Fix benefit cap exemption defects (#1818)

[vahid-ahmadi]

Fixes defects in the benefit cap exceptions, found while investigating #1818 (the model caps 222k UC households against DWP's 111k for Great Britain, November 2025). None of them is the main cause of the over-capping, which is believed to lie in calibration and rents.

Reproducer: policyengine_uk/tests/policy/baseline/gov/dwp/benefit_cap/is_uc_benefit_cap_exempt_earnings.yaml::The 2026-27 earnings threshold rises to 881 a month

axiom: TheAxiomFoundation/rulespec-uk#399 queued (UC reg 82 threshold and earned income)

Main is merged in at ce3e0bc. Three of this PR's original fixes have since landed on main and are kept as they are there:

• Fix benefit-cap statutory-pay earnings exemption #2399 measures the earnings exception with each assessed claimant's uc_individual_earned_income_before_mif (which since Model Universal Credit reg 77 for owner-managers of a company #1965 includes company owner earnings).
• Lift the benefit cap only for the support component and the people HB reg 75F and UC reg 83 name, and cap only the family's benefits #2076 rewrote the specified-benefit exceptions, removing the dead code.
• Give Universal Credit and Housing Benefit their own benefit cap and LHA shared-accommodation rules #2089 split the benefit cap by scheme. Its is_uc_benefit_cap_exempt_specified_benefit counts armed forces independence payment (AFIP) for a claimant only (UC Regs 2013 regs 2 and 83(1)(c)), and is_housing_benefit_benefit_cap_exempt_specified_benefit counts it for the claimant, partner or a young person (HB Regs 2006 reg 75F(1)(ea)). That is the AFIP fix this PR made in the shared is_benefit_cap_exempt_health_disability, which Give Universal Credit and Housing Benefit their own benefit cap and LHA shared-accommodation rules #2089 deleted, so this PR no longer changes that code.

What this PR still changes relative to main is the earnings threshold, plus regression tests.

The earnings threshold was hardcoded

is_uc_benefit_cap_exempt_earnings on main compares earned income with a literal 10_152 (£846 a month, the 2025-26 value). It is now a dated monthly parameter, gov.dwp.benefit_cap_earnings_exemption: 16 hours at the reg 4 rate × 52 ÷ 12, with the fraction of a pound disregarded (UC Regs 2013 reg 82).

Tax year	NLW	× 16 × 52 ÷ 12	Threshold
2020-21	£8.72	£604.59	£604
2021-22	£8.91	£617.76	£617
2022-23	£9.50	£658.67	£658
2023-24	£10.42	£722.45	£722
2024-25	£11.44	£793.17	£793
2025-26	£12.21	£846.56	£846
2026-27	£12.71	£881.23	£881

Each matches the figure DWP published. 2025-26 is unchanged; 2026-27 rises to £881, which caps more households earning between the two figures.

test_benefit_cap_lha_scheme_split_properties.py (from #2089) re-adds the old shared earnings formula as a reference, with the 10_152 literal, and compares it with the new schemes at 2026. Its reference now reads the same parameter, so it keeps testing the split rather than the threshold.

Tests

• policyengine_uk/tests/policy/baseline/gov/dwp/benefit_cap/is_uc_benefit_cap_exempt_earnings.yaml, 10 cases: the same earnings either side of the 2025-26 and 2026-27 thresholds, unearned income and couples, and the earnings regressions from the earlier review (tax on a private pension; one partner's pension contributions against the other's pay; contributions after 75 in a mixed-age couple), which pass on main too. Run against main's earnings formula, the reproducer fails and the other 9 pass.
• benefit_cap_specified_benefits.yaml: the paired final-award AFIP cases from the earlier review, now in the scheme-split names (a young person's AFIP on UC leaves the award capped at £22,054.80; on legacy HB it is paid in full at £30,000). They pass on main and pin the award, which main's flag-only case does not.
• test_uc_single_claim_properties.py generates AFIP for both members and keeps the partner-only AFIP example, asserted against the UC and HB specified-benefit variables.

