Correct AK SNAP ABAWD HR1 date for good-faith exemption

[daphnehanse11]

Fixes #8821

Summary

Corrects Alaska's hr1_in_effect date from 2025-11-01 to 2026-11-01. PR #8861 modeled both Hawaii and Alaska as implementing the HR1 SNAP ABAWD changes on 2025-11-01, but that date is only correct for Hawaii: FNS approved Alaska's good-faith exemption under 7 U.S.C. 2015(o)(7) (added by OBBBA §10102(c)) effective November 1, 2025 through October 31, 2026, deferring the HR1 ABAWD rules in all areas except the Municipality of Anchorage (Alaska Department of Health, "H.R. 1 – AK Impacts").

Modeling decisions

• Statewide approximation: the parameter models all of Alaska as exempt through 2026-10-31 (HR1 in effect 2026-11-01). The Anchorage carve-out (HR1 from 2025-11-01) requires borough-level geography, which is out of scope here and tracked in SNAP ABAWD: model Alaska borough-level waiver geography #8822. This matches the majority of the state's population; the trade-off is that Anchorage's earlier adoption is not captured.
• The 2026-11-01 date assumes the current exemption is not renewed; renewals are permitted through 2028-12-31. Documented in the parameter comment.

Changes

• gov/states/ak/dpa/snap/work_requirements/abawd/hr1_in_effect.yaml: true from 2026-11-01 (was 2025-11-01); comment and references updated (adds 7 U.S.C. 2015(o)(7)).
• is_snap_abawd_hr1_in_effect.yaml tests: AK cases updated for the exemption window (false at 2025-11 and 2026-01; true at 2026-11 and 2027-01).
• meets_snap_abawd_work_requirements.yaml tests: behavioral regressions — a 60-year-old in AK is ABAWD-exempt during the exemption window (pre-HR1 age threshold) and in-band and failing after it expires (post-HR1 threshold 65).

Validation

• SNAP work requirements suite: 110 passed
• Full SNAP YAML suite: 372 passed
• make format clean

🤖 Generated with Claude Code

Renewal trigger note

The good-faith exemption under 7 U.S.C. 2015(o)(7)(D)(i) may be renewed once granted, but any renewal expires no later than December 31, 2028 (an expiry ceiling, not an entitlement). The 2026-11-01 flip date encoded in hr1_in_effect.yaml assumes the exemption is not renewed; if Alaska's exemption is renewed, that parameter date should be extended (bounded by 2028-12-31).

[DTrim99]

PR #8972 — Correct AK SNAP ABAWD HR1 date for good-faith exemption

Program Review — via /review-program (multi-agent audit)

PR: daphnehanse11, branch hi-ak-snap-abawd-hr1 — flips gov.states.ak.dpa.snap.work_requirements.abawd.hr1_in_effect from 2025-11-01: true to 2026-11-01: true, reflecting FNS approval of Alaska's good-faith-effort exemption under 7 U.S.C. 2015(o)(7) (effective 2025-11-01 through 2026-10-31; all of AK except Anchorage — modeled statewide as an approximation). Updates comments, adds a statutory reference, revises/adds tests (is_snap_abawd_hr1_in_effect Cases 9/11/13 revised, 14-15 added; meets_snap_abawd_work_requirements Cases 44-45 appended), plus changelog fragment. 149-line merge-base diff across 4 files.

