[MaxGhenis]

Fixes #10138.

Summary

Apply 20 CFR 416.1163(e)(2): SSI with spousal deeming is the lesser of the couple computation and the individual benefit after the claimant's own countable income. Apply the cap inside uncapped_ssi, so state supplements and other downstream readers receive the corrected federal shortfall.

Identify SSI couples by the existing resource-deeming test for marriage and co-residence, rather than tax-unit head/spouse roles. Joint claims, couple income/resources, deemed income and facility checks use the identified spouse. The methodology predicate retains main's citizenship treatment because CAPI, Medicaid and supplements depend on it. is_ssi_eligible_spouse remains a tax-unit-spouse label; payment calculations treat both spouses alike.

Review

Round 6: the hub's independent review at the new head, pending

Round 5 fixes

• The shared state retained-pair helper now preserves SSI's parent-link and tax-dependent exclusions across households. Existing parent-link resolution runs within the marital unit, including across tax units, while preserving a known co-resident parent identity if an ID repeats in another household. A possible default two-person unit cannot turn known nonspouses into a retained marriage. CO, MO, OR, SD, VA and IA receive this shared correction; genuine recorded marriages apart retain their program-specific budgets.
• Iowa's separated 96-year-old parent and 66-year-old dependent child, with explicit parent IDs and $1,800 resources each, receive individual RCF budgets with omitted marital units and explicit singletons alike: $1,192.57 each. Generated parent-only and dependent-only cases independently test both exclusions across the five states.
• Both requested 2026 SSI: deemed benefit is capped at the individual rate instead of the no-deeming amount, and couples are identified by tax roles #10138 cases are pinned: claimant age 67 with $12,000 wages, spouse age 45 with self-employment income of $6,000 (deeming active) or $5,900 (inactive), annual SSI $6,438 in both. The independent oracle now varies claimant wages; generated cases check continuity at $496/$497/$498 monthly spouse income.
• Three narrow tests call the shared helpers through actual simulation populations, isolating separated-marriage retention, cross-household parent exclusion and co-resident parent identity when an ID repeats across households. The first two isolate the requested mutation checks.

Main merge and added SSI readers

Merged main cbe322b51c6c9f791a318a32f0b093abee683fab as 3765b00cde without conflicts: 554 files, +36,583/-627 lines since the previous merged main cce379ea00. Main brings per-person California CAPI, California food-preparation defaults and blind-rate corrections, eight heating programs (LA/ME/MO/MT/NV/PA/VT/WI), South Carolina TANF SSI exclusions, state tax-role invariance, Social Security PIA and student-loan changes, plus test/performance updates. #10532 is not pre-merged.

The new CAPI readers now use the identified spouse for facility/category/payment-standard counts, CAPI membership, countable income and spouse SSI/SSP receipts. Six third-adult cases isolate each reader. The existing California supplement attribution assumes spouses share an SPM unit and retains its pro rata child-share approximation. New energy programs intentionally count household SSI receipts; SC TANF intentionally excludes individual SSI recipients and their income/resources. Corrected capped payments feed those readers consistently. Montana LIHEAP's spouse categorical gate and SSI-standard counts retain an inherited default-unit identity limitation: holding reported SSI receipt flags fixed reproduces it unchanged on main. This independent limitation remains outside the supplement/Medicaid fixes; follow-up is proposed below.

Medicaid aggregation sweep

These non-MAGI budgets follow spouse financial responsibility under 42 CFR 435.602. Pair totals replace whole-marital-unit sums; existing medically needy parental expense/deeming treatment is retained. All regressions distinguish the unrelated adult's tax/family/SPM units while omitting marital units.