Results at ce3e0bc (fresh venv):

• Ruff: format and check clean on the changed Python files.
• Benefit cap YAML: 105 passed. gov/dwp YAML: 606 passed. Full YAML policy suite: 3,587 passed (run before the last, non-overlapping main merge).
• The 28 pytest files that touch the benefit cap or AFIP, including test_uc_single_claim_properties.py, test_benefit_cap_specified_benefits_properties.py and test_benefit_cap_lha_scheme_split_properties.py: 351 passed, 1 xfailed (the existing strict one for Marriage Allowance is booked on the recipient as the transferor's unused allowance, not the fixed s.55B transfer #1947).
• .github/check_fix_pr.py against this body and head: PASS.

Refs #1818.

🤖 Generated with Claude Code

[vahid-ahmadi]

@MaxGhenis review request — benefit cap exemption defects found while investigating #1818.

The thing to decide is timing, not correctness. The main fix (using earnings-attributable tax rather than total income tax in the reg. 82 test) reduces capping and moves with the known over-capping. But parameterising the stale earnings threshold raises it from GBP 846 to GBP 881/month for 2026-27, which removes exemptions and increases capping — correct by law, wrong direction for #1818. 2025 is unchanged, since the new parameter resolves to GBP 846 exactly.

So this makes 2026 capping modestly worse while #1818 is open. It is still right, but you may prefer it lands alongside the calibration work rather than before it.

Worth knowing: AFIP produces no numeric change. The variable has no formula and nothing in policyengine-uk-data populates it. The exemption is legally correct to add, but nobody should expect the number to move. I would rather that were stated than discovered later.

The dead-code removal keeps return semantics byte-identical apart from AFIP, and no pre-existing policy test changed (1118 → 1130, all additions) — which was the signal I asked to watch for. Gated Test job passes.

[MaxGhenis]

Coordination with #1944 (mixed-age couples on UC: benefit cap and CTR pensioner status). The UC benefit cap has no age exception that reaches a mixed-age couple. UC Regs 2013 regs 79, 82 and 83 turn only on earnings and benefit receipt; the one age-based exception, SI 2014/1230 reg 60C, covers only claims where every claimant has reached the qualifying age; and GOV.UK says "If you're part of a couple and one of you is under State Pension age, the cap may apply."

#1944 therefore replaces has_pensioner in is_benefit_cap_exempt_other with a new housing_benefit_pension_age_regulations_apply. Under HB Regs 2006 reg 5, pension-age HB has no cap, and a family on UC is never under it.

Two effects on this PR:

1. is_benefit_cap_exempt_other.py conflicts. Keep Apply the benefit cap to mixed-age couples on Universal Credit, and keep them off pensioner Council Tax Reduction #1944's formula; the AFCS and contributory-ESA terms are unchanged.
2. This PR's new comment in is_benefit_cap_exempt_earnings says households with pension income are "generally already exempt through the state pension age exception". After Apply the benefit cap to mixed-age couples on Universal Credit, and keep them off pensioner Council Tax Reduction #1944, that no longer holds for a mixed-age couple on UC, which is capped unless an earnings, health or disability exception applies. Please reword it when rebasing.

#1944 leaves the dead has_pensioner blocks in _earnings and _health_disability for this PR to remove. Your "State pension age excepts from the benefit cap" case (single, aged 70) still passes on #1944.

🤖 Generated with Claude Code

[MaxGhenis]

Coordination with #1999 (fixes #1986), which rewrites is_benefit_cap_exempt_earnings. It covers this PR's defects 1 and 2:

• Earned income (benefit_cap_earned_income): per person, earnings less their own uc_income_tax_on_earnings and uc_national_insurance_on_earnings (from Deduct only each person's own tax and NI on earnings from UC earned income #1949) and relievable pension contributions, summed over the unit, without minimum-income-floor income (reg 82(4)).
  • earned_income_tax still includes tax on pensions, and national_insurance includes voluntary Class 3. Benefit cap, targeted childcare and local CTR use the wrong UC earned income figure #1986's counterexample is a partner's £923 of Class 3 removing a couple's exception.
• Threshold (benefit_cap_earnings_threshold): 16 × the reg 4 NLW × 52 / 12, with the fraction of a pound disregarded (reg 6(1A)(za)), × 12.
  • It reproduces your table (604, 617, 658, 722, 793, 846, 881) from the minimum wage parameter, and adds £430 before 1 April 2017 (S.I. 2017/138) and 2017-19 (520, 542, 569).
  • Use each rule's own Universal Credit earned income in the benefit cap, targeted childcare and local CTR #1999 also adds the missing 2023-24 NLW (£10.42).

