanics by which this occurs is the focus of additional work that is planned.