Sources

Verified in the regulatory pass:

• AK DoH "H.R. 1 - AK Impacts" page (https://health.alaska.gov/en/education/hr-1-ak-impacts/ — the parameter's href): states verbatim that FNS approved Alaska's Good Faith Exemption waiver effective November 1, 2025 through October 31, 2026, covering all Alaska census areas and boroughs except the Municipality of Anchorage. Confirms the window, the carve-out, and that Anchorage faces HR1 from 2025-11-01.
• P.L. 119-21 (OBBBA), Sec. 10102(c) (congress.gov PDF, 139 STAT. 82-83): redesignates old (o)(7) as (o)(8) and inserts the new "(7) EXEMPTION FOR NONCONTIGUOUS STATES" — so the PR's 7 U.S.C. 2015(o)(7) cite is exact. Noncontiguous-State definition includes AK and HI, excludes Guam/USVI.
• 7 U.S.C. 2015(o)(7)(D)(i): any exemption "shall expire not later than December 31, 2028, and may not be renewed beyond that date" (a ceiling, not a renewal entitlement — see Should Address Basic prototype #1).
• USDA FNS OBBBA ABAWD Waivers Implementation Memo (Oct 3, 2025): could not be fetched directly (fns.usda.gov connection reset; archive blocked), but the AK DoH page and the enacted statute independently corroborate every load-bearing claim.

Branch Status: ⚠ 173 commits behind main; overlaps files reworked by merged #8872 — rebase REQUIRED before merge

The branch is 1 ahead / 173 behind main, and merged PR #8872 (2026-08-04) reworked the SNAP ABAWD waiver files, including meets_snap_abawd_work_requirements.yaml, which this PR also touches. The tests reviewer performed a hunk-by-hunk collision analysis against the post-#8872 working tree:

• is_snap_abawd_hr1_in_effect.yaml: Model California county and litigation-reinstated statewide SNAP ABAWD waivers #8872 did not touch Cases 9/11/13; all three PR-targeted hunks and the header hunk match the working-tree content exactly. Clean apply, no renumbering.
• meets_snap_abawd_work_requirements.yaml: the working tree still ends at Case 43 (Model California county and litigation-reinstated statewide SNAP ABAWD waivers #8872's additions — new cases up to ~32, Cases 28-30 moved to 2026-03 — sit entirely above the append point). The PR's pure-append hunk for Cases 44-45 matches trailing context and applies via line-offset fuzz; case numbering aligns with no collision.
• Model California county and litigation-reinstated statewide SNAP ABAWD waivers #8872's waived_states / waived_county_fips parameter files are not touched by this PR.

Verdict: no genuine case collision and no semantic divergence — this is a rebase/staleness note, not a correctness finding. Still, at 173 commits behind, the branch should be rebased onto main and CI re-run before merge so the merge commit is validated against current ABAWD logic, not a two-month-old tree.

Critical (Must Fix)

None. The date correction, statutory citation (2015(o)(7)), effective window (2025-11-01 → 2026-10-31), Anchorage carve-out, and the semantics of every new/revised test case were all independently verified. Both reviewers confirmed the new cases compute correctly against the actual formulas (Case 44: exemption window → pre-HR1 age-55+ exemption → age 60 exempt; Case 45: post-expiry → post-HR1 age-65 threshold → age 60 fails; verified against exempted.yaml brackets and the where(hr1_in_effect, post, pre) split in meets_snap_abawd_work_requirements.py:38-42, 111-115).

Should Address

1. Renewal comment overstates the statute. The new parameter comment says renewals "are permitted through 2028-12-31." 7 U.S.C. 2015(o)(7)(D)(i) actually sets a ceiling — an exemption "shall expire not later than December 31, 2028, and may not be renewed beyond that date" — it does not affirmatively grant a renewal right (renewal remains a discretionary FNS act; the statute also allows early termination for reporting failures). Reword to track the statute, e.g. "any renewal is capped at (and may not extend beyond) 2028-12-31 per 7 U.S.C. 2015(o)(7)(D)(i)."

2. Missing 2026-10 boundary test (last exempt month). The parameter toggles false → true at 2026-11-01. Tests cover the ON side (Case 14, 2026-11 → true) but never the last-exempt month 2026-10 → false; a regression shifting the toggle one month early would pass every existing case. Add an AK 2026-10 → false case in is_snap_abawd_hr1_in_effect.yaml (and ideally in meets_snap_abawd_work_requirements.yaml, which tests 2026-01 and 2026-12 but not 2026-10). Note: 2026-10 is a non-January month, so inputs must be year-keyed (state_code: {2026: AK}), matching the Case 45 pattern.

