Skip to content

Date Medicaid adult-group limits from state coverage starts - #10826

Merged
MaxGhenis merged 10 commits into
mainfrom
medicaid-adult-group-start-dates
Nov 15, 2026
Merged

MaxGhenis merged 10 commits into
mainfrom
medicaid-adult-group-start-dates

Conversation

@MaxGhenis

@MaxGhenis MaxGhenis commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

Date Medicaid adult-group income limits in Maine, Idaho and Nebraska from their coverage starts. Maine covers the modeled childless adult in 2019 and 2020, and Idaho covers that adult in 2020. Nebraska's October 2020 start changes no annual result, since the model reads the limit on January 1.

#10874 already dated Virginia's and Utah's limits. We add state sources for both and test all six transitions.

The adult-group parameter matches main from January 1, 2021 onward. We retain the other states' limits and annual evaluation.

Closes #10791

State Coverage start and effective FPL limit Official sources
Maine July 2, 2018: 1.38 CMS SPA 18-0006, page 2; CMS approval letter, April 3, 2019, page 1; 22 MRSA 3174-G(1)(H)
Idaho January 1, 2020: 1.38 CMS SPA ID-19-0007, page 3; DHW strategic plan, page 12; House Health and Welfare Committee minutes, January 29, 2020, page 3
Nebraska October 1, 2020: 1.38 CMS Heritage Health Adult approval, October 20, 2020, page 4; SFY2020 Medicaid annual report, page 4

For Virginia and Utah, we add these state sources beside the CMS approvals that #10874 cited:

State Coverage start and effective FPL limit State sources
Virginia January 1, 2019: 1.38 October 2018 eligibility manual, M0310.136(A), PDF page 43
Utah April 1, 2019: 1.00 July 2019 expansion chart, page 1; SB 96, section 26-18-415(1)(b)(i)
Utah January 1, 2020: 1.38 2020 provider training, page 38

Invariants and tests

  • Each limit changes on its coverage start date and retains its prior value through the preceding day. Python tests check all six daily transitions in the five states.
  • From 2018 through 2026, each of the five states' limits never falls and takes only its permitted values. A Hypothesis property checks this invariant.
  • Each of the five states retains 1.38 from 2021 through 2035. A second Hypothesis property checks this invariant.
  • Household tests cover Kentucky, Virginia, Maine, Idaho, Utah and Nebraska in 2019, 2020, 2021 and 2026. All existing cases remain, including the three spouse cases from Apply each state's rule for a Medicaid applicant whose spouse is not aged, blind or disabled #10640.
  • The Maine 2019, Maine 2020 and Idaho 2020 cases expect the adult group and no PTC eligibility. On main, the adult group stays closed in Maine and Idaho until January 1, 2021.
  • Additional 2026 households earn $20,000, above the PTC income floor. In the six expansion states, these adults qualify for the adult group and do not qualify for PTCs; the Texas adult qualifies for PTCs and no Medicaid category.

The 48 Medicaid category cases and eight Python tests, including two Hypothesis properties with 30 examples each, pass in CI.

Hand calculations use the checked-in poverty guidelines. For $15,000 of wages, the Medicaid income ratios equal 15,000 / 12,490 = 1.201 in 2019 and 15,000 / 12,760 = 1.176 in 2020.

For the additional 2026 households, 20,000 / 15,960 = 1.253, below 1.38, using the 2026 HHS guideline. We calculate 2026 PTC eligibility with the 2025 HHS guideline: floor(100 × 20,000 / 15,650) / 100 = 1.27, above 1.00.

Under 26 USC 36B(c)(1)(A) and (c)(2)(B)(i), these adults need at least 100% FPL in 2026 and cannot claim PTCs while eligible for Medicaid.

