Complete school meal coverage: state policies, child-specific eligibility, and separate breakfast/lunch treatment #9789

School-meal coverage remains incomplete beyond the North Carolina copay correction in PR #9756, which addresses #9755. The model needs additional state policies and a way to represent different benefits by meal, child, and participating school. This tracking issue records the research and implementation work for separate, reviewable PRs.

The code observations below are based on PR #9756 at commit c6e71b41a8cface1f6bcfe24c93f4c15eb38c4f3. Legal research was checked on October 4, 2026, primarily against school year 2026–27. Historical effective dates and the remaining jurisdictions still need a systematic audit before declaring national coverage complete.

Current behavior and concrete gaps

• The daily subsidy calculation already has separate breakfast (sbp) and lunch (nslp) reimbursement schedules, but applies one SPM-unit school_meal_tier to both. It cannot represent free breakfast with reduced-price or paid lunch.
• The state copay parameter enables only NC in the reviewed branch. Additional confirmed policies are listed below; they cannot all be represented by one unconditional boolean.
• Categorical eligibility aggregates across the SPM unit. Its source list includes was_in_foster_care and is_head_start_eligible. This conflates historical foster status with qualifying foster status, Head Start eligibility with enrollment, and child-specific eligibility with household-wide eligibility. These are code-inspection findings; add regression cases before fixing them.
• State residence and K–12 enrollment currently stand in for access to a covered school. CEP, district opt-ins, public/private restrictions, and siblings attending different schools require more precise treatment or documented assumptions.
• Annual parameter lookup needs review against school-year effective dates. Income guidelines, reimbursement schedules, state policy changes, and appropriation periods can change within a calendar year.
• The annual valuation uses reimbursement differences relative to the paid tier, multiplied by K–12 children and 180 days. Preserve a clear distinction between this modeled benefit value, federal/state spending, and actual family out-of-pocket savings.

Federal authorities for the eligibility work

• 7 CFR 245.2: definitions of family, foster child, Head Start child, and categorical eligibility.
• 7 CFR 245.6(b)(7)–(8): SNAP/FDPIR/TANF categorical eligibility extends to all children in the family; foster, homeless, migrant, runaway, and Head Start eligibility does not automatically extend to siblings. The same section also addresses certification and duration of eligibility.
• USDA income eligibility guidelines: school-year schedules effective July 1–June 30; validate the guideline year, household size, income frequency, and rounding against published amounts.

State research to carry forward

The current universal breakfast-and-lunch list is CA, CO, ME, MA, MI, MN, NM, NY, and VT, subject to covered-school and participation conditions. This matches the current national inventory. Preserve the legal references in the model and verify historical implementation and funding periods; a universal-meals state flag does not establish access for every school or resident child.

The following policies were confirmed outside those nine states. This is a research starting point, not a completed negative finding for every unlisted state. “Both” means breakfast and lunch for otherwise reduced-price-eligible students at covered schools. FPG means federal poverty guidelines. Do not backfill current policy into earlier years without evidence.

Suggested implementation sequence and acceptance criteria

• Complete a 50-state/DC policy matrix with meal type, income limit, child/grade and school restrictions, start/end dates, funding conditions, and authoritative references. Distinguish “not researched” from “no applicable state policy”; record verified historical periods.
• Correct categorical eligibility in a focused PR. Separate household-wide assistance-program eligibility from child-specific categories; reuse existing enrollment/current-status inputs where suitable. Test a foster child with a nonqualifying sibling, historical foster status alone, and Head Start enrollment versus eligibility.
• Add straightforward, verified both-meal copay policies using shared calculations and dated parameters. Do not set a single coverage flag for meal-specific, opt-in, or potentially partial subsidies without representing the relevant conditions.
• Separate breakfast and lunch treatment while keeping federal eligibility distinct from state-funded assistance. Preserve existing results for unaffected cases before adding mixed-meal policies. Support free breakfast with reduced/paid lunch and the converse.
• Add expanded state income thresholds separately from federal thresholds. Audit countable income and the SPM-unit approximation to the program's family definition; do not replace school-meal income with SNAP income without policy verification.
• Handle school-year/calendar-year transitions explicitly, including the correct income-guideline year, published threshold rounding, reimbursement changes, and state start/expiry dates. Document any approximation and distinguish annual entitlement modeling from administrative certification history.
• Add school participation/CEP and child-level aggregation where supported by inputs or data. Reuse existing inputs, add only necessary distinctions, and document assumptions when school information is unavailable. Address local programs and pre-statewide historical coverage through the related work below.
• Keep benefit valuation and funding attribution clear, prevent overlap/double counting, and handle appropriation-limited policies using supported implementation guidance. Do not invent a household-level rationing rule from a statewide funding cap.
• Add threshold, effective-date, mixed-household, mixed-school, and vectorized cases in the corresponding variable-named YAML test files. Check household benefit totals and API behavior; assess partner contract impacts rather than updating expected outputs merely to make tests pass.
• Protect dependent programs: WIC reads the federal school-meal reduced-price income limit; CSFP (including Texas) and a California rebate calculation reuse school_meal_fpg_ratio. Audit callers before changing shared definitions and add appropriate cross-program checks.
• Update programs.yaml and documentation to reflect implemented scope and verified years. Mark unsupported or approximate coverage explicitly instead of claiming complete coverage prematurely.

The current age-based school-enrollment default (5–17), 180-day assumption, attendance, and meal take-up can be refined separately; they should not block clearly supported state-policy additions.

Related existing issues to coordinate with rather than duplicate:

• Clarify income sources for free and reduced price school meals #4601 — Clarify income sources for free and reduced price school meals.
• Replace school meal income source parameter with SNAP #597 — Replace school meal income source parameter with SNAP (the proposed equivalence still requires verification).
• Universal free school meals in NYC #2324 — Universal free school meals in NYC (including historical coverage before the statewide policy).
• Adjust school meals by attendance rates #278 — Adjust school meals by attendance rates.
• Change school year to 179 days #672 — Change school year to 179 days.
• Add federal vs. state cost attribution for benefit programs (Medicaid FMAP, SNAP OBBBA, CHIP eFMAP, etc.) #8075 — Federal versus state cost attribution for benefit programs.

This issue concerns regular NSLP/SBP school meals. Summer EBT and CACFP are separate programs; administrative precision beyond the model's household-benefit purpose should be scoped explicitly rather than bundled into the first PR.