carers_allowance (variables/gov/dwp/carers_allowance.py) and carer_support_payment (variables/gov/social_security_scotland/carer_support_payment.py) pay anyone with care_hours >= min_hours or a reported award, gated on would_claim_carers_allowance. So two partners who each enter 35+ hours of care both get the allowance, even when they care for the same person. A synthetic couple aged 40 and 38 with 35 and 40 hours each get £4,495.40 CA in England (2026), or £5,103.80 CSP in Scotland.
The law allows only one:
- SSCBA 1992 s.70(7ZA): "Where, apart from this subsection, two or more persons would have a relevant entitlement for the same day in respect of the same severely disabled person, one of them only shall have that entitlement and that shall be such one of them— (a) as they may jointly elect in the prescribed manner; or (b) as may, in default of such an election, be determined by the Secretary of State in the Secretary of State's discretion."
- SSI 2023/302 reg 5(3): "No individual can be entitled to Carer Support Payment in respect of a cared for person where another individual is entitled to— (a) Carer Support Payment, (b) Carer's Allowance, or (c) the carer element of Universal Credit, in respect of that cared for person."
#2001 adds partners_care_for_same_severely_disabled_person (BenUnit) for the legacy carer premium. Unless supplied, it is true unless both the claimant and the partner report a Carer's Allowance award. The two allowances do not read it yet.
Possible fix: pay CA/CSP to at most one of the claimant and partner when the flag is true. Choosing which one is the couple's election under s.70(7ZA)(a), so pick a deterministic default, such as the one with a reported award, otherwise the claimant. The rule also applies across benefit units, but the model cannot see who a non-dependant cares for, so scope the fix to the couple. Enhanced FRS impact should be about zero, because the dataset carries no caring hours, only reported awards, and two reported awards mean two different cared-for people. The change matters for household calculations. Axiom parity: TheAxiomFoundation/rulespec-uk#403 already cites s.70(7ZA).
carers_allowance(variables/gov/dwp/carers_allowance.py) andcarer_support_payment(variables/gov/social_security_scotland/carer_support_payment.py) pay anyone withcare_hours >= min_hoursor a reported award, gated onwould_claim_carers_allowance. So two partners who each enter 35+ hours of care both get the allowance, even when they care for the same person. A synthetic couple aged 40 and 38 with 35 and 40 hours each get £4,495.40 CA in England (2026), or £5,103.80 CSP in Scotland.The law allows only one:
#2001 adds
partners_care_for_same_severely_disabled_person(BenUnit) for the legacy carer premium. Unless supplied, it is true unless both the claimant and the partner report a Carer's Allowance award. The two allowances do not read it yet.Possible fix: pay CA/CSP to at most one of the claimant and partner when the flag is true. Choosing which one is the couple's election under s.70(7ZA)(a), so pick a deterministic default, such as the one with a reported award, otherwise the claimant. The rule also applies across benefit units, but the model cannot see who a non-dependant cares for, so scope the fix to the couple. Enhanced FRS impact should be about zero, because the dataset carries no caring hours, only reported awards, and two reported awards mean two different cared-for people. The change matters for household calculations. Axiom parity: TheAxiomFoundation/rulespec-uk#403 already cites s.70(7ZA).