When rebasing, keep #1999's is_benefit_cap_exempt_earnings.py, and drop benefit_cap_earnings_exemption.yaml and the earnings cases that duplicate #1999's benefit_cap_earnings_exception.yaml. AFIP and the dead-code removal in _health_disability and _other still stand. #1999 leaves those two files alone, and #1944 rewrites _other.

🤖 Generated with Claude Code

[MaxGhenis]

Coordination with #2076, stacked on #2027.

#2076 removes the unused benefit sums from is_benefit_cap_exempt_other.py and is_benefit_cap_exempt_earnings.py. It also moves the afcs and ESA terms out of is_benefit_cap_exempt_other into is_benefit_cap_exempt_health_disability, limited to the claimant and partner. Before, any member's AFCS or contributory ESA lifted the cap.

Whichever lands second: keep this PR's age or earnings test, and drop the afcs and esa_support_component terms and the dead QUAL_* lists from _other and _earnings.

[MaxGhenis]

Coordination with #2089 (each scheme its own benefit cap), stacked on #2084.

#2089 deletes the shared exception files this PR edits (is_benefit_cap_exempt*.py) and gives each scheme its own:

• UC: universal_credit/benefit_cap/is_uc_benefit_cap_exempt_earnings.py (reg 82) and _specified_benefit.py (reg 83);
• HB: housing_benefit/benefit_cap/is_housing_benefit_benefit_cap_exempt_*.py (75E, 75F, reg 5).

This PR's earnings-test fix overlaps #1999, which ports the same way, into the UC earnings file. Its other defect fixes belong in whichever scheme's file the regulation names.

[juaristi22]

Review of #1820 — Fix benefit cap exemption defects (#1818)

Base repository: PolicyEngine/policyengine-uk
PR number: 1820
Reviewed head SHA: 35efb46
Merge base SHA: 3d12d26
Mode: full
Scope: changed behavior and affected dependencies
Source evidence: Official sources are linked below; local source copies and validation artifacts were retained.
Review status: COMPLETE

Source Documents

• current-policy-source-s8.html.
• current-policy-source-s12.html.
• current-policy-source-s23.html.
• current-policy-source-s24.html.
• current-policy-source-s25.html.

Critical

None.

Should Address

None.

Suggestions

None.

Evidence Gaps

No material gaps reported.

Notes

• Snapshot imports verified; no private population run.
• Property-income capital premise is unestablished; sensitivity only, not a confirmed finding.
• Changed threshold/earnings/AFIP rules corroborated; inherited mixed-age rule and residual pension tax are explicit scope notes.

Validation Summary

187 passed, 11.34s pytest /15.06s CLI; actual ordinary threshold/pension and asymmetric partner AFIP/member-order cases through final UC/HB/net income; named source-base earnings control completed (sensitivity only).

Original statutory/government/JRF sources read with SHA-bound extracts. current-pr-1820-review-sources.json; current-pr-1820-review-source-cache-check.json. No duplicate calculation diagnostics; original PDF extraction via bundled pypdf. Canonical source-manifest validator exited 0.

CI at 35efb46: Install + smoke-import (py3.9)=SUCCESS, Install + smoke-import (py3.10)=SUCCESS, Install + smoke-import (py3.11)=SUCCESS, Install + smoke-import (py3.12)=SUCCESS, Install + smoke-import (py3.13)=SUCCESS, Install + smoke-import (py3.14)=SUCCESS, Lint=SUCCESS, Test=SUCCESS.

Detailed code, policy, source and reproduction artifacts were retained locally.

Timing

elapsed seconds: 1151.70s; setup seconds: 272.25s

Review Severity

APPROVE. Open findings: 0 critical, 0 should address, 0 suggestions.

[MaxGhenis]

Coordination note on #1999 (fixes #1986), replacing the 1 October note above.