3. Anchorage statewide approximation: quantify it and pin it. The comment documents the direction of the approximation (all of AK modeled exempt, though Anchorage actually faces HR1 from 2025-11-01) but not the magnitude — Anchorage is ~40% of Alaska's population (~286k of ~740k), a material over-statement of exemption coverage during the window. (a) Add the population share to the parameter comment; (b) consider a pinned test that explicitly documents the approximation (an in-window AK case where post-HR1 rules would change the outcome, annotated # statewide approximation: Anchorage differs) so a future "fix" cannot silently diverge from the documented behavior. Case 44 partially covers this but is framed as an age-band test. The borough-level Anchorage split is tracked separately and needs sub-state geography.

Suggestions

4. Baseline convention for the 2026-11-01 resumption — defensible; document the follow-up trigger. Hard-coding 2026-11-01: true assumes non-renewal, which is the correct PolicyEngine baseline convention (only the one-year exemption has been approved; no projecting discretionary future FNS action). The comment already states the assumption. Just ensure the PR description notes that a future renewal (possible up to 2028-12-31) would require a parameter update.

5. Reference anchor precision. The new 2015(o)(7) href resolves to the whole of §2015 (uscode.house.gov has no sub-paragraph anchors — acceptable). The congress.gov PDF #page=81 reference (which contains the enacted (o)(7) text) is arguably the more precise primary anchor; keep the uscode link as the "current codified text" pointer. Reference format (title + href tuples) is otherwise correct.

6. Stale TODO in the consuming variable (out of scope, flag for cleanup). meets_snap_abawd_work_requirements.py:86-87 still carries # TODO: HI/AK delayed adoption (2025-11-01) to be handled in a follow-up PR — now stale (HI/AK are handled via the state hr1_in_effect params) and its 2025-11-01 date mis-states AK after this PR. Not blocking.

7. Header comment vs. modeled behavior. The is_snap_abawd_hr1_in_effect.yaml header notes Anchorage applies HR1 from 2025-11-01 — accurate to reality but not what the statewide-approximation parameter models. Acceptable as ground-truth documentation; consistent with Implement SNAP rules at the national level #3.

Verified-good items (no action): year-keyed inputs for all non-January periods handled correctly (Cases 14, 45 keyed; January cases correctly bare); changelog fragment present with correct .fixed type; month-boundary toggle date aligned with the month-period parameter; boundary coverage on both edges of the window otherwise complete.

Validation Summary

Validator	Result
Regulatory & sources	✅ Pass — 2015(o)(7) cite exact vs enacted P.L. 119-21 text; window and Anchorage carve-out confirmed against AK DoH page; FNS memo unfetchable but corroborated by statute + state page; 2 wording/documentation items (findings 1, 3)
Tests & patterns	✅ Pass — new-case semantics verified against formulas; #8872 conflict analysis: clean apply, no collision; year-keyed period gotcha handled; 1 boundary-coverage gap (finding 2)
CI	✅ All 33 checks pass (on the stale base — re-run after rebase)

Review Severity: COMMENT

No critical findings; the fix is correct, well-sourced, and well-tested. Approve after: (a) rebase onto main (173 behind; overlap with #8872 verified clean but CI should re-validate on current main), (b) the three Should Address items — renewal wording, 2026-10 boundary case, Anchorage-approximation quantification.

Next Steps

• Run /fix-pr 8972 to apply the Should Address items (rebase onto main; reword the renewal comment to the statutory ceiling; add the AK 2026-10 → false boundary case with year-keyed inputs; add the ~40% Anchorage population share to the parameter comment and optionally a pinned approximation test).
• Re-run CI post-rebase; confirm all 33 checks still pass against post-Model California county and litigation-reinstated statewide SNAP ABAWD waivers #8872 main.
• Optional cleanups: swap primary reference to the congress.gov #page=81 anchor; remove the stale TODO in meets_snap_abawd_work_requirements.py (separate PR).

