Does lengthening the school day increase students’ academic achievement? Results from a natural experiment in Chile

Chilean SIMCE tenth-grade cohorts, 2001 pre-reform against 2003 post-reform

Chilean tenth graders in schools moving from a four-hour half day to a six- or seven-hour full day

Economics of Education Review 28(5), 629-640. READ AT ABSTRACT DEPTH ONLY - the article is paywalled and the primary text was not obtained, so the numbers below come from the abstract plus a later review. Difference-in-differences exploiting the staggered rollout of Chile's Jornada Escolar Completa, which raised annual instructional time at high-school level by roughly 27%. Grade C rather than B because schools entered the reform in a sequence that was not random and the record read here is too thin to assess the parallel-trends evidence.

A roughly 27% increase in annual instructional time bought 0.05 to 0.07 SD in language, an estimate the author reports as 'not sensitive to control for covariates, different control groups, and historical trends', and 0.00 to 0.12 SD in mathematics, which was much more specification-sensitive - a later review treats 0.07 as the most defensible figure. THAT IS THE DOSE-RESPONSE THAT MATTERS FOR THIS TOPIC: adding more than a quarter to the school day for two years produced roughly what a few extra weekly subject-hours produce in the cross-country estimates. Effects were larger for rural pupils, public-school pupils, and - the equity point - for pupils who were ALREADY HIGHER-ACHIEVING.

Medium and not assessable from the material read. The reform was rolled out school by school on administrative criteria rather than by family choice, which protects against child-level selection, but schools that converted earlier may differ systematically. The finding that gains accrued to already-higher-achieving pupils is descriptive.

Unreplicated for Chile, but the magnitude sits squarely inside the large-dose extended-time record: Massachusetts added 300 hours a year for five years and found nothing, Florida's extra literacy hour gave 0.05 SD decaying to zero, and a Dutch randomised trial adding five weekly hours found nothing. It also matches Lavy's concavity - the marginal hour is worth much less at high totals.