The child welfare system plays an important role in the lives of a large number of children who are particularly at risk of poor life outcomes. Over two million children are investigated for abuse and neglect in the U.S. each year, and abused children are far more likely to commit crimes, dropout of school, enter the homeless population, join welfare, experience substance abuse problems, and die in childhood. While many descriptions of this population are available, little is known about the effects of foster care - the removal of abused children from home while families are rehabilitated or adoptive homes are sought- on child outcomes. The proposed projects involve empirical research on two issues involving the effectiveness of the child welfare system: (1) the causal effect of foster care placement on child outcomes; and (2) the types of child welfare agencies that provide the highest quality foster care for a given cost. In both projects, a remarkably wide range of child outcomes will be considered using a database that links children across administrative datasets in Illinois. These outcomes include: school performance and educational attainment; preventive medical care utilization, teen pregnancy, and treatment for sexually transmitted diseases; juvenile delinquency and adult criminal convictions; and earnings and employment. The key innovation in the first project considers the effective randomization of families to case managers during the child abuse investigation. Some case managers appear more likely to place children in foster care than others. By focusing on the variation in foster care placement induced by this random assignment, it is possible to estimate a causal effect of foster care on the wide range of child outcomes available in Illinois. The second project considers the introduction of performance-based contracting in Illinois. Agencies were financially rewarded contingent on child outcomes, and all new foster care entrants were randomized to agencies in an effort to equitably distribute cases. This randomization controls for child characteristics so that child outcomes can be compared across agencies characterized as public versus private, large versus small, and faith-based versus secular. Characteristics of agency caseworkers can also be compared, such as measures of experience, education, and language skills.
Intellectual Merit These projects demonstrate the ability of administrative datasets, coupled with an understanding of institutional details, to yield creative ways to answer important questions. For example, the effective randomization of children to case managers, or children to child welfare agencies, provides a way to mimic randomized social experiments. Specifically, these projects attempt to estimate the causal effect of foster care on child outcomes for the first time, as well as a new understanding of the effectiveness of private social service agencies.
Broader Impacts These projects will inform child welfare policy and social service policy more generally. A better understanding of the response of private child welfare agencies to financial incentives can inform the growing use of private organizations to supervise social services. Further, the overrepresentation of former foster children in prison populations, homeless populations, and in poverty, suggests that a better understanding of the development of these children can have a broad impact.