🤖 Generated by /review-program (Claude Code multi-agent review)

[DTrim99]

Fixes Applied

Should-Address Issues Fixed

• Renewal wording (hr1_in_effect.yaml): reworded to state 7 U.S.C. 2015(o)(7)(D)(i) accurately — a granted good-faith exemption may be renewed but expires no later than December 31, 2028 (an expiry ceiling on renewals, not a renewal entitlement). The comment now also notes the encoded 2026-11-01 date assumes no renewal.
• Anchorage approximation quantified (hr1_in_effect.yaml): the comment now states the exemption determination is area-based, Anchorage (~40% of Alaska's population) is the only area where the HR1 rules currently apply, and the statewide parameter approximates the remaining ~60% as exempt through the window.
• Missing boundary test: added Case 16 (period: 2026-10, AK → false) to is_snap_abawd_hr1_in_effect.yaml, completing the boundary pair with the existing 2026-11 → true case around the corrected flip date.
• Rebase: branch rebased onto current main (was 173 commits behind) and force-pushed.

Suggestions Applied

• Reference anchor: repointed the citation to the enrolled-bill PDF — with a correction to the suggestion itself. The proposed #page=81 anchor does not contain the exemption text (page 81 is an energy-title section in the enrolled PDF, and NPS rescissions in the public-law PDF). The verified location of the noncontiguous-states good-faith exemption duration clause is file page 12, and the enacting subsection is Section 10102(c) (not 10102(a)). Reference now reads BILLS-119hr1enr.pdf#page=12, Section 10102(c).
• PR description: appended a "Renewal trigger" note documenting the renewal assumption behind the 2026-11-01 date.

Skipped

• Stale TODO in meets_snap_abawd_work_requirements.py — explicitly scoped to a separate PR by the reviewer.
• Pinned Anchorage-approximation test — the model keys only on state_code == AK with no borough-level geography, so a two-scenario test would be byte-identical inputs documenting nothing; the approximation is documented in the parameter comment instead.

Verification

• Tests: 16/16 passing in is_snap_abawd_hr1_in_effect.yaml, re-verified after the rebase onto main.
• No formula or variable changes; parameter wording, one reference, and one test case only.

🤖 Generated with Claude Code

[hua7450]

The 2026-11-01 date is the correct end of Alaska's approved good-faith exemption window, but it should not replace Alaska's statewide hr1_in_effect date.

PR #9030 already models the geographic part correctly: the 29 boroughs/census areas outside Anchorage are waived through 2026-10-31, Anchorage is excluded, and the county waiver list clears on 2026-11-01. Moving this broader toggle to 2026-11-01 additionally puts all of Alaska—including Anchorage—on the complete pre-HR1 rule set during the window.

That produces incorrect combined behavior because hr1_in_effect: false also suppresses HR1's new Alaska Native/Indian exception and restores every pre-HR1 exception, instead of only the temporary categories Alaska was approved to retain. Alaska's current guidance describes a mixed regime: the non-Anchorage area waiver plus retained exceptions for specified groups, while the new Alaska Native/Indian exception is in effect.

Requested fix:

1. Keep Alaska's general HR1 effective date at 2025-11-01.
2. Leave #9030's Alaska waived_counties entries unchanged (29 areas through 2026-10-31, Anchorage excluded, empty from 2026-11-01).
3. Add a separate Alaska good-faith-exemption-window parameter: false before 2025-11-01, true from 2025-11-01, and false from 2026-11-01.
4. During that window, add only the approved temporary exceptions to the post-HR1 branch: the published age group, households with children aged 14–17, veterans, people experiencing homelessness, and former foster youth aged 18–24. Keep is_snap_abawd_indian_exempt active independently as the new permanent HR1 exception.
5. Replace the current behavioral regression with Anchorage-specific cases (county_fips: "02020"). Without county input, Alaska falls back to waived Aleutians East Borough, so the age-60 case passes through the area waiver and does not exercise this date change.
6. Add Alaska boundary cases for 2026-10 and 2026-11, plus Anchorage cases for a healthy adult, the temporary retained groups, and an Alaska Native/Indian person.

