Farm to School (FTS) programs generate two competing e!ects on household food demand: school meal provisioning substitutes for home purchasing in the short run, while repeated exposure builds taste capital that complements demand over time. Linking Nielsen Home Scan and USDA FTS Census data (2012–2023), we estimate household demand for different food categories via the Exact Affine Stone Index (EASI) system and exploit staggered FTS adoption with differential exposure by child age to identify both mechanisms. In the short run, households with school-age children in FTS districts exhibit substitution in school-provided food categories. In mature programs (8+ years), this reverses: household spending on fruits and vegetables increases by 22 percent, with the crossover at approximately 11–12 years. FTS programs also generate community spillovers of 0.42 percentage points for childless households, and offset 80 percent of the child-driven fresh produce deficit in households with
children. The main takeaways are: short-run evaluations substantially understate program benefits; programduration, not intensity, is the key impact parameter.
