[juaristi22]

Refs #2050.

CTC paid the £545 family element even when no responsible child or qualifying young person was born before 6 April 2017. For a continuing zero-income 2024 claim with one child born on that date, modeled Child Tax Credit was £4,000; it is now £3,455. A child born the prior day retains £4,000.

The family element now uses the existing CTC responsibility mask and DOB helper. Separate dated family-condition and cutoff parameters preserve earlier supported years and independent reforms. The flat amount, individual-element cutoff, claim and income tests, explicit amount overrides and 2025 closure are unchanged.

This is a partial resolution: #1925 already removed uprating; the pre-2011 baby addition and 2003–2014 coverage remain open under #2050. The 2015 false activation is a model-coverage anchor, not legal commencement. Existing April annual sampling, age/month fallback when DOB is missing, looked-after/own-right-benefit proxies and statutory daily award-rounding limitations remain. No historical coverage, population impact or encoded parity is certified. Draft #2057 and ready #2061 retain their separate State Pension/PC work and fixture dependencies.

Sources: CTC Regulations 2002 reg 7(2)(a), SI 2017/387 reg 4(b), commencement reg 1, and Tax Credits Act 2002 s 9(2)(a).

Reproducer: policyengine_uk/tests/policy/baseline/gov/dwp/tax_credits/ctc_family_element.yaml::A child born on the family cutoff receives no family element

Validation: natural-cash RED 2 failed/1 control passed on unchanged production (initial PyYAML/absent-source setup attempt retained separately, zero tests). One bounded focused batch passed 68 unique cases: 23 new and 45 unchanged regressions, with JUnit and per-case JSONL; 62.61s pytest/70.24s CLI. Both post-fix final CTC and aggregate tax credits match the independent cash oracles. Native HTML build passed in 10.57s with existing Node warnings/public web-asset bootstrap retained. All 2,968 protected tracked files stayed byte-identical; no old-fixture changes were needed. Publication requires passing pinned formatting and independent captured-head review, recorded in the coordinator’s review report.

Local runtime: Python3.13.14/Core3.32.27/Hypothesis6.168.2/pytest8.4.1 versus locked Core3.33.2/Hypothesis6.169.0; offline/no-sync, exact checkout imports, external caches, no private inputs or full-suite run. The original 15 background source examples are not a count of executed tests.

axiom: needed

[vahid-ahmadi]

Looks right, fine to approve.

@juaristi22 thanks. The rule matches CTC Regulations 2002 reg 7(2)(a) as amended by SI 2017/387 reg 4(b): the family element is included only if a claimant is responsible for a child or qualifying young person born before 6 April 2017. The strict < on birth_day gets the on-the-day case right, and benunit.any gives one family element per claim. Using Refs #2050 rather than Fixes is right, since the pre-2011 baby addition and pre-2015 coverage stay open.

1. nit: parameters/gov/dwp/tax_credits/child_tax_credit/eligibility/family_element_born_before.yaml:1 holds the same 20170406 date as the existing child_tax_credit/limit/born_before. A separate lever is a fair choice for reforms, but a comment saying the two dates are the same in law, and set separately on purpose, would stop someone "tidying" them later.
2. nit: family_element_born_before.yaml:9-12 cites reg 7 and s.9 as amended, but not the amending SI. Adding SI 2017/387 reg 4(b) there, as in the ..._condition_applies.yaml file, would make the source of the date clear.
3. nit: family_element_birth_date_condition_applies.yaml:10 titles the first reference "regulations 1 and 4(b)" but links to reg 4 only, and the second entry links reg 1 again. Titling the first "regulation 4(b)" would be tidier.
4. nit: docs/book/programs/gov/dwp/tax-credits.md:67-98: the new section reads more like an internal note than user docs ("override contract", "own-right-benefit approximations", "certify"). The first two paragraphs are enough for most readers.
5. nit (process): #2472 also edits docs/book/programs/gov/dwp/tax-credits.md. The hunks are next to each other but git merge-tree merges the two branches cleanly. No open PR of Vahid's touches these files.

What I checked

• Read the issue, PR body and full diff. Checked reg 7(2)(a) and (3) on legislation.gov.uk (current text, £545 family element and £3,455 child element), and SI 2017/387 reg 1 (in force 6 April 2017).
• Hand-computed the YAML cases: zero income, 2024, child element £3,455 plus family £545 gives £4,000; a child born on or after 6 April 2017 gives £3,455.
• Ran policyengine-core test .../ctc_family_element.yaml (3 passed) and pytest test_ctc_family_element.py (20 passed) on the PR head. With CTC_family_element.py reverted to main, 2 of 3 YAML cases and 9 of 20 pytest cases fail.
• Checked that parameters("2017") reads the condition as true and parameters("2016") as false.
• ruff check and ruff format --check on the changed Python files: clean. Hosted CI is green, including Fix evidence. The Reproducer and axiom lines are present.
• Did not run the full suite or any population dataset.

🤖 Generated with Claude Code

[vahid-ahmadi]

Approving, per my review above. Nits are optional.