File	Can an outsider enter the former sum?	Treatment and regression
medicaid_optional_senior_or_disabled_countable_income.py	Yes, premiums.	SSI couple computation now deducts only pair premiums. CA pensions $36,000/$24,000 plus an unrelated $12,000 premium give $29,880 per spouse.
is_optional_senior_or_disabled_income_eligible.py	Yes, countable income.	Pair-only SSI responsibility; WV spouses at $8,946 each stay eligible beside unrelated $50,000 income.
is_optional_senior_or_disabled_asset_eligible.py	Yes, resources; the third adult also removed the couple limit.	Pair resources and recognized-spouse limit. WV spouses with $1,200 each qualify beside unrelated $50,000 resources. MSP standards are unchanged.
is_209b_ssi_recipient_income_eligible_for_medicaid.py	Yes, income and expenses.	Pair-only spenddown: IL couple income after expenses stays $30,000, above the supplied $25,000 limit, despite an unrelated $12,000 expense.
is_medically_needy_for_medicaid.py	Yes, income/expenses/resources, plus couple-limit identity.	SSI pair totals and spouse recognition, retaining responsible-parent treatment. MA couple income $7,000/resources $2,000 qualify beside unrelated $50,000 income/resources.
medicaid_optional_senior_or_disabled_income_deemed_from_ineligible_spouse.py	A third singleton is already excluded by the SSI spouse-status mask; another couple's ineligible spouse could enter.	MO's married-couple gross-income exception now uses the identified pair. Third-singleton case pins $2,004; a four-person/two-couple case pins separate $2,004/$17,364 deeming. The SSI earned/unearned inputs already read the identified spouse and need no change.
medicaid_optional_senior_or_disabled_income_limit.py	No sum; a third adult changed the unit-size gate.	Necessary companion fix uses recognized marriage. WV limits stay $17,892/$17,892/$11,928.
medicaid_household_income.py	Its sums implement MAGI households.	Unchanged: follows MAGI membership under 42 CFR 435.603.

Iowa facts and false defaults

ia_ssa_rcf_shares_room_with_spouse and ia_ssa_rcf_received_ssi_before_entry both default to false. No repository data builder sets or imputes either. Same-month entry does not establish a shared room; SSI awarded after separation does not establish receipt before entry. These defaults implement the supplied Iowa rulings and 441 IAC 51.4(3) / 6-B, printed p.60.

For co-resident legacy situations, current or reported SSI still approximates pre-entry receipt. Separated spouses require the explicit prior-SSI fact.

January 2026, both spouses aged 70, 22 covered entry-month days:

Scenario	Previous main cce379ea00	This revision
Co-resident, monthly pensions $1,800/$100; neither fact supplied	Couple budget; $771.84 each	Individual budgets; care income $1,800/$100, payments $667.34/$846.34
Same, explicit shared room	—	Couple budget; care income $950 each, payments $771.84 each
Separate households, pensions $175/$100; neither fact supplied	Individual budgets; care income $793/$718, payments $846.34 each	Individual budgets; care income $1,014 each, payments $846.34 each
Same, explicit prior SSI	Input absent	Retained couple budget; care income $755.50 each, payments $846.34 each

The retained entry calculation uses couple SSI $1,491 - ($175 + $100 - $20) = $1,236 once, giving ($275 + $1,236) / 2 = $755.50 care income each. The later entrant retains the couple budget in their own qualifying entry month while the spouse already resident remains individual, as the supplied staggered-entry ruling directs.

Validation

Validation head: 4e4136e6ea66deab720b03677ac2e34c6895476a.

464 unique tests passed in serial foreground runs, one file per process. The full invariants file, every listed state-supplement file and the supplied probes passed again after the final co-resident-parent identity correction. The Medicaid, CAPI and joint-claim files passed before that narrow correction. Every run used the supplied 4 GB/1,800-second cap, load-at-start gate of 100, thread limits of four and OMP_WAIT_POLICY=PASSIVE. make format passes. No microsimulation or full local suite ran.

