Education Savings Accounts and School Vouchers

What the evidence says

The clearest empirical signal in these sources is a null result on test scores. In Chile, comparing municipalities where private school voucher enrollment expanded rapidly against municipalities where it stayed flat, panel/difference-in-differences analysis found no measured effect on test score growth: regression coefficients ranged from approximately −0.05 to +0.08 standard deviations and were not statistically significant. The same study reports this null result held across four alternative measures of school competition, and found that parental education was sufficient to explain most of the observed test score growth in the sample — meaning the growth that did occur is attributed to a factor other than voucher-driven competition, not to the voucher expansion itself.

The same Chilean panel/DiD study also tested whether voucher-driven school competition narrowed test score inequality between high- and low-income students, on the theory that expanded choice should let lower-income students access higher-performing schools. It found little association between changes in school competition and changes in test score inequality on its primary voucher-share measure. It also examined three other competition measures; on those, it found significant effects in the opposite direction from what would be expected if choice reduced inequality — that is, in the direction of increased test score inequality.

A separate line of evidence, an interpretive historical study of multiple school choice reforms (including vouchers) affecting black students in the United States, reaches a compatible conclusion through an entirely different method: it finds no significant increase in academic performance for the majority of black children, and specifically not for children from low-income families, across the reforms it examines collectively. This study does not isolate the voucher component from the other reforms it reviews, and its method is interpretive historical analysis rather than a quasi-experimental design, so it carries lower confidence in both direction and magnitude than the Chilean panel study.

Across these three claims, no source reports a positive effect of voucher expansion on test scores or on narrowing income-based achievement gaps. The sources are silent on other outcomes commonly discussed alongside vouchers — there is no claim here about educational attainment (graduation, college enrollment), parental or student satisfaction, effects on the public schools students leave behind, or program cost. Those are simply absent from the evidence assembled on this page.

Where the evidence splits

Within the Chilean panel study, the inequality analysis produces an internal split: the primary measure (voucher share) shows no significant association with test score inequality, but the study's other competition measures show significant effects in the opposite direction from the inequality-reducing hypothesis — i.e., toward increased inequality. The claim does not explain why the alternative measures diverge from the primary measure; it only notes that individual-level longitudinal data would be needed to resolve the question further. This split is unexplained by the sources on this page.

Thresholds and dose-response

No threshold established in the current sources.

Open empirical questions

• Whether voucher effects on test score inequality would differ using individual-level longitudinal data, as the Hofflinger (2020) study itself flags as needed to resolve its own internal split.
• Whether vouchers have a measurable effect on outcomes other than test scores, such as educational attainment, graduation, or college enrollment — not addressed by any source here.
• Whether the voucher-specific effect can be isolated from the bundled reforms examined in Hurie (2021).
• Whether these findings, drawn from Chile and from studies of black students in the United States, generalize to other populations or voucher program designs.