Not All That Glitters: Assessing the Performance of Private Foster Care Providers

Background and Purpose:  Public child welfare agencies often rely on private providers to deliver services to vulnerable children and youth, including foster care (Flaherty, Collins-Camargo, & Lee, 2008).  In recent years, reliance on the private sector has increased significantly (Flaherty et al., 2008; General Accounting Office, 1997).  Kansas and Florida, prime examples of the trend toward privatization, turned over their entire child welfare systems (except for protective investigations) to private entities (Flaherty et al., 2008).  Although the rationale behind an expanded role for the private sector is varied, the most frequently cited reasons are a desire to improve outcomes, improve quality, and stimulate competition (Westat and Chapin Hall, 2002).

Despite the desire to improve services through privatization, there is almost no research that specifically addresses the performance of private foster care agencies relative to other private agencies.  Mason and colleagues (2003) examined outcomes six months post-discharge from private agencies, but they did not consider between-provider variation in outcomes.  More recently, Curtis McMillen and colleagues (2008) considered risk adjustment methods at the provider level but did not specifically report whether provider performance varied from one to the other.  Given the significant public investment pouring into the private sector, the gaps in knowledge are too large to ignore.

Methods:  We address the shortcomings in the literature by assessing private agencies with respect to four indicators of performance: reunification, adoption, placement stability, and reentry.  We rely on a unique database that disaggregates time spent by children in foster care into unique, agency-specific segments.  There are 40 plus agencies and more than 15,000 children placed with those agencies in the analysis.  We use multilevel, longitudinal models to partition the variance into individual and agency level components.  The model residuals reveal whether some agencies are significant outliers, after controlling for child level characteristics. The Empirical Bayes estimates, as the residuals are called, are used in health care and education to profile hospital and school performance. This is the first time these methods have been used to assess the performance of private foster care providers.

Results:  As expected, the results show that individual level factors are a significant source of variation.  For example, children with elevated CANS scores stay longer in care than children with lower scores.  That said, the results also show that how long children spend in foster care clearlydepends on the agency with which the child is placed.  Finally, we are able to identify a small number of agencies that lag behind the performance of their peer organizations across all measures of performance.

Conclusions and Implications:  Investments in the private sector make good sense provided the children served benefit.  Our study indicates that performance within the private sector varies considerably, even after adjusting for the attributes of children.  Privatization has strong resonance in this country, but our results suggest that not all that glitters is gold.  Moreover, oversight of the private sector has to be strengthened if we hope to improve care for foster children through privatization.