#1999 is now stacked on this PR at ea2b87cec and depends on it: it stays a draft and will not merge before this PR does. It keeps this PR's threshold parameter, the AFIP exception and the exception-file changes as they are. On top it adds only:

• the reg 82 earned income measured per claimant, so voluntary Class 3 contributions and tax on pension income are not deducted (the Benefit cap, targeted childcare and local CTR use the wrong UC earned income figure #1986 counterexample);
• threshold values for 2013-14 to 2019-20 in this PR's parameter, with a test that each value from April 2017 equals 16 hours at the national living wage;
• the Benefit cap, targeted childcare and local CTR use the wrong UC earned income figure #1986 targeted childcare and local CTR changes, which do not touch this PR's files.

Once this PR is on main, #1999 merges main and its diff reduces to those.

🤖 Generated with Claude Code

[vahid-ahmadi]

Rebased onto main at 4a5174c. The threshold parameter (benefit_cap_earnings_exemption.yaml), the AFIP exception and the exception files keep the same names and paths, so #1999 can merge main cleanly.

[MaxGhenis]

Independent review (GPT-6.1 Sol, UK hub, 2026-10-12): REQUEST_CHANGES at 4a5174c8c.

@vahid-ahmadi: two blockers, each reproduced with a household (earnings floored after combining both partners' deductions; pension contributions deducted after 75), and one should-fix (AFIP recipient scope). The full review is below. Would you like to fix these yourself, or should the hub push the fixes as new commits on your branch, without rewriting yours? Say which here. If there's no reply by tomorrow, the hub will open them as a separate PR against your branch for you to merge or not. CI: the shard a runner shutdown killed has been re-run.

VERDICT: REQUEST_CHANGES
Head reviewed: 4a5174c

1. [blocker] is_benefit_cap_exempt_earnings.py:42 floors earnings after combining claimants’ deductions. In 2026, one spouse earning £10,600 and a self-employed spouse earning £100 with £3,000 pension contributions produce £7,700 here. Legally, the latter’s earnings floor at zero, leaving £10,600 and an exemption. Reproduced with £5,638/year wrongly capped. Sum individually floored earnings, preferably using uc_individual_earned_income_before_mif with the assessed-claimant mask, and add a cross-partner deduction property test. UC reg. 57(2).

2. [blocker] is_benefit_cap_exempt_earnings.py:40 deducts pension contributions after age 75, although these are not relievable. A mixed-age couple with the older partner earning £11,000 and contributing £1,000 wrongly loses its 2026 exemption; reproduced with £16,631/year capped. Apply the existing relievability age guard, or reuse the individual earnings helper, and add a mixed-age regression. Use each rule's own Universal Credit earned income in the benefit cap, targeted childcare and local CTR #1999’s planned correction does not fix this head. Finance Act 2004 s.188(3)(a).

3. [should-fix] is_benefit_cap_exempt_health_disability.py:46 applies AFIP to every unmasked member. UC defines AFIP as attendance allowance, whose paid-receipt exception requires a claimant. A dependent 17-year-old’s AFIP incorrectly exempts the family. Give AFIP separate statutory recipient scopes for UC and HB; replace the child-positive test:151, and add AFIP to the single-claim masking property tests. UC definition, reg. 83(1)(c).

4. [nit] Correct two documentation claims: 1818.fixed.md:1 says pension-income tax no longer erodes earnings, although that approximation remains; remove “and pension.” test_uc_income_from_capital_properties.py:38 says property income remains in the earnings-tax base, although taxable_property_income is explicitly excluded; correct that explanation.

The £846/£881 thresholds match DWP’s table. The rebase retains reg. 3(3) masking and routes AFIP through it. 20 benefit-cap YAML cases and all 6 capital-income Python tests passed, serially after checking memory. The removed strict xfail passes for the intended dividend-tax correction; its assertions were preserved. No skips or new xfails were added.

The saved impact outputs reconcile to +£10.248m cap reduction, +4,017.7 capped benefit units and −£10.256m UC, supporting the hub’s rounded figures. Both runs used the same dataset, and their cap formulas match this head.

When #2076 merges main, preserve the dated threshold and corrected earnings calculation—including the fixes above. Its current earnings rewrite restores £10,152 and total income tax. Retain AFIP and claimant masking while tightening recipient scopes, move AFCS/ESA out of _other, and rerun the cap and single-claim property tests.

gh could not connect; the diff and comments were read through the GitHub connector. No files were edited or GitHub writes made.

