[hua7450]

Closes #10385

Stacked on #10361. This branch merges #10361 at b5e2f0256c (its approved partner-case update), which includes main at 4a1a3cdeb7, so until #10361 merges the diff also shows its commits. This PR's own changes are commits 08cfcae (the change), 4e94e07 (review fixes), 084e198 (drops the Maine spouse-deeming case), 372bb19 (SSI income test at or below the rate) and f765258 (round-1 review); the other merge commits bring in #10361 and main.

What and why

Arkansas: ARSeniors covers only aged people

is_optional_senior_or_disabled_for_medicaid accepted anyone who is is_ssi_aged_blind_disabled. Arkansas's group for this pathway, ARSeniors, covers only people aged 65 or older.

• New parameter senior_or_disabled.covers_blind_or_disabled, by state. It is false for Arkansas from 2002-10-01, when ARSeniors started (Policy Directive MS 02-07), and true elsewhere from 2018-01-01, the date the other senior_or_disabled files use.
• is_optional_senior_or_disabled_for_medicaid: where the flag is false, the person must be is_ssi_aged (65 or older).
• Noncitizen exception. Turn off the Medicaid optional aged/disabled pathway in 14 states that cover only SSI recipients #10361 keeps noncitizens who fail the Medicaid immigration test in this pathway, so emergency Medicaid still has a category for them (42 U.S.C. 1396b(v)(2)(B) waives only the receipt of SSI). In Arkansas a blind or disabled noncitizen under 65 would get emergency Medicaid through the SSI group, not ARSeniors. So they keep the pathway only if they also meet SSI's income test (ssi_countable_income at or below the SSI rate: "not more than" the benefit rate under 42 U.S.C. 1382(a)(1)(A), and POMS SI 02005.001 C.3) and resource test (meets_ssi_resource_test). The aged-only rule applies to everyone who passes the immigration test. People kept only through this exception take SSI's income test instead of the pathway's: for a couple, the SSI couple rate ($17,892 in 2026) is above 80% of the guideline for two ($17,312).

Maine: savings exclusion

Maine excludes up to $8,000 of savings for an individual and $12,000 for a couple, and counts only the excess toward the $2,000 / $3,000 limit. The model counted all of ssi_countable_resources.

• New parameters gov.states.me.dhhs.medicaid.savings_exclusion.individual ($8,000) and .couple ($12,000), from 2002-01-01, the date in 22 M.R.S. § 3174-AA.
• is_optional_senior_or_disabled_asset_eligible subtracts the exclusion from the marital unit's resources in Maine, floored at zero, before the limit test. The couple amount applies when the marital unit has two people: an eligible couple, or an eligible person with an ineligible spouse, whose assets are deemed (Part 7, § 7-1, Section 1.3).

Definition of savings. Part 16, § 16-4, Section 4.47 counts as savings: savings and checking accounts, IRAs, Keoghs, annuity cash value, stocks, bonds, mutual funds and life insurance cash surrender value. One exclusion covers all of them together. ssi_countable_resources adds bank_account_assets, stock_assets and bond_assets, all of which are savings, so the exclusion applies to the whole of it and no new inputs are needed. The model does not count IRAs, Keoghs, annuities or life insurance as resources at all.

Maine: couple income disregard

Chart 3.1 sets a $20 federal disregard plus a state disregard of $55 for an individual and $80 for a couple, effective July 1, 1987. The couple value changes from $75 to $100, dated 1987-07-01. The individual value ($75) was already right; its date moves from 2015-01-01 to 1987-07-01, which changes no output. The ME lines are their own hunks; #10364 edits only the CA, CT and SC lines and appends its references after the Connecticut entries, while the Maine references sit after the KFF entry.

The couple value also feeds medicaid_optional_senior_or_disabled_income_deemed_from_ineligible_spouse, so when SSI deeming applies the couple budget there uses $100 too. That deeming does not follow Maine's rule for an eligible person with an ineligible spouse: Maine first excludes the Chart 3.5 ineligible spouse standard from the spouse's income (Part 7, Section 2.2.2.I.D), which the model does not apply (#10474). No test here pins a deemed amount.

Sources

Each was read on 2026-10-22 or 2026-10-26.

Rule	Source	Text
Other states covered	42 CFR 435.210; 42 U.S.C. 1396a(m)(1)(A); KFF 50-State Survey (2018), Appendix Table 1	1396a(m)(1)(A) covers people "who are 65 years of age or older or are disabled individuals"; KFF note 2: "AR extends coverage beyond SSI limits for seniors only." These support the true values for every other state.
AR aged only	Arkansas Medical Services Policy Manual, B-321 ARSeniors (September 2026 edition)	"This group consists of individuals sixty-five (65) years of age or over whose income is equal to or below eighty percent (80%) of the Federal Poverty Levels (FPL)."
AR start date	Ark. Code R. 016.20.02-014, Policy Directive MS 02-07, MS 23110	"Effective October 1, 2002 ..."; "The individual must be Aged to qualify for ARSeniors."
AR aged only	AR DHS Health Care Eligibility Quick Reference (January 2025)	"ARSeniors is only for individuals aged 65 or older."
Noncitizen exception	42 U.S.C. 1396b(v)(2)	"(B) such alien otherwise meets the eligibility requirements for medical assistance under the State plan ... (other than the requirement of the receipt of ... supplemental security income benefits under subchapter XVI ...)"
ME savings	22 M.R.S. § 3174-AA(2), PL 2001, c. 450	"Beginning January 1, 2002 ... Savings. An amount up to $8,000 for an individual and up to $12,000 for a household of more than one person."
ME savings	10-144 C.M.R. ch. 332, Part 16, § 16-4, Section 4.47	"Up to $8,000 of savings for an individual, $12,000 for an assistance unit of two or more. Any amount over the excluded amount is counted toward the asset limit." It applies to the SSI-related categories, except State Supplement.
ME savings	2023 MaineCare Eligibility Guidelines	The Aged and Disabled rows' footnote says the same.
ME savings	2026 MaineCare Eligibility Guidelines	Same footnote and $2,000 / $3,000 limits.
ME disregard	10-144 C.M.R. ch. 332, Chart 3.1	"Disregards (eff. 7/1/87) $20.00 Federal disregard $55.00 State disregard for Individual $80.00 State disregard for Couple"
ME disregard	10-144 C.M.R. ch. 332, Part 7, § 7-4, Section 4.3	"Subtract the State Disregard for two ($80.00) ... If countable income is equal to or below 100% of the Federal Poverty Level for two ... the individual or couple is Medicaid eligible."

