Nutrition and cognitive achievement: An evaluation of the School Breakfast Program

NAEP 2003 grade-4 math sample 53,430 students; ECLS-K grade-5 (2004) sample 3,040 students

US public-school children in grades 4-5; identification concentrated on schools near state free/reduced-price-eligibility thresholds, i.e. relatively high-poverty US schools. Well-nourished by global standards; food-insecure at the margin.

Exploits state mandates requiring schools to offer the School Breakfast Program once the share of free/reduced-price-eligible students crosses a state-specific threshold. Both difference-in-differences across the threshold and a regression discontinuity at the cutoff. First stage is strong - schools above threshold are ~33 percentage points more likely to offer SBP (~98% vs ~45% participation in ECLS-K). Outcomes are NAEP and ECLS-K assessments, independently administered and not tied to state accountability. Identifies the effect of program AVAILABILITY, not of a nutrient - the channel could be nutrition, calories, earlier arrival at school, household income relief, or attendance.

Availability of the School Breakfast Program raises achievement by roughly 0.05-0.12 SD. NAEP: math +0.077 SD (SE 0.032) DD and +0.091 SD (SE 0.044) RD; reading +0.054 SD (SE 0.031) DD and +0.122 SD (SE 0.050) RD. ECLS-K DD estimates are the same order but imprecise: math +0.096 SD (SE 0.084), reading +0.119 SD (SE 0.093). This is an attainment-adjacent domain-skill effect on standardized tests, not a claim about g.

Low. The mandate thresholds are set by state statute and are not chosen by families; comparison is across schools on either side of an administrative cutoff, so parental traits cannot select into treatment at the discontinuity. Residual risk is that threshold-crossing schools differ in other compensating programmes.