[vahid-ahmadi]

Pushed 70e426c, fixing two test expectations found while reviewing #2076, and the formula behind one of them.

A child's AFIP no longer lifts the cap. The test expected true because AFIP was summed over every member of the benefit unit, so the formula was wrong as well as the test. HB Regs 2006 reg 75F(1)(ea) names "the claimant, the claimant's partner or a young person", not a child. For Universal Credit, AFIP sits inside "attendance allowance" (UC Regs 2013 reg 2), and reg 83(1)(c) excepts only where "a claimant is receiving" it. AFIP now counts for the claimant, the partner or a Housing Benefit young person. The 10-year-old case expects false, and a new case covers a 17-year-old qualifying young person (true). The old formula fails the corrected child test.

AFCS test. is_benefit_cap_exempt_other: true was an output assertion, not an input stub, so the test did run the formula. But #2076 moves AFCS out of _other into _health_disability, so that line would break when it lands. It now asserts _health_disability and is_benefit_cap_exempt, with a comment citing UC reg 83(1)(d)-(e) and HB reg 75F(1)(d) and 75F(2)(a) (a guaranteed income payment is a war pension). AFCS still counts for any member here except a non-joint-claimant partner; #2076 limits it to the claimant and partner.

Changelog updated to match.

Tests: benefit cap YAML 21 passed, dwp YAML 458 passed, full YAML policy suite 3071 passed, the 17 pytest files that touch the benefit cap 210 passed and 1 xfailed. Ruff clean.

@juaristi22 could you review when you have a moment?

[juaristi22]

Reviewed 70e426c01768ceb08c379cd877154c0d6bc31819. I am requesting changes on two benefit-cap exceptions:

1. is_benefit_cap_exempt_health_disability.py:67-75 puts the young-person AFIP route into the shared exemption used for Universal Credit. HB reg. 75F(1)(ea) allows a qualifying young person, but UC reg. 83(1)(c) requires a claimant receiving attendance allowance (which includes AFIP under reg. 2). A UC claimant whose dependent young person alone receives AFIP is therefore wrongly exempted. Please separate the UC and HB routes and test both final awards; the new YAML case currently asserts the shared result.

2. is_benefit_cap_exempt_earnings.py:37-42 subtracts earned_income_tax, which includes tax on private pension income. UC regs. 55(5)(b) and 57(2) require tax in respect of employment or self-employment. On this head, £10,300 wages plus £5,000 private pension produces £546 earned_income_tax versus £0 uc_income_tax_on_earnings, incorrectly removing the earnings exception. Please use the earnings-specific tax measure and add a regression case.

The changed benefit-cap YAML suite passed 13 cases locally, but it does not catch these outcomes. Please also add the required axiom: status to the PR description. Full hosted CI was pending at review time.

[vahid-ahmadi]

@MaxGhenis thanks, I'll fix these myself on this branch today.

[vahid-ahmadi]

Pushed e9b20e5 on top of 70e426c (no rebase, as the branch merges cleanly). I checked each point against legislation.gov.uk and agree with all of them, so there are no push-backs on the law.

@juaristi22

1. AFIP young-person route in the UC exception. Fixed in e9b20e5. AFIP now has a separate route for each scheme. For a family with a UC award, only an assessed claimant's AFIP counts, because AFIP is "attendance allowance" under UC reg 2 and reg 83(1)(c) needs "a claimant is receiving" it. Otherwise the HB route applies: claimant, partner or young person (HB reg 75F(1)(ea)). I picked the route on universal_credit_pre_benefit_cap > 0, as benefit_cap_couple does in Lift the benefit cap only for the support component and the people HB reg 75F and UC reg 83 name, and cap only the family's benefits #2076. Two new cases check the final awards for the same family, where only the 17-year-old gets AFIP. On UC the cap applies, cutting the award from £52,792.44 to £22,054.80 (this fails on 70e426c). On legacy HB the cap does not apply and the full £30,000 is paid. That case passes both before and after, as a control. Without the AFIP it would be capped by £7,945.
2. earned_income_tax includes tax on private pension income. Fixed in e9b20e5. The exception now adds up each assessed claimant's uc_individual_earned_income_before_mif, which deducts uc_income_tax_on_earnings and uc_national_insurance_on_earnings (so no voluntary Class 3 either). Your case is a regression with two children in London in 2025: £546 against £0, as you found. With the earnings exception the family's UC is now £33,876.72. On 70e426c it was capped by £10,805.32.
3. axiom: line. Added to the description. It points to rulespec-uk Bump version #399 (reg 82 threshold and earned income) and Add semver actions #425 (reg 83 and reg 75F, including AFIP), both queued.