Methodology choices

  • Use coverage effective dates. CMS approved Maine's amendment on April 3, 2019, effective July 2, 2018.
  • Leave Virginia's and Utah's dates and limits as Correct 2018–2020 Medicaid parent limits in UT, TN and VA and adult-group starts in UT and VA #10874 set them, including Utah's 1.00 from April 1, 2019: 95% + 5% = 100%. Utah set the 95% standard in SB 96, section 26-18-415(1)(b)(i) and published the effective 100% limit in its expansion chart.
  • Preserve the existing convention of storing the 133% standard as 1.38, including the five-point disregard.
  • Preserve annual January 1 evaluation. Maine's July 2, 2018 start first reaches an annual result in 2019. Nebraska's October 1, 2020 start reaches none, since main already holds 1.38 from January 1, 2021. Daily Python tests exercise both dates.
  • Retain main's formulas and the other jurisdictions' limits.

Microsimulation impact

None from 2021 on, by a code read. The PR moves the adult group's start in Maine, Idaho and Nebraska from January 1, 2021 to dates in 2018 and 2020, and every limit equals main's from January 1, 2021. The default dataset holds 2024 records, so we report no aggregate for 2018 to 2020.

axiom: TheAxiomFoundation/rulespec-us#2267 queued (42 CFR 435.119 adult group starts in Maine, Idaho and Nebraska; the corpus holds SPAs ME-18-0006, ID-19-0007 and NE-19-0002 outside rulespec-us's pinned release and lacks the 2019 to 2021 poverty guidelines, so the signed encoder can't run yet; TheAxiomFoundation/rulespec-us#2248 covers Utah and Virginia)

🤖 Generated with Claude Code

MaxGhenis and others added 3 commits November 12, 2026 00:27
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MaxGhenis added a commit that referenced this pull request Nov 14, 2026
Virginia began covering the adult group, with income under 133% of the
poverty guideline, on January 1, 2019 (CMS approval of July 9, 2020,
page 8). Utah covered it to 95% from April 1, 2019 and to 133% as of
January 1, 2020 (CMS approval of December 16, 2020, pages 7 and 19).
These rows match those in #10826.

The Virginia 2019-2020 and Utah 2020 parents that fail the lower parent
standard now qualify through the adult group, so the three household
cases expect eligibility. Two new Utah 2020 cases cover a parent at
$9,500 and one at $25,000; the second qualifies only under 1.38.

Utah's rule read 95% until DAR 52479, so the January 2020 value goes
into the candidates file as an open item, with captures of the four
sources.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MaxGhenis and others added 2 commits November 14, 2026 22:46
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…floor

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MaxGhenis and others added 4 commits November 15, 2026 11:41
On main, #10874 dated Utah's and Virginia's adult-group limits and cited
CMS approvals. This branch dated the same limits and cited state sources.
The values agree: Utah 1.00 from April 1, 2019 and 1.38 from January 1,
2020, and Virginia 1.38 from January 1, 2019. Each date keeps one value
and the references of both sides.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
#10874 reported Virginia's and Utah's starts in the 2.135.4 changelog.
Against main, this branch still changes three starts: Maine on July 2,
2018, Idaho on January 1, 2020 and Nebraska on October 1, 2020.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Idaho's SFY 2020-2024 strategic plan says the Department of Health and
Welfare will implement expansion by January 1, 2020. On January 29,
2020, the department's deputy director told the House Health and
Welfare Committee that coverage began on that date.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Since the last merge, main added #10002, #10924 and #10976. None of
them changes this branch's four files.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@MaxGhenis
MaxGhenis marked this pull request as ready for review November 15, 2026 17:22
@MaxGhenis MaxGhenis closed this Nov 15, 2026
@MaxGhenis MaxGhenis reopened this Nov 15, 2026
@MaxGhenis MaxGhenis closed this Nov 15, 2026
@MaxGhenis MaxGhenis reopened this Nov 15, 2026
@MaxGhenis
MaxGhenis merged commit fa3c3d3 into main Nov 15, 2026
76 of 80 checks passed
@MaxGhenis

Copy link
Copy Markdown
Contributor Author

US + core hub merge audit, us#10826 at 72349633f88139aab0fdaf69e35e7d7aa97eb17e (squash): Date Medicaid adult-group limits from state coverage starts.

