Association for Public Policy Analysis & Management

Over 20,000 children age out of the foster care system annually; aging out of the foster care system has been associated with a plethora of negative health and human capital outcomes. In large part, this struggle is linked to the lack of permanency, stability, and resources during the formative years of childhood development. In light of this public health issue, many states have extended their legal foster care age beyond 18 years in order to continue support and resources for this extremely vulnerable population. In specific, under the Fostering Connections Act (FCA), between the years 2010 and 2017, 25 states extended the foster care emancipation age from 18 to 19-21. By exploiting temporal variation in the passage of state-level legislation, and state-level variation in age cutoffs, I examine the effect of foster care age-extensions on adult outcomes such as incarceration, substance abuse, employment, and homelessness. For this analysis, I use a rich national longitudinal survey data set that is conducted over three waves at age 17, 19, and 21---two separate cohorts were surveyed one starting in 2011 and the other in 2014---which can be linked to administrative data pertaining to case characteristics (such as the number of foster care placements), and services received in foster care (such as career advising). Preliminary evidence indicates that after the passage of an age-extension law foster youth were over 40 percent more likely to remain in the foster care system. Given that the laws increase retention in the child welfare system, I find that extending foster care to 19 leads to improvement in educational attainment, employment, and health outcomes---both at age age 19 and 21. In addition, results indicate that further extending the foster care to 21 leads to additional improvements in these outcomes. Last, I find that the provision of additional services augment the benefits of age-extension to health and human capital outcomes. Due to concerns over both state-level and individual-level selection into treatment, I examine the robustness of my results to alternative matching analysis.