@MaxGhenis (hub review at 4a5174c)

1. Earnings floored after combining deductions (blocker). Fixed in e9b20e5 by the same change as María's point 2. As you suggested, I used uc_individual_earned_income_before_mif with the is_uc_assessed_claimant mask, so each person's earnings are floored at nil before the couple's are added (UC reg 57(2)). It is read before the minimum income floor, as reg 82(4) requires. New case: £10,600 of wages plus a self-employed partner with £100 of profit and £3,000 of contributions, 2026. This is now exempt and pays UC of £31,074.08. On 70e426c it was capped by £8,088.48.
2. Pension contributions after 75 (blocker). Fixed in e9b20e5. The helper already applies the pension_contributions_relief_age_limit guard (FA 2004 s.188(3)(a)). New mixed-age case: the 76-year-old earns £11,000 and contributes £1,000, with a 40-year-old partner, 2026. The test also checks that both are assessed claimants and that the pension-age HB route does not apply, so it passes for the right reason. It is now exempt and pays UC of £42,066.40. On 70e426c it was capped by £19,080.80.
3. AFIP recipient scope (should-fix). Covered by María's point 1 above. The child-positive case was already replaced in 70e426c. test_uc_single_claim_properties.py now generates AFIP for both members, includes it in PARTNER_OWN, and adds a partner-only AFIP example asserting it does not lift a single claim's cap but does lift the joint claim's. This is coverage rather than a reproducer: the reg 3(3) mask already handled it.
4. Documentation (nit). test_uc_income_from_capital_properties.py no longer says property income can switch the cap on, since it cannot now. In the changelog I kept "pension" rather than removing it, because that claim is now true. It now names the helper and describes the per-person floor and the age-75 rule.

Reproducer. The description has a Reproducer: line naming seven cases in is_benefit_cap_exempt.yaml: the four new ones above plus the original defect 1-3 cases. Main's .github/check_fix_pr.py passes locally against the PR diff and this head: "Fix PR has valid reproducer evidence. Changed executable tests: 3."

Coordination. The earnings test is now what #1999 planned (per-claimant earned income, no Class 3, no tax on pensions), so #1999's diff on top of this should shrink to the pre-2020 threshold values and the #1986 childcare and CTR changes. One thing for #2076: its HOUSING_BENEFIT_YOUNG_PERSON_BENEFITS comment still gives AFIP the HB young-person scope "either scheme", which is the case María flagged. When it merges main, keep the UC route here.

Tests (fresh venv, policyengine-core 3.33.2, worktree import path confirmed):

• Ruff: ruff format --check . clean (1,617 files), ruff check clean on the changed files.
• Benefit cap YAML: 25 passed (21 before). On 70e426c the four new reproducers fail and the other 13 cases in the file pass.
• gov/dwp YAML: 462 passed (458 before).
• Full YAML policy suite: 3,075 passed (3,071 before).
• The 19 pytest files that touch the benefit cap or AFIP: 219 passed, 1 xfailed. The xfail is the existing strict one in test_uc_state_pension_properties.py (Marriage Allowance, Marriage Allowance is booked on the recipient as the transferor's unused allowance, not the fixed s.55B transfer #1947).

@juaristi22 @MaxGhenis could you re-review?

[MaxGhenis]

Merge-order note from the UK hub (2026-10-13). #1965 (UC reg 77 for company owner-managers) merged today as de511929d. Its hand-off asked that whichever of #1965 and this PR lands second checks one thing: is_benefit_cap_exempt_earnings should read uc_individual_earned_income_before_mif and add no separate company term. That variable now includes company earnings.

