Update 2026-27 Child Benefit rates

PolicyEngine UK currently appears to use forecast-indexed 2026 Child Benefit weekly rates instead of the published 2026-27 rates.

Repro with policyengine[uk]==4.11.0 (policyengine-uk@2.88.20):

from policyengine.tax_benefit_models.uk import uk_latest
for name in [
    "gov.hmrc.child_benefit.amount.eldest",
    "gov.hmrc.child_benefit.amount.additional",
]:
    for value in uk_latest.get_parameter(name).parameter_values:
        if value.start_date.year == 2026:
            print(name, value.value, value.start_date, value.end_date)

Actual 2026 values:

• gov.hmrc.child_benefit.amount.eldest: 26.935709699992934
• gov.hmrc.child_benefit.amount.additional: 17.836506423219888

Published 2026-27 weekly rates are:

• Eldest or only child: 27.05
• Additional children: 17.90

Sources:

• GOV.UK rates and allowances table: https://www.gov.uk/government/publications/rates-and-allowances-tax-credits-child-benefit-and-guardians-allowance/tax-credits-child-benefit-and-guardians-allowance
• GOV.UK Child Benefit page: https://www.gov.uk/child-benefit/what-youll-get/
• SI 2026/232 PDF says the enhanced rate increases from £26.05 to £27.05 and the other rate from £17.25 to £17.90: https://www.legislation.gov.uk/uksi/2026/232/pdfs/uksi_20260232_en.pdf

This surfaced while comparing Axiom RuleSpec UK against PolicyEngine over local Enhanced FRS. Positive child-benefit rows return annual respective amounts that imply weekly rates of about 26.9357 and 17.8365, while the generated RuleSpec from the 2026 instrument returns 27.05 and 17.90.

Expected: update the Child Benefit amount parameters for 2026-27, with the correct effective date semantics for the UK tax year / benefit week.

Search the UK parameter catalog for gov.hmrc.child_benefit.amount.eldest and gov.hmrc.child_benefit.amount.additional, then run the reproduction snippet from the issue. Check the 2026-27 effective dates against the linked GOV.UK sources and confirm that the reported weekly values and UK tax-year or benefit-week semantics are correct.