File	Passed
ia_ssa.yaml	121
test_ssi_spousal_deeming_invariants.py	79
medicaid_ssi_couple_three_adult.yaml	6
ca_capi_three_adult.yaml	6
ca_capi_person.yaml	15
ca_capi_payment_standard.yaml	20
mo_ssp.yaml	35
va_ag.yaml	32
sd_ssp.yaml	39
or_ssp.yaml	38
ssi_state_supplement_three_adult.yaml	37
ssi_claim_is_joint.yaml	9
co_oap_deemed_income.yaml	20
earlier-probes.yaml (supplied probes)	7

All four Iowa default scenarios above were freshly recomputed at this revision; none of the reviewed values moved. CI: Pull request run 41177118616 concluded success at the validation head, with all 33 checks passed and no failed or cancelled checks. The latest exact-head Pull request run was verified as completed/success.

Mutation checks

Each mutant runs one direct helper test against a real simulation population, under the supplied 2 GB/900-second cap. Both tests pass without mutation. Both mutants were rerun and killed after the final co-resident-parent identity correction; each patch was then reversed and the source was verified clean.

Mutant	Selected test	Result
Co-resident-only shared helper	test_retained_state_spouse_helpers_keep_separated_marriage	Killed: the declared separated marriage becomes false for both people; assertion failure, exit 1, 83 seconds.
Drop parent-link exclusion	test_retained_state_spouse_helpers_exclude_cross_household_parent	Killed: the cross-household parent/child pair becomes married; assertion failure, exit 1, 164 seconds.

Limits and follow-ups

• Impact remains for the hub to recompute. Residence/separation timing, sponsored-noncitizen allocations, married-minor rate treatment, mixed-citizenship income methodology, Medicaid citizenship gating, the facility deeming exception and the SSI two-month lag retain the disclosed limitations from the earlier rounds.
• MSP's own whole-unit budgets and Florida's OSS maximum are unchanged. Ordinary state supplement budgets use the actual co-resident pair; CO/MO/OR/SD/VA/IA retain only their existing program-specific separated-spouse rules. Missing temporary-absence/event facts remain an inherited representation limit.
• A lone married pair with one spouse inferred as the other's tax dependent remains indistinguishable from a possible default group; explicit appropriate tax roles are required. Unlabelled default groups can still infer a marriage; parent/dependent facts now rule out known nonspouses even apart.
• Proposed follow-up (not filed): Identify actual spouses in Montana LIHEAP categorical and SSI-standard budgets. The newly merged readers use whole marital-unit SSI recipient counts, permitting nonspouse coverage or incorrect standards in omitted-unit situations. Audit ARM 37.70.401 and manual sections 3.10/3.11 and 9.5, then add omitted-unit and third-adult regressions.

Impact

impact: pending (hub)

axiom: TheAxiomFoundation/rulespec-us#1706 queued (20 CFR 416.1163(d)-(f), 416.1801(c), 416.1806, 416.1160(d); same corpus-release blocker as #1626). Resource deeming: TheAxiomFoundation/rulespec-us#1626 queued.

[MaxGhenis]

Local results at 2924cd90da (macOS, not Linux CI):

• policyengine_us/tests/core/test_ssi_spousal_deeming_invariants.py: 8 passed. 19 s user CPU, 1.0 GB peak RSS.
• test_ssi_resource_deeming_invariants.py and test_ssi_state_supplement_medicaid_dependency.py: 24 passed. 73 s user CPU, 4.9 GB peak RSS (mostly the dependency module).
• All 33 YAML files under tests/policy/baseline/gov/ssa/ssi, run one process at a time: pass, after the three is_ssi_spousal_resource_deeming_applies.yaml updates described in the body. The new cases in ssi_if_takes_up.yaml (10), ssi_claim_is_joint.yaml (10) and the deemed-income file (7) also pass.
• The same invariant module, with the new variable replaced by the oracle's no-deeming amount, fails 7 of 8 tests on main.

A real microsimulation of the impact (merge base b2aaef3391 against this head, 2026, default dataset) is queued under the shared heavy lock and will be added to the body.