This PR's head e9b20e5d4 already does that, so nothing needs changing for #1965. Merge main in when you next push, so CI runs against the combined code.

One more thing to be aware of: main's release tests currently fail on test_uc_statutory_pay_properties.py::test_award_never_rises_with_a_claimants_statutory_pay. Statutory sick pay can end the cap's earnings exemption in a family with pension income. A hub job is fixing that in the same variable; it will keep its change consistent with this PR and link it here.

🤖 Generated with Claude Code

[vahid-ahmadi]

@MaxGhenis thanks for the merge-order note. I've merged main into this branch (01e90fe, a merge, no rebase or force-push), so CI now runs on the combined code.

Conflicts

• is_benefit_cap_exempt_earnings: Fix benefit-cap statutory-pay earnings exemption #2399 had already made the change this PR proposed (summing each assessed claimant's uc_individual_earned_income_before_mif), so I kept main's formula and documentation. Since Model Universal Credit reg 77 for owner-managers of a company #1965 that variable includes company owner earnings, so there is no separate company term. The only addition from this PR is the dated gov.dwp.benefit_cap_earnings_exemption parameter in place of 10_152, plus the reg 57 and FA 2004 s.188 references.
• is_benefit_cap_exempt_health_disability: I took Lift the benefit cap only for the support component and the people HB reg 75F and UC reg 83 name, and cap only the family's benefits #2076's version as it is and added one change. On a Universal Credit award only a claimant's AFIP counts (UC reg 2 and reg 83(1)(c)). Without one, Housing Benefit's young person still counts (HB reg 75F(1)(ea)). Lift the benefit cap only for the support component and the people HB reg 75F and UC reg 83 name, and cap only the family's benefits #2076's dead-code removal supersedes this PR's.
• is_benefit_cap_exempt: main's label.
• test_uc_single_claim_properties.py: main's AFCS assertions plus this PR's partner-only AFIP example.

Two of main's tests changed because they assumed AFIP had the same scope in both schemes. The benefit_cap_specified_benefits.yaml case "A young person's armed forces independence payment lifts the cap (HB reg 75F(1)(ea))" now enters universal_credit_pre_benefit_cap: 0, so it tests the Housing Benefit route its name cites. The reference in test_benefit_cap_specified_benefits_properties.py now takes the family's scheme.

What the PR still changes relative to main: the threshold parameter (2025-26 stays at £846 a month, 2026-27 rises to £881), the AFIP split by scheme, and the 17 regression cases in is_benefit_cap_exempt.yaml. Run on main's code, two of them fail (the 2026-27 threshold and the young person's AFIP on UC), so the Reproducer: line now names just those two. I've updated the description and changelog fragment to match.

Tests (fresh venv, at 01e90fe): benefit cap YAML 110 passed, gov/dwp YAML 571 passed, full YAML policy suite 3,429 passed. The 33 pytest files touching the cap or AFIP: 497 passed, 1 skipped (needs a dataset token), 1 xfailed (#1947). That includes test_uc_statutory_pay_properties.py, so the release-test failure you mentioned passes on the combined code. Ruff is clean on the changed files, and .github/check_fix_pr.py passes against the new description.

[juaristi22]

Re-reviewed 01e90fe28badede6d11e2fe51e679e08f4b94fee. Both blockers from my earlier change request are resolved, and I am approving.

• A qualifying young person’s AFIP no longer exempts a UC claimant from the cap, while the separate Housing Benefit route remains. The paired final-award regressions pass: UC £22,054.80 and legacy HB £30,000. This matches UC reg.83(1)(c) and HB reg.75F(1)(ea).
• The earnings exception preserves main’s earnings-specific tax/NI helper, including company earnings and statutory pay, without a separate company term. The £10,300 wages/£5,000 pension case now deducts £0 tax on earnings and pays UC £33,876.72.
• All seven 2020–2026 thresholds match the DWP table; below/at/above-threshold final-UC checks pass. The Reproducer and axiom lines are present.

Validation:174 focused tests and diagnostics passed at this head; all captured hosted CI checks pass. I inspected the changed Hypothesis tests but did not execute them locally because the existing environment lacks Hypothesis.

