The Effect of ESA Policies on Public School Enrollment and Composition

• In Event: School Choice, Screening, and Stratification: Evidence on Enrollment and Composition

Abstract

Education Savings Accounts (ESAs) have emerged as the fastest-growing and most expansive school choice mechanism in the United States with over a million students enrolled in these programs by the fall of 2025. They provide families with public funds for educational expenses, including private school tuition, tutoring, teaching supplies, and homeschooling materials. Nineteen states have recently adopted or expanded ESA programs, often phasing eligibility from targeted groups toward universal access. Despite this rapid diffusion, little evidence exists on how ESAs affect public school enrollment or whether their consequences vary across districts. This paper provides the first multi-state causal analysis of ESA adoption on public school enrollment and student composition. It addresses three questions: whether ESAs substantially affect public school enrollment; what drives concentrated enrollment shifts in some districts, such as the availability of private school alternatives; and which students pursue ESAs, including variation by grade level and sociodemographic characteristics. The study draws on publicly available district-level data from the National Center for Education Statistics, spanning 2010 to 2024. It focuses on five states, Arizona, Florida, West Virginia, Iowa, and Arkansas, whose ESA programs vary substantially in eligibility rules, timing, allocation amounts, and pace of expansion. This variation allows us to assess both average effects and how program design shapes public school responses across district types, including by urbanicity, private school presence, and the availability of other choice programs. We employ a two-part empirical strategy. First, we estimate staggered difference-in-differences models (Callaway & Sant'Anna, 2021) comparing ESA-adopting states to a matched pool of non-adopting states. Rather than relying on a binary policy indicator, we construct a continuous treatment dosage based on state-reported ESA expenditures and average award amounts, capturing variation in treatment intensity across states and over time. Second, within adopting states, we estimate two-way fixed effects models exploiting district-level differences in private school proximity, urbanicity, and baseline demographic composition to examine heterogeneous impacts. Preliminary evidence suggests that average enrollment effects are modest, but impacts vary meaningfully across student entry points. Stronger responses appear among incoming kindergarten cohorts, consistent with ESAs reshaping initial enrollment decisions more than inducing exits from public schools mid-enrollment. These findings contribute to debates over school choice, educational stratification, and public finance. Modest average effects may mask substantial local consequences if enrollment losses concentrate among particular districts or student populations. As more states consider ESA expansion, understanding where and for whom these programs alter public school participation is essential for designing equitable and fiscally sustainable education systems.

Author

• Abigail Francis, Boston University
  Presenting Author