The official text of 10-144 C.M.R. ch. 332 (filing 2025-101, effective April 29, 2025) on maine.gov matches the Cornell copies. The AR DHS site returns 403 to scripts; the Quick Reference was read through a Wayback capture of 2025-10-30 and the manual with browser headers.

Readers checked

Reader	Effect
medicaid_category	ADULT comes before SENIOR_OR_DISABLED. So blind or disabled Arkansans under 65 who leave the pathway are mostly people on Medicare, who are outside the adult group. They fall to MEDICALLY_NEEDY where they qualify (#10374 switches Arkansas's group on), or to NONE. In Maine, more people with savings, and some couples, now land in SENIOR_OR_DISABLED.
is_209b_ssi_recipient_for_medicaid	Reads is_optional_senior_or_disabled_asset_eligible, but neither Arkansas nor Maine is a 209(b) state. Unchanged.
medicaid_group, is_medicaid_eligible, is_emergency_medicaid_eligible and their readers (ACA PTC, Lifeline, CHIP and others)	Follow the category. Microsimulation results change for AR and ME.

No partner test has an Arkansas or Maine household. No existing YAML case asserts an output that these changes move.

Tests

• is_optional_senior_or_disabled_for_medicaid.yaml (Cases 6-9 are Turn off the Medicaid optional aged/disabled pathway in 14 states that cover only SSI recipients #10361's Alabama cases)
  • Case 10: a disabled Arkansan aged 50 with $900 a month of SSDI who doesn't take up SSI. Income-eligible ($10,560 countable, under 80% of the guideline, $12,768) but not covered.
  • Case 11: an Arkansan aged 70 with the same income is covered.
  • Case 12: a disabled undocumented Arkansan aged 50 with no income or resources keeps the group: they fail the immigration test and meet SSI's tests ($0 below the $994 rate; $0 within $2,000).
  • Case 13: a disabled permanent resident 10 years after entry passes the immigration test, so the aged-only rule applies and they are not covered.
  • Case 14: a disabled undocumented Arkansan with $5,000 in the bank is under Arkansas's pathway asset limit but over SSI's $2,000 limit, so they are not covered.
  • Case 15: a disabled permanent resident 2 years after entry (in the five-year bar, so failing the Medicaid immigration test) with $1,014 a month of SSDI has (1,014 - 20) x 12 = $11,928 countable, equal to the 2026 SSI rate ($994 x 12), so they keep the group.
  • Case 16: the same person with $1,015 a month has $11,940 countable, over the SSI rate though under Arkansas's 80% of the guideline, so they are not covered.
  • Case 17: a blind Arkansan aged 50 who is not disabled is not covered.
  • Case 18: an Arkansas couple aged 70 and 50 (disabled) is covered only for the aged spouse.
  • Case 19: a disabled Arkansas couple in the five-year bar with $735 a month of SSDI each has $8,700 countable each, within half the SSI couple rate ($745.50 a month), but $17,400 together is over 80% of the guideline for two ($17,312). They keep the group through the SSI test.
• is_optional_senior_or_disabled_asset_eligible.yaml
  • A Mainer with $9,000 in the bank is eligible: $1,000 counts, under $2,000.
  • A Mainer with $10,000 is ineligible: $2,000 counts, which is not under $2,000 (Part 7, § 7-1: assets "must be under" the limits).
  • A Maine couple with $9,000 in the bank and $5,000 in stocks is eligible: $2,000 counts, under $3,000. The individual exclusion would leave $6,000.
  • A Maine couple with $10,000 in the bank and $5,000 in stocks is ineligible: one $12,000 exclusion leaves $3,000, not under $3,000. An $8,000 exclusion per spouse would leave $2,000.
• medicaid_optional_senior_or_disabled_countable_income.yaml: a Maine couple aged 72 and 70 with $1,100 and $790 a month of Social Security has (1,890 - 100) x 12 = $21,480 countable, under the 2026 guideline for two ($21,640). With a $75 disregard it would be $21,780, over the guideline.
• Tests were not run locally; CI validates.