[MaxGhenis]

#2089 merged on 2026-10-17 as 0583e23ec, so each scheme now has its own benefit cap exceptions on main. As promised on #2089, here is what that means for this PR:

• The earnings fix follows the move. git merge-tree of this PR's head 01e90fe28 with Give Universal Credit and Housing Benefit their own benefit cap and LHA shared-accommodation rules #2089's head applies benefit_cap_earnings_exemption in variables/gov/dwp/universal_credit/benefit_cap/is_uc_benefit_cap_exempt_earnings.py, on Fix benefit-cap statutory-pay earnings exemption #2399's measure, with no conflict in that file.

• Three files will conflict when this PR merges main:

  • is_benefit_cap_exempt_health_disability.py, which Give Universal Credit and Housing Benefit their own benefit cap and LHA shared-accommodation rules #2089 deleted;
  • test_benefit_cap_specified_benefits_properties.py;
  • test_uc_single_claim_properties.py.

  Each fix in the deleted file goes into the scheme its regulation names: universal_credit/benefit_cap/is_uc_benefit_cap_exempt_specified_benefit.py (UC reg 83) or housing_benefit/benefit_cap/is_housing_benefit_benefit_cap_exempt_specified_benefit.py (HB reg 75F).

• The £10,152 literal is still on main in the UC earnings file until this PR lands.

[vahid-ahmadi]

@MaxGhenis thanks for the #2089 note. I've merged main into this branch (44d696c, then ce3e0bc for the later commits; merges, no rebase or force-push), so it is mergeable again.

The three conflicts

• is_benefit_cap_exempt_health_disability.py: dropped. Give Universal Credit and Housing Benefit their own benefit cap and LHA shared-accommodation rules #2089's is_uc_benefit_cap_exempt_specified_benefit already counts AFIP for a claimant only (UC Regs 2013 regs 2 and 83(1)(c)), and is_housing_benefit_benefit_cap_exempt_specified_benefit counts it for the claimant, partner or a young person (HB Regs 2006 reg 75F(1)(ea)). That is the fix this PR made, so there was nothing left to carry over.
• test_benefit_cap_specified_benefits_properties.py: main's version, which already has the per-scheme AFIP reference.
• test_uc_single_claim_properties.py: main's version, plus this PR's partner-only AFIP example, now asserted against the UC and HB specified-benefit variables.

What follows from that

• The earnings threshold parameter is in universal_credit/benefit_cap/is_uc_benefit_cap_exempt_earnings.py on Fix benefit-cap statutory-pay earnings exemption #2399's measure, as you expected, so the £10,152 literal goes when this lands.
• test_benefit_cap_lha_scheme_split_properties.py re-adds the old shared earnings formula with the 10_152 literal and compares it with the new schemes at 2026. Its reference now reads the same parameter, so it keeps testing the split rather than the threshold.
• The earnings YAML cases moved to is_uc_benefit_cap_exempt_earnings.yaml under the new names. The paired final-award AFIP cases (UC £22,054.80, legacy HB £30,000) are in benefit_cap_specified_benefits.yaml beside main's flag-only case.
• The changelog no longer claims the AFIP fix, and the PR body's reproducer is now only the 2026-27 threshold case. It fails against main's earnings formula.

Benefit cap YAML 105 passed, gov/dwp YAML 606 passed, full YAML policy suite 3,587 passed, and the 28 benefit cap and AFIP pytest files 351 passed with 1 xfailed (#1947). check_fix_pr.py passes on the new body and head.

🤖 Generated with Claude Code

[juaristi22]

Re-reviewed ce3e0bccc01879c5eed826d6d7756b4fb1b0286c.

The main merge and #2089 conflict resolutions preserve the previously approved benefit-cap behavior. The remaining dated earnings parameter is unchanged from the verified 2020–2026 table. The renamed earnings test aggregates claimant earnings before the minimum income floor and annualises the monthly threshold once.

162 focused tests pass, including the pension-tax and young-person AFIP final-award regressions. Current-head hosted checks pass. The changed property tests were inspected; Hypothesis was unavailable in the selected local runtime, so their execution evidence is hosted CI. I am renewing approval for this commit.

Review status: COMPLETE. No open critical findings.