Please also verify the approval letter's exact age boundary before pinning it: Alaska's public page says ages 56–64, which is not identical to restoring the entire pre-HR1 age threshold.

Sources:

• Alaska Department of Health: https://health.alaska.gov/en/education/hr-1-ak-impacts/
• USDA ABAWD exceptions implementation memorandum: https://www.fns.usda.gov/snap/obbb-abawd-exemptions-implementation

[DTrim99]

Reworked per @hua7450's review — good-faith window as an overlay, not a toggle flip

Thanks for the detailed review. You're right that flipping the statewide hr1_in_effect toggle to 2026-11-01 was over-broad: it put all of Alaska (including Anchorage) on the complete pre-HR1 rule set during the window and suppressed HR1's new Alaska Native/Indian exception. Implemented the mixed regime exactly as requested.

What changed

1. Kept Alaska's HR1 effective date at 2025-11-01 (statewide, same as Hawaii) — so HR1 and its new Alaska Native/Indian exception are in effect throughout.
2. Left the geographic layer to Statewide ABAWD waiver map (FNS FY2025 Q1) + Sec. 10102 discretionary-cut parameter #9030 — no change to waived_counties (29 boroughs through 2026-10-31, Anchorage excluded, cleared 2026-11-01).
3. Added a separate good-faith-exemption-window parameter (gov/states/ak/dpa/snap/work_requirements/abawd/good_faith_exemption/in_effect: false / 2025-11-01 true / 2026-11-01 false) plus an is_snap_abawd_in_good_faith_exemption_window resolver.
4. During the window, layered ONLY the approved temporary categories on top of the post-HR1 set: ages 56–64, households with children 14–17 (pre-HR1 dependent threshold restored to 18 in meets_snap_work_requirements_person), veterans, homeless, and former foster youth 18–24. is_snap_abawd_indian_exempt stays active independently as the permanent HR1 exception.

Age boundary — verified per your note

Confirmed against the Alaska DOH page (verbatim, twice): the retained band is 56–64, not the pre-HR1 55+ threshold. Encoded as a retained_age_minimum of 56 (ages 65+ are already exempt post-HR1, so a single lower bound reproduces the 56–64 band). Test Case 47/48 pin exactly this: age 55 is not exempt during the window, age 56 is.

Tests

Replaced the area-waiver-masked regression cases with Anchorage-specific cases (county_fips: "02020"), since without county input Alaska falls back to the waived Aleutians East Borough and never exercises the date logic (Case 58 keeps that regression explicitly). Added window boundary cases (2026-10 exempt → 2026-11 not), a healthy non-working adult, each retained category (in-window exempt / after-window not), the child 14–17 threshold restoration, and an Alaska Native/Indian person exempt both in and after the window. All affected suites pass locally (ABAWD 58, work-requirements 19, hr1-resolver 16, window-resolver 7, waived-area 32, snap integration 11).

Note on the renewal ceiling (good-faith exemptions renewable but expiring no later than 2028-12-31): the USDA memo page was WAF-blocked to automated fetches, so I sourced it from secondary snippets — it's immaterial to the encoded window (which ends 2026-10-31), but flagging it for a primary-source check if you'd like.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (bb81609) to head (2d1f58a).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #8972   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            1         3    +2     
  Lines           17        68   +51     
  Branches         1         0    -1     
=========================================
+ Hits            17        68   +51     

Flag	Coverage Δ
unittests	100.00% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.