Not changed here

What	Source	Why
Maine's ineligible spouse standard	Part 7, Section 4.3 and Section 2.2.2.I.D; Chart 3.5	Maine excludes the Chart 3.5 ineligible spouse standard (half the SSI individual rate, $497 a month in 2026) from the spouse's income, unearned income first, uses the $80 couple disregard whether or not anything is deemed, and tests the total against 100% of the guideline for two. The model deems only when the spouse's income is above the SSI couple-minus-individual differential, and then deems all of it. For example, $1,000 of Social Security with a spouse's $600 pension is $12,036 a year under Maine's rule and $18,000 in the model. Follow-up issue #10474.
Order of Maine's disregards	Part 7, Section 2.1	Maine subtracts the $20 federal disregard, then the $65 and one-half earned income exclusion, then the $55 / $80 state disregard. The model subtracts the combined $75 / $100 as the SSI general exclusion before the earned income exclusion. For unearned income the results match; for earners the model's countable income is up to $27.50 a month higher for an individual and $40 for a couple. After #10364 merges, Maine can be encoded as disregard $20 plus #10364's additional_disregard $55 / $80. Follow-up issue #10474 (b).
Maine blind-only applicants	Part 7, Section 3.3 Note and Section 4.3 Note; Chart 3.4	They must meet the State Supplement income standard (the SSI rate plus $10 for an individual or $15 for a couple living alone or with others), strictly below it, not 100% of the guideline. Follow-up issue #10474.
Maine parent deeming	Part 7, Section 3.4	Maine takes only the $20 federal disregard from the parents' income and the state disregard once, from the child's countable income; the model applies Maine's $75 to the parents' income. Follow-up issue #10474.
Maine's savings exclusion in the medically needy test	22 M.R.S. § 3174-AA; Part 16, Section 4.47; Part 10, § 10-3, Section 3.1	The statute covers "all individuals and families subject to an asset test", but is_medically_needy_for_medicaid tests assets without the exclusion (and over the tax unit, with <=). Follow-up issue #10474 (e).
$80 disregard for an applicant with an ineligible spouse, without deeming	Part 7, Section 4.3	The model picks the disregard by ssi_couple_computation_applies, so this applicant gets the $75 individual amount. Follow-up issue #10474 (f).
Zero countable income for some noncitizens	—	On main, ssi_countable_income and medicaid_optional_senior_or_disabled_countable_income are 0 for anyone who is not an SSI-eligible individual or spouse, for example an undocumented person, whatever their income. Until that changes, the SSI income test here binds only for noncitizens SSI treats as qualified; the resource test binds for everyone. #10240 (draft) counts that income.
ARSeniors' Medicare condition	B-321	A person entitled to Medicare must enroll in it. Not modeled before or after.
Arkansas's asset limit	—	#10364 updates it.
New York	18 NYCRR 360-4.6(a)(2)(vii)	Its deductions, income level and asset limit are a follow-up issue.

Merge order

Merge after #10361, then #10364.

Test merges of this branch at f765258 (0 commits behind main) on 2026-10-27:

PR	Head	Result
#10361	45d7868	Already merged in
#10364	216aa26	Conflicts in is_optional_senior_or_disabled_asset_eligible.yaml and medicaid_optional_senior_or_disabled_countable_income.yaml, both cases appended at the end of the file by each branch. Keep both. The disregard files and is_optional_senior_or_disabled_asset_eligible.py merge cleanly.
#10240	7555f0f	Conflict in medicaid_optional_senior_or_disabled_countable_income.yaml, cases appended at the end by each branch. Keep both.
#10370	d70717e	Clean
#10373	72e4a4e	Clean
#10374	c1b120a	Clean
#10378	928c372	Conflicts in income/limit/couple.yaml, income/limit/individual.yaml and medicaid_optional_senior_or_disabled_income_limit.py, all from #10361 (#10361 and #10378 conflict with each other directly); this PR changes none of them.

Test plan

• YAML tests with expected values from the Maine statute and manual, the Arkansas manual and rule, and 42 U.S.C. 1396b(v) (amounts from the 2026 poverty guideline and SSI rate already in the model)
• Source-conflict register: not applicable (the Maine statute and manual agree; the Arkansas sources agree)
• ruff format
• Changelog fragment in changelog.d/

axiom: TheAxiomFoundation/rulespec-us#1930 queued

🤖 Generated with Claude Code

[PavelMakarchuk]

Program Review

PR #10386, "Limit Arkansas's Medicaid aged, blind and disabled pathway to aged people, and apply Maine's savings exclusion and couple disregard" (hua7450). Round 1, full review.

IDs: C = critical, SA = should address, G = suggestion.

Paths: F = policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/senior_or_disabled/covers_blind_or_disabled.yaml, D = the income/disregard/ folder beside F, S = policyengine_us/parameters/gov/states/me/dhhs/medicaid/savings_exclusion/, V and A = is_optional_senior_or_disabled_for_medicaid.py and is_optional_senior_or_disabled_asset_eligible.py in policyengine_us/variables/gov/hhs/medicaid/eligibility/categories/, TV and TA = their test files in policyengine_us/tests/policy/baseline/gov/hhs/medicaid/eligibility/categories/. Line numbers refer to the reviewed head.

Source Documents

• Reviewed head SHA: 372bb19
• Mode: full
• Scope: this PR's own commits; stacked on #10361
• Own commits 08cfcae, 4e94e07, 084e198, 372bb19: 11 files, +374/-3.
• PDF (read as images): Arkansas MS Manual, B-321 (page 60); Arkansas Quick Reference, January 2025 (page 2); 2023 and 2026 MaineCare Eligibility Guidelines (page 3 each).
• Web: Ark. Code R. 016.20.02-014; 22 M.R.S. § 3174-AA; 10-144 C.M.R. ch. 332 (§ 7-1 to § 7-4, § 10-3, § 16-4, Chart 3.1); 42 CFR 435.210; 42 U.S.C. 1396a(m), 1396b(v), 1382(a); KFF 2018 survey.
• Year: 2026.

Branch Status

• This branch carries #10361, so CI shows #10361's partner contract failure (ssi_earned_income_at_breakeven_ineligible (federal)); that item is in the #10361 review and is not counted here. After #10361 merges, merge main into this branch.
• The branch is 19 ahead of and 21 behind main. A trial merge with current main is clean.
• Merge order: #10361, then #10364, then this PR. A trial merge with #10364 (60ef70c) gives two append/append conflicts, in TA and in medicaid_optional_senior_or_disabled_countable_income.yaml. Keeping both sides passes the own test files, 90 of 90. Maine keeps the strict <, and Arkansas Case 13 stays under #10364's MSP limit.
• Process note: the body's "Merge order" heads are stale (#10364 is at 60ef70c and also edits MN lines; #10240 is at c4145b0). The conflict sets do not change.

What the PR gets right

• Arkansas. All three Arkansas sources limit ARSeniors to people 65 or older from 2002-10-01. is_ssi_aged gives false at 64 and true at 65.
• Noncitizen exception. 1396b(v)(2)(B) waives only the receipt of SSI. The <= test matches "not more than" in 1382(a)(1)(A), and ssi_amount_if_eligible carries the couple, deeming and reduced rates. Cases 11-15 pin each condition.
• Maine savings. $8,000 / $12,000 from 2002-01-01 match § 3174-AA(2), Section 4.47 and both guidelines. The exclusion comes after the marital-unit sum and before the strict < of § 7-1.
• Maine disregard. 75 / 100 from 1987-07-01 match Chart 3.1. Four mutants of the couple step are killed.
• Scope. All 13 new references resolve. "Not changed here" names each Maine gap. No partner file changes.

Critical (Must Fix)

C1. F cites sources only for Arkansas. The 50 true values have none.

• Where: F:8-14 (references), F:16-118 (values). The file is new in this PR.
• Evidence: the three references each support only AR: 2002-10-01: false. Nothing supports the 50 2018-01-01: true values. The sibling flags cite the federal side: covers_non_ssi_recipients.yaml cites 42 CFR 435.210 and 1396a(m), and uses_ssi_federal_benefit_rate.yaml cites 435.210 and KFF. 435.210(b) covers groups "who meet the income and resource requirements of SSI". 1396a(m)(1)(A) covers people "who are 65 years of age or older or are disabled individuals". KFF Appendix Table 1, note 2: "AR extends coverage beyond SSI limits for seniors only."
• Why critical: a missing reference is critical under the criteria. The values are right.
• Fix: put these entries before the Arkansas ones. A head copy with them (and the G2 and G3 code) passes the own test files, medicaid_category.yaml and the SA1 cases (54/54), and gov/hhs/medicaid (714/714).

    - title: 42 CFR 435.210 - Optional eligibility for individuals who meet the income and resource requirements of the cash assistance programs
      href: https://www.ecfr.gov/current/title-42/section-435.210
    - title: 42 U.S. Code § 1396a(m)(1)(A) - Poverty-level aged and disabled group
      href: https://www.law.cornell.edu/uscode/text/42/1396a#m_1_A
    - title: KFF, Medicaid Financial Eligibility for Seniors and People with Disabilities - Findings from a 50-State Survey (2018), Appendix Table 1, note 2
      href: https://www.kff.org/report-section/medicaid-financial-eligibility-for-seniors-and-people-with-disabilities-findings-from-a-50-state-survey-appendix-tables/

Should Address

SA1. Three edge cases are missing, and three mutants survive the whole Medicaid test folder.

• Where: the Arkansas gate at V:44-48 (Cases 9-15, TV:157-261) and the Maine exclusion at A:28-31 (TA:152-195).
• Evidence: each mutant passes the own test files plus medicaid_category.yaml (51 cases) and gov/hhs/medicaid (714 cases). Each fails on one case below.
  • is_ssi_aged | is_blind in place of is_ssi_aged: every Arkansas case uses meets_ssi_disability_criteria, and none has a blind person who is not disabled. Case 16 kills it.
  • person.marital_unit.any(is_ssi_aged): no couple has one aged and one younger spouse, so nothing proves that the gate is per person. Case 17 kills it.
  • $8,000 per spouse before the marital sum: the Maine couple case (9,000 + 5,000) passes under both rules, but Section 4.47 says "The exclusion is not applied to each account." The Maine case kills it.
• Why should address: missing edge cases for variables that are already tested.
• Fix: the three cases pass at the head. Append Cases 16 and 17 after TV:261, and the Maine case after TA:195 (after #10364's cases once you rebase; its output does not change).

- name: Case 16, blind Arkansan under 65 who is not disabled is not covered, since ARSeniors covers only aged people.
  period: 2026
  input:
    age: 50
    state_code: AR
    is_blind: true
    social_security_disability: 10_800
    takes_up_ssi_if_eligible: false
  output:
    is_optional_senior_or_disabled_income_eligible: true
    is_optional_senior_or_disabled_for_medicaid: false

- name: Case 17, Arkansas couple with an aged spouse and a disabled spouse under 65 covers only the aged spouse.
  period: 2026
  input:
    people:
      person1:
        age: 70
        social_security_retirement: 6_000
        takes_up_ssi_if_eligible: false
      person2:
        age: 50
        meets_ssi_disability_criteria: true
        social_security_disability: 6_000
        takes_up_ssi_if_eligible: false
    marital_units:
      marital_unit:
        members: [person1, person2]
    households:
      household:
        members: [person1, person2]
        state_code: AR
  output:
    is_optional_senior_or_disabled_income_eligible: [true, true]
    is_optional_senior_or_disabled_for_medicaid: [true, false]

- name: Maine couple with 15,000 of savings is ineligible, since one 12,000 exclusion covers the couple.
  period: 2026
  input:
    people:
      person1:
        age: 72
        bank_account_assets: 10_000
      person2:
        age: 70
        stock_assets: 5_000
    marital_units:
      marital_unit:
        members: [person1, person2]
    households:
      household:
        members: [person1, person2]
        state_code: ME
  output:
    # 15,000 minus 12,000 leaves 3,000, not under 3,000. An 8,000
    # exclusion per spouse would leave 2,000 + 0 = 2,000.
    is_optional_senior_or_disabled_asset_eligible: [false, false]

Suggestions

G1. Maine's $55 / $80 state disregard comes before the earned income exclusions.

• Where: D/individual.yaml:63-65, D/couple.yaml:63-65, used as the SSI general exclusion at policyengine_us/variables/gov/hhs/medicaid/income/_apply_medicaid_optional_senior_or_disabled_exclusions.py:73-79.
• Evidence: § 7-2 Section 2.1 deducts "in this order": the $20, IRWE, the $65 and one-half, then the $55 / $80. In 2026, $1,000 a month of wages gives an individual $5,160 at the head and $4,830 under Maine's order; $1,500 gives a one-earner couple $8,010 against $7,530. Eligibility flips only for wages of about $2,800-$2,855 (individual) and $3,772-$3,852 (couple) a month.
• Why a suggestion: the combined encoding predates the PR, and the individual result is unchanged. For couples the PR narrows the gap from $52.50 to $40 a month. "Not changed here" lists it, and #10474 tracks it. The head has no slot after the earned income exclusions; #10364 adds one.
• Fix, after the rebase on #10364 (run on the #10364 merge tree): set Maine's disregard to 20 and #10364's additional_disregard to 55 / 80, and move the Maine references with them. This gives exactly $4,830 and $3,765 each, and the own test files pass 90/90. The spouse-deeming variable reads only disregard.*, so its amount moves too; #10474 covers that path. You can also leave this to #10474.

# D/individual.yaml and D/couple.yaml
ME:
  # The $20 federal disregard. Maine's state disregard comes after the
  # earned income exclusions (additional_disregard).
  1987-07-01: 20
# additional_disregard/individual.yaml (couple.yaml: 80, "for a couple")
ME:
  # Maine's $55 state disregard for an individual (Chart 3.1).
  1987-07-01: 55

G2. Maine's savings exclusion is not applied in the medically needy asset test.

• Where: policyengine_us/variables/gov/hhs/medicaid/eligibility/categories/medically_needy/is_medically_needy_for_medicaid.py:28, :43. Not in this PR.
• Evidence: § 3174-AA covers "all individuals and families subject to an asset test", and § 10-3 Section 3.1 applies the $2,000 / $3,000 limits to the SSI-related medically needy groups. A Mainer aged 70 with $1,600 a month of Social Security and $20,000 of medical expenses is MEDICALLY_NEEDY with $1,500 in the bank. With $9,000, the pathway asset test passes, but the medically needy test fails and the category is NONE.
• Why a suggestion: the merge base gives the same results, so the gap predates the PR. But the S description ("under the Medicaid aged, blind, and disabled program") is broader than its one reader, and #10474 does not mention medically needy.
• Fix: add a "Not changed here" row and a #10474 item. Optional code, passed at the head, inserted before under_limits: the $9,000 case becomes MEDICALLY_NEEDY, the $1,500 case does not change, and the medically needy tests (8/8) and gov/hhs/medicaid (714/714) pass.

        me = parameters(period).gov.states.me.dhhs.medicaid.savings_exclusion
        me_exclusion = where(is_joint, me.couple, me.individual)
        assets = max_(assets - where(state == "ME", me_exclusion, 0), 0)

G3. The Arkansas noncitizen exception still applies the ARSeniors couple income limit.

• Where: V:41-43, V:49-53.
• Evidence: the SSI group tests a couple against the couple rate (1382(a)(2)(A)), and 80% of the 2026 guideline for two ($17,312) is below it ($17,892). Take an Arkansas couple, both 50, disabled, in the five-year bar and Medicare-eligible, with no SSI take-up and an emergency condition. With $8,820 of SSDI each, ssi_countable_income is $8,700 each, under 12 × $745.50, so the SSI test passes. But the pathway income test fails, the category is NONE, and emergency Medicaid is [false, false]. With $8,760 each, both are covered.
• Why a suggestion: the merge base gives the same results, so the PR does not create this $580-a-year band. For an individual the limit does not bind ($12,768 is above $11,928).
• Fix (optional, passed at the head): replace V:41-43 and V:49-51 as below; V:44-48 and the return stay. The $8,820 couple becomes SENIOR_OR_DISABLED with emergency Medicaid, the $8,760 couple does not change, and gov/hhs/medicaid passes 714/714.

        state_covers = p.covers_blind_or_disabled[state].astype(bool)
        meets_ssi_tests = meets_ssi_income_test & meets_ssi_resource_test
        on_exception_path = ~state_covers & ~immigration_eligible & meets_ssi_tests
        covers_blind_or_disabled = state_covers | on_exception_path
        # ... is_senior_or_disabled (V:44-48) unchanged ...
        pathway_income_eligible = person(
            "is_optional_senior_or_disabled_income_eligible", period
        )
        exception_only = on_exception_path & ~person("is_ssi_aged", period)
        income_eligible = where(exception_only, True, pathway_income_eligible)

G4. Maine's exclusion is a state literal, and it depends on the SSI resource list.

• Where: A:25-31; policyengine_us/parameters/gov/ssa/ssi/eligibility/resources/countable.yaml:5-9.
• Evidence: the other state values of this pathway are state_code breakdowns under senior_or_disabled/; A:30 uses state == "ME". The exclusion comes off all of ssi_countable_resources, which is right only while that list holds savings (bank, stock and bond assets).
• Why a suggestion: the folder has precedent ("MO", "CT"), and the output is correct today.
• Fix (no output change; not run): a senior_or_disabled/assets/savings_exclusion/{individual,couple}.yaml breakdown, 0 elsewhere, removes the literal and also serves G2. At least add a NOTE in countable.yaml that Maine's exclusion assumes each listed resource is savings.

G5. Cite the current Maine guidelines and a stable Quick Reference copy.

• Where: S/individual.yaml:13-14, S/couple.yaml:13-14, F:13-14.
• Evidence: the 2023 guidelines are three editions old. The 2026 edition, page 3, has the same footnote and limits. The Quick Reference URL is unversioned.
• Fix: add this entry after the 2023 entry in each S file. At F:14, use the body's Wayback capture: https://web.archive.org/web/20251030205724/https://humanservices.arkansas.gov/wp-content/uploads/Quick-Reference.pdf#page=2.

    - title: 2026 MaineCare Eligibility Guidelines, Age 65 or Older and Individuals with a Disability
      href: https://web.archive.org/web/20260903154955/https://www.maine.gov/dhhs/sites/maine.gov.dhhs/files/inline-files/2026%20MaineCare%20Eligibility%20Guidelines%206.23..26_0.pdf#page=3

G6. A Maine aged person with an ineligible spouse gets the $75 disregard, not the $80.

• Where: policyengine_us/variables/gov/hhs/medicaid/income/medicaid_optional_senior_or_disabled_countable_income.py:42-51.
• Evidence: the disregard follows ssi_couple_computation_applies, but § 7-4 Section 4.3 applies $80 to "an eligible individual with an ineligible spouse" too. An aged Mainer with $1,100 a month of Social Security and a non-disabled spouse aged 50 gets (1,100 - 75) × 12 = $12,300.
• Why a suggestion and fix: the body defers this budget to #10474. Make sure #10474 names the disregard too.

G7. Readability. At V:41, the local covers_blind_or_disabled mixes the parameter with the exception (the G3 code splits them). At A:31, the max_(..., 0) floor never binds. At V:37-39, a comment can say that January's rate holds all year.

Investigated and cleared

• Earlier years. The 2002 and 2018 start dates break no year.
• Maine assets with an ineligible spouse. $13,000 gives [true, true] and $15,000 gives [false, false], as § 7-1 Sections 1.1 and 1.3 require.
• KFF Maine value. KFF gives one $75 (the individual amount). The Maine sources govern. Not a source conflict.
• axiom: line. TheAxiomFoundation/rulespec-us#1930 queued points to an open pe-parity issue. Process note only.

Out of scope / follow-up

• #10474. It tracks Maine's ineligible spouse standard, disregard order, blind-only standard and parent deeming. Add medically needy (G2).
• Medically needy. The variable sums assets over the tax unit and uses <= where Maine says "must be under". Both predate this PR.
• Noncitizen income. ssi_countable_income is 0 for people that SSI treats as not qualified (#10240).

Validation Summary

Check	Result
CI	32 pass, 3 skip, 1 fail (from #10361, not counted)
Local tests	Own 3 test files 37/37; gov/hhs/medicaid 714/714; SA1 cases 3/3
Partner contract tests	None edited. SSI partner file: 14/14 at the merge base, 13/14 at the #10361 head and this head
References	13 new entries resolve. No source for the 50 true values (C1)
Regulatory	Arkansas rule and date, Maine amounts and dates correct. G1-G3 predate the PR
Code patterns	ruff clean. "ME" literal (G4)
Mutation	17 mutants: 13 killed, 3 survive (SA1), 1 equivalent
Merge trials	#10364: two append conflicts, both sides kept 90/90. main: clean
PR body	Valid axiom: line. Stale merge-order heads (process note)

Review Severity: REQUEST_CHANGES

One critical item (C1, no source for the 50 true values) and one should-address item (SA1, three missing edge cases) are open.

Next Steps

To auto-fix issues: run the fix-pr workflow for this PR.

1. Add the three references in C1 to F.
2. Add the three cases in SA1 to TV and TA.
3. Take the suggestions that you agree with. For G2, at least add the "Not changed here" row and the #10474 item.
4. Run make format, then the own test files and the gov/hhs/medicaid folder.
5. After #10361 merges, merge main into this branch. Merge after #10364, and keep both sides of the two appended test files.
6. Refresh the body's "Merge order" heads.

[hua7450]

Thanks, Pavel. Round 1 is applied at f765258, which also merges #10361 at 45d7868 and main at 86b04f3 (0 behind). No local tests were run (CI validates); every new expected value was derived by hand. ruff format and ruff check are clean. #10361 added a Case 9, so the Arkansas cases are now Cases 10-16, and the new ones are 17-19.

Item	Disposition
C1	Fixed. F now cites 42 CFR 435.210, 42 U.S.C. 1396a(m)(1)(A) and KFF's 2018 50-state survey, Appendix Table 1 note 2 ("AR extends coverage beyond SSI limits for seniors only", checked on the page), ahead of the Arkansas sources.
SA1	Fixed. Your three cases are appended as given: Case 17 (blind Arkansan under 65, not disabled), Case 18 (Arkansas couple aged 70 and 50, only the aged spouse covered), and the Maine couple with $10,000 + $5,000 (one $12,000 exclusion leaves $3,000, not under $3,000).
G1	Not in this PR. It needs #10364's additional_disregard, which isn't on this branch. Follow-up #10474 (b) records the encoding you gave (Maine disregard $20 plus additional_disregard $55 / $80), and the body's "Not changed here" row says so.
G2	Not in this PR; added to #10474 (e) and as a "Not changed here" row. is_medically_needy_for_medicaid is edited by #10370 and #10374, and its asset unit (tax unit) and operator (<=) are also off for Maine, so the savings exclusion is better fixed together with those.
G3	Applied. People kept only through the SSI-group exception now take SSI's income test instead of the pathway's (`exception_only
G4	Partly. I kept the Maine-only parameter under gov.states.me.dhhs.medicaid, following the "MO" and "CT" branches in this pathway; a state_code breakdown would add 50 zero entries for one state's rule. I added your NOTE to gov/ssa/ssi/eligibility/resources/countable.yaml: Maine's exclusion assumes each listed resource is savings.
G5	Applied. Both S files cite the 2026 MaineCare Eligibility Guidelines (Wayback copy; page 3 has the same footnote and $2,000 / $3,000 limits), and F now links the Wayback Quick Reference at #page=2.
G6	Added to #10474 (f). Its "Done when" list now includes the $80 for an eligible person with an ineligible spouse, whether or not anything is deemed.
G7	Mostly applied. The flag and exception are split (with G3), and the comment says January's SSI rate stands for the year. I kept the max_(..., 0) floor in A: it doesn't change the < result, but it keeps assets a countable amount that can't go negative for the limit tests that follow it once #10364 merges.
Merge order heads	Refreshed in the body at this head.

rulespec-us#1930 (the axiom: line) is updated for this head, including the SSI-income replacement for the exception path (companion case 13).

[PavelMakarchuk]

Program Review

PR #10386, "Limit Arkansas's Medicaid aged, blind and disabled pathway to aged people, and apply Maine's savings exclusion and couple disregard" (hua7450). Round 2, incremental review of 372bb19..88851f0.

IDs: C = critical, SA = should address, G = suggestion. New items continue the numbering after the highest prior ID in each class.

Paths as in round 1: F = policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/senior_or_disabled/covers_blind_or_disabled.yaml, S = policyengine_us/parameters/gov/states/me/dhhs/medicaid/savings_exclusion/, V and A = is_optional_senior_or_disabled_for_medicaid.py and is_optional_senior_or_disabled_asset_eligible.py in policyengine_us/variables/gov/hhs/medicaid/eligibility/categories/, TV and TA = their test files in policyengine_us/tests/policy/baseline/gov/hhs/medicaid/eligibility/categories/. Line numbers refer to the reviewed head.

Source Documents

• Reviewed head SHA: 88851f0
• Mode: incremental (round 2)
• Scope: one own commit, f765258 (the round-1 fixes), in 7 own files: F (+6/-1, references), policyengine_us/parameters/gov/ssa/ssi/eligibility/resources/countable.yaml (a NOTE), both S files (+2 each), TA (+22, one case), TV (Cases 17-19, and renumbering after #10361's new Case 9), and V (G3, G7). Merges: #10361 at 45d7868, 4f631a5 and b5e2f02; main at 86b04f3 and 4a1a3cd. The #10361 hunks (covers_non_ssi_recipients.yaml, income/limit/*, medicaid_optional_senior_or_disabled_income_limit.py, the unit test, Case 9 and the partner file) belong to the #10361 review.
• PDF: 2026 MaineCare Eligibility Guidelines, Wayback copy, page 3: the savings footnote ($8,000 / $12,000) and the $2,000 / $3,000 limits. Arkansas Quick Reference, Wayback 2025-10-30, page 2: ARSeniors is only for people aged 65 or older.
• Web: 42 CFR 435.210(b); 42 U.S.C. 1396a(m)(1)(A) (the m_1_A anchor exists); KFF 2018 survey, Appendix Table 1, note 2: "AR extends coverage beyond SSI limits for seniors only." Issue #10474. The #10240 diff (ssi_countable_income, medicaid_optional_senior_or_disabled_countable_income).

Branch Status

• The branch is 28 ahead of and 34 behind main. main changed none of the 12 own files since the merge base (4a1a3cd). GitHub reports the PR as mergeable (blocked only by review).
• CI at head: 34 of 34 checks pass. The round-1 partner failure from #10361 is gone.
• Partner file: the full diff edits policyengine_us/tests/policy/baseline/partners/analytics_coverage/edge_cases/federal/cash/ssi/federal.yaml (ssi_earned_income_at_breakeven_ineligible (federal)). The edit comes from #10361's b5e2f02, not from this PR's own commits. The model change is #10361's: Texas covers aged, blind and disabled people at home only if they receive SSI. Before and after: medicaid 13,708.50 to 0, is_medicaid_eligible true to false, is_aca_ptc_eligible false to true, aca_ptc and premium_tax_credit 0 to 18,344.29. The #10361 review owns that rule and the partner gate. The file passes 14/14 at this head.
• Merge order: #10361 first (open; I approved its round 2 at b5e2f02, which this head contains), then #10364 (open, now at 45120fa, a main merge after 216aa26). #10364 touches the same five own files as before, so the body's conflict notes still hold.
• Process notes: the body's merge-order table still lists #10361 at 45d7868 and #10364 at 216aa26, and its test merges are from f765258. The axiom: line is a process note only.

Round 1 items

ID	Status	Evidence
C1	RESOLVED	F:9-14 cite 42 CFR 435.210, 1396a(m)(1)(A) and KFF Appendix Table 1 note 2, before the Arkansas entries. All three resolve and support the true values.
SA1	RESOLVED	Cases 17 and 18 (TV:287-321) and the Maine $15,000 couple (TA:197-217) are in, as given. Each kills its mutant: is_ssi_aged | is_blind fails Case 17, marital_unit.any(is_ssi_aged) fails Case 18, and an $8,000 exclusion per spouse fails the Maine case. Arithmetic: (900 - 20) x 12 = 10,560, at or below 12,768; 12,000 - 240 = 11,760, at or below 17,312; 15,000 - 12,000 = 3,000, not under 3,000.
G1	DEFERRED, accepted	#10474 (b) gives Section 2.1's order, the $27.50 / $40 gaps and the disregard $20 plus additional_disregard $55 / $80 encoding. #10364, which adds additional_disregard, is still open, so the reason holds. The body has the row.
G2	DEFERRED, accepted	#10474 (e) and a "Done when" line cover the savings exclusion in the medically needy test, the tax-unit sum and the <=. The reason holds: #10370 and #10374 also edit that variable. The body has the row. Small point: the title and first paragraph of #10474 still name four rules, but the issue now has six, (a) to (f).
G3	RESOLVED	V:41-59 split the state flag from the exception path and apply the suggested code. Case 19 (TV:323-359) checked by hand and by probe: ssi_countable_income is 8,700 each and ssi_amount_if_eligible is 745.50, so 725 is at or below 745.50; the pathway sum is 17,400, above 17,312; the result is [true, true]. Without the income change, Case 19 fails, so the case is functional. Couples at $9,066 and $9,072 of SSDI each give covered and not covered, so the SSI <= edge also works for a couple. The change adds an OR, so it drops nobody. It adds only Arkansans under 65 who are blind or disabled, fail the immigration test, and pass SSI's income and resource tests and the pathway asset test. The SSI group reaches each of them, so none is wrongly covered. #10240 makes both countable-income variables count the income of non-qualified noncitizens, so it does not open a gap here. But my G3 code also carved out aged people, and that part was wrong (SA2).
G4	PARTIAL, accepted	The "ME" literal stays (precedent: "MO", "CT"). The NOTE is at countable.yaml:6-8. That was the minimum asked.
G5	RESOLVED	Both S files cite the 2026 guidelines (S/individual.yaml:15-16, S/couple.yaml:15-16). The Wayback copy resolves, and page 3 has the footnote and the limits. F:20 uses the Wayback Quick Reference, and page 2 supports the rule.
G6	DEFERRED, accepted	#10474 (f) gives the $12,300 against $12,000 example and a "Done when" line. The body has the row.
G7	RESOLVED	state_covers and on_exception_path are separate (V:41-47), and V:36 says January's rate stands for the year. The max_(..., 0) floor at A:31 stays. It changes no result, and the reason is acceptable.

New findings

Critical

None.

Should address

SA2. Aged noncitizens on the Arkansas exception path do not get SSI's income test. This withdraws part of my round-1 G3 code.

• Where: V:56, exception_only = on_exception_path & ~person("is_ssi_aged", period).
• Evidence: 42 U.S.C. 1396b(v)(2)(B) waives only the receipt of SSI. So the SSI group reaches aged noncitizens in the same way as blind or disabled ones. Take the Case 19 couple at age 70, with $8,820 of Social Security retirement each. The SSI test passes: $725 is at or below $745.50 a month each. The pathway test fails: $17,400 is above $17,312. At head, is_optional_senior_or_disabled_for_medicaid is [false, false], medicaid_category is NONE, and emergency Medicaid is [false, false] with an emergency condition. The disabled couple aged 50 with the same income is covered (Case 19). The band is above $8,776 and up to $9,066 a year of Social Security each.
• Why should address: the new line at V:56 states a rule that 1396b(v)(2)(B) does not support. It is not a regression: at PRIOR_HEAD and at the merge base, the pathway test was the only income test, so the result was the same NONE. The band is narrow, and it changes only emergency Medicaid. But the fix is 2 lines and one case. The PR followed my G3 code, which had the aged carve-out, so the error is mine.
• Fix (run on a head copy): replace V:53-59 with the code below, and append Case 20 after TV:359. Case 20 fails at head and passes with the fix. With the fix, gov/hhs/medicaid passes 736/736. In the body, change "People kept only through this exception take SSI's income test instead of the pathway's" to say that people on this exception path, aged or not, also pass with SSI's income test.

        # People on the SSI-group exception path pass with its income test:
        # for a couple, the SSI couple rate is above 80% of the poverty
        # guideline for two. This holds for aged people too, since 42 U.S.C.
        # 1396b(v)(2)(B) waives only the receipt of SSI.
        income_eligible = on_exception_path | person(
            "is_optional_senior_or_disabled_income_eligible", period
        )

# The Case 19 income for a couple aged 70 with Social Security retirement.
# Aged noncitizens who fail the immigration test also reach the SSI group
# for emergency Medicaid, so the SSI couple test applies to them too.
- name: Case 20, aged Arkansas couple in the five-year bar under the SSI couple rate but over 80% of the poverty guideline keeps the group.
  period: 2026
  input:
    people:
      person1:
        age: 70
        is_tax_unit_head: true
        immigration_status: LEGAL_PERMANENT_RESIDENT
        years_since_us_entry: 2
        social_security_retirement: 8_820
        takes_up_ssi_if_eligible: false
      person2:
        age: 70
        is_tax_unit_spouse: true
        immigration_status: LEGAL_PERMANENT_RESIDENT
        years_since_us_entry: 2
        social_security_retirement: 8_820
        takes_up_ssi_if_eligible: false
    marital_units:
      marital_unit:
        members: [person1, person2]
    households:
      household:
        members: [person1, person2]
        state_code: AR
  output:
    is_medicaid_immigration_status_eligible: [false, false]
    is_optional_senior_or_disabled_income_eligible: [false, false]
    is_optional_senior_or_disabled_for_medicaid: [true, true]

• Out of scope: the same band exists outside Arkansas, for any immigration-ineligible couple in a state whose pathway couple limit is below the SSI couple rate. The same aged couple gets NONE in VT, WV and ID; NY covers it. This comes from the #10361 design, and a general fix needs care in 209(b) states. It belongs to #10361 or a follow-up issue, not to this PR.

Suggestions

None.

Tests run

• policyengine-core test on TV, TA, policyengine_us/tests/policy/baseline/gov/hhs/medicaid/income/medicaid_optional_senior_or_disabled_countable_income.yaml and medicaid_category.yaml: 56/56 pass.
• policyengine-core test policyengine_us/tests/policy/baseline/gov/hhs/medicaid: 736/736 pass.
• Partner file partners/analytics_coverage/edge_cases/federal/cash/ssi/federal.yaml: 14/14 pass.
• pytest policyengine_us/tests/unit/test_medicaid_abd_ssi_rate_standard_properties.py (#10361's file): 108 pass.
• ruff format --check on V and A: clean.
• Mutation, 6 mutants: 5 killed. Without the G3 income change: Case 19 fails. is_ssi_aged | is_blind: Case 17 fails. marital_unit.any(is_ssi_aged): Case 18 fails. An $8,000 Maine exclusion per spouse: the $15,000 couple fails. A carve-out on is_ssi_aged_blind_disabled: Case 19 fails. One survives gov/hhs/medicaid (736/736): dropping the aged carve-out, which is the SA2 fix, not a test gap.
• Probes: Case 19 intermediates; disabled couples at $8,760, $9,066 and $9,072 each; the aged couple in SA2 at head and with the fix; the same aged couple in VT, WV, ID and NY.

Verdict

REQUEST_CHANGES: C1 and SA1 are resolved, and #10474 tracks G1, G2 and G6 with reasons that hold. One new should-address item is open: SA2, the aged carve-out that came from my G3 code. The 2-line fix and Case 20 are above. Merge after #10361.