  • Reviews:
    • The PR has no GitHub reviews. Three independent review rounds ran outside GitHub, with a fix round after each of the first two.
    • The first round, at 5a61023db4, read Maine's, Idaho's, Nebraska's, Utah's and Virginia's statutes and agency records on government sites. medicaid.gov returned 403 for the CMS approvals, so that round verified none of their pages. It ran 45 households on the head and on the merge base 935b24ff4f.
    • Six households moved from no Medicaid category to the adult group and lost premium tax credit eligibility: Virginia and Maine in 2019 and 2020, and Idaho and Utah in 2020. Every 2021, 2024 and 2026 household matched the merge base.
    • The YAML file passed 38 cases at that head. On the merge base's code it failed those six cases and passed the other 32. The Python file passed eight tests. Against the base parameter, its six boundary tests failed and its two Hypothesis properties passed. A mutant that moved Maine's start to 2019 failed the Maine boundary test alone.
    • It found a merge conflict with Apply each state's rule for a Medicaid applicant whose spouse is not aged, blind or disabled #10640 in medicaid_category.yaml and an empty impact section. Its minor points covered CMS citations dated after the starts, a page list inside a Utah reference title, and 2026 PTC assertions that couldn't fail at 0.940 of the guideline. It also noted that two midyear starts never reach an annual result.
    • The fix round, from d7ccdcc453 to d0f61a0faa, merged main twice and kept main's 17 cases and the PR's 38. It added a state source beside each CMS approval, and seven 2026 cases at $20,000, among them a Texas adult who qualifies for PTCs. It cut run narration from the description. No log of its local test runs survived.
    • The second round, at d0f61a0faa, found that Correct 2018–2020 Medicaid parent limits in UT, TN and VA and adult-group starts in UT and VA #10874, merged on November 15, had dated the same Utah and Virginia limits. The head therefore conflicted with main in income_limit.yaml, and the changelog and description claimed those two states as this PR's fixes.
    • Its minor points covered an Idaho source that looked forward, a validation paragraph that reported local runs, an axiom: line in the external-contributor form and a Maine row without a state-hosted copy of CMS's letter. It read the sources on government sites, recomputed the $20,000 and Texas cases, and compared both merges case by case as parsed YAML.
    • The second fix round merged main at e7505bd0ec in 61c66a9074. It resolved three reference hunks in income_limit.yaml by keeping both sides' references, since Utah's and Virginia's values matched on both sides. Parsed as YAML, the merged file keeps every jurisdiction and dated value of both sides, and its values differ from main's only in Maine, Idaho and Nebraska.
    • It rewrote the changelog fragment to name those three states, added the Idaho House Health and Welfare minutes of January 29, 2020, and merged main at 8dfe48b240 without conflict. It rewrote the description's opening, removed the validation paragraph and marked the PR ready.
    • The third round, at the head, found the merge resolution and CI right. Run 43740734253 passed all 37 jobs, with eight Python tests and 48 Medicaid category cases passing. Main's 17 cases survive unchanged, among them Apply each state's rule for a Medicaid applicant whose spouse is not aged, blind or disabled #10640's three spouse cases. It fetched the Idaho minutes and the Utah CMS approval and found both support their dates.
    • It asked for parity evidence in place of axiom: needed and for a sentence on the CI result, and, as a minor point, for removal of the description's generated-with footer. I made the parity and CI corrections before merge: the axiom: line names a queued parity issue, and the tests section cites CI. The footer stays, as on the merged descriptions of Correct 2018–2020 Medicaid parent limits in UT, TN and VA and adult-group starts in UT and VA #10874, Fix New York child care categorical routes and family shares #10817 and Apply each state's rule for a Medicaid applicant whose spouse is not aged, blind or disabled #10640.
  • Methodology choices for maintainers:
    • The limits start on each amendment's effective date. CMS approved Maine's on April 3, 2019, effective July 2, 2018; Idaho's on November 13, 2019, effective January 1, 2020; and Nebraska's on March 10, 2020, effective October 1, 2020.
    • Each limit stores the 133% standard as 1.38, with the five-point disregard of 42 CFR 435.603(d)(4), following main's convention.
    • is_adult_for_medicaid_fc, the only reader of the parameter, reads it on January 1. Maine's July 2, 2018 start first changes a 2019 result. Nebraska's October 1, 2020 start changes no annual result, since 1.38 already applied from January 1, 2021. The Python tests read the parameter on each start date and the day before.
    • Virginia's and Utah's dates and values stay as Correct 2018–2020 Medicaid parent limits in UT, TN and VA and adult-group starts in UT and VA #10874 set them, including Utah's 1.00 from April 1, 2019. This PR adds state sources and boundary tests for both.
    • 22 MRSA 3174-G(1)(H) set a deadline for Maine's coverage, and the model takes the start date from the CMS-approved amendment.
  • Gates checked live at merge:
    • gh pr checks exits 0;
    • the latest Pull request run at the head, 43763100903, passed and tested this head merged into main at b0edcee095;
    • MERGEABLE;
    • not a draft;
    • no reviewer's latest review requests changes;
    • the semantic-overlap guard: since b0edcee095, main has changed no area this PR touches. Between the PR's merge base 8dfe48b240 and main at b0edcee095, main changed 105 files and none of the 4 this PR touches. It changed one value in fpg.yaml, Hawaii's 2018 amount for each added person (Correct Hawaii's 2018 poverty guideline increment #10970), and every household in this PR has one person outside Hawaii. Use Mississippi parent Medicaid dollar limits for 2017–2026 #10884 added Mississippi's parent limits, which the adult-group reader doesn't read. Main also changed Indiana's CHIP child limits for 2017 to 2020, and no household here lives in Indiana. The branch merges into main at b0edcee095 without conflict;
    • the merge pins the head.
  • Impact: none from 2021 on, by a code read. The PR replaces main's January 1, 2021 entries for Maine, Idaho and Nebraska with the same 1.38 from July 2, 2018, January 1, 2020 and October 1, 2020. For Utah and Virginia it adds only references. Every limit therefore matches main's from 2021. The default dataset holds 2024 records, so we report no aggregate for 2018 to 2020. No microsimulation pair ran.
  • Next:
  • Axiom: Encode the Medicaid adult group start dates in Maine, Idaho and Nebraska (42 CFR 435.119; SPAs ME-18-0006, ID-19-0007 and NE-19-0002) (PE-US #10826) TheAxiomFoundation/rulespec-us#2267 queued, a new pe-parity issue for the 42 CFR 435.119 adult group start dates in Maine, Idaho and Nebraska, with companion rows from this PR's YAML. The corpus holds SPAs ME-18-0006, ID-19-0007 and NE-19-0002 in a state-plan scope outside the release rulespec-us pins. It lacks the 2019 to 2021 poverty guideline notices, so the signed encoder waits on a release that carries both. Encode the 2018-2020 Medicaid parent and caretaker relative income standards in Utah, Tennessee and Virginia, and the Utah and Virginia adult group start dates (42 CFR 435.110 and 435.119) (PE-US #10874) TheAxiomFoundation/rulespec-us#2248 covers Utah and Virginia.

🤖 Generated with Claude Code

@MaxGhenis
MaxGhenis deleted the medicaid-adult-group-start-dates branch November 15, 2026 23:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

The model denies Medicaid to expansion-group adults in Virginia, Maine, Idaho and Utah in 2019 or 2020, when those states covered them

1 participant