Association for Public Policy Analysis & Management

As school choice options in cities have expanded, so has the debate over how to fairly allocate public funds across traditional, magnet, and charter schools. Are children enrolled in choice schools fundamentally different (and potentially less costly to educate) than those in traditional public schools? This paper leverages a unique linked dataset to investigate how early childhood interactions with various human services predict school choice enrollment in a midsize U.S. city. One indicator proves particularly powerful in prediction: whether a student was linked to a child welfare investigation in early childhood (approximately 15% of students). These children are significantly less likely to enroll in charter, magnet, or private schools, even after controlling for other forms of early childhood disadvantage like Medicaid-eligibility. Moreover, concentrating these children in traditional public schools carries implications for school-level outcomes. In elementary school, they miss more school and receive more suspensions than children who were not linked with a child welfare investigation in early childhood, and there is evidence of peer effects